Molecular Regulation of Granulocyte Macrophage Colony-Stimulating ...

9 downloads 0 Views 216KB Size Report
Molecular Regulation of Granulocyte Macrophage Colony-Stimulating. Factor in Human Lung Epithelial Cells by Interleukin (IL)-1ß, IL-4, and. IL-13 Involves ...
Molecular Regulation of Granulocyte Macrophage Colony-Stimulating Factor in Human Lung Epithelial Cells by Interleukin (IL)-1␤, IL-4, and IL-13 Involves Both Transcriptional and Post-Transcriptional Mechanisms Martin Bergmann, Peter J. Barnes, and Robert Newton Franz-Volhard Clinic at Max-Delbrück Center, Charité, Humboldt University, Berlin, Germany; and Department of Thoracic Medicine, National Heart and Lung Institute, Imperial College School of Medicine, London, United Kingdom Interleukin (IL)-1␤ stimulates the release of granulocyte macrophage colony-stimulating factor (GM-CSF) from lung epithelial cells. To investigate the molecular mechanisms underlying GM-CSF regulation, we studied GM-CSF production, messenger RNA (mRNA) expression levels, and GM-CSF promoter activity in A549 human alveolar carcinoma cells stimulated with IL-1␤. Coincubation with IL-4 or IL-13 dose-dependently inhibited IL-1␤–induced GM-CSF release. Time-course studies of intracellular and extracellular protein release and mRNA expression indicated tight coupling of protein and mRNA synthesis within 6 h after stimulation. IL-4 and IL-13 both inhibited expression of GM-CSF mRNA and protein by 2 h after stimulation. Stable transfection of A549 cells, with GM-CSF promoter/ enhancer constructs containing up to 3.3 kb upstream of the transcription start site, revealed maximal activation by IL-1␤ and phorbol 12-myristate 13-acetate (PMA) with a reporter containing the proximal promoter (⫺627 to ⫹35). This excludes sequences further upstream from a major regulatory role in GM-CSF promoter activation by IL-1␤ or PMA in these cells. IL-4 and IL-13 downregulated promoter activation but had no effect on GM-CSF mRNA half-life. However, IL-1␤ activation of all constructs was far less pronounced than in Jurkat T cells, suggesting a requirement for additional mechanisms, possibly post-transcriptional, to potentiate the observed transcriptional induction.

In addition to their role as a physical barrier between the lung and the exterior environment, bronchial epithelial cells are also involved in the pathogenesis of inflammatory disease such as asthma by releasing proinflammatory cytokines (1). Accumulation of eosinophils in the respiratory mucosa is characteristic of asthma and is directed by a number of cytokines, including granulocyte macrophage colony-stimulating factor (GM-CSF), which prolongs eosinophil survival. Conditioned media from in vitro–cultured human bronchial epithelial cells promotes eosinophil survival, and this effect is antagonized by neutralizing anti– GM-CSF antibodies (2). Interleukin (IL)-1␤ stimulation increased GM-CSF release from primary human bronchial

(Received in original form August 9, 1999 and in revised form December 1, 1999 ) Address correspondence to: Dr. R. Newton, Dept. of Thoracic Medicine, National Heart and Lung Institute, Imperial College School of Medicine, Dovehouse Street, London SW3 6LY, UK. E-mail: [email protected] Abbreviations: base pair(s), bp; complementary DNA, cDNA; conserved lymphokine element 0, CLE0; enzyme-linked immunosorbent assay, ELISA; full-length, FL; glyceraldehyde-3-phosphate dehydrogenase, GAPDH; granulocyte macrophage colony-stimulating factor, GM-CSF; interleukin, IL; messenger RNA, mRNA; nuclear factor, NF; polymerase chain reaction, PCR; phorbol 12-myristate 13-acetate, PMA; reverse transcribed, RT; standard deviation, SD; signal transducer and activator of transcription, STAT; tumor necrosis factor, TNF. Am. J. Respir. Cell Mol. Biol. Vol. 22, pp. 582–589, 2000 Internet address: www.atsjournals.org

epithelial cells, supporting a role for GM-CSF upregulation in the context of inflammation (3). In addition, increased numbers of GM-CSF messenger RNA (mRNA) positive staining cells are found in the bronchial mucosa of patients with asthma compared with nonasthmatic subjects (4). Transcriptional regulation of GM-CSF production by T cells has been studied extensively. Activation of the proximal promoter (⫺620 to ⫹34) by the phorbol ester phorbol 12-myristate 13-acetate (PMA) and ionophore is significantly reduced by mutation of the conserved lymphokine element 0 (CLE0) (⫺54 to ⫺31) or the nuclear factor (NF)-␬B binding region (⫺85 to ⫺76) (5). Transcription factors that bind to the CLE0 element include NF-AT, activator protein (AP)-1, surfactant protein (SP)-1, and YY-1 (5, 6). A second NF-␬B binding site (⫺2,002 to ⫺1,984) is also implicated in PMA induction of the GM-CSF promoter and has been shown to bind c-Rel and p65 (7). Further, in response to T-cell receptor activation, an enhancer region, which contains functional NF-AT/AP-1 binding sites and is located 3.3 kb upstream, seems to act in conjunction with the proximal promoter (8, 9). In addition to trancriptional control, GM-CSF mRNA and protein release have been linked to modulation of cytoplasmic mRNA half-life (10). PMA-induced GM-CSF release from the mouse T-cell line EL-4 is predominantly regulated at the post-transcriptional level (11). The 3⬘untranslated region (3⬘UTR) of GM-CSF contains eight AUUUA motifs, which direct rapid mRNA turnover (12). These AU-rich elements can regulate translational efficiency independently of mRNA destabilization (13). An imbalance toward increased levels of T helper (Th)2–like cells in the lung is believed to contribute to the pathogenesis of asthma (1). The cytokines IL-4 and IL-13, released by this subset of T cells, induce a humoral immune response by switching human B cells to produce immunoglobulin (Ig)G4 and IgE (1). Likewise, release of various proinflammatory cytokines from monocytes (14) and alveolar macrophages (15), which promote the cellular immune response, is downregulated by IL-4 and IL-13. The similar biologic properties of these cytokines is explained by the fact that they share a receptor subunit and therefore activate many signal transduction pathways in common (16– 18). IL-4 was shown to interfere with the transcriptional upregulation of IL-2 (19) as well as the IL-1␤–induced transcriptional upregulation of macrophage (CSF) from primary human monocytes (20). Further, IL-4 inhibits the synthesis of nitric oxide from a murine macrophage cell line stimulated with interferon-␥ by the inhibition of inducible nitric oxide synthase mRNA formation, possibly via interference with the activation of the protein kinase

