Transcriptional Mechanisms Regulating Alveolar Epithelial Cell-specific CCL5 Secretion in Pulmonary Tuberculosis*

CCL5 (or RANTES (regulated upon activation, normal T cell expressed and secreted)) recruits T lymphocytes and monocytes. The source and regulation of CCL5 in pulmonary tuberculosis are unclear. Infection of the human alveolar epithelial cell line (A549) by Mycobacterium tuberculosis caused no CCL5 secretion and little monocyte secretion. Conditioned medium from tubercu-losis-infected human monocytes (CoMTB) stimulated significant CCL5 secretion from A549 cells and from primary alveolar, but not upper airway, epithelial cells. Differential responsiveness of small airway and normal human bronchial epithelial cells to CoMTB but not to conditioned medium from unstimulated human monocytes was specific to CCL5 and not to CXCL8. CoMTB induced CCL5 mRNA accumulation in A549 cells and induced nuclear translocation of nuclear factor (cid:1) B (NF (cid:1) B) subunits p50, p65, and c-rel at 1 h; nuclear binding of activator protein (AP)-1 (c-Fos, FosB, and c-Jun) at 4–8 h; and binding of NF-interleukin (IL)-6 at 24 h. CCL5 promoter-reporter analysis using deletion and site-specific mutagenesis constructs demonstrated a key role for AP-1, NF-IL-6, and NF (cid:1) B in driving CoMTB-induced promoter activity.

CCL5 1 is a member of the CC chemokine subfamily and a chemoattractant for CD4ϩ memory T lymphocytes, monocytes, and eosinophils (1,2). Originally described as restricted to activated T lymphocytes, CCL5 is expressed by many cell types including fibroblasts, renal and pulmonary epithelium, endothelium, and airway smooth muscle (3)(4)(5)(6). Tuberculosis is principally a pulmonary disease causing 3 million deaths each year (7), yet the regulation of cellular influx to pulmonary granuloma is poorly defined. Tuberculous granulomas contain cell types potentially recruited by CCL5, such as antigen-specific T lymphocytes and cells of the monocyte/macrophage lineage. Macrophages from human tuberculous lymph node granulomas express both CCL5 protein and gene (8). Furthermore, anti-CCL5 antibodies decrease pulmonary granuloma lesion size in Mycobacterium bovis BCG strain-infected mice, suggesting a functional role for CCL5 in murine mycobacterial granulomas (9). CCL5 concentrations in BALF from infected patients rise acutely, correlating with BALF CD4ϩ T lymphocyte counts, and fall during convalescence (10,11). However, in vitro studies have demonstrated that human alveolar macrophages and monocytes infected with Mycobacterium tuberculosis secrete only low concentrations of CCL5 (11).
The pulmonary epithelium is the initial barrier to respiratory infection by M. tuberculosis. Epithelial cells, covering the entire alveolar surface area of the lung of ϳ70 m 2 (12), may contribute to host defense by chemokine production (13), by adhesion molecule expression (14), and possibly by antigen presentation via HLA-DR expression (15). CCL5 may be secreted by pulmonary epithelial cells in response to proinflammatory cytokines (5,13,16,17), hyperosmolarity (18), or pathogens such as RSV (19,20). However, virulent M. tuberculosis, strain H37-Rv, which stimulates epithelial cell CXCL8 (IL-8) secretion (21), does not cause CCL5 release (22) despite invasion and replication of the pathogen within alveolar epithelial cells (23). Thus, direct infection of alveolar macrophages, monocytes, or epithelial cells by tuberculosis cannot explain the source of CCL5 detected in the BALF of patients with tuberculosis. We hypothesized that cellular networks between lung epithelium and monocytes involving proinflammatory cytokines were central to CCL5 secretion in pulmonary tuberculosis. Monocyte-derived IL-1␤ may be critical in such networks as polymorphisms in the IL-1 locus can affect relative IL-1␤ and IL-1ra monocyte responses to M. tuberculosis (24), and IL-1␤ and IL-1ra imbalances may affect outcomes of chronic diseases (25).
The mechanisms regulating CCL5 secretion in lung epithelium are not completely understood but may be cell type-and stimulus-specific. Transcriptional regulation of the CCL5 gene is critical in T lymphocytes (1,2,26,27). The CCL5 promoter contains response elements for the transcriptional activators NFB, AP-1, and NF-IL-6 (26,27) and for late acting factors restricted to T cells (28). NFB comprises a family of Relrelated proteins retained in the cytoplasm bound to the inhib-itor IB␣ and is a pivotal regulator of proinflammatory responses (29). Following cellular activation, IB␣ is phosphorylated on specific serine residues (30), ubiquitinated, and degraded (31), allowing NFB to pass into the nucleus where binding to CCL5 B binding sites may occur. NF-IL-6, originally identified as a nuclear factor binding to an IL-1response element in the human IL-6 gene, and the AP-1 complex, comprised of Jun/Fos heterodimers or Jun/Jun homodimers, are other key regulators in cytokine gene expression (32)(33)(34). RSV-induced CCL5 gene expression in lung epithelium is associated with the nuclear translocation of NF-IL-6 (33). However, little data exists on the functional role of NF-IL-6 or AP-1 in cytokine-induced CCL5 gene activation, although NFB may be required (16,27,35).
We demonstrate for the first time that pulmonary epithelial cells are an important source of CCL5 following in vitro infection of human monocytes by live, virulent M. tuberculosis. The stimulus for CCL5 secretion is IL-1␤, released after monocyte infection by M. tuberculosis and not direct epithelial infection. Such CCL5 secretion required new gene transcription and translocation of NFB subunits p50, p65, and c-rel to the nucleus of human primary alveolar epithelium. Promoter-reporter analysis of the CCL5 gene demonstrated that NFB, AP-1, and NF-IL-6 binding was involved in cytokine-stimulated CCL5 gene transcription in pulmonary epithelium together with a distal region of the promoter not identified previously as a key regulatory site. Primary pulmonary epithelial CCL5, but not CXCL8, secretion was differentially responsive to the monocyte-derived IL-1␤ depending on the origin of the epithelium. Primary alveolar epithelium of the distal airways secreted CCL5 in response to the monocyte-derived IL-1␤, but the upper airway cells of the trachea and main bronchi did not. Thus, the observed networking effects causing CCL5 secretion were due to monocyte-derived IL-1␤ acting on primary alveolar epithelium. Our data indicate that the vast distal alveolar epithelial cell surface is the major cellular source of CCL5 in pulmonary tuberculosis and that the mechanisms controlling this CCL5 gene expression depend on AP-1 and NF-IL-6 activation as well as NFB.

EXPERIMENTAL PROCEDURES
Human Epithelial Cell and Monocyte Culture-The human pulmonary type II alveolar epithelial A549 cell line (36) (European Collection of Animal Cell Cultures 86012804) was maintained in a humidified 5% CO 2 atmosphere in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal calf serum (endotoxin Ͻ 0.02 ng/ml), 2 mM glutamine, and 10 mg/ml ampicillin (an antibiotic with no significant antimycobacterial activity at this concentration).
SAEC and NHBE were purchased from Clonetics (San Diego, CA) as cryopreserved cells. They were regenerated and maintained according to the supplier's instructions. NHBE were maintained and seeded in bronchial epithelium growth medium containing 13 mg/ml bovine pituitary extract, 0.5 mg/ml hydrocortisone, 0.5 g/ml human recombinant epidermal growth factor, 0.5 mg/ml epinephrine, 10 mg/ml transferrin, 5 mg/ml insulin, 6.5 g/ml triiodothyronine, 50 mg/ml gentamicin, 50 g/ml amphotericin-B. Small airway epithelial growth medium contained the above but less bovine pituitary extract (7.5 mg/ml) and in addition 50 mg/ml bovine serum albumin (fatty acid-free). The medium was changed every other day, and cells were subcultured when 60 -80% confluent. To subculture, the medium was aspirated, and the monolayer was rinsed with Hanks' solution; trypsin/EDTA was added for 5 min and neutralized with trypsin-neutralizing solution. Centrifuged cells were resuspended in medium prior to cell counting and seeding. Cells were discarded after 15 population doublings.
Human monocytes were prepared from pooled buffy coats (North Thames Blood Transfusion Service, Colindale, UK) by density gradient centrifugation on a Ficoll-Paque (Amersham Biosciences). The monocytes were purified by adhesion to tissue culture plastic for 2 h and maintained in RPMI 1640.
Mycobacterial Culture-The virulent strain of M. tuberculosis, H37-Rv (National Collection of Type Cultures, Colindale, UK), was maintained in enriched Dubos medium (Difco). Before use, 1 ml of a mid-logarithmic phase mycobacterial suspension was sonicated (20 s) and left to stand for 10 min, and the top 750 l was used, thus generating single cell mycobacterial suspensions (confirmed by modified Kinyoun staining). Viable mycobacteria were quantitated by colony counting in triplicate on Middlebrook 7H10 plates.
Cytokine Assays-CCL5 or IL-1␤ (R&D Systems) was assayed in cell culture supernatants by ELISA. The lower limit of sensitivity was 15 pg/ml for the CCL5 assay and 7.8 pg/ml for the IL-1␤ assay. CCL5 concentrations are expressed in pg/10 6 cells, and IL-1␤ concentrations are expressed as ng/ml.
Electrophoretic Mobility Shift Assays-Nuclear extracts were prepared using a modified version of an established protocol (41). Following stimulation, cells were washed with cytoplasmic extraction buffer (10 mM Tris-HCl, 60 mM KCl, 1 mM EDTA, 1 mM dithiothreitol) containing protease inhibitors and with 0.15% Nonidet P-40 added. After centrifugation (500 ϫ g), the nuclear pellet was resuspended in nuclear extract buffer (20 mM Tris-HCl, pH 8.0, 400 mM NaCl, 1.5 mM MgCl 2 , 1.5 mM EDTA, 25% glycerol, 1 mM dithiothreitol, and protease inhibitors), incubated on ice, and recentrifuged. The protein was quantitated by Bradford assay (42). 5 g of nuclear extract was mixed with 0.7 ng of ␥-32 P-end-labeled double stranded DNA probe with specific activity greater than 1 ϫ 10 8 cpm/g in DNA binding buffer (10 mM Tris, pH 7.5, 50 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol, 0.25 mg/ml bovine serum albumin, 3 mg of poly(dI-dC), and 5% glycerol) and was resolved on 4% polyacrylamide gels. The oligonucleotides contained the consensus sequence corresponding to the NFB (5Ј-AGTTGAGGGGACTTTCCCAGG-3Ј), AP-1 (5Ј-CTAGTGA-TGAGTCAGCCGGATC-3Ј), or NF-IL-6 (5Ј-TGCAGATTGTGCAATG-TACG-3Ј) binding sites (43). Consensus sequences allow concurrent investigation of the four different NFB binding sites and the four TPA-response elements. Dried gels were autoradiographed at Ϫ70°C overnight. Binding specificity was demonstrated by competition assays using a 10-fold excess of unlabeled probe and by failure to compete out the signal with an excess of unlabeled, irrelevant probe. For supershift assays, 1 g of antibodies either to the NFB subunits p50, p52, p65, Rel B, and c-rel or to c-Fos (a subunit of the unrelated transcription factor AP-1) (Santa Cruz Biotechnology, Santa Cruz, CA) was added to the binding mix before electrophoresis.
ELISA-based Transcription Factor Assay-To investigate the complex binding of the multiple subunits of AP-1, a transcription factor ELISAbased assay (TransAM TM , Active Motif North America, Carlsbad, CA), which is 5-fold more sensitive than EMSA, was performed. Nuclear extracts were added to a 96-well plate containing immobilized oligonucleotides including a TPA-responsive element (TRE, 5Ј-TGAGTCA-3Ј). AP-1 dimers in the nuclear extract bind to this TPA-responsive element and were detected using antibodies against c-Fos, FosB, Fra-1, Fra-2, c-Jun, JunB, or JunD. A secondary antibody conjugated to horseradish peroxidase was added, and the color change was detected by spectrophotometry at 450 nm. Competition experiments demonstrated specificity of binding by adding 20 pmol/well either wild type or mutated AP-1 oligonucleotide before assaying with the c-Fos antibody.

Effect of CoMTB on CCL5 Secretion from Pulmonary Epithelial Cells Compared with Effects of Direct Infection by M. tuberculosis-
CoMTB was used to stimulate A549 cells over 24 h at dilutions of 1:10, 1:100, and 1:1000 in serum-free Dulbecco's modified Eagle's medium. A549 cells stimulated with CoMTB (1:10) secreted 20,342 Ϯ 2142 pg of CCL5/10 6 cells within 4 h and 53,250 Ϯ 2696 pg of CCL5/10 6 cells after 24 h (Fig. 1A). Such CoMTB-induced CCL5 secretion at 24 h was 12-fold greater than from tuberculosis-infected human monocytes, which as expected secreted only 4310 Ϯ 816 pg of CCL5/10 6 cells at 24 h (Fig. 1B). A549 cells directly infected by M. tuberculosis (MOI of 10, causing no significant cytotoxicity (44)) did not release CCL5 unlike the positive control TNF␣ (Fig. 1C). To confirm our findings with CoMTB in primary cells, NHBE were stimulated with CoMTB, but no CCL5 release was detected. Because tuberculosis is primarily an infection of the alveolar airway, we investigated the effect of CoMTB on SAEC, which are harvested from terminal bronchi of 2-mm diameter and below including alveoli consisting of type I and II pneumocytes. In contrast to NHBE, SAEC stimulated with CoMTB did secrete significant CCL5 at concentrations of 809 Ϯ 141 pg/10 6 cells (Fig. 1D), suggesting a differential responsiveness of upper and lower airway epithelium to CoMTB. For this reason, further experiments focused on distal airway cells.
The finding that lower but not upper airway epithelial cells respond to CoMTB by secreting CCL5 appears specific. Thus, both NHBE and SAEC stimulated with CoMTB secrete CXCL8, although the magnitude of the response in the two cell types is different (Fig. 1E). CXCL8 secretion after CoMTB stimulation increased from background levels (because of monocyte-derived CXCL8 in CoMTB) to 16,911 Ϯ 3381 pg/10 6 NHBE. This was over 3.5-fold greater than that found in NHBE after stimulation for 24 h with CoMControl. The absolute concentration of CoMTB-induced CXCL8 secretion from lower airway epithelial cells was higher at 57,004 Ϯ 325 pg/10 6 SAEC, although this represented just a 4.5-fold increase compared with CoMControl at 24 h.
CoMTB Induces Early CCL5 mRNA Accumulation in A549 Cells-Northern analysis demonstrated CCL5 mRNA accumulation by 2 h, peaking at 8 h and present at 24 h post-CoMTB stimulation of A549 cells ( Fig. 2A), with none seen in controls (Fig. 2B). These kinetics are consistent with those for the CCL5 protein secretion. CoMTB exhibited a dose-response effect in stimulating CCL5 mRNA accumulation (Fig. 2C) demonstrating that even at a dilution of 1:100 the CoMTB was capable of inducing significant CCL5 mRNA up-regulation at 24 h. This is consistent with the observed dose-response effect of CoMTB on CCL5 secretion (Fig. 1A).
Activation of NFB and Degradation of IB␣ in CoMTBstimulated A549 Cells-The role of NFB in the control of CCL5 gene expression in pulmonary epithelial cells was analyzed by EMSA. Cell nuclear extracts of A549 and SAEC were stimulated by CoMTB or CoMControl for over a 24-h period. CoMTB stimulated strong NFB nuclear binding in A549 cells at 1-2 h, which decreased at 4 h consistent with the kinetics of onset of CCL5 gene expression (Fig. 3A, upper panel). NFB activation was not observed with CoMControl (Fig. 3A, lower  panel). CoMTB-stimulated SAEC demonstrated specific NFB binding at 1 h post-CoMTB (Fig. 3B, upper panel), which persisted for longer, at least to 8 h, than in A549 cells. Supershift assay showed that the p50 and p65 NFB subunits were strongly activated and that c-rel was weakly activated in the SAEC stimulated by CoMTB after 1 h. No supershift was seen with an irrelevant antibody c-Fos (Fig. 3C). Rapid degradation of IB␣ following A549 cell stimulation by CoMTB was shown by Western analysis (Fig. 3D) with kinetics in keeping with the early but transient nuclear localization of NFB. IB␣ protein disappeared within 5 min, reappeared within 90 min, and was completely resynthesized by 180 min.
Nuclear Translocation of AP-1 and NF-IL-6 in Response to CoMTB Stimulation of A549 and SAEC-Nuclear extracts from A549 and SAEC stimulated by CoMTB or CoMControl for 0, 2, 4, 8, and 24 h were analyzed by EMSA using oligonucleotides corresponding to the NF-IL-6 or AP-1 binding sites. To prevent excess constitutive AP-1 expression, cells were prepared in serum-containing medium (45). Although AP-1 binding was constitutively present in A549 cells, increased nuclear translocation was evident 4 h post-CoMTB stimulation, persisting at lower levels between 8 and 24 h (Fig. 4A). A similar pattern of increase in AP-1 binding was demonstrated with CoMTB-stimulated SAEC (Fig. 4B). The kinetics of NF-IL-6 binding was very different compared with AP-1 or NFB in A549 cells. NF-IL-6 nuclear translocation was seen only after 24 h poststimulation in A549 cells (Fig. 4D) with a relatively low signal intensity. Specificity of binding of each transcription factor was confirmed by competition studies (Fig. 4, A, B, and E). Competition for AP-1 binding was complete in A549 cells; however, in SAEC it appears that there is a small amount of total binding that is nonspecific and is not competed out by a specific probe, although the band intensity is much reduced compared with experiments involving an irrelevant probe (Fig. 4B, 6th and 7th lanes). In terms of specific AP-1 subunits, c-Jun, FosB, and to a lesser extent c-Fos had low level constitutive activity and were activated following the CoMTB but not the CoMControl at 4 h (Fig. 4C). Other members of the Fos/Jun family (Fra-2, JunB) were constitutively activated but were not significantly altered by CoMTB or CoMControl. Signal specificity was shown by an excess of WT AP-1 oligonucleotide CoMTB Activates the CCL5 Promoter and Requires Intact NFB-, AP-1-, and NF-IL-6-response Elements-Nuclear binding of transcription factors provides indirect evidence of their involvement in regulating CCL5 gene activation. To investigate the functional effects of CoMTB, promoter-reporter gene analysis was performed (Fig. 5). Kinetic studies with the WT (Ϫ961) CCL5 promoter demonstrated that CoMTB-induced activation in A549 cells occurred by 2 h (2.5 Ϯ 0.6-fold increase over CoMControl) with maximal activity at 8 h (21.7 Ϯ 5.0-fold increase) (Fig. 5B) and continued until 24 h (10.2 Ϯ 0.8-fold increase). These data are consistent with EMSAs showing binding of first NFB, then AP-1 at 4 h, and finally NF-IL-6 at 24 h. To identify which transcription factor binding sites within the promoter were functionally important, deletion constructs were used (Fig. 5, A and  C). Deletion Ϫ737 reduced CCL5 reporter activity significantly (to 37.48 Ϯ 7.49% of wild type activity). Few consensus binding sites have been identified in this area; some are T cell-specific, whereas others, such as a CCAAT/enhancer-binding protein (C/ EBP) site, are ubiquitous (26). A greater deletion at Ϫ431, which includes removal of the single distal NFB binding site at position Ϫ579, did not reduce promoter activity more than the Ϫ737 deletion (41.9 Ϯ 15.7% versus 37.5 Ϯ 7.5% of wild type, respectively) (Fig. 5C). With the Ϫ181 deletion construct, which has also lost the AP-1-response elements, a further reduction in reporter activity to 20.9 Ϯ 2.9% of WT occurred. With the smallest construct, having 79 bp upstream of the transcriptional start site intact and no longer containing the NF-IL-6 binding site but containing two intact NFB sites, there was a reduction to 5.5 Ϯ 1.9% maximal luciferase activity values, similar to those obtained with the CoMControl stimulus (5.9 Ϯ 1.8% of maximal luciferase activity). These data indicate that AP-1 and NF-IL-6 binding sites may be required for CoMTB-induced CCL5 promoter activity.
Site-directed mutagenesis, however, showed that proximal NFB binding sites (B1 and B2) were important in CCL5 promoter activity with activity decreasing, respectively, to 49.5 Ϯ 7.8% and 20.4 Ϯ 1.72% of WT (Fig. 6). This was not evident from the analysis with deletion constructs presumably because of the significant loss of upstream transcription factor binding sites. ⌬CD28RE/B2/B1 reduced activity to 19.0 Ϯ 4.2%, which is no different from the effect of the ⌬B2 mutation alone and is consistent with the data using the deletion reporter constructs. ⌬NF-IL-6 and ⌬TRE3/4 (Ϫ327) reduced promoter-reporter activity to 49.4 Ϯ 5.9% and 30.8 Ϯ 16.4% of WT, respectively, whereas TRE1/2 had no effect (108.0 Ϯ 32.2%). Mutation of all four TRE sites did not have any additional effect over that of ⌬TRE3/4.
Next CoMTB IL-1␤ concentrations were measured and found to be 2.05 Ϯ 0.50 ng/ml (Fig. 8A). As CoMTB was used in experiments at dilutions of 1:10, the effective IL-1␤ stimulus to epithelial cells will be 0.2 ng/ml. Recombinant IL-1␤ at 0.2-2 ng/ml induced levels of CCL5 secretion from SAEC (Fig. 8B) comparable with levels of secretion from CoMTB at a 1:10 dilution (Fig. 1D). This confirms IL-1 as the critical mediator in the monocyte epithelial cell network resulting in CCL5 secretion from pulmonary epithelial cells. DISCUSSION In this study we have demonstrated that the large alveolar epithelial cell surface area is a significant contributor to production of the chemokine CCL5 during the pulmonary immune response to M. tuberculosis. Epithelial cell CCL5 secretion occurs as a result of monocyte-derived IL-1 released following phagocytosis of M. tuberculosis. We confirmed that epithelial cells infected directly by M. tuberculosis do not secrete significant concentrations of CCL5 (22). In contrast, when the alveolar epithelial cell line A549 is stimulated by CoMTB, high level CCL5 secretion occurs. A similar response of lower magnitude in human primary alveolar epithelial cells was observed. Alveolar macrophages and monocytes are an important source of immunoregulatory mediators in pulmonary tuberculosis (8,47,48). Although there are relatively few macrophages/alveolus, monocytes are recruited early to sites of tuberculous infection in significant numbers. We found that monocytes infected with M. tuberculosis at an MOI of 10:1 secrete CCL5 at only slightly  7th lane), competed out NFB binding using SAEC nuclear extracts. C, SAEC nuclear extracts harvested at 1 h were analyzed by supershift assay using antibodies to NFB subunits p65, p50, p52, c-rel, and Rel B and to the c-Fos subunit of AP-1 (an irrelevant antibody). The p65 and p50 NFB subunits were supershifted strongly, and c-rel was supershifted weakly, with no supershift seen using the irrelevant c-Fos antibody. D, cytoplasmic proteins from CoMTB-or CoMControl-stimulated A549 cells, stimulated over 180 min, were analyzed for IB␣ protein expression by Western blotting. IB␣ was rapidly degraded at 5 min and reformed at 90 min. No IB␣ degradation was seen in the control. higher concentrations than from CoMTB-stimulated SAEC. Even lower concentrations of monocyte-derived CCL5 secretion occurs in response to lower MOIs of M. tuberculosis such as are likely to be present in infected patients (11). The very large surface area of alveolar epithelium suggests therefore that the predominant source of CCL5 in the BALF during pulmonary tuberculosis will be epithelial cells exposed to secreted mediators from infected monocytes. The CCL5 secreted by pulmonary epithelium will stimulate the influx of T lymphocytes and further monocytes, critical for mycobacterial elimination and granuloma formation. CCL5 secretion from CoMTB-stimulated epithelial cells was high at 24 h consistent with the kinetics of CCL5 protein expression in pulmonary granulomas in murine models of mycobacterial inflammation (9).
We found that CCL5 secretion by CoMTB-stimulated pulmonary epithelium requires new gene transcription. In contrast, studies have shown that CCL5 secretion from tuberculosisinfected monocytes involved release of preformed, stored chemokine (11). In addition, posttranscriptional regulation of CCL5 gene expression, which we have not studied, may be important in RSV-induced CCL5 activation (49,50). Differential control of CCL5 gene activation depends on transcriptional regulators and may occur in a cell type-and stimulus-specific fashion. The transcription factor NFB comprises a family of at least five Rel-related subunits within the cytoplasm: p65, p50, p52, c-rel, and Rel B (29). Differing subunit combinations have stimulatory and inhibitory effects on gene promoter regions. CoMTB induced rapid nuclear translocation of NFB in A549 cells and primary alveolar epithelium with activation of p65, p50, and c-rel subunits. In contrast, using adenoviral vectors overexpressing IB␣ we have shown that RSV-induced CCL5 expression from pulmonary epithelium is NFB-dependent and involves the p50/p65 heterodimer of NFB but not the c-rel subunit (19). The kinetics of CCL5 gene expression production differed from those of NFB nuclear translocation in A549 cells. NFB binding to the CCL5 promoter peaked at 1 h or earlier consistent with rapid IB␣ degradation and the early generation of CCL5 mRNA transcripts at 2 h. CCL5 mRNA was still detected at 8 -24 h even though NFB binding was virtually absent by 4 h in A549 cells. Late detection of mRNA may be because of accumulation of mRNA that has undergone posttranscriptional stabilization rather than because of new gene transcription. RSV increases CCL5 mRNA half-life from 0.8 to 6.8 h in airway epithelial cells (49). Other transcription factors may assume greater importance in late promoter activation, and we have shown that AP-1 and NF-IL-6 are translocated at later time points than NFB. Interestingly, in SAEC, NFB nuclear translocation may be more critical in late CCL5 gene expression because NFB translocation persisted until at least 8 h. Relatively few studies have directly investigated transcriptional regulation of CCL5 in lung epithelium in response to cytokine stimulation. CCL5 gene expression was shown to be NFB-dependent in TNF␣-stimulated A549 cells (16,51) and in CD40-stimulated airway epithelial cells (52). However, NFB may be necessary but not sufficient for CCL5 gene activation as indicated by the studies with our maximally truncated promoter containing the two proximal NFB binding sites. IL-1␤ has been shown previously to activate p65 and p50 subunits in epithelial cell lines (53), and the fact that it drives IL-8 secretion, which is usually NFB-dependent, suggests that this transcription factor is activated in NHBE. In addition, in another study involving primary epithelial cells in which the source is not clearly specified and may be a mixture of upper and lower airway cells, CCL5 gene expression and NFB nuclear binding were demonstrated, although no CCL5 secretion was reported (35).
The CCL5 promoter contains four potential NFB binding sites, Ϫ30, Ϫ44, Ϫ213, and Ϫ579 relative to the transcription start site (27). However, the Ϫ213 site is a nuclear factor of activated T cells (NF-AT) binding site with only a weak NFB affinity, and the distal (Ϫ579) site also serves as a CD28responsive element (26,27). Transient transfection assays using a WT CCL5 promoter-reporter expression plasmid confirmed that the CoMTB stimulus activated the CCL5 promoter, evident at 4 h and peaking at 8 h poststimulation of A549 cells.
The observation that deletion Ϫ737 reduced CCL5 reporter activity and that deletion Ϫ431 has no further effect indicated that the most distal NFB binding site (Ϫ579) is of little functional significance in the alveolar epithelial cell response to CoMTB. Site-directed mutagenesis did not demonstrate a key role for this distal site but did confirm that the two proximal B binding sites are important with promoter activity reduced to 20.4% of WT in studies using ⌬B2 construct. ⌬NFB1 and ⌬CD28RE markedly reduced human immunodeficiency virusinduced CCL5 promoter activity (27), whereas ⌬B1/2 reduced both TNF␣-(54) and RSV-induced activity (55), demonstrating stimulus specificity in CCL5 transcription factor binding site activation. However, in the distal promoter between Ϫ961 and Ϫ737, there is a cis-acting element for CCL5 promoter activity in alveolar epithelial cells. Previously distal negative rather than positive regulatory sites have been identified in T cells (27), but this region has not been extensively investigated. In addition, in other cell types the proximal part of the promoter alone controls activity. For example, in IL-1␤-stimulated astrocytoma cells (56) and T cells (26), deletions to 400 base pairs have little effect on CCL5 promoter activity. This was also the situation in alveolar epithelial cells stimulated by either RSV or TNF␣, where a Ϫ220 deletion of the CCL5 promoter retained near wild type activity (54,55).
An interesting finding was the involvement and functional significance of AP-1 and NF-IL-6 in the control of pulmonary epithelial cell CCL5 secretion in response to CoMTB. The CCL5 promoter contains four AP-1 TPA-response elements (26,27). Evidence exists to indicate a role for AP-1, often involving cooperative interactions with NFB, in the regulation of chemokine genes other than CCL5 such as CXCL8 gene activation in RSV-stimulated A549 cells and CCL2 (monocyte chemoattractant protein-1) gene activation in IL-1␤-stimulated endothelial cells (34,43,45,57). We demonstrate for the first time in epithelial cells a functional role for AP-1 in CCL5 gene activation, both in CoMTB-stimulated A549 cells and in human primary alveolar epithelial cells. CoMTB induced the nuclear translocation of AP-1 in both A549 and SAEC, exhibiting delayed kinetics compared with NFB binding with maximal binding at 4 -8 h poststimulation. The AP-1 family members binding to the TRE site were identified as c-Jun, FosB, and c-Fos in contrast with the c-Fos/JunD heterodimer, which binds to AP-1 sites in the CXCL8 gene promoter of A549 cells activated by TNF␣ (58). RSV infection of alveolar cells causes heterodimers involving c-Jun to bind to a cAMP-response element within the CCL5 promoter (55). These data suggest that there may be stimulus specificity in the profile of AP-1 family members that are activated to bind the CCL5 promoter in different diseases (59). Early promoter reporter studies in an erythroleukemic cell line do suggest a role for AP-1 in CCL5 promoter activation (26), but little data exists on whether AP-1 may functionally activate the CCL5 promoter in pulmonary epithelium. We found a significant amount of reporter activity in experiments using the Ϫ421 deletion construct containing all four TREs (41.9 Ϯ 15.7% of WT activity). This was reduced by ϳ50% (to 20.9 Ϯ 2.9% of WT activity) by a truncation of the promoter region to remove all AP-1-response elements, suggesting that these sites contribute to CCL5 promoter activity. Site-specific analysis revealed that the ⌬TRE3/4 region was the most critical, and independent or concurrent mutation of TRE1/2 did not further reduce promoter-reporter activity. This is a stimulus-specific effect of CoMTB because the minimal CCL5 promoter truncated at Ϫ220 base pairs retained activity similar to a full-length promoter of Ϫ974 base pairs in cytokine-and RSV-stimulated epithelial cells (54,55). In addition, there is cell specificity because in an astrocytoma line, a CCL5 promoter with a Ϫ278 deletion was activated to a similar extent to a full-length promoter by IL-1␤ in part as a consequence of translocation of p50/p65 but not c-rel-containing NFB complexes (56).
NF-IL-6 is a human basic domain leucine zipper-containing transcription factor important in inducible gene expression in acute inflammatory responses and is able to regulate promoter activity of the CXCL8 gene (60). There is a single NF-IL-6 binding site within the CCL5 promoter at position Ϫ92 relative to the transcriptional start site. NF-IL-6 protein expression in the lung is unique because in the lung, in contrast to tissues such as liver or to macrophages, NF-IL-6 translation is constitutively repressed (33,61). In normal lung tissue unlike elsewhere there is an abundance of NF-IL-6 mRNA, but protein is almost undetectable, suggesting posttranscriptional modifications are regulating protein expression in a tissue-specific fashion. RSV-induced NF-IL-6 expression in pulmonary epithelium has been shown to involve de novo synthesis of the protein without any increases in NF-IL-6 mRNA expression assessed by Northern analysis (33). The role of NF-IL-6 in cytokine-induced CCL5 promoter activity in pulmonary epithelium is emerging. To our knowledge, we show for the first time that an IL-1-mediated stimulus, CoMTB, induces CCL5 gene activation in pulmonary epithelium by stimulating NF-IL-6 DNA binding. NF-IL-6 nuclear translocation was evident only 24 h poststimulus. This late binding of NF-IL-6 may be of functional significance as CoMTB-induced CCL5 secretion increases over a 24-h period; moreover, WT CCL5 promoter activity was present 24 h poststimulus. Furthermore, the CCL5 promoter deletion construct (Ϫ79), which removes the NF-IL-6 binding site at position Ϫ92, causes almost a 75% reduction in the activity obtained with the Ϫ181 deletion construct (from 20.9 Ϯ 2.9% to 5.5 Ϯ 1.9% of WT). Site-directed mutation of the NF-IL-6 binding site confirmed a central role for this region with promoter activity falling to 49.4% of WT. This is a stimulus-specific response; in studies using a minimal CCL5 promoter (up to Ϫ220 bp), NF-IL-6 had a very limited influence on RSV-induced CCL5 secretion, and the interferon regulatory factor binding site was key (55). NF-IL-6 was more critical in control of epithelial cell-derived CCL5 secretion responses to TNF␣, FIG. 7. IL-1ra inhibits CCL5 secretion, mRNA accumulation, and promoter activity in CoMTB-stimulated epithelial cells. A, A549 cells were stimulated with CoMControl (1st lane), CoMTB (2nd lane), CoMTB after the A549 cells had been preincubated with 200 ng/ml IL-1ra for 2 h (3rd lane), CoMTB that had been preincubated with neutralizing anti-TNF␣ (50 g/ml) (4th lane), and finally CoMTB that had been preincubated with neutralizing anti-TNF␣ (50 g/ml) and after the A549 cells had been preincubated with 200 ng/ml IL-1ra for 2 h (5th lane). Cell culture supernatants were assayed for CCL5 by ELISA after 24 h of CoMTB stimulation. Data is expressed as mean Ϯ S.E. of three experiments. B, A549 cells were stimulated with CoMTB or CoMControl for 2 h (1st-5th lanes, conditions as described in A). CCL5 mRNA accumulation was assessed by Northern analysis. An autoradiograph representative of three independent experiments is shown with mean Ϯ S.E. of the densitometric analysis of the CCL5 mRNA signal. C, A549 cells were transfected with the WT CCL5 promoter luciferase expression plasmid and were stimulated for 8 h with CoMTB or CoMControl or were preincubated with IL-1ra prior to CoMTB stimulation. The cells then were assayed for luciferase activity. RANTES, regulated upon activation, normal T cell expressed and secreted. D, SAEC were stimulated by CoMTB or CoMControl (1st-5th lanes, conditions as described in A), and CCL5 secretion was assayed from cell culture supernatant.
where NF-IL-6 activity was detected on gel shifts within 1 h, much earlier than we found in response to CoMTB (54). In addition, the effects of NF-IL-6 activation are gene-specific because NF-IL-6 site mutations have no effect on CXCL8 promoter activity (62). Taken together the EMSA and promoterreporter data suggest that it is possible that NFB is more important in early activation of CCL5 mRNA accumulation and secretion and that AP-1 is responsible for the increased promoter activity between 4 and 8 h with NF-IL-6 contributing to late stage maintenance of CCL5 gene expression.
We demonstrated that IL-1␤ is the predominant active constituent involved in monocyte-epithelial cell networks in pulmonary tuberculosis. CCL5 secretion and gene expression from CoMTB-stimulated A549 and SAEC are inhibited by IL-1ra but not neutralizing anti-TNF␣. Furthermore, IL-1ra inhibited CoMTB-induced WT CCL5 promoter activity to 29.0 Ϯ 1.4% suggesting that IL-1ra inhibited IL-1-mediated CCL5 gene activation. We determined that CoMTB contained IL-1␤ and that recombinant IL-1␤ at 200 -500 pg/ml stimulated SAEC CCL5 secretion similar to that of CoMTB (1:10 dilution). These results suggest the existence of a paracrine cytokine network involving very low concentrations (200 pg/ml) of monocytederived IL-1␤. This highlights a potential critical role for IL-1␤ in the immune response to pulmonary tuberculosis and for IL-1ra in modifying these responses. IL-1␤ is secreted from tuberculosis-infected human monocytes (24,63) and is found both in tuberculous granulomas of infected lymph nodes (64) and the BALF of patients with active disease (65). We show that exogenous IL-1ra can inhibit epithelial CCL5 responses during tuberculosis infection. However, airway epithelium can secrete endogenous IL-1ra to modulate IL-1 bioactivity in the airway microenvironment (37). Interestingly polymorphisms at the IL-1 locus have been shown to influence the molar ratios of IL-1ra/IL-1␤ secreted by tuberculosis-infected monocytes (24), and a proinflammatory haplotype for this locus (higher IL-1␤ and lower IL-1ra expression) can affect disease phenotype in tuberculosis (24). Such host genetic factors, by influencing net bioactive IL-1␤, may influence monocyte epithelial cell networks and subsequent CCL5 secretion. Cytokine networks involving lipopolysaccharide-stimulated monocytes stimulate pulmonary epithelial CXCL8 secretion, but in contrast to the present findings, both TNF␣ and IL-1 are involved (66). We previously have found that conditioned medium from tubercu-losis-infected monocytes may stimulate primary pulmonary epithelial CXCL8 secretion; however, unlike for CCL5, this epithelial response is much less than that from the tuberculosisinfected monocytes themselves (21).
Finally, the data demonstrate differential responsiveness of NHBE and SAEC to monocyte-derived IL-1␤. Such differences between similar but distinct cell types within a single tissue are likely to be important in immune responses to different pathogens. Primary alveolar epithelium of the small airways secreted CCL5 in response to the monocyte-derived IL-1␤ and to recombinant IL-1␤, whereas upper airway cells of the trachea and main bronchi did not, consistent with data showing that IL-1␤ stimulates secretion of only picomolar CCL5 concentrations in these cells (5). Upper airways require both TNF␣ and interferon ␥ to stimulate CCL5 secretion (5,13). There remains one report that demonstrates up-regulation of CCL5 mRNA and NFB nuclear binding in human bronchial epithelial cell cultures stimulated by IL-1␤, although no CCL5 secretion was observed (35). SAEC are a better primary cell correlate than NHBE for the human type II alveolar cell carcinoma line A549 because they are derived from terminal bronchioles (2 mm and below) and include type I and II pneumocytes. We have found that this differential responsiveness to IL-1 is specific for CCL5. CXCL8 is secreted by CoMTB-stimulated NHBE and SAEC, although CoMControl also induces relatively low level CXCL8 secretion from SAEC but not NHBE. Thus, small airway cells may be primed to secrete CXCL8 in the absence of exogenous stimulus, although whether this is found in vivo is not known. The full significance of this differential responsiveness of the upper and lower airway is unclear except that pulmonary tuberculosis is primarily a disease of the lower airway, a site where monocyte epithelial cell networks stimulating CCL5 secretion will be functional.
This study defines a new contextual function for monocytederived IL-1␤ and a new role for the lung epithelium in host response to tuberculosis. The large surface area of pulmonary epithelial cells generates high levels of CCL5 during the pulmonary immune response to M. tuberculosis. Investigation into the mechanism of regulation of the CCL5 gene expression in the pulmonary epithelium has demonstrated that the two proximal NFB binding sites are required for gene activation and not the Ϫ579 CD28RE/NFB binding site, although the distal area of the promoter has a regulatory role in CoMTB-stimu- lated respiratory epithelium. This study has also defined functional roles for the transcription factors AP-1 and NF-IL-6 in CCL5 responses in our cellular model of tuberculosis. Very low IL-1 concentrations are identified as mediating this TNF-independent monocyte epithelial cