Extracellular nucleotides regulate CCL20 release from human primary airway epithelial cells, monocytes and monocyte-derived dendritic cells

Reagents

ATP, ATPγS, ADP, ADPβS, UTP, UDP, UDP-glucose, suramin, LPS, PD98059, SB203580 were purchased from Sigma (Bornem, Belgium). Caffeic acid phenethyl ester (CAPE) was purchased from Biomol International, LP (Exeter, United Kingdom). INS365 (Up₄U, P¹, P⁴-di(uridine 5′-) tetraphosphate) and INS48823 (2′,3′(phenylacetaldehyde acetal)Up₃U) were a gift from Dr. B. Yerxa (Inspire, Durham, NC). Recombinant human IL-8 and CCL20 were purchased from R&D Systems (Abingdon, United Kingdom). Human TNF-α was a gift from Dr. B. Robaye (ULB, IRIBHM, Belgium).

Subjects/tissue samples

Inferior turbinates were collected from 18 non-CF patients and one CF patient (homozygous deltaF508) who required surgical turbinectomy for nasal obstruction or septoplasty (Dr. S. Hassid, ORL Department, Erasme Hospital). Asthmatic or allergic patients were excluded from the study. All procedures were approved by the Ethics Committee of the School of Medicine of the Université Libre de Bruxelles.

Isolation and culture of human nasal epithelial cells (HNECs)

The primary cultures of human epithelial cells derived from nasal mucosa of inferior turbinates were performed by a method adapted from Wu et al. (1985). After excision, the nasal mucosa of inferior turbinates was immediately immersed in HBSS (Ca²⁺/Mg²⁺-free; Invitrogen, Merelbeke, Belgium)-HEPES (25 mM, Gibco product, Invitrogen, Merelbeke, Belgium) culture medium containing 100 U/ml penicillin (Gibco, Invitrogen), 100 µg/ml streptomycin (Gibco, Invitrogen), 100 µg/ml of gentamicin sulfate (Gibco, Invitrogen), and 2.5 µg/ml amphotericin B (Gibco, Invitrogen). After repeated washings with cold HBSS-HEPES medium, tissue was cut off in 1–3 mm large specimens using sterile scalpels and then digested with 0.1% Pronase (Sigma) at 4°C overnight under continuous rotation. After incubation, fetal calf serum (FCS) was added to a concentration of 10%, and nasal epithelial cells were detached from the stroma by gentle agitation. The cell suspensions were filtered through 100 µm cell strainer nylon filter (Falcon, Becton Dickinson, Franklin Lakes, NJ), centrifuged at 150g for 10 min at 4°C, and the pellet was resuspended in 10% FCS-HBSS-HEPES medium and centrifuged again. The second cell pellets were then suspended in the serum-free Dulbecco's modified Eagle's (DMEM)/F12 (1:1) medium (Invitrogen) containing insulin (5 µg/ml; Sigma), apo-transferrin (5 µg/ml; Sigma), epidermal growth factor (10 ng/ml; VWR International, Leuven, Belgium), epinephrine (5.5 µM, Sigma), hydroxycortisone (0.4 µg/ml; Sigma), 3,3′,5-triiodothyronine (2 × 10⁻⁹ M, Sigma), endothelial cell growth supplement (4 µg/ml, Sigma), retinoic acid (300 nM, Sigma), amphotericin B (2.5 µg/ml), streptomycin (100 µg/ml), penicillin (100 U/ml) gentamicin sulfate (50 µg/ml), and L-glutamine (2 mM).

The cells were plated (10⁶/cm²) on collagen-coated Transwell® permeable supports (6.5 mm diameter; 0.4 µm pore size; ELSCOLAB, Belgium) and incubated in a humidified atmosphere of 5% CO₂ at 37°C. The culture medium was changed daily. Monolayer cell confluence was achieved after 7–10 days of culture. Morphological observation using a phase contrast microscope showed that HNECs consisted primarily of epithelial cells. The epithelial nature of the cultured cells was confirmed using immunocytochemistry with cytokeratin immunofluorescence labeling (Cytokeratin Cocktail (Ab-4) monoclonal, Oncogene Research Products, EMD Biosciences, Inc., Darmstadt, Germany), showing 95% of positive cells. Once HNECs reached confluence, they were cultured in an air–liquid interface (ALI) for 2 weeks. During the air–liquid culture, EGF was removed from the culture medium and retinoic acid was added at the final concentration of 50 nM. Tested agents were added in general on the apical side for the last 24 h in 300 µl of culture media. Samples (300 µl of apical medium and 2 ml of basolateral medium) were aliquoted and stored at −80°C before analysis. Experiments were conducted on cultures of transepithelial electrical resistance >300 ohm/cm².

Microarray experiments

HNECs cultured at an ALI were apically stimulated by UTP (300 µM, Sigma) for 3 h. The stimulation was stopped by addition of TRIZOL® reagent (Life Technologies, Groningen, Netherlands) directly on the cells. Total RNAs were then purified on a Qiagen RNeasy kit column (Westburg, Leusden, Netherlands) according to the manufacturer's instructions. Microarray experiments were then made at the VIB microarray facility of K.U.L.-Leuven using slides containing 21,000 human sequences as fully described at www.microarrays.be/service.htm. Arrays were scanned at 532 and 635 nm using a Generation III scanner (Amersham BioSciences, Diegem, Belgium). Image analysis was performed with ArrayVision (Imaging Research, Inc., Ontario, Canada).

Reverse transcription polymerase chain reaction experiments

Total RNA was isolated as described above and reverse transcribed using random hexamers and Superscript-II Reverse Transcriptase according to manufacturer's instructions (Invitrogen). To rule out any genomic DNA contamination, total RNA was incubated with RNase-free DNase (Invitrogen) and experiments were also performed without reverse transcriptase (RT) to control for the cDNA origin of the amplified fragment. To amplify target cDNAs, a set of specific primers for the monoexonic P2Y₂, P2Y₄, and P2Y₆ genes receptors and the P2Y₁₁ receptor were designed and synthesized (Eurogentec, Seraing, Belgium): P2Y₂ sense, 5′-GCTTCAACGAGGACTTCAAG-3′ and P2Y₂ antisense primer, 5′-CACGCTGATGCAGGTGAGGA-3′; P2Y₄ sense, 5′-TGGCATTGTCAGACACCTTGTATGTG-3′ and P2Y₄ antisense, 5′-AAGCAGACAGCAAAGACAGTCAGCAC-3′; P2Y₆ sense 5′-CCCTGCTGGCCTGCTACTGTCTCCTG-3′ and P2Y₆ antisense, 5′-CTAATTCTCCGCATGGTTTGGGGTTGG-3′; P2Y₁₁ sense, 5′-CCCCCGCTGGCCGCCTACCTCTATCC-3′ and P2Y₁₁ antisense, CGCAGCCCAACCCCGCCAGCACCAG-3′. PCR experiments were carried out using primers (500 pM each) with 2.5 U of the Taq DNA polymerase (Invitrogen) in 1.5 mM MgCl₂. The PCR amplification conditions were: 94°C for 5 min, 1 cycle; 94°C for 1 min, 50°C for 1 min, 72°C for 1 min, 35 cycles; 72°C for 7 min, 1 cycle. PCR products were analyzed using 1.5% agarose gel electrophoresis and ethidium bromide coloration. As a negative control, PCR was carried out as before, except that water was added instead of DNA template.

Real-time RT-PCR expression analysis

One microgram of total RNA was isolated and reverse transcribed as described above. Diluted cDNA (5 µl) was analyzed using 2× SYBR Green PCR Master Mix (Applied Biosystems, Lennik, Belgium) by a 7500 Fast Real-Time PCR System and 7500 Fast software (Applied Biosystems) following the manufacturer's protocol. Gene-specific primers were designed according to sequences to cover the conserved peptide sequence region and, when it is possible, in interexonic gene sequences. PCR primers used were: human CCL20 forward: CTGGCTGCTTTGATGTCAGT; human CCL20 reverse: CGTGTGAAGCCCACAATAAA; human IκB-α forward: CTTCAGATGCTGCCAGAGAGT; human IκB-α reverse: GCCTCCAAACACACAGTCATC; human β-actin forward: AGAAAATCTGGCACCACACC; human β-actin reverse: GGGGTGTTGAAGGTCTCAAA. The PCR were conducted in 96-well optical reaction plates. Briefly, each well contained a 25-µl reaction mixture that contained 12.5 µl of SYBR Green PCR Master mix, 1 µl of each forward and reverse primer, 5.5 µl of water and 5 µl of cDNA samples. The SYBR Green dye was measured at 530 nm during the extension phase. The threshold cycle (Ct) value reflects the cycle number at which the fluorescence generated within a reaction crosses a given threshold. The Ct value assigned to each well thus reflects the point during the reaction at which a sufficient number of amplicons have been calculated. The relative mRNA amount in each sample was calculated based on its Ct in comparison to the Ct of the two most stable housekeeping genes chosen by the software geNorm® from a set of nine tested candidate reference genes. From this, a gene expression normalization factor can be calculated for each sample based on the geometric mean of a defined number of reference genes. The purity of amplified product was determined as a single peak of dissociation curve. Real-time PCR were conducted in duplicate for each sample from 2 to 4 different donors, and the mean value was calculated. This procedure was performed in at least two independent experiments. The results were calculated as follows (2^{(Ct of CCL20 − Ct of housekeeping gene)}) × (normalization factor) and expressed in arbitrary units. A mean of values obtained in each experiment performed in different donors were calculated and expressed as a ratio of stimulated values/control values.

Inositol phosphate measurements

HNECs were seeded (4 × 10⁶ per well) on 6-well dishes and labeled for 24 h with 10 µCi/ml [myo-D-2-³H]inositol in DMEM/F12 (1:1) medium containing 5% FCS, antibiotics and amphotericin. Cells were incubated for 2 h in Krebs–Ringer–HEPES (KRH) buffer (124 mM NaCl, 5 mM KCl, 1.25 mM MgSO₄, 1.45 mM CaCl₂, 1.25 mM KH₂PO₄, 25 mM HEPES, pH 7.4, and 8 mM D-glucose). The cells were then incubated in presence of the tested compounds (nucleotides, from Sigma or INS365, a gift from Inspire) together with LiCl (10 mM) for 20 min. The incubation was stopped by the addition of 1 ml of an ice-cold 3% perchloric acid solution. Total inositol phosphates (IPs) were extracted and separated on Dowex columns as previously described (Pirotton et al., 1991). The radioactivities of the different IP fractions (IP₁, IP₂, IP₃, IP₄) were added and normalized to that in phosphatidylinositol.

Cytokine quantification by ELISA

Cytokine concentrations were measured by ELISA (CCL20 and SLPI: R&D Systems; IL-8 and IL-10: Flexia Biosource, Nivelles, Belgium) according to manufacturer's instructions. Apical and basolateral secretions were recovered by directly collecting the apical (300 µl) and basolateral (2 ml) culture media. After a brief centrifugation these collected media were stored at −80°C before ELISA analysis. ELISA experiments were run in triplicate to ensure reproducibility.

Isolation of human neutrophils

Heparinized human venous blood was obtained from healthy volunteers and mixed with an equal volume of 3% dextran T500 in saline. Erythrocytes were allowed to sediment for 20 min, and the leukocyte-rich plasma was then centrifuged on Ficoll-Paque at 400g for 45 min. The pellet was collected and the contaminating erythrocytes were removed by hypotonic lysis. Isolated neutrophils were resuspended at 10⁶ cells/ml in Hanks' balanced salt solution (HBSS) containing glucose (10 mM). Flow cytometric and videomicroscopy analyses revealed an homogeneous population of polymorphonuclear cells. Cell viability was checked by the Trypan blue exclusion method and was routinely found to be 95%.

Isolation of CD4⁺ T cells

CD4⁺ T cells were isolated from buffy coats of healthy volunteers using a CD4⁺ cell isolation kit II (Miltenyi Biotech, Bergisch, Gladbach, Germany) that depletes non-CD4⁺ cells (CD8, CD14, CD16, CD19, CD36, CD56, CD123, TCR γ/δ, and glycophorin A) by an immunomagnetic separation. This method gives more than 95% of CD4⁺ T cells. T cells were then incubated for 24 h at 37°C with complete medium in absence or in presence of agonists.

Preparation of human monocytes and monocyte-derived dendritic cells

Human monocytes and immature DCs were derived from adherent peripheral blood monocytes of normal donors as previously described (Romani et al., 1994). In brief, PBMCs were isolated from leukocytes-enriched buffy coats by standard density gradient centrifugation using Lymphoprep solution from Nycomed (Oslo, Norway), resuspended in complete medium and 2.5 × 10⁸ PBMC were allowed to adhere during 2 h at 37°C at 5% CO₂ in air in 75-cm² cell culture flasks. Non-adherent cells were removed and rinsed-adherent cells were cultured in 15 ml of FCS-containing medium to obtain an homogenous population of monocytes (95% of CD14 positive cells). To obtain DCs, monocytes were supplemented with 800 U/ml of GM-CSF and 500 U/ml of IL-4. GM-CSF and IL-4 were added twice a week. The purity of each cell preparation was evaluated by checking expression of CD1a, HLA-DR, CD14, CD83, and CD3 (90–95% of the DCs were CD1a and HLA-DR positive). The medium for monocytes or DCs culture was RPMI 1640 supplemented with 10% heat-inactivated fetal bovine serum (Hyclone, Logan, UT), 25 mM HEPES buffer, 2 mM L-glutamine, 1 mM sodium pyruvate, 50 mg/ml gentamicin, all purchased from Gibco BRL (Ghent, Belgium), and 5 × 10⁻⁵ M β-mercaptoethanol (Merck, Darmstadt, Germany).

Flow cytometric analysis

Cell staining was performed using FITC-conjugated anti-human HLA-DR, CD83 and CD3 and PE-conjugated anti-human CD1a and CD14, supplied by PharMingen (San Diego, CA). Staining was made on 2 × 10⁵ cells in 100 µl of PBS with 0.1% sodium azide for 30 min in the dark at 4°C. After washing with 2 ml of PBS, samples were analyzed on a FACSort® (Becton Dickinson). Data were analyzed and presented using Cellquest software (Becton Dickinson); the number of gated events was at least 10,000.

Chemotaxis assay

Migration assays for neutrophils and CD4⁺ lymphocytes were performed using 96-wells (Millipore®, Brussels, Belgium) chambers separated by a polycarbonate filter with a pore size of 5-µm. The lower chamber was filled with 130 µl of supernatant of basolateral HNECs plus 0.1% FCS. Neutrophils or CD4⁺ T cells (10⁵/well) were placed into the upper chamber and incubated at 37°C for 2 h. After this period, migrated cells were quantified using the luminescence ATP detection assay system (ATPlite™, PerkinElmer, Zaventem, Belgium) according to the manufacturer's instructions. Briefly, migrated cells in the lower chamber were lyzed, substrate solution (luciferase and D-luciferin) was added and the production of light caused by the reaction of cellular ATP with added luciferase and D-luciferin was measured using the microplate luminometer LB96V (EG&G Berthold Technologies, Vilvoorde, Belgium). Luminescence values were converted in number of migrated cells using an internal linearized standard curve of neutrophils or CD4⁺ T cells performed for each experiment.

P2Y receptors responsive to uridine nucleotides are functionally expressed in primary human nasal epithelial cells (HNECs)

We first tested mRNA expression of uridine nucleotide-sensitive P2Y receptors in HNECs. As shown in Figure 1A, PCR products were amplified from cDNAs (n = 2 donors) of HNECs cultured in ALI (see Methods Section) with P2Y₂ (331-bp), P2Y₆ (455-bp) but not P2Y₄ or P2Y₁₁ primers. In addition, HNECs responded to uridine nucleotide stimulations (UTP and the UTP derivative INS365, 100 µM, n = 3 donors) by activating the phosphoinositide signaling pathway (Fig. 1B).

UTP-regulated expression of CCL20 mRNA in HNECs

Since P2Y₂ agonists are potential therapeutic agents in CF, we focused our investigation on the immunomodulatory properties of UTP on HNECs. Using microarray technology, we determined the effect of UTP (300 µM, 3 h) on gene expression in HNECs cultured in ALI. A few weak regulations were observed in response to UTP but were not reproduced with a second preparation of HNECs except for CCL20 which was up-regulated more than twofold in both experiments. We further focused our study on CCL20 which displays a major interest in the context of airway inflammation. First, we conducted real-time RT-PCR analysis for CCL20 expression in UTP-treated and TNF-α-treated samples. As shown in Figure 2A, CCL20 transcripts were elevated 2.4 ± 0.25-fold by UTP 300 µM after 3 h (n = 4), and they had returned to the basal level after 24 h. On comparison, CCL20 transcripts were enhanced 7 ± 0.3-fold by TNF-α after 3 h (n = 2) and returned to the basal level after 24 h of stimulation. One possible mechanism of chemokine regulation involves mRNA stability as investigated in several works for CCL20 mRNA regulation (Scapini et al., 2002; Kao et al., 2005). To elucidate whether UTP could alter stability of CCL20 mRNA, cultures treated with or without UTP or TNF-α for 3 h were exposed to the RNA synthesis inhibitor, actinomycin D (Fig. 2B). The ratio of the relative mRNA abundance of CCL20 and hIκB-α transcripts, well-known to be short-lived, was calculated from two independent experiments, normalized with the stable housekeeping gene β-actin and expressed as percent. These experiments showed that 3 h following the addition of actinomycin D (2 µg/ml), the level of UTP (3 h)-induced CCL20 transcripts was reduced by 49 ± 9%. On comparison the TNF-α-induced transcripts were also reduced by 52 ± 3% in the same conditions. As control, we showed that the hIκB-α transcripts were much less stable than β-actin or CCL20 and were dramatically reduced by actinomycin D treatment (Fig. 2B). These data indicate that CCL20 mRNA stability was similar between UTP- and TNF-α-induced responses.

UTP regulated the release of CCL20 from HNECs

We quantified CCL20 levels (ELISA, Quantikine kit, R&D) in supernatants from untreated and UTP-treated primary HNECs cultured in ALI. After 24 h of incubation we observed a constitutive release of CCL20 that was higher on the apical than on the basolateral side. The mean quantities of CCL20 constitutively secreted apically (n = 11 donors) and basolaterally (n = 10 donors) were 23 ± 5.6 and 8 ± 2.3 ng/mg of proteins, respectively. Since the apical and basolateral volumes are 300 µl and 2 ml respectively, the mean experimental concentrations of apical and basolateral CCL20 obtained were 1.92 ± 0.46 and 0.11 ± 0.03 ng/ml, respectively. The apically secreted protein is physiologically concentrated in the small volume of the airway surface liquid (ASL). Because the depth of ASL in cultures grown at the ALI is ∼20 µm (Jayaraman et al., 2001) and the surface area of the epithelial sheet is 4.6 cm², the apically secreted CCL20 would be physiologically diluted in 9.4 µl of ASL, resulting in an estimated apical concentration of 61 ± 14 ng/ml (n = 11). Because the real basolateral volume of distribution is unknown, the in vivo concentration of CCL20 in the subepithelial space cannot be precisely estimated. Then, we investigated the effects of apical stimulation by UTP on CCL20 release by HNECs. Figure 3A,B shows that apical UTP (300 µM, 24 h) increased significantly the apical CCL20 release by 3.2 ± 1.8-fold to 72.4 ± 40 ng/mg of proteins (giving an experimental concentration of 6.1 ± 3.3 ng/ml or a theoretical apical concentration of 192.2 ± 108 ng/ml, n = 11, P < 0.01) and the basolateral release by 2.85 ± 1.3-fold to 22.8 ± 10.4 ng/mg of proteins (giving an experimental concentration of 0.31 ± 0.14 ng/ml, n = 10, P < 0.01). As control, basolateral application of UTP failed to provoke any effect on CCL20 level (data not shown). The effect of UTP on the apical and basolateral release of CCL20 was mimicked by another P2Y₂ agonist, ATPγS (100 µM, 24 h) and by UDP (300 µM, 24 h), a product of UTP degradation by ectonucleotidases and agonist of the P2Y₆ receptor (Fig. 3C). It was also reproduced by the dinucleotide derivatives INS365 (Up₄U, 100 µM, 24 h) and INS48823 (a Up₃U derivative, 100 µM, 24 h) that have a prolonged half-life and are agonists of P2Y₂ and P2Y₆ receptors, respectively (Pendergast et al., 2001) (Fig. 3B,C). Finally, since P2Y₂ agonists may have therapeutic value in CF, we tested the effects of UTP on CCL20 release in HNECs isolated from one CF patient. UTP (300 µM, 24 h) was also able to double the apical and basolateral CCL20 release in CF HNECs (data not shown).

UTP-prevented TNF-α- and LPS-induced CCL20 release from HNECs

In airway obstructive diseases like CF, airway epithelium is often submitted to inflammatory and infectious environments. Thus, we tested the effects of pro-inflammatory stimuli (TNF-α and LPS) on CCL20 release. Apical application of LPS alone (100 ng/ml, 24 h) failed to evoke a rise in apical CCL20 secretion whereas it stimulated by threefold the basolateral release of CCL20 (P < 0.01) (Fig. 4A,B). The apical CCL20 increase induced in response to apical application of UTP was not significantly different in the presence or in the absence of apical LPS (Fig. 4A,B; P > 0.05). On the other hand, apical application of TNF-α (30 ng/ml, 24 h) increased the apical release of CCL20 by 3.9 ± 1.6-fold (n = 4; P < 0.05) and the basolateral one by 4.1 ± 0.8-fold (n = 4, P < 0.001). Apical UTP (300 µM, 24 h) blocked both the TNF-α- and LPS-induced CCL20 release (n = 3–4; P < 0.05) (Fig. 4A,B) and in these conditions the level of CCL20 was not significantly different from the one obtained with UTP alone (Fig. 4A,B). This inhibitory effect of UTP on TNF-α-induced CCL20 release was also observed in CF HNECs (data not shown). Furthermore, using real-time PCR, we observed that CCL20 transcript level was similar between TNF-α and TNF-α plus UTP after 3 h of stimulation whereas TNF-α-induced CCL20 protein release was reduced by UTP after 3 h of stimulation (data not shown). Then, we found, using actinomycin D experiments as before, that UTP had no influence on the decay of the CCL20 transcripts induced by TNF-α (data not shown).

Effects of inhibitors of NF-κB, ERK1/2 and p38 pathways on UTP- and TNF-α-stimulated CCL20 release in HNECs

To elucidate the possible signaling pathways involved in UTP-enhanced CCL20 release, inhibitors of various signaling pathways were used (Fig. 4C). Recently, it has been shown that P2Y₆ receptor activation in osteoclasts could induce NF-κB pathway activation (Korcok et al., 2005). Furthermore, CCL20 expression regulation has been shown to involve a JAK-independent but p42/44 extracellular signal responsive kinase (ERK1/2) and NF-kappaB-dependent signaling pathways (Kao et al., 2005). Then, we first tested the effects of the CAPE, an inhibitor of the activation of the NF-κB pathways (Natarajan et al., 1996), on the UTP- or TNF-α-induced CCL20 release in HNECs. As shown in Figure 4C, CAPE (100 µM, 24 h) blocked (∼80%) the TNF-α-induced CCL20 release but had no suppressive effect on UTP-induced CCL20 release. Since P2Y receptor activation in AECs and in monocytes could trigger ERK1/2 pathway (Santiago-Perez et al., 2001; Yang et al., 2004), we tested the effects of the ERK1/2 inhibitor PD98059 (25 µM, 24 h). Figure 4C shows that PD98059 strongly inhibited (79 ± 15%, n = 2) the release of CCL20 induced by UTP without affecting TNF-α response. Since it has been shown that P2Y receptors could activate p38 mitogen-activated protein kinase (MAPK) in neutrophils (Meshki et al., 2004), we also tested the p38 inhibitor SB203580. SB203580 (25 µM, 24 h) had no effect on TNF-α response but inhibited by 50% of the UTP response (Fig. 4C). The inhibitory effect of UTP on TNF-α-induced CCL20 release was still observed in the presence of PD98059 or SB203580 (data not shown) suggesting an involvement of other pathways than ERK1/2/p38/MAPK.

Effects of UTP stimulation on the release of other cytokines from HNECs

Several major inflammatory proteins such as IL-8 or IL-10 were not present on the tested microarray slides. We therefore analyzed their potential regulation by UTP using specific ELISAs. Our studies in primary cultures showed that HNECs secrete constitutively large amounts of IL-8. The mean IL-8 quantities secreted apically and basolaterally in absence of stimulus were 281 ± 56.4 and 1514 ± 394 ng/mg of proteins, respectively (n = 16). Apical stimulation by UTP (300 µM, 24 h) failed to provoke any significant increase in IL-8 secretion (296 ± 62 and 1,341 ± 218 ng/mg of proteins on apical and basolateral sides, respectively; P > 0.05; n = 16) (Fig. 5). We also tested the effects of UTP treatment on the secretion of the major anti-inflammatory cytokine IL-10. In our experiments, we failed to detect any variation of IL-10 level after apical stimulation of HNECs by UTP (300 µM, 24 h) (n = 3; data not shown). Finally, it has been reported that SLPI release, an elastase inhibitor which has anti-inflammatory and antimicrobial properties (Bals and Hiemstra, 2004), was promoted by nucleotides in a human tracheal epithelial cell line (Merten et al., 1996). We therefore tested whether UTP may induce SLPI secretion in human primary HNECs cultured in ALI. Although we observed a strong constitutive release of SLPI by HNECs on apical and basolateral sides (range of experimental concentrations from 28 to 160 ng/ml on apical against 10 to 60 ng/ml on basolateral side), we failed to detect any effect of UTP (n = 3, data not shown).

UTP prevented TNF-α- and LPS-induced IL-8 release from HNECs

IL-8 is an important pro-inflammatory cytokine in airways and plays a crucial role in many airway diseases like CF. Since we observed an inhibitory effect of UTP on TNF-α-induced CCL20 increase, we tested whether UTP could also affect the TNF-α-induced IL-8 release. TNF-α (30 ng/ml, for 24 h) increased by 2.8 ± 0.9-fold and 2.25 ± 0.3-fold respectively the apical (n = 4; P < 0.01) and basolateral (n = 4; P < 0.001) concentrations of IL-8 (Fig. 6A,B). As shown in Figure 6A,B, LPS (100 ng/ml, 24 h) failed to induce an apical release of IL-8 (n = 3; P > 0.05) whereas it increased by 2.3 ± 0.8-fold the basolateral IL-8 secretion (n = 3; P < 0.05). Figure 6A,B also show that UTP abolished the TNF-α- and LPS-induced IL-8 release (n = 3–4; P < 0.01). The same inhibitory effect of UTP was also observed in CF HNECs (data not shown).

Pre-treatment of HNECs with UTP inhibited the neutrophils and CD4⁺ T lymphocytes chemotaxis induced by inflammatory stimuli

Since IL-8 is the most potent neutrophil chemoattractant by acting on CXCR1 and CXCR2 receptors and CCL20 is known to attract CCR6-expressing immune cells, we tested the effect of UTP on the capacity of HNECs basolateral supernatants to induce neutrophils or CD4⁺ lymphocytes chemotaxis. Flow cytometric analysis revealed that neutrophils (n = 4) failed to express the CCR6 receptor at their cell surface whereas 60 ± 5% of CD4⁺ T lymphocytes (n = 4) expressed CCR6 receptors (data not shown). Figure 7 shows that, neutrophil or CD4⁺ T lymphocyte chemotaxis induced by basolateral supernatants of HNECs pre-treated with TNF-α (30 ng/ml, 24 h) or LPS (100 ng/ml) was abolished when UTP (300 µM, 24 h) was added together with TNF-α or LPS (n = 3; P < 0.01). Basolateral medium of HNECs apically stimulated with UTP for 24 h failed to promote a significant increase in the chemotaxis of neutrophils or CD4⁺ T lymphocytes (n = 4; P > 0.05). We observed the same inhibitory effect of UTP on neutrophil or CD4⁺ T lymphocytes chemotaxis with the basolateral supernatants of CF HNECs treated with TNF-α (data not shown).

Extracellular nucleotides induced CCL20 release from monocytes and DCs

DCs are immune cells playing a key role in mucosal immune response which can intimately interact with epithelial cells (Rimoldi et al., 2005). Therefore, it could be hypothesized that nucleotides released by epithelial cells may affect DC or monocyte functions by a paracrine pathway. As shown in Figure 8A, we observed that adenine triphosphate nucleotides (ATP, 300 µM and ATPγS, 100 µM) and UDP (300 µM) but neither UTP nor INS365 significantly (P < 0.001) stimulated CCL20 release from DCs. ATPγS was the most efficient nucleotide: 44 ± 13 pg/ml for control versus 631 ± 191 pg/ml for ATPγS (n = 10; P < 0.001) (Fig. 8A). Figure 8B shows that ATPγS-enhanced CCL20 release was dependent on the ATPγS concentration with a half-maximal effective concentration (EC₅₀) value of 12.8 ± 2.8 µM. The ATPγS-induced CCL20 release was partially (∼50%) prevented by suramin, an antagonist of P2Y₁₁ receptors and other P2 receptors (data not shown). In addition, ADP and the ADP derivative ADPβS failed to evoke any CCL20 increase (data not shown). Furthermore, UDP (609 ± 209 pg/ml, n = 7; P < 0.001) but neither UTP, nor INS365 (Fig. 8) nor UDP-glucose (not shown) promoted release of CCL20 from DCs. Figure 8C shows that, like UDP, the P2Y₆ agonist INS48823 was able to increase the secretion of CCL20 in a concentration-dependent manner, with an EC₅₀ value of 12.7 ± 4.4 µM. Figure 8D shows that LPS (100 ng/ml, 24 h) strongly increased the CCL20 release up to 50 times (1892 ± 674 pg/ml; n = 6). Finally, we tested whether nucleotides may alter the LPS-induced CCL20 release as previously observed in AECs. Our results showed that contrary to UDP which failed to have any effect, ATPγs and to a lesser extent ATP potentiated the LPS-induced CCL20 release (Fig. 8D).

Effects of inhibitors of NF-κB, ERK1/2, and p38 pathways on nucleotide-stimulated CCL20 release in DCs

As previously done for HNECs, we tested the possible signaling pathways involved in ATPγS- and UDP-enhanced CCL20 release in DCs using specific inhibitors (Fig. 9). As shown in Figure 9, CAPE (50 µM, 24 h) blocked both ATPγS- and UDP-induced CCL20 release (respectively by 83 ± 4% and 95 ± 3%; n = 3), as well as the one induced by LPS (100 ng/ml, 24 h). Whereas PD98059 (25 µM, 24 h) blocked ATPγS response (70 ± 6%, n = 3), it inhibited only partially (30 ± 3.6%; n = 3) the UDP response. On the other hand, the p38/MAPK inhibitor SB203580 (25 µM, 24 h) reduced by 60–70% of the ATPγS, UDP, and LPS responses. These data suggest that in DCs ATPγS-enhanced CCL20 release involved NF-κB, ERK1/2 as well as p38/MAPK pathways while UDP-induced CCL20 increase mainly involved NF-κB and p38/MAPK pathways. Moreover, LPS-induced CCL20 release was mainly due to NF-κB and p38/MAPK pathways.

UDP-induced CCL20 release from monocytes

Finally, we tested whether nucleotides may affect CCL20 release from other immune cells. Figure 10 shows that adenine nucleotides slightly increased CCL20 secretion from human monocytes while UDP, but not UTP, strongly up-regulated the CCL20 release from monocytes (n = 3). On the contrary, we failed to detect any CCL20 release in response to any nucleotide (100 µM, 24 h) in freshly isolated human neutrophils or T CD4⁺ lymphocytes (data not shown).

Nucleotides modulate CCL20 and IL-8 release from HNECs

Previous works have shown that the airway epithelium released the chemokine CCL20 in response to various inflammatory mediators (Reibman et al., 2003; Starner et al., 2003; Kao et al., 2005). In the present work, we first observed that HNECs constitutively secreted CCL20. Then, we demonstrated that in basal conditions, UTP up-regulated CCL20 expression and release by normal HNECs as well as by CF HNECs. CCL20 mRNA level was increased by UTP after 3 h of stimulation and returned to basal level within 24 h. Following actinomycin D treatment, the rate of decay of CCL20 transcripts induced by UTP and TNF-α was similar. These data indicate that UTP-induced CCL20 mRNA increase was not due to stabilization of CCL20 mRNA but rather involved transcriptional activation.

Since UTP may be hydrolyzed in UDP by ectonucleotidases and subsequently activate P2Y₆ receptors, we tested INS365, a P2Y₂ and P2Y₄ agonist with prolonged half-life and used in clinical trials in CF (Pendergast et al., 2001), and ATPγS (a slowly hydrolysable P2Y₂ and P2Y₁₁ agonist). Both agonists reproduced the effect of UTP on CCL20 release, suggesting an involvement of the P2Y₂ receptor, according to the absence of P2Y₄ and P2Y₁₁ mRNAs in our HNECs cultures. Furthermore, we show that both UDP and INS48823 (a P2Y₆ agonist with prolonged half-life) increased CCL20 release by HNECs indicating also an involvement of P2Y₆ receptors. Taken together, our results indicate that activation of both P2Y₂ and P2Y₆ receptors led to an increase in CCL20 release by HNECs. Our inhibitors experiments suggest that P2Y₂ and/or P2Y₆ receptor activation induced CCL20 release through an ERK1/2/p38 based mechanism but not through NF-κB signaling pathways in HNECs. Since we still observed the inhibitory effect of UTP on TNF-α-induced CCL20 release in the presence of PD98059 or SB203580, that suggested that this inhibitory effect of UTP observed in pro-inflammatory conditions may involve other pathways than ERK1/2/p38/MAPK. In addition, we failed to detect any effect of UTP on CCL20 mRNA expression level or stability in TNF-α-stimulated cells after 3 h of stimulation. Our data rather suggest that the inhibitory effect of UTP on TNF-α-induced CCL20 release might depend on post-transcriptional mechanisms.

A previous work has shown that P2Y₂ receptor activation evokes only a modest increase in IL-8 release from bronchial epithelial cells but potentiates the IL-8 release induced by a mucopurulent supernatant of CF patients (Ribeiro et al., 2005). Here, we show that mucosal UTP alone failed to evoke regulation of IL-8 secretion by HNECs. Furthermore, we show that, in pro-inflammatory conditions, mucosal UTP was capable of significantly preventing TNF-α or LPS-induced CCL20 and IL-8 releases. The difference of effects induced by UTP on IL-8 release in pro-inflammatory conditions between Ribeiro et al. (2005) and our data may come from the airway origin (upper airway in our study but lower airway for Ribeiro et al.), and the inflammatory conditions (TNF-α/LPS in our experiments but a pool of mucopurulent multi-infected lung materials of CF patients in Ribeiro's study). In CF, production of pro-inflammatory cytokines like IL-8 is up-regulated in AECs resulting in an elevated neutrophil infiltration in airways (Tabary et al., 1998). Our data suggest that in pro-inflammatory conditions, UTP may prevent an excessive inflammation by reducing IL-8 secretion by airway epithelium.

CCL20 is known as an antimicrobial chemokine

CCL20 is a bi-functional peptide which also exerts antimicrobial activity against a wide spectrum of mainly Gram-negative bacteria (Starner et al., 2003; Yang et al., 2003). Recombinant CCL20 at 0.3 µg/ml kills ∼90% f E. coli bacteria (Hoover et al., 2002; Yang et al., 2003). In our experiments, HNECs generated in response to UTP about 0.5 ng of CCL20 per well (∼20 ng/mg of total proteins) in ASL, resulting in a theoretical mean apical concentration of ∼0.35 µg/ml. According to Hoover et al. (2002), our data suggest that CCL20 level induced by nucleotides might be high enough to exert direct antimicrobial activity in particular microenvironments such as ASL or the paracellular space.

Effects of UTP on the interaction of HNECs with immune cells

CCL20 is secreted by airway epithelium in response to inflammatory stimuli and attracts immune cells at inflammation sites (Reibman et al., 2003). CCR6, the CCL20 receptor, is expressed on memory T cells, NK cells and some DC subtypes and plays an important role in DC localization and lymphocyte homeostasis in mucosal tissues, as demonstrated in CCR6 knockout mice (Lukacs et al., 2001; Lugering et al., 2003). Nevertheless, it has been shown that CCL20 had no effect on monocyte-derived immature DCs (Dieu et al., 1998) but rather attracted Langherans-type DCs that express CCR6 receptors or other leukocyte populations (memory T cells, γ/δ T cells) (Banchereau et al., 2000). In our experiments, monocyte-derived DCs and neutrophils failed to express CCR6 whereas about 60% of CD4⁺ T cells expressed CCR6. On the other hand, IL-8 is the most potent neutrophil chemoattractive chemokine. We therefore tested whether HNECs basolateral supernatants were capable of attracting neutrophils and CD4⁺ T cells. We found that CD4⁺ T cells and neutrophils chemotaxis was up-regulated ∼2-fold by basolateral supernatants of normal and CF TNF-α-treated HNECs. In presence of LPS or TNF-α, one of the factor produced during airway inflammatory diseases, UTP acted as an anti-inflammatory agent by preventing neutrophils and CD4⁺ T cells chemotaxis. This was correlated to the inhibition of CCL20 and IL-8 release. We could speculate that in CF therapy where CF airways are inflamed and infected, uridine nucleotides may act as anti-inflammatory agents by inhibiting CCL20 and IL-8 production induced by CF inflammatory environment and by preventing an excessive neutrophil or CCR6-expressing immune cell recruitment in CF airways.

Nucleotides induced CCL20 release from DCs and monocytes

A dense network of DCs has been described in the airways from nasal to alveolar mucosa. In airways, DCs sample and process pathogens in order to prime naïve T cells, either locally or within the draining lymph nodes (Lambrecht and Hammad, 2003). Here, we hypothesized that the stimulatory effect of nucleotides on CCL20 secretion by HAECs could be amplified by a similar effect on airway DCs. Epithelial cells and DCs can interact through cell–cell interactions and release of soluble mediators (Lambrecht and Hammad, 2003). One class of mediators which are abundantly released by epithelia during inflammation is extracellular nucleotides (Homolya et al., 2000).

Here, we showed that external nucleotides promoted CCL20 release not only by airway epithelia but also by DCs or monocytes but not by T CD4⁺ lymphocytes or neutrophils. To our knowledge, this is the first clear demonstration that DCs or monocytes could secrete CCL20. However, some caution is necessary since DCs differentiated from monocytes may not be perfectly representative of endogenous DCs residing in the airway mucosa in vivo. The release of CCL20 in response to ATPγS and UDP, but not in response to UTP/INS365, suggested the involvement of P2Y₁₁ and P2Y₆ receptors in DCs. Due to the lack of specific agonists or antagonists to efficiently discriminate P2X and adenine nucleotide P2Y receptors, we could not exclude a concomitant involvement of P2Y₁₁ and P2X receptors in ATPγS-induced CCL20 release. On the other hand, since only UDP-induced CCL20 secretion in monocytes, the P2Y₆ receptor seems to be preferentially involved in monocytes. Our data are in agreement with the up-regulation of the P2Y₁₁ receptor during monocytes to DC differentiation (Wilkin et al., 2001). Monocytes and DCs have already been described to secrete IL-8 in response to P2Y₆ receptor stimulation (Warny et al., 2001; Idzko et al., 2004) whereas ATP-induced P2Y₁₁ receptor activation promoted thrombospondin-1 release in DCs (Marteau et al., 2005). The pharmacological differences that we observed between HAECs and DCs or monocytes on CCL20 release certainly depend on different expression of P2Y receptors in the distinct cell types. Our inhibitor experiments performed in DCs suggested that NF-κB, ERK1/2 as well as p38/MAPK pathways play a critical role in ATPγS-enhanced CCL20 release while UDP-induced CCL20 increase mainly involved NF-κB and p38/MAPK pathways.

Likewise, we found that LPS, a microbial agent that induces DC maturation (Banchereau et al., 2000), evoked CCL20 release by activating NF-κB and p38/MAPK pathways. Moreover, we observed that ATPγS but not UDP-potentiated CCL20 release from LPS-matured DCs. This could be explained by the observations that LPS induces nucleotide release and P2Y₆ receptor activation in monocytes (Warny et al., 2001) and by the fact that the P2Y₆ receptor is responsive to submicromolar concentrations of UDP whereas higher concentrations of ATP are required to activate the P2Y₁₁ receptor (Communi et al., 1996, 1999). Finally, the P2Y₁₁ receptor is coupled to both the cAMP and the phosphoinositide pathways while the P2Y₆ receptor activation mainly triggers the Ca²⁺ signaling pathway (Communi et al., 1996, 1999). Thus, we could speculate that ATPγS-induced P2Y₁₁ activation by stimulating the cAMP pathway may have synergistic effects with the LPS-induced signaling pathway.

In conclusion, we demonstrate for the first time that nucleotides, previously known as regulator of the mucociliary clearance, may also act as connecting molecules between the airway epithelium and the immune system by regulating inflammatory mediator release. UTP or analogs used in aerosol therapy in pulmonary and inflammatory diseases like CF might have anti-inflammatory properties that could be of interest for further clinical investigations.