Bergmann, Barnes, and Newton: Regulation of GM-CSF in Human Lung Epithelial Cells

(PK) C isoform ␧ (21). Recently, inhibition of tumor necrosis factor (TNF)-␣ stimulated E-selectin gene transcription by IL-4 was attributed to the antagonism of IL-4–induced signal transducer and activator of transcription (STAT) 6 with the transcription factor NF-␬B (22). In addition to the interference with transcriptional upregulation, IL-4 increases mRNA turnover of IL-1␤ and TNF-␣ in human monocytes and mouse macrophages, respectively (23, 24). In the present study, human alveolar A549 cells, which were recently shown to express high-affinity IL-4 binding sites (25), were used as a model for human alveolar epithelial cells to examine the mechanisms of GM-CSF induction by IL-1␤ and the effects of IL-4 and IL-13.

Materials and Methods Cell Culture and Drugs Human A549 type II alveolar cell carcinoma cells (ECACC code 86012804) were grown to confluency in six-well plates as previously described (26). Before treatment, cells were incubated overnight in serum-free media and stimulated with IL-1␤ (R&D Systems, Abingdon, UK) or PMA (Sigma, Poole, UK) at 1 ng/ml and 1 ⫻ 10⫺6 M, respectively. IL-4 and IL-13 (R&D Systems) were used at 20 and 10 ng/ml, respectively. Actinomycin D and cycloheximide (Sigma) were used at 10 ␮g/ml. The Jurkat T-cell line E6.1 (ECACC No. 88042803) was cultured in supplemented RPMI 1640 at a cell density of 5 to 10 ⫻ 105 cells/ml. At 24 h before treatment, cells were washed and cultured in fresh medium at 1 ⫻ 106 cells/ml. Where not otherwise stated, drugs and cytokines were added simultaneously.

Enzyme-Linked Immunosorbent Assay for GM-CSF Matched antibody pairs to GM-CSF were purchased from PharMingen (Cambridge Bioscience, Cambridge, UK), and enzymelinked immunosorbent assay (ELISA) was performed essentially as described by the manufacturer. For measurement of intracellular GM-CSF levels, cells were lysed in reporter lysis buffer (Promega, Southampton, UK) and subjected to one freeze–thaw cycle before ELISA.

RNA Extraction, Semiquantative Reverse Transcriptase/ Polymerase Chain Reaction, and Southern Blotting Cells were harvested and RNA extracted using the guanidine thiocyanate–phenol–chloroform method (27). RNA was reverse transcribed (RT) and polymerase chain reaction (PCR) performed as previously described (28). Amplification primers were: GM-CSF: 5⬘-ATG TTT GAC CTC CAG GAG CCG A-3⬘ (sense), 5⬘-CCA TTC TTC TGC CAT GCC TG-3⬘ (antisense); and glyceraldehyde-3-phosphate dehydrogenase (GAPDH): 5⬘CCA CCC ATG GCA AAT TCC ATG GCA-3⬘ (sense), 5⬘-TCT AGA CGG CAG GTC AGG TCC ACC-3⬘ (antisense). Primers were separated by at least one intron, allowing identification of contaminating genomic DNA. Amplification conditions for GMCSF were: 94⬚C 30 s, 68⬚C 30 s, 72⬚C 30 s; and for GAPDH: 94⬚C 30 s, 58⬚C 30 s, 72⬚C 30 s. Product size were 422 and 598 base pairs (bp), respectively. For each experiment, the exponential phase of amplification, where starting complementary DNA (cDNA) is proportional to final product, was determined by performing PCR at various cycle numbers on an “average” sample created by combining aliquots of all cDNA samples in an experiment as described previously (28). The cycle number needed to first visualize a product on a 1.5% agarose gel stained with ethidium bromide was then used for amplification of each individual sample. Using this method,

583

cycle numbers for GM-CSF ranged between 32 and 36 and for GAPDH between 22 and 26, and the optical densities were within the linear range. Aliquots, 10 ␮l, were size-fractionated on agarose gels and subjected to Southern hybridization to confirm identity as described before (28). In addition, 5 ␮l of each PCR reaction were dot-blotted onto Hybond-N (Amersham, Little Chalfont, UK) and hybridized to GM-CSF or GAPDH cDNA probe labeled with [␣-32P]–deoxyadenosine triphosphate. Dot blots were excised and radioactivity was measured by Cerenkov counting. Data are expressed as a ratio of GM-CSF to GAPDH.

Northern Blotting Total cellular mRNA was used for mRNA half-life studies and 200 to 400 ␮g was PolyA⫹ selected using the PolyA Track System IV (Promega). RNA was fractionated by electrophoresis on 1% agarose/formaldehyde gels in 1⫻ 3-(N-Morpholino)propanesulfonic acid. Capillary blotting and hybridization was performed according to standard procedures (29). A 776-bp GM-CSF hybridization probe was generated by cloning a cDNA fragment generated by RT-PCR using primers 5⬘-AGT ACA CAG AGA GAA AGG CTA AAG-3⬘ (sense) and 5⬘-TAG AAG CAT ATT TTT AAT TAT TAC-3⬘ (antisense) into pGEM-T (Promega); the hybridization probe for GAPDH was as described (28). Densitometry was performed on autoradiographs and results expressed as ratios of GM-CSF to GAPDH.

