Airway epithelial cells are a rich source of eosinophil-selective C-C chemokines. We investigated whether cytokines and the topical glucocorticoid budesonide differentially regulate RANTES, monocyte chemoattractant protein-4 (MCP-4), and eotaxin mRNA and protein expression in the human bronchial epithelial cell line BEAS-2B and in primary human bronchial epithelial cells by Northern blot analysis and ELISAs. Eotaxin and MCP-4 mRNA expression induced by TNF-α alone or in combination with IFN-γ was near-maximal after 1 h, peaked at 4 and 8 h, respectively, remained unchanged up to 24 h, and was protein synthesis independent. In contrast, RANTES mRNA was detectable only after 2 h and slowly increased to a peak at 24 h, and was protein synthesis dependent. Induction of eotaxin and MCP-4 mRNA showed a 10- to 100-fold greater sensitivity to TNF-α compared with RANTES mRNA. IL-4 and IFN-γ had selective effects on chemokine expression; IL-4 selectively up-regulated the expression of eotaxin and MCP-4 and potentiated TNF-α-induced eotaxin, while IFN-γ markedly potentiated only the TNF-α-induced expression of RANTES. Although budesonide inhibited the expression of chemokine mRNA to a variable extent, it effectively inhibited production of eotaxin and RANTES protein. Budesonide inhibited both RANTES- and eotaxin promoter-driven reporter gene activity. Budesonide also selectively accelerated the decay of eotaxin and MCP-4 mRNA. These results point to IL-4 as a possible mediator by which Th2 cells may induce selective production of C-C chemokines from epithelium and indicate that glucocorticoid inhibit chemokine expression through multiple mechanisms of action.
It is now well established that airway inflammation plays a major role in causing clinical manifestations of allergic diseases, such as reversible airway obstruction and increased airway responsiveness to several stimuli in asthma (1). Eosinophils and their products, including cytotoxic granule proteins and de novo synthesized leukotrienes, are important effectors in the pathophysiology of airway allergic inflammation, causing destruction of airway epithelium, sensitization of airway nerve terminals, vascular leakage, and other pathological changes (2, 3). The mechanisms underlying the selective recruitment of eosinophils is a complex, multistep process, probably mediated by the cooperative action of cytokines that cause eosinophil priming and increased survival (IL-3, IL-5, GM-CSF) with those that activate endothelium (IL-1, TNF-α, IL-4, IL-13) and eosinophil-selective chemoattractant molecules, especially C-C chemokines (4). Several members of the C-C branch of the chemokine family display potent and/or selective chemoattractant and activating properties toward eosinophils, basophils, monocytes, and T lymphocytes, cell types associated with allergic reactions, while being very poor neutrophil chemoattractants. The potent, eosinophil-selective C-C chemokines include RANTES, eotaxin, eotaxin-2, monocyte chemoattractant protein (MCP)3-3, and MCP-4 (5, 6, 7, 8, 9, 10). Cooperation between IL-5 and chemokines in promoting eosinophil migration has been described both in vitro (11) and in animal models (12, 13). Cutaneous injection of RANTES or eotaxin has been shown to induce eosinophil accumulation (14, 15, 16, 17). Interestingly, injection of RANTES induced eosinophil accumulation with a profoundly faster time course in allergic individuals than in nonallergic individuals (15).
Within the last few years, numerous studies have documented increased expression of mRNA and/or protein for several C-C chemokines, including RANTES, eotaxin, MCP-3, and MCP-4, in diseases characterized by tissue eosinophilia, such as nasal polyposis, rhinitis, and asthma (16, 18, 19, 20, 21, 22, 23, 24), as well as after experimental Ag challenge in human skin (25) and lung (26). Studies in murine models of allergic inflammation provided crucial information on the nonredundant role of chemokines at different stages of the development of an inflammatory infiltrate (17, 27, 28). Work by Gonzalo et al. showed that challenge-induced recruitment of specific leukocyte types in the lung correlates with a distinct pattern of chemokine expression (17). Individual blockade of eotaxin, MCP-5, macrophage inflammatory protein-1α, RANTES, and MCP-1 in OVA-induced airway allergic inflammation in mice has revealed distinct roles for each chemokine in inducing inflammatory cell recruitment and bronchial hyper-reactivity (27). The nonredundant role of chemokines has also been suggested by studies conducted with a gene disruption strategy. The macrophage inflammatory protein-1α knockout mice were unable to mount an inflammatory response to influenza virus and showed resistance to Coxsackievirus-induced myocarditis (29). Disruption of the eotaxin gene has been shown by Rothenberg et al. to suppress the early eosinophil recruitment after Ag challenge, but not the late eosinophil influx (30), although in a different eotaxin knockout system, the eosinophil influx in the bronchoalveolar lavage induced by OVA challenge was not modified compared with that in wild-type mice (31).
Several lines of evidence indicate that epithelial cells play a crucial role in regulating chemokine production and function in the airways. Localization of infiltrating cells to the epithelial region (18, 32) strongly implicates epithelial cells as a relevant source of these chemoattractants. In fact, immunohistochemistry studies of RANTES, MCP-1, MCP-4, and eotaxin show that epithelium is among the most heavily staining cell types, if not the most heavily staining cell type in biopsies of both upper and lower airways of humans and mice (17, 18, 19, 22, 26, 33, 34). Also, numerous in vitro studies have demonstrated that airway epithelial cells produce substantial quantities of RANTES and other C-C chemokines (9, 35, 36, 37). Epithelial cells are also considered to be among the most relevant cellular targets of topically delivered glucocorticoids (GC).
The aim of the present study was to test the hypothesis that the epithelial-derived expression of the C-C chemokines RANTES, eotaxin, and MCP-4 can be differentially regulated by inflammatory and immunomodulatory cytokines as well as by GC, leading to distinct patterns of expression.
Materials and Methods
The following reagents were purchased from the indicated sources: Ham’s F-12 (BioWhittaker, Walkersville, MD), LHC-8 (Biofluids, Rockville, MD), DMEM, Ca2+- and Mg2+-free HBSS, heat-inactivated FCS, vitrogen (Collagen Biomaterials, Palo Alto, CA), l-glutamine, penicillin/streptomycin solution, agarose, formamide, random priming labeling kit (Life Technologies, Gaithersburg, MD), Opti-MEM (Life Technologies, Gaithersburg, MD), RNAzol B (Tel-Test, Friendswood, TX), chloroform, isopropanol, formaldehyde (Fisher Scientific, Fernwood, NJ), DMSO, erythrosin B, cycloheximide, actinomycin D, (Sigma, St. Louis, MO), GeneScreen Plus membranes and [α-32P]dATP (DuPont-New England Nuclear, Boston, MA), FuGene (Boehringer Mannheim, Indianapolis, IN), luciferase detection kit (Analytical Luminescence Laboratory, Ann Arbor, MI), Bio-ray reagent (Bio-Rad, Hercules, CA). Human recombinant TNF-α and IL-4 (R & D, Minneapolis, MN), IFN-γ (Genzyme, Cambridge, MA), and ExpressHyb solution (Clontech, Palo Alto, CA). Budesonide was a gift from Drs. Per Andersson and Ralph Brattsand (Astra Draco, Lund, Sweden) and was stored as a 0.1-M stock in DMSO at −20°C.
Isolation of primary bronchial epithelial cells (PBEC)
Culture of BEAS-2B cells and PBEC
The BEAS-2B cell line, derived from human bronchial epithelium transformed by an adenovirus 12-SV40 hybrid virus (40), was supplied by Dr. Curtis Harris. BEAS-2B cells and PBEC were cultured in 75-cm2 tissue culture flasks uncoated (BEAS-2B) or coated with collagen (PBEC) and maintained in F-12/DMEM medium containing 5% heat-inactivated FCS, 1% l-glutamine, penicillin (100 U/ml), and streptomycin (100 mg/ml). This medium is referred to as complete medium. BEAS-2B cells were used from passages 36–45. Each batch of PBEC was seeded in multiple T-75 flasks and used only at their first passage. Cells were cultured at 37°C with 5% CO2 in humidified air. When the cells reached 80–90% confluence, they were stimulated according to the experimental protocols, then harvested with 0.02% trypsin/EDTA in HBSS and washed twice in HBSS, with a recovery of ∼5–8 × 106 cells/flask. The viability of both PBEC and BEAS-2B cells, assessed by staining with erythrosin B, was consistently >95% of the cells harvested.
Northern blot analysis
Total RNA was extracted using the RNAzol B reagent according to a previously described method (41). Northern blot analysis was performed as previously described (9, 36). To compare the expression of eotaxin, MCP-4, and RANTES mRNA from the same cells, three aliquots, 20 μg each, of the total RNA samples were run in parallel on the same 1% agarose/6% formaldehyde gel. Particular effort was made to maintain equal amounts of RNA among the aliquots by measuring the RNA concentration spectrophotometrically and comparing the results with the intensity of ethidium bromide-stained 28S and 18S RNA bands observed with 1 μg (based on the optical density reading) of total RNA for each condition loaded on a 1% agarose gel. After separation by electrophoresis, RNA samples were blotted onto a single piece of GeneScreen Plus nylon membrane by a 1-h transfer with a positive pressure blotter (Stratagene, La Jolla, CA). After the transfer, the membrane was carefully cut into three pieces, each containing one of the three mRNA aliquots. The three membranes were prehybridized for 45 min and then hybridized for 90 min with the three 32P-labeled cDNA probes (for eotaxin, MCP-4, and RANTES) using Expresshyb hybridization solution. The same amounts of eotaxin, MCP-4, and RANTES cDNA probes (100 ng) were labeled using the random priming method. This procedure yields 32P-labeled probes with comparable specific activities, and the same amount of radioactive probe (1 × 106 cpm/ml of hybridization buffer) for each chemokine was used. After hybridization, membranes were washed with 2× SSC/0.2% SDS at room temperature (twice, 15 min each time), then with 2× SSC/0.2% SDS at 65°C (four times, 15 min each time), and finally with 0.2× SSC/0.2% SDS at 65°C (twice, 10 min each time). Membranes were then exposed to the same x-ray film for equal exposure time. Autoradiographs were quantified by video densitometry using a gel documentation system configured by UVP (San Gabriel, CA) interfaced with a Macintosh Centris 610 containing Image 1.60 software (National Institutes of Health Public Software, Bethesda, MD). Results are shown as the ratio of chemokine/GAPDH (housekeeping gene) densitometric units. BEAS-2B cells and PBEC showed low, but detectable, constitutive levels of eotaxin and MCP-4 mRNA in 6 of 20 and 3 of 20 experiments, respectively, whereas constitutive RANTES mRNA was never detected. The probes used were a PCR-amplified fragment spanning 290 bp of the coding region of RANTES, an EcoRI-XhoI fragment spanning 312 bp of the coding region of MCP-4, a BamHI fragment spanning 260 bp of the coding region of eotaxin, and a cDNA probe (1100 bp) for GAPDH (Clontech).
Eotaxin and RANTES protein levels in cell supernatants were assayed by commercially available ELISA kits (R & D Systems). The limits of detection in the assays were 5 and 2 pg/ml for eotaxin and RANTES, respectively.
RANTES and eotaxin promoter constructs
The pRANTES/Luc-884 luciferase reporter construct has been prepared by subcloning bp −884 to +64 of human RANTES from the plasmid PCR 1000 (a gift from Dr. Tom Schall) into the HindIII-XhoI sites of pGL2-Basic (Promega, Madison, WI). The promoter region of eotaxin was amplified by a modification of 5′ rapid amplification of cDNA ends using the Genome Walker Kit (Clontech, Palo Alto, CA). PCR product was generated from adaptor-ligated genomic DNA fragments using a 5′ adaptor primer and a primer complementary to a sequence from the first exon of eotaxin (5′-TAGCAGCTGCCTTCAGCCCCCAGGGG-3′) (16), then used as template for a nested PCR reaction using the adaptor primer and a primer derived from the eotaxin 5′-upstream sequence (5′-ACTTCTGTGGCTGCTGCTCATAG-3′). A 1.4-kb product was cloned into a TA vector (Invitrogen, San Diego, CA), and its sequence was verified by the dideoxy method (Johns Hopkins DNA Analysis Facility, Baltimore, MD). The resulting plasmid (referred to as TA-Eotax. 1400 Vector) was used as template to amplify a region of the eotaxin promoter spanning bp −1363 to −1 using a 5′ primer containing a restriction site for MluI (5′-ACTATAGGGCACGCGTGGT-3′) and a 3′ primer containing a restriction site for BglII (5′-GAAGATCTCAGCCTCTCTGCTCCTC-3′). The PCR product was cloned into the MluI and BglII sites of pGL3-Basic (Promega, Madison, WI), and the resulting plasmid is referred to as pEotax/Luc-1363.
Cell transfections and analysis of reporter gene expression
BEAS-2B cells were subcultured in six-well plates in complete medium at a density of ∼200,000–250,000/well, then transfected 24 h later, at about 50–60% confluence, using the nonliposomal cationic vehicle FuGene. Briefly, 3 μl of FuGene was resuspended (5 min at room temperature) in serum-free Opti-MEM, then allowed to complex with 1 μg of plasmid DNA for 15 min at room temperature. The plasmid/FuGene mixture (100 μl/well) was overlayed on the cells in a final volume of 2 ml complete medium. After incubation for 4–24 h at 37°C, cells were treated according to the experimental protocols indicated in the text, then washed twice with 1× PBS and lysed using a 1× lysis buffer. The total protein content was assayed in 10 μl of each lysate using a modified Bradford reagent. Luciferase activity in each sample was measured by light emission and expressed as relative luciferase units normalized to total protein.
Analysis of data was performed using StatView II software (Abacus Concepts, Berkeley, CA). Data are expressed as the mean ± SEM. Statistical analysis between pairs was performed using the nonparametric Mann-Whitney U test. A p value <0.05 was considered significant.
Kinetics and concentration dependence of induction of eotaxin, MCP-4, and RANTES mRNA by TNF-α and IFN-γ
To characterize and compare the kinetics of expression of chemokines by epithelial cells, we stimulated epithelial cells with TNF-α and IFN-γ (100 ng/ml each) for increasing periods of time and analyzed mRNA levels by Northern blot. Upon stimulation with TNF-α and IFN-γ, eotaxin and MCP-4 mRNA were very rapidly up-regulated, being detectable as early as 30 min after stimulation (not shown) and reaching maximum expression levels within the first 2–8 h of stimulation, respectively (Fig. 1,A). In contrast, induction of RANTES mRNA displayed a slow kinetic, being detectable after 2–4 h and reaching maximum expression only at 18–24 h. Similar results were obtained for eotaxin and RANTES with freshly isolated PBEC (Fig. 1,B) and using TNF-α alone as the stimulus (n = 3; data not shown). To further explore the response to TNF-α and IFN-γ, we performed concentration-response experiments, incubating BEAS-2B cells with increasing concentrations of TNF-α (1, 10, and 100 ng/ml) alone or in combination with a fixed concentration (100 ng/ml) of IFN-γ for 8 h, a time point at which all three TNF-α-induced chemokine mRNAs were detectable. The results shown in Fig. 2 reveal that TNF-α was 10- to 100-fold more potent in inducing eotaxin and MCP-4 than RANTES. In fact, 1 ng/ml of TNF-α was near maximal for both eotaxin and MCP-4, but 100 ng/ml was required for maximal induction of RANTES. The same concentration-response curve to TNF-α observed for RANTES mRNA expression after 8 h was maintained after 24 h of TNF-α treatment (36). IFN-γ alone induced small amounts of mRNA for eotaxin and MCP-4, but not RANTES. We have previously reported (36) that stimulation of BEAS-2B cells with IFN-γ induces small amounts of RANTES mRNA and protein, but it requires longer (18–24 h) incubation times. As expected (9, 36), IFN-γ produced a strong potentiation of TNF-α-induced RANTES and MCP-4 mRNA (Fig. 2), whereas potentiation of eotaxin expression by IFN-γ was less pronounced.
RANTES and eotaxin protein secretion induced by TNF-α and IFN-γ
The characteristics of RANTES and eotaxin protein secretion, assessed by ELISA in the supernatants of BEAS-2B cells, were similar. The data in Fig. 3 demonstrate that secretion of eotaxin induced by TNF-α preceded RANTES secretion, becoming significant at 4 h after stimulation, whereas RANTES secretion was significantly elevated only after 18 h of incubation. The addition of IFN-γ to TNF-α induced a faster production of RANTES and resulted in a 3-fold increase in RANTES secretion. In contrast, IFN-γ did not affect the kinetics or the amount of eotaxin produced by BEAS-2B cells in response to TNF-α stimulation.
Protein synthesis requirement for eotaxin, MCP-4, and RANTES mRNA expression
The time- and stimulus-dependent differences in chemokine expression reported above indicate potential differences in mechanisms regulating their synthesis. To evaluate regulatory mechanisms of chemokine production, we stimulated cells with TNF-α for 8 h (100 ng/ml) in the absence or the presence of a 2-h preincubation with the protein synthesis inhibitor cycloheximide (5 μg/ml). Induction of mRNA for eotaxin and MCP-4 was only weakly inhibited after cycloheximide treatment, while RANTES mRNA production was significantly reduced (Fig. 4). These results suggest that induction of an intermediate protein (e.g., a transcription factor) may be necessary for expression of RANTES mRNA, but not for that of eotaxin or MCP-4.
Differential regulation of eotaxin, MCP-4, and RANTES by IL-4
Incubation of airway epithelial cells with IL-4 induced the expression of eotaxin and, to a lesser extent, of MCP-4 mRNA (Fig. 5,A). Interestingly, in the same experiments we observed no induction of RANTES mRNA by IL-4. This result has been reproduced in freshly isolated PBEC (Fig. 5,B). Concentration-response experiments in BEAS-2B cells (10–100 ng/ml; not shown) revealed that chemokine mRNA expression was detectable after challenge with 10 ng/ml of IL-4, as documented by others in human fibroblasts (42), and was maximal when 50 ng/ml of IL-4 was used. We then incubated epithelial cells in the presence of TNF-α (100 ng/ml), alone or in combination with IL-4 (50 ng/ml), for 18 h. Densitometric analysis revealed that IL-4 consistently induced an increase in TNF-α-induced eotaxin mRNA levels (36 ± 15%) and MCP-4 mRNA (50 ± 38%), while reducing by 27 ± 7% RANTES mRNA, although these changes were not statistically significant. A more striking differential effect of IL-4 on eotaxin and RANTES production was found at the protein level (Fig. 6). When used alone as a stimulus for 18 h, IL-4 induced a significant release of eotaxin from BEAS-2B cells (274 ± 95 pg/ml), but not RANTES, while TNF-α stimulated the secretion of both eotaxin and RANTES (582 ± 313 and 8689 ± 3155 pg/ml, respectively). Furthermore, IL-4 caused a striking (7.6-fold) increase in TNF-α-induced eotaxin release (4402 ± 1903 pg/ml), while it did not affect RANTES secretion. IFN-γ alone did not induce eotaxin or RANTES protein release (n = 2; not shown), but it selectively potentiated TNF-α-induced RANTES release (Fig. 6; see also Fig. 3). Taken together, these results suggest that the Th2-derived cytokine IL-4 up-regulates eotaxin, but not RANTES, while the Th1-derived cytokine IFN-γ enhances the up-regulation of RANTES, but not that of eotaxin.
Effect of budesonide on epithelial-derived eotaxin, MCP-4, and RANTES
Studies from our laboratory and others have demonstrated that epithelial production of chemokines is inhibited by GC (9, 36). We first compared the effect of the potent topical GC budesonide on the epithelial expression of eotaxin, RANTES, and MCP-4 mRNA. Epithelial cells were treated for 24 h with a concentration of budesonide (10−7 M) previously shown to maximally suppress epithelial-derived RANTES and MCP-4 (9, 36) and then were stimulated with TNF-α (100 ng/ml), alone or in combination with IFN-γ or IL-4 for 18 h. Treatment of cells with the budesonide diluent, DMSO, tested at an amount equivalent to that contained in the budesonide preparation, did not alter basal or TNF-α-induced chemokine expression (n = 2; data not shown). Kinetic experiments, in which BEAS-2B cells were pretreated with 10−7 M budesonide at various times before, simultaneously with, and after stimulation with TNF-α revealed that inhibition of chemokine mRNA expression was similar even if budesonide was added up to 6 h after cytokine stimulation (data not shown). Budesonide inhibition of the expression of eotaxin, MCP-4, and RANTES mRNA varied substantially. TNF-α-induced RANTES mRNA appearance was inhibited to the greatest extent (80.9 ± 3.3% inhibition); MCP-4 mRNA was also inhibited significantly, although to a lesser extent (43.5 ± 4.9% inhibition), while eotaxin mRNA was reduced only by 25.5 ± 8.2% (Fig. 7,A). In cells activated with TNF-α and IFN-γ, the effect of budesonide was less marked, although the same rank of sensitivity was observed among the three chemokines (data not shown). In contrast to TNF-α, when IL-4 was used as a stimulus, eotaxin mRNA expression was substantially diminished by GC treatment (Fig. 7,B). Similar results to those in Fig. 7, A and B, were obtained in PBEC (n = 2; data not shown). Despite the relative resistance of the expression of eotaxin mRNA to budesonide, eotaxin protein production was well inhibited by the GC (Fig. 8).
Effects of cytokines and budesonide on RANTES and eotaxin promoter activity and chemokine mRNA decay
To further dissect the underlying molecular mechanisms responsible for the differential regulation of epithelial chemokines by cytokines and budesonide, we compared the effects of cytokine stimulation and treatment with budesonide in BEAS-2B cells transfected with either RANTES or eotaxin promoter-driven luciferase reporter constructs (pRANTES/Luc-884 and pEOTX/Luc-1363). Transfected cells were incubated in the presence of medium containing the DMSO diluent or budesonide (10−7 M) and challenged with increasing concentrations of TNF-α (0.1–100 ng/ml), IL-4 (50 ng/ml), or IFN-γ (10 ng/ml), alone or combined with TNF-α (10 ng/ml). In cells transfected with pEOTX/Luc-1363 (Fig. 9, left panel), TNF-α elicited a 2-fold increase in luciferase levels at 0.1 ng/ml, with only a minimal further increase observed at 1 and 10 ng/ml (2.2- and 2.3-fold increases, respectively). TNF-α produced a concentration-dependent increase in RANTES promoter-driven luciferase production as well (Fig. 9, right panel). As observed with Northern blot analysis, higher concentrations of TNF-α (∼10- to100-fold) were required to maximally activate the RANTES promoter than the eotaxin promoter construct. Also in agreement with Northern blot and ELISA results, IL-4 stimulation produced an increase (2.1-fold) in luciferase in cells transfected with pEOTX/Luc-1363, while it did not affect luciferase levels in cells transfected with pRANTES/Luc-884. Furthermore, IL-4 increased TNF-α-induced luciferase production in cells transfected with pEOTX/Luc-1363 (>5-fold), but not in RANTES/Luc-884-transfected cells. In fact, IL-4 significantly inhibited TNF-α-induced luciferase production in cells transfected with pRANTES/Luc-884 (Fig. 9). Finally, as observed with Northern blot and ELISA, IFN-γ significantly potentiated TNF-α-induced luciferase production in cells transfected with pRANTES/Luc-884, but not in cells transfected with pEOTX/Luc-1363. The sensitivities to inhibition by GC of the cytokine-induced response driven by the eotaxin and RANTES promoter also differed. Budesonide treatment inhibited the TNF-α-mediated RANTES promoter activity more markedly than the eotaxin promoter activity (Fig. 9).
GC have been shown to influence the rate of decay of mRNA for several cytokines. To determine the effect of cytokines and GC on chemokine mRNA stability, we treated BEAS-2B cells with budesonide (10−7 M) or DMSO diluent for 24 h before stimulation with TNF-α and IFN-γ (100 ng/ml each) for 8 h. Cells were then either harvested as controls or treated with the transcription inhibitor actinomycin D (3 μg/ml) for increasing times to determine the decay of mRNA for eotaxin, MCP-4, and RANTES. For all three chemokines, the cytokine-induced mRNA were very stable, with >70% of mRNA still detectable after 8-h incubation with actinomycin D in cells activated by cytokines but not treated with budesonide (Fig. 10). Budesonide treatment significantly accelerated the decay of eotaxin and MCP-4 mRNA, but not that of RANTES mRNA. The same results were obtained when TNF-α alone was used as the stimulus (data not shown).
Numerous studies have documented an increased production of several eosinophil-active C-C chemokines at the site of allergic inflammatory reactions (9, 10, 16, 18, 19, 20, 21, 22, 23, 24, 26, 33, 43). Although the respective biological roles of these molecules are still unknown, mounting evidence indicates that these chemokines are quite distinct with respect to their profiles of activity. For example, RANTES, eotaxin, and MCP-4 are all potent eosinophil chemoattractants, but they differ in receptor usage, target cell effects, and cellular sources. RANTES can induce migration of cells expressing CCR1, CCR3, and CCR5. Eotaxin appears to be CCR3 selective, and MCP-4 can activate cells expressing CCR2 as well as CCR3 (44). Chemokine expression in the airways varies significantly in the timing, amount, localization, and regulatory mechanisms (17, 28, 45). In particular, Gonzalo et al. found that the kinetics of accumulation of monocytes/macrophages, T lymphocytes, and eosinophils correlated with distinct patterns of expression of chemokine mRNA in the lung in a mouse model of allergic inflammation. The early influx of macrophages correlated with the expression of MCP-1/JE, while RANTES and eotaxin expression was concomitant with the subsequent infiltration of lymphocytes and eosinophils (17). Thus, it is conceivable that during allergic inflammatory reactions, chemokine production is tightly regulated to produce specific chemokine patterns.
Although it is now well established that multiple C-C chemokines are produced by airway epithelial cells, little is known about the differences in their specific profiles of activation. In this study, we obtained evidence for distinct regulatory pathways for three epithelial-derived, eosinophil-active C-C chemokines: eotaxin, MCP-4, and RANTES. We found remarkable differences in the rate of expression of mRNA for these three chemokines. Eotaxin and MCP-4 were rapidly induced by TNF-α and IFN-γ and reached near-maximal mRNA expression within 2 h, while RANTES mRNA induction was much slower, reaching peak expression only after 18–24 h. The rapid induction of eotaxin mRNA levels is in agreement with findings using the eotaxin knockout mouse model, in which the earliest recruitment of eosinophils after allergen challenge in sensitized animals is selectively ablated (30). We also found significant differences in the sensitivity to TNF-α among the three chemokines, with eotaxin and MCP-4 mRNA being almost maximally induced by 1 ng/ml of TNF-α, whereas RANTES required 10- to 100-fold higher concentrations to reach maximal induction. It is conceivable, therefore, that in vivo, low concentrations of TNF-α could selectively induce eotaxin and MCP-4. In contrast, the degree of synergism of TNF-α and IFN-γ in inducing RANTES was much greater than that observed for eotaxin and MCP-4, especially at protein level. Another factor distinguishing the expression of RANTES mRNA from that of eotaxin and MCP-4 in epithelial cells is the protein synthesis requirement. The relative lack of de novo protein synthesis requirement together with the rapid kinetics of activation indicate that eotaxin and MCP-4 are immediate-early genes, whereas RANTES, being expressed late after cell activation and requiring de novo protein synthesis, is not.
A remarkable feature distinguishing RANTES from eotaxin and MCP-4 production is the response to IL-4. We found that both eotaxin and MCP-4 were up-regulated by incubation of epithelial cells with IL-4, whereas RANTES was not. Furthermore, IL-4 induced a striking potentiation of TNF-α-induced eotaxin, but not RANTES release. In further support of our findings, IL-4-induced eosinophil accumulation in various mouse models appears to be at least partially mediated by the endogenous production of eotaxin. Eosinophil accumulation induced by intradermal injection of IL-4 was significantly inhibited by anti-eotaxin Ab treatment (46), anti-IL-4 treatment of mice with Schistosoma (type 2) granulomas in the lungs impaired the local expression of eotaxin mRNA (47), and administration of IL-4, IL-13 and Th2 supernatants intranasally to naive mice induced lung eosinophilia and eotaxin expression in airway epithelium (34). In vitro, IL-4 induces the selective production of eotaxin in human fibroblasts (42). The divergent regulation of eotaxin and RANTES by IL-4 may have a relevant role in the establishment of chronic allergic inflammation. It can be hypothesized that in vivo, epithelial chemokine expression might be determined by the profile of cytokine released in the microenvironment. The coordinated release of TNF-α from macrophages and mast cells and of IL-4 from adjacent cells, particularly Th2 lymphocytes, may act as a strong regulatory signal for the selective expression of eotaxin from epithelium. In support of this hypothesis, MacLean et al. found that the increased eotaxin production induced in the lungs by OVA challenge is blocked by anti-CD3 Ab treatment of the animals, suggesting that eotaxin production might be driven by Ag-specific T cells (28). The Th1 cytokine, IFN-γ, had a profile of activity in stark contrast with that of IL-4. Although IFN-γ profoundly potentiated the production of RANTES protein and mRNA induced by TNF-α, it failed to increase the activity of the eotaxin promoter luciferase construct, failed to increase eotaxin protein production induced by TNF-α, and only modestly increased mRNA detected by Northern blot. Taken together, these data indicate that the Th2 cell-derived cytokine IL-4 may lead to induction of eotaxin expression (and perhaps even suppression of RANTES; see Fig. 9 and Ref. 48), while the Th1 cytokine IFN-γ has the opposite effect, namely, potentiation of RANTES expression with little or no effect on eotaxin expression.
Several potential cis-regulatory elements in the promoter regions of the RANTES and eotaxin genes have been identified (49, 50, 51, 52). The human RANTES promoter contains four NF-κB binding sites (53), and disruption of any of these elements dramatically reduces RANTES promoter activity in T cells (50). The critical involvement of NF-κB in RANTES expression in epithelial cells is consistent with the exquisite GC sensitivity of this gene. The progressive accumulation of RANTES mRNA over at least 24 h after stimulation and the protein synthesis requirements of the response are suggestive of the later involvement of additional factors mediating sustained transcription of the gene. One possible candidate responsible for the delayed maximal expression of RANTES mRNA is STAT-1, which has been shown to cooperate with NF-κB in the synergistic activation of the RANTES gene by TNF-α and IFN-γ in murine fibroblasts (54). The higher sensitivity to TNF-α stimulation observed with eotaxin and MCP-4 may be due to different transcriptional activation patterns, such as differences in the affinity of transcription factors for cis-regulatory elements within the chemokine promoter regions. Regarding the ability of IL-4 to stimulate eotaxin expression, it is noteworthy that at least four potential STAT-6 binding elements can be recognized within the proximal 1 kb of the human eotaxin promoter (51, 52).
The inhibitory effect of topical GC on epithelial-derived inflammatory mediators is increasingly viewed as an important feature of the clinical efficacy of inhaled GC. Our data suggest that the GC inhibition of epithelial chemokine expression is a complex process, possibly involving multiple pathways with different sensitivities to GC as well as different mechanisms of gene repression. It is now known that GC can down-regulate gene expression by at least three different molecular mechanisms: binding of ligand-activated GC receptor to negative GC-responsive elements present in the promoter regions of target genes; interference with the DNA binding activity and/or trans-activation potential of transcription factors, such as NF-κB and AP-1, by protein-protein interaction with ligand-activated GC receptor; and posttranscriptional regulatory mechanism(s), via destabilization of mRNA transcripts and degradation of protein products via mechanisms yet to be fully elucidated (55). Each of these mechanisms has been implicated or can be speculated for the GC inhibition of chemokines produced by epithelium. A GRE is present in the promoter region of eotaxin (51, 52), but its influence is unknown. The transcription factors AP-1 and NF-κB are known to be involved in the up-regulation of several chemokines induced in epithelial cells, including RANTES (50) and eotaxin (56). Recently, GC-mediated inhibition of AP-1 and NF-κB reporter gene constructs has been reported in BEAS-2B cells (57). Inhibition of IL-8 expression by GC has been shown to occur through GR-mediated repression of NF-κB function (58). Expression of C-C chemokines induced by TNF-α, such as RANTES, MCP-4, and eotaxin, could also be inhibited by GC through this process. Based on our results, it can be hypothesized that the GC effect on epithelial-derived chemokines is mediated by multiple inhibitory mechanisms, acting at both transcriptional and posttranscriptional levels, that contribute to a different degree to the suppression of each chemokine. We have previously demonstrated that the expression of RANTES mRNA and protein in epithelial cells was significantly inhibited by GC without an effect on RANTES mRNA half-life, leading us to hypothesize that suppression occurs mainly at the transcriptional level (36). This hypothesis is supported by the finding in the present study that budesonide suppressed luciferase expression from a construct driven by the RANTES promoter. We did find, however, that budesonide selectively destabilized eotaxin and MCP-4 mRNA in epithelial cells, in contrast to RANTES. In the case of eotaxin, one possible explanation of these results may be the influence of destabilizing sequences, known as AU-rich elements (ARE), in the 3′ untranslated region (UTR) of the transcripts (59), acting as binding sites for trans-acting factors (60, 61, 62, 63). It has been shown that GC can influence mRNA decay via AREs (64). Based upon published sequences, the eotaxin mRNA 3′ UTR contains two AUUUA motifs (37). Interestingly, no ARE are present in the 3′ UTR of MCP-4 and RANTES (10, 65). Other regulatory cis-elements present in the MCP-4 3′ UTR or in other regions of the unspliced, nuclear form of the mRNA may be targets of GC regulation. Non-ARE mRNA destabilizing elements have recently been identified (66, 67). Although the effect of budesonide on eotaxin mRNA level measured by Northern blot was modest, it suppressed transcription of luciferase in the eotaxin promoter construct and destabilized eotaxin mRNA. Together these effects led to a profound suppression of eotaxin protein release in epithelial cells.
Taken together, our results indicate the existence of independent mechanisms of regulation of epithelial chemokine production that can lead to different profiles of chemokines according to the nature and the timing of the stimulation. Furthermore, we propose that GC suppress the expression of these important genes by both transcriptional and posttranscriptional mechanisms.
We thank Mary E. Brummet, James R. Plitt, and Curt Reynolds for skilled technical assistance, and Dr. Vincenzo Casolaro for critical discussions of the manuscript.
This work was supported by National Institutes of Health Grants RO1AI44885 and RO1AR31891 and a gift from ASTRA Pharmaceutical Co.
Abbreviations used in this paper: MCP, monocyte chemoattractant protein; ARE, AU-rich elements; GC, glucocorticoids; PBEC, primary bronchial epithelial cells; UTR, untranslated region; pRANTES, RANTES promoter plasmid; pEOTX, eotaxin promoter plasmid; BUD, budesonide.