Plasmid Construction GM-CSF promoter fragments were amplified from human genomic DNA extracted from peripheral blood mononuclear cells using primers based on previously published sequences (as discussed later, in Figure 5). The reactions were supplemented with 1 U of Taq extender polymerase (Stratagene, Cambridge, UK). BglII and HindIII restriction enzyme sites were incorporated into primers (underlined); these are part of the genomic sequence apart from the HindIII site in the antisense primer for the promoter fragment. For the enhancer element (9, 30) (Acc. no. L07488), primers were: GM-A: 5⬘-CGA AGA TCT CAG GTC CCC CAG AGA T-3⬘ (sense) and GM-B: 5⬘-CGA AGA TCT GGC AGC GGT ACA TGT-3⬘ (antisense). For the promoter element (Acc. no. X03021 [31], M13207 [32]), primers were: GM-C: 5⬘-CAG AAG CTT GCT GAG AGT GGC TG-3⬘ (sense) and GM-D: 5⬘-CAG AAG CTT CCT CCA GAG AAC TTT AGC CT-3⬘ (antisense). For the 3.3-kb construct, primers GM-A and GM-D were combined. The resulting fragments were cloned into pGEM-T (Promega). Both the promoter and enhancer elements were sequenced and their identities confirmed. In addition, the 3.3-kb construct was compared with a previously published distal 5⬘-fragment (33) (Acc. no. U31279). Identity was greater than 99%. The sequence of the full-length (FL) 3,286-bp GM-CSF promoter/enhancer fragment starting from bp ⫺1 (31) was submitted to EMBL Nucleotide Sequence Database (Acc. no. AJ224149). The fragments described earlier were excised using the BglII and HindIII sites, resulting in a 663-bp HindIII promoter fragment (⫺627 to ⫹35), a 716-bp BglII enhancer fragment (⫺3.3 to ⫺2.6 kb), and a 3.3-kb BglII/HindIII FL promoter/enhancer fragment. These DNA fragments were then subcloned into the respective sites of pGL3basic (Promega) upstream of the luciferase gene, and the resulting vectors were checked for correct orientation of the inserts. pGL3.GM.P contains the 663-bp promoter fragment, pGL3.GM.PEnh contains the 716-bp enhancer fragment upstream of the promoter fragment, and pGL3.GM.FL contains the whole 3.3-kb DNA promoter/enhancer fragment (termed FL). Neomycin resistance was conferred by cloning an XhoI/SalI fragment from pMC1neoPolyA (Stratagene) containing the neomycin resistance gene into the Sal1 site of the pGL3

584

AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 22 2000

vectors downstream of and in the same orientation as the luciferase gene to give pGL3.GM.P.neo, pGL3.GM.PEnh.neo, and pGL3.GM.FL.neo. The NF-␬B–dependent reporter, 6NF-␬Btkluc.neo, has three tandem repeats of the sequence 5⬘-AGC TTA CAA GGG ACT TTC CGC TGG GGA CTT TCC AGG GA-3⬘, which contains two copies of the consensus NF-␬B binding site (underlined) upstream of a minimal thymidine kinase promoter (⫺105 to ⫹51) driving a luciferase gene as previously described (34).

Stable Transfection Before transfection, 8 ␮g of plasmid was incubated with 20 ␮l of the Tfx50 (Promega) in serum-free medium for 15 min at room temperature. Preconfluent A549 cells were washed with serumfree medium and incubated with medium containing plasmid and Tfx50 for 2 h. Subsequently, cells were cultured in fresh medium for 16 h before adding 0.5 mg/ml G-418 (GIBCO Life Technology, Paisley, UK). Foci of stable transfected cells developed after approximately 21 d of culture in the presence of G-418. Clones were harvested to create a heterogeneous population with regard to integration site, and were maintained in medium containing 0.5 mg/ml G-418. Before treatment, cells were seeded into 24well plates, grown to confluency, and incubated overnight in serum-free medium. Cells were stimulated with IL-1␤ and PMA at 1 ng/ml and 1 ⫻ 10⫺6 M, respectively. Cells were harvested 12 h later and luciferase activity was measured in a luminometer (Turner Design; Steptech, Stevenage, UK) using a commercially available luciferase reporter gene assay (Promega). Total protein content of cell lysates as determined by Bradford total protein assay (Bio-Rad, Abingdon, UK) varied less than 10% between samples.

Electroporation Before electroporation, Jurkat T cells were resuspended at 2 ⫻ 107 cells/ml and 500-␮l aliquots were incubated for 5 min with 10 ␮g of plasmid DNA. Electroporation was carried out with a Bio-Rad Gene Pulser II at 260 V and 975 ␮F in cuvettes with a 0.4-cm gap width, resulting in a ␶ of 22 ⫾ 3.2 ms. Cells were resuspended in 2 ml of serum-free medium and treated for 12 h before harvesting for luciferase activity measurement.

Effect of IL-4 and IL-13 on IL-1␤–Induced GM-CSF Protein Release

Figure 1. Effect of actinomycin D, cycloheximide, IL-4, and IL13 on IL-1␤–induced GM-CSF release. A549 cells were treated with IL-1␤ (1 ng/ml) for 24 h and GM-CSF release into the supernatant was measured by ELISA. (A) Cells were cultured in the presence or absence of IL-1␤ (1 ng/ml), actinomycin D (10 ␮g/ ml), and cycloheximide (10 ␮g/ml), as indicated. Data from four independent experiments are expressed as picograms per milliliter (pg/ml) and represent means ⫾ SD. Dose response of (B) IL-4 (max: 20 ng/ml) and (C) IL-13 (max: 10 ng/ml) on IL-1␤– induced GM-CSF release. Data from four independent experiments are experiments are expressed as a percentage of IL-1␤– treated cells as means ⫾ SD. **P ⭐ 0.01.

IL-1␤ stimulation of A549 cells released GM-CSF, whereas levels were undetectable without stimulation. Coincubation with the transcriptional blocker actinomycin D and the translational blocker cycloheximide totally inhibited GM-CSF production after IL-1␤ stimulation (Figure 1A), indicating a requirement for de novo mRNA and protein synthesis. Coincubation with the cytokines IL-4 (Figure 1B) and IL-13 (Figure 1C) dose-dependently inhibited IL1␤–induced GM-CSF release with a ⬎ 50% inhibition at 20 and 10 ng/ml (P ⬍ 0.01, 452 ⫾ 76 to 176 ⫾ 40 ng/ml and 485 ⫾ 82 to 146 ⫾ 35 ng/ml, respectively). Preincubation with IL-4 and IL-13 up to 2 h before stimulation, as well as

incubation with both cytokines together, did not change the level of inhibition. Time-course studies were performed to further characterize the GM-CSF protein release after IL-1␤ stimulation. GM-CSF was detectable in the supernatant only 6 h after stimulation and accumulated over the subsequent time points (Figure 2A); intracellular levels reached a maximum 2 to 6 h after stimulation (Figure 2B) and then rapidly declined in parallel with the accumulation of GMCSF in the supernatant. Downregulation of both intracel-

Statistical Analysis Data are presented as means ⫾ standard derivation (SD). Statistical analysis of GM-CSF release was performed by the Wilcoxon U-test. Luciferase activity was analyzed using Student’s t test. Statistic were performed with BMDP New System 1.0 software (BMDP Statistical Software Ltd., Cork, Ireland).

Results

Bergmann, Barnes, and Newton: Regulation of GM-CSF in Human Lung Epithelial Cells

585

Figure 3. Time course for GM-CSF mRNA. A549 cells were stimulated with IL-1␤ (1 ng/ml) for the indicated times before harvesting for total mRNA and RT-PCR analysis. Representation Southern blots of GM-CSF and GAPDH amplification products are shown (A) and data (n ⫽ 4) plotted as percentages of IL1␤–treated cells at 2 h as means ⫾ SD (B).

in inhibiting GM-CSF production only within the first 2 h after stimulation. Inhibition of GM-CSF by IL-4 and IL-13 was reduced by only about 50% when added 2 h after IL-1␤ stimulation (421 ⫾ 43 to 293 ng/ml and 465 ⫾ 81 to 313 ⫾ 66 ng/ml, respectively).

Figure 2. Time course of GM-CSF protein levels in total cellular lysates and supernatants; effect of IL-4, IL-13, and actinomycin D added with or after IL-1␤. Cell lysates (A) and supernatants (B) from IL-1␤ (1 ng/ml)–stimulated A549 cells coincubated with IL-4 (20 ng/ml) or IL-13 (10 ng/ml) were harvested at the times indicated. (In A and B: blackened bars, IL-1␤; cross-hatched bars, IL-1␤ plus IL-4; striped bars, IL-1␤ plus IL-13.) Data (n ⫽ 3) are expressed as means ⫾ SD as a percentage of the GM-CSF level at 2 h for cell lysates and at 24 h for supernatants. (C) Cells were treated with IL-1␤ (1 ng/ml) for 24 h and supernatants harvested for GM-CSF ELISA. Actinomycin D (blackened bars), IL-4 (cross-hatched bars), or IL-13 (striped bars) was added at the indicated time points with or after IL-1␤. Data (n ⫽ 3) are expressed as a percentage of GM-CSF levels at 24 h after IL-3 stimulation (means ⫾ SD).

lular and extracellular IL-1␤–induced GM-CSF by IL-4 and IL-13 was observed at all time points, indicating a rapid mechanism of inhibition. Effect of IL-4, IL-13, and Actinomycin D When Added with or after IL-1␤ When added up to 6 h after IL-1␤ stimulation, the inhibitory effect of the transcriptional blocker actinomycin D on GM-CSF protein levels was progressively lost (Figure 2C). This suggests that suppression of transcription is effective

Effect of IL-4 and IL-13 on GM-CSF mRNA Treatment with IL-1␤ rapidly induced GM-CSF mRNA levels, which peaked at 2 h before declining by 6 h (Figure 3). This kinetic profile mirrors the intracellular GM-CSF protein level (Figure 2B), indicating tight coupling of mRNA expression and protein synthesis. Northern blot analysis of polyA⫹selected mRNA revealed downregulation of GM-CSF mRNA at 2 h by IL-4 and IL-13 (Figure 4). To assess the role of mRNA half-life, classical actinomycin D transcriptional arrest experiments were performed. Cells were stimulated by IL-1␤ in the presence or absence of IL-4 and IL-13. After 2 h, actinomycin D was added and total mRNA harvested at various time points. GM-CSF mRNA degradation was complete about 6 h after addition of actinomycin D, as anticipated from the mRNA time-course studies. Neither IL-4 nor IL-13 significantly enhanced the degradation of GM-CSF mRNA. GM-CSF Promoter Activation GM-CSF promoter constructs were cloned and stably transfected into A549 cells (Figure 5A). All three constructs gave rise to similar levels of inducibility (Figures 5B and 6A–6C). Induction of these constructs by IL-1␤ was only 1.5- to 2-fold and induction by PMA was 3- to 6-fold (Figure 5), whereas parallel measurements of GM-CSF protein were similar to that for wild-type cells (data not shown). These results were confirmed by a second set of stably transfected A549 cells with all three vectors under

586

AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 22 2000

Coincubation of IL-1␤ or PMA with IL-4 or IL-13 resulted in downregulation of luciferase expression from all three promoter constructs (Figures 6A–6C). When expressed as a percent of stimulation, IL-4 or IL-13 downregulated PMA and IL-1␤–dependent luciferase induction of each construct by 25 ⫾ 11% and 15 ⫾ 8%, respectively, indicating that the target for IL-4– and IL-13–dependent transcriptional repression lies within the proximal promoter region. For PMA stimulation, the effect of IL-4 and IL-13 was statistically significant with all constructs (P ⬍ 0.05). However, with IL-1␤ stimulation statistical significance of downregulation by IL-4 and IL-13 was variable and probably related to the low induction levels. One possible target for this effect is the NF-␬B binding site present in the proximal promoter fragment. This hypothesis was examined using A549 cells stably transfected with an NF␬B–dependent reporter construct (35). As shown in Figure 6D, IL-4 and IL-13 did not inhibit IL-1␤–induced or PMAinduced NF-␬B–dependent reporter activity. Effect of PMA on GM-CSF mRNA Synthesis, Reporter Gene Activation, and Protein Release To test the relevance of PMA stimulation, A549 cells were incubated with either IL-1␤ or PMA. In marked contrast to IL-1␤, PMA resulted in no detectable GM-CSF protein in the supernatant after stimulation for either 12 or 24 h (Figure 7B) as well as a later time point (48 h) (data not shown). However, GM-CSF mRNA was induced by PMA, but with a delayed time course in comparison with IL-1␤ (Figure 7A). This effect may indicate that additional, posttranscriptional/translational mechanisms that are activated by IL-1␤, but not PMA, are necessary for GM-CSF protein expression. Figure 4. IL-4 and IL-13 downregulate IL-1␤–induced GM-CSF: effect on mRNA half-life. (A) Representative Northern blots of GM-CSF and GAPDH mRNA expression in polyA-extracted mRNA harvested 2 h after cotreatment with IL-1␤ (1 ng/ml) and IL-4 (20 ng/ml) or IL-13 (10 ng/ml) are shown. (B) After densitometric analysis, data from four independent experiments are expressed as the ratio of GM-CSF/GAPDH and plotted as percentages of IL-1␤–treated cells as means ⫾ SD. (C) Cells were treated with IL-1␤ (1 ng/ml) in the presence or absence of IL-4 (20 ng/ml) or IL-13 (10 ng/ml). IL-1␤, filled squares; IL-1␤ plus actinomycin D, filled circles; IL-1␤ plus IL-4 plus actinomycin D, open squares; IL-1␤ plus IL-13 plus actinomycin D, inverted triangles. After 2 h, actinomycin D (10 ␮g/ml) was added as indicated, and total RNA extracted at time points indicated. After densitometric analysis, data (n ⫽ 3) are expressed as the ratio of GM-CSF/GAPDH and plotted as a percentage of each treatment at 2 h as means ⫾ SD.

investigation. The marked difference in induction compared with the levels reported from T cells was not due to artifacts of the reporter gene cloning process, inasmuch as transient transfection of our vectors into Jurkat T cells resulted in a 10-fold induction after PMA stimulation (pGL3.GM.P 9.7 ⫾ 1.13-fold and pGL3.GM.PEnh 12.09 ⫾ 3.03-fold). These data suggest that up to 3.3 kb of GMCSF promoter is not by itself sufficient to result in strong induction of GM-CSF mRNA.

Discussion Although GM-CSF release from T lymphocytes has been studied extensively (5, 6, 9), only limited data are available on the mechanisms controlling GM-CSF release from other cell types, including epithelial cells. In this study, IL-1␤ induced GM-CSF mRNA and protein synthesis in human alveolar epithelial–like A549 cells. With regard to cytokine production, these cells behave similarly to human airway epithelial cells (36). Both transcriptional and translational inhibitors totally blocked GM-CSF release from A549 cells, indicating the need for de novo transcription and translation. Time-course studies, on extracellular and intracellular GM-CSF protein as well as mRNA levels, revealed tight coupling of protein synthesis with mRNA induction. Because GM-CSF protein was not present intracellularly or extracellularly before IL-1␤ treatment, and accumulation in the supernatant paralleled the decrease in intracellular levels, release of stored, preformed protein does not appear to be important in IL-1␤–dependent induction of GM-CSF. By 6 h after stimulation, mRNA and intracellular protein levels were declining, indicating negative feedback control of GM-CSF production. Due to these kinetic studies, further experiments were focused on transcriptional regulation of GM-CSF in A549 cells. Stable transfection of the proximal GM-CSF promoter (⫺627 to ⫹35) construct showed inducibility by

Bergmann, Barnes, and Newton: Regulation of GM-CSF in Human Lung Epithelial Cells

587

Figure 5. Activity of the GM-CSF promoter/enhancer in stably transfected A549 cells. (A) Structure of GM-CSF promoter/enhancer fragments in pGL3basic. Transcription factor binding sites previously reported to be involved in inducible GM-CSF regulation (5, 8) are shown. (B) Stably transfected A549 were stimulated with IL-1␤ (1 ng/ml) or PMA (1 ⫻ 10⫺6 M) in serum-free media. Cells were harvested for luciferase assay after 12 h and data from four independent experiments performed are plotted as mean relative light units (RLU) ⫾ SD. *P ⭐ 0.001.

PMA and IL-1␤ that was not potentiated by the enhancer element (⫺3.3 to ⫺2.6 kb). Likewise, the 3.3-kb reporter construct (⫺3,281 to ⫹35) produced similar levels of induction with PMA. However, inducibility by IL-1␤ was even less pronounced with the 3.3-kb fragment or the promoter/ enhancer construct, possibly explained by negatively regulating sites located in the distal promoter fragments. However, studies on previously known distal GM-CSF promoter fragments using transient transfection assays did not describe negatively regulating elements (8). PMA induction of this vector was similar to the promoter/enhancer construct both in A549 and Jurkat cells. Therefore, the effect may be related to artifacts introduced by the stable integration of this large vector. In summary, the upstream enhancer region, which is important for transcriptional induction in T cells (8), or other regions outside the proximal promoter do

not seem involved in activation of GM-CSF transcription in these cells. In PMA-treated T cells, mutations of the CLE0 element or the NF-␬B site located within the proximal promoter reduced activation (5). Induction of NF-␬B by IL-1␤ and PMA in these cells is consistent with a major role for this factor in induction of GM-CSF (35, 37). The degree of reporter activation described herein is similar to the GM-CSF transcription rate observed in human fibroblasts by nuclear run-on experiments (38), yet these are both strikingly different from the levels of gene activation reported in T cells (11). These results are not due to functional differences in the reporter plasmids, inasmuch as reporter constructs transfected into Jurkat T cells resulted in a 10-fold induction by PMA. Because we used a 3.3-kb construct, which confers full inducibility in T cells, it seems unlikely that binding sites further upstream account

Figure 6. Effect of IL-1␤, PMA, IL-4, and IL-13 on the activity of GM-CSF promoter constructs stably transfected into A549 cells. A549 cells stably transfected with the promoter constructs (A) pGL3.GM.P.neo, (B) pGL3.GM.PEnh.neo, and (C) pGL3.GM.FL.neo were exposed to either IL-1␤ (1 ng/ml) or PMA (1 ⫻ 10⫺6 M) (blackened bars) coincubated with IL-4 (20 ng/ml) (open bars) or IL-13 (10 ng/ml) (striped bars) for 12 h before harvesting for luciferase activity determination. (D) A549 cells stably transfected with 6NF-␬Btkluc.neo were stimulated with IL-1␤ (1 ng/ ml) or PMA (1 ⫻ 10⫺6 M) and coincubated with IL-4 (20 ng/ml) or IL-13 (10 ng/ml) for 12 h before harvesting for luciferase activity determination. In each case, data from four independent experiments are expressed as fold activation. *P ⭐ 0.05; **P ⭐ 0.01.

588

AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 22 2000

Figure 7. Comparison of PMA-induced and IL-1␤–induced GMCSF mRNA and protein levels. (A) Cells were incubated with IL1␤ (1 ng/ml) or PMA (1 ⫻ 10⫺6 M) for the times indicated and semiquantative RT-PCR for GM-CSF and GAPDH was performed. Ethidium bromide–stained gels representative of four such experiments are shown. (B) Supernatant of cells used for mRNA measurement in A were harvested at 9 and 24 h and GMCSF release was determined by ELISA. Nonstimulated (NS), open bars; IL-1␤, darkened bars; PMA, hatched bars. Data are plotted as means ⫾ SD.

for GM-CSF mRNA upregulation. Further, our studies used stable transfection analysis, which eliminates many of the problems of transient transfection techniques. However, nuclear run-on experiments previously described using this cell line failed to detect a signal for GM-CSF (39) (data not shown). This result was probably due to the relatively low levels of GM-CSF mRNA expression in A549 cells. One explanation for the discrepancy between the level of promoter activation and the degree of mRNA induction is that rapidly activated post-transcriptional mechanisms substantially potentiate the observed transcriptional induction. This complex regulation for GM-CSF involving both transcriptional and post-transcriptional mechanisms has been suggested previously both in fibroblasts and T cells (10, 11, 38). Recently, IL-4– and IL-13–mediated inhibition of TNF-␣ synthesis in mouse macrophages was attributed to interference with translational activation of TNF-␣ mRNA directed by its 3⬘UTR (40). In the present study, IL-4 and IL13 both inhibited GM-CSF release from IL-1␤–stimulated A549 cells. In contrast to the earlier study, a similar level of inhibition was seen on the mRNA level in the present work. This may reflect different mechanisms in macrophages and epithelial cells. Inhibition of reporter activity was less pronounced than inhibition on the protein level, yet reached statistical significance with all constructs after PMA stimulation in this study. Activation of PKC isoenzymes by PMA in part mimics IL-1␤ signal transduction pathways (41). Interestingly, the repressive effects must occur very rapidly because intracellular levels of GM-CSF

are reduced within 2 h of IL-1␤ stimulation. At this time point, GM-CSF mRNA levels are also reduced, yet mRNA half-life is not significantly changed. Recently, competitive binding of IL-4–induced STAT6 to an overlapping STAT6/ NF-␬B binding site was implicated in IL-4–mediated downregulation of E-selectin expression in endothelial cells (22). However, no sequence matching the STAT6 consensus sequence (TTCNNNNGAA) (22, 42) was found in the proximal GM-CSF promoter fragment, and NF-␬B–dependent transcription was unaffected by IL-4 or IL-13, indicating that other mechanisms are involved in IL-4– and IL-13– dependent transcriptional repression in this system. This result contrasts with recently published data from the monocytic U937 cell line, where IL-13 potently inhibited TNF-␣–induced NF-␬B activation (43). In conclusion, the data presented here suggest a complex regulation of GM-CSF release from A549 cells. In contrast to human T cells, the proximal GM-CSF promoter is sufficient to confer maximal inducibility with IL1␤ and PMA stimulation. The physiologic stimulus IL-1␤ is less potent than PMA on promoter activation but substantially more effective at inducing protein release. These data strongly suggest a major role for post-transcriptional and translational mechanisms in the IL-1␤–induced upregulation of GM-CSF release in A549 cells. PMA stimulation, which is frequently used as a model for promoter activation, was insufficient to induce GM-CSF protein release despite inducing GM-CSF mRNA, suggesting that additional signaling pathways are necessary for efficient GMCSF expression. In this study the Th-2 cytokines IL-4 and IL-13 inhibited GM-CSF release in part by interference with transcriptional upregulation. However, this effect is significantly less pronounced than the inhibition observed at the protein level. We therefore propose a model for GM-CSF regulation whereby relatively minor changes in promoter activity are substantially amplified by post-transcriptional and/or translational mechanisms. Acknowledgments: One author (M.B.) holds a Deutsche Forschungsgemeinschaft scholarship.

References 1. Barnes, P. J. 1994. Cytokines as mediators of chronic asthma. Am. J. Respir. Crit. Care Med. 150(Suppl.):S42–S49. 2. Cox, G., T. Ohtoshi, C. Vancheri, J. A. Denburg, J. Dolovich, J. Gauldie, and M. Jordana. 1991. Promotion of eosinophil survival by human bronchial epithelial cells and its modulation by steroids. Am. J. Respir. Cell Mol. Biol. 4:525–531. 3. Marini, M., M. Soloperto, M. Mezzetti, A. Fasoli, and A. Mattoli. 1991. Interleukin-1 binds to specific receptors on human bronchial epithelial cells and upregulates granulocyte/macrophage colony-stimulating factor synthesis and release. Am. J. Respir. Cell Mol. Biol. 4:519–524. 4. Humbert, M., S. Ying, C. Corrigan, G. Menz, J. Barkans, R. Pfister, Q. Meng, D. J. Van, G. Opdenakker, S. R. Durham, and A. B. Kay. 1997. Bronchial mucosal expression of the genes encoding chemokines RANTES and MCP-3 in symptomatic atopic and nonatopic asthmatics: relationship to the eosinophil-active cytokines interleukin (IL)-5, granulocyte macrophage-colony-stimulating factor, and IL-3. Am. J. Respir. Cell Mol. Biol. 16:1–8. 5. Jenkins, F., P. N. Cockerill, D. Bohmann, and M. F. Shannon. 1995. Multiple signals are required for function of the human granulocyte-macrophage colony-stimulating factor gene promoter in T cells. J. Immunol. 155:1240–1251. 6. Ye, J., X. Zhang, and Z. Dong. 1996. Characterization of the human granulocyte-macrophage colony-stimulating factor gene promoter: an AP1 complex and an Sp1-related complex transactivate the promoter activity that is suppressed by a YY1 complex. Mol. Cell. Biol. 16:157–167. 7. Ghiorzo, P., M. Musso, M. Mantelli, C. Garre, R. Ravazzoio, and G. Bian-

Bergmann, Barnes, and Newton: Regulation of GM-CSF in Human Lung Epithelial Cells

8.

9.

10.

11. 12. 13. 14.

15.

16.

17.

18.

19. 20.

21. 22.

23. 24. 25.

chi-Scarra. 1997. cRel and p65 subunits bind to an upstream NF-kappaB site in human granulocyte macrophage-colony stimulating factor promoter involved in phorbol ester response in 5637 cells. FEBS Lett. 418:215–218. Cockerill, P. N., A. G. Bert, F. Jenkins, G. R. Ryan, M. R. Shannon, and M. A. Vadas. 1995. Human granulocyte-macrophage colony-stimulating factor enhancer function is associated with cooperative interactions between AP-1 and NFATp/c. Mol. Cell. Biol. 15:2071–2079. Cockerill, P. N., M. F. Shannon, A. G. Bert, G. R. Ryan, and M. A. Vadas. 1993. The granulocyte-macrophage colony-stimulating factor/interleukin 3 locus is regulated by an inducible cyclosporin A-sensitive enhancer. Proc. Natl. Acad. Sci. USA 90:2466–2470. Razanajaona, D., C. Maroc, M. Lopez, P. Mannoni, and J. Gabert. 1992. Shift from posttranscriptional to predominant transcriptional control of the expression of the GM-CSF gene during activation of human Jurkat cells. Cell Growth Differ. 3:299–305. Bickel, M., R. B. Cohen, and D. H. Pluznik. 1990. Post-transcriptional regulation of granulocyte-macrophage colony-stimulating factor synthesis in murine T cells. J. Immunol. 145:840–845. Akashi, M., G. Shaw, M. Hachiya, E. Elstner, G. Suzuki, and P. Koeffler. 1994. Number and location of AUUUA motifs: role in regulating transiently expressed RNAs. Blood 83:3182–3187. Kruys, V., O. Marinx, G. Shaw, J. Deschamps, and G. Huez. 1989. Translational blockade imposed by cytokine-derived UA-rich sequences. Science 245:852–855. de Waal Malefyt, R., C. G. Figdor, R. Huijbens, S. Mohan Peterson, B. Bennett, J. Culpepper, W. Dang, G. Zurawski, and J. E. de Vries. 1993. Effects of IL-13 on phenotype, cytokine production, and cytotoxic function of human monocytes: comparison with IL-4 and modulation by IFN-gamma or IL-10. J. Immunol. 151:6370–6381. Yanagawa, H., S. Sone, T. Haku, K. Mizuno, S. Yano, Y. Ohmoto, and T. Ogura. 1995. Contrasting effect of interleukin-13 on interleukin-1 receptor antagonist and proinflammatory cytokine production by human alveolar macrophages. Am. J. Respir. Cell Mol. Biol. 12:71–76. Palmer, C. R., C. C. Hughes, and J. S. Pober. 1996. IL-4 and IL-13 activate the JAK2 tyrosine kinase and Stat6 in cultured human vascular endothelial cells through a common pathway that does not involve the gamma c chain. J. Clin. Invest. 98:604–609. Keegan, A. D., J. A. Johnston, P. J. Tortolani, L. J. McReynolds, C. Kinzer, J. J. O’Shea, and W. E. Paul. 1995. Similarities and differences in signal transduction by interleukin 4 and interleukin 13: analysis of Janus kinase activation. Proc. Natl. Acad. Sci. USA 92:7681–7685. Kohler, I., P. Alliger, A. Minty, D. Caput, P. Ferrara, N. B. Holl, G. Rank, and E. P. Rieber. 1994. Human interleukin-13 activates the interleukin-4-dependent transcription factor NF-IL4 sharing a DNA binding motif with an interferon-gamma-induced nuclear binding factor. FEBS Lett. 345:187–192. Schwarz, E. M., P. Salgame, and B. R. Bloom. 1993. Molecular regulation of human interleukin 2 and T-cell function by interleukin 4. Proc. Natl. Acad. Sci. USA 90:7734–7738. Gruber, M. F., C. C. Williams, and T. L. Gerrard. 1994. Macrophage-colony-stimulating factor expression by anti-CD45 stimulated human monocytes is transcriptionally up-regulated by IL-1 beta and inhibited by IL-4 and Il-10. J. Immunol. 152:1354–1361. Sands, W. A., V. Bulut, A. Severn, D. Xu, and F. Y. Liew. 1994. Inhibition of nitric oxide synthesis by interleukin-4 may involve inhibiting the activation of protein kinase C epsilon. Eur. J. Immunol. 24;2345–2350. Bennett, B. L., R. Cruz, R. G. Lacson, and A. M. Manning. 1997. Interleukin-4 suppression of tumor necrosis factor alpha-stimulated E-selectin gene transcription is mediated by STAT6 antagonism of NF-kappaB. J. Biol. Chem. 272:10212–10219. Donnelly, R. P., M. J. Fenton, J. D. Kaufman, and T. L. Gerrard. 1991. Il-1 expression in human monocytes is transcriptionally and posttranscriptionally regulated by IL-4. J. Immunol. 146:3431–3436. Suk, K., and K. L. Erickson. 1996. Differential regulation of tumour necrosis factor-alpha mRNA degradation in macrophages by interleukin-4 and interferon-gamma. Immunology 87:551–558. Brinckmann, R., M. S. Top, I. Zalan, D. Heydeck, P. Ludwig, H. Kuhn,

26.

27. 28.

29. 30.

31. 32.

33.

34.

35.

36.

37. 38. 39.

40.

41. 42. 43.

589

W. E. Berdel, and J. R. Habenicht. 1996. Regulation of 15-lipoxygenase expression in lung epithelial cells by interleukin-4. Biochem. J. 318:305–312. Newton, R., L. M. Kuitert, M. Bergmann, I. M. Adcock, and P. J. Barnes. 1997. Evidence for involvement of NF-kappaB in the transcriptional control of COX-2 gene expression by IL-1beta. Biochem. Biophys. Res. Commun. 237:28–32. Chomczynski, P., and N. Sacchi. 1987. Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal. Biochem. 162:156–159. Newton, R., L. M. Kuitert, D. M. Slater, I. M. Adcock, and P. J. Barnes. 1997. Cytokine induction of cytosolic phospholipase A2 and cyclooxygenase-2 mRNA is suppressed by glucocorticoids in human epithelial cells. Life Sci. 60:67–78. Rousell, J., E. B. Haddad, J. C. Mak, and P. J. Barnes. 1995. Transcriptional down-regulation of m2 muscarinic receptor gene expression in human embryonic lung (HEL 299) cells by protein kinase C. J Biol. Chem. 270:7213–7218. Cockerill, G. W., A. G. Bert, G. R. Ryan, J. R. Gamble, M. A. Vadas, and P. N. Cockerill. 1995. Regulation of granulocyte-macrophage colony-stimulating factor and E-selectin expression in endothelial cells by cyclosporin A and the T-cell transcription factor NFAT. Blood 86:2689–2698. Miyatake, S., T. Otsuka, T. Yokota, F. Lee, and K. Arai. 1985. Structure of the chromosomal gene for granulocyte-macrophage colony stimulating factor: comparison of the mouse and human genes. EMBO J. 4:2561–2568. Kaushansky, K., P. J. O’Hara, K. Berkner, G. M. Segal, F. S. Hagen, and J. W. Adamson. 1986. Genomic cloning, characterization, and multilineage growth-promoting activity of human granulocyte-macrophage colony-stimulating factor. Proc. Natl. Acad. Sci. USA 83:3101–3105. Fiorentini, P., M. Musso, S. Penco, R. Giuffrida, V. Pistoia, C. Garre, and G. Bianchi-Scarra. 1995. Characterization of a distal 5⬘-flanking region (⫺2010/⫺630) of human GM-CSF. Biochem. Biophys. Res. Commun. 214: 1015–1022. Israel, N., U. Hazan, J. Alcami, A. Munier, S. F. Arenzana, F. Bachelerie, A. Israel, and J. L. Virelizier. 1989. Tumor necrosis factor stimulates transcription of HIV-1 in human T lymphocytes, independently and synergistically with mitogens. J. Immunol. 143:3956–3960. Bergmann, M., L. Hart, M. Lindsay, P. J. Barnes, and R. Newton. 1998. 1␬B␣ degradation and nuclear factor-␬B DNA binding are insufficient for IL-1␤ and TNF-␣-induced ␬B-dependent transcription: requirement for an additional activation pathway. J. Biol. Chem. 273:6607–6610. Kwon, O. J., B. T. Au, P. D. Collins, I. M. Adcock, J. C. Mak, R. R. Robbins, K. F. Chung, and P. J. Barnes. 1994. Tumor necrosis factor-induced interleukin-8 expression in cultured human airway epithelial cells. Am. J. Physiol. 267:L398–L405. Newton, R., I. M. Adcock, and P. J. Barnes. 1996. Superinduction of NF-␬B by actinomycin D and cycloheximide in epithelial cells. Biochem. Biophys. Res. Commun. 218:518–523. Nimer, S. D., M. J. Gates, H. P. Koeffler, and J. C. Gasson. 1989. Multiple mechanisms control the expression of granulocyte-macrophage colonystimulating factor by human fibroblasts. J. Immunol. 143:2374–2377. Newton, R., D. A. Stevens, L. A. Hart, M. Lindsay, I. M. Adcock, and P. J. Barnes. 1997. Superinduction of COX-2 mRNA by cycloheximide and interleukin-1beta involves increased transcription and correlates with increased NF-kappaB and JNK activation. FEBS Lett. 418:135–138. Mijatovic, T., V. Kruys, D. Caput, P. Defrance, and G. Huez. 1997. Interleukin-4 and -13 inhibit tumor necrosis factor-alpha mRNA translational activation in lipopolysaccharide-induced mouse macrophages. J. Biol. Chem. 272:14394–14398. Dinarello, C. A. 1996. Biological basis for interleukin-1 in disease. Blood 87: 2095–2147. Hou, J., U. Schindler, W. J. Henzel, T. C. Ho, M. Brasseur, and S. L. McKnight. 1994. An interleukin-4-induced transcription factor: IL-4 Stat. Science 265;1701–1706. Manna, S. K., and B. B. Aggarwal. 1998. IL-13 suppresses TNF-induced activation of nuclear factor-kappa B, activation protein-1, and apoptosis. J. Immunol. 161:2863–2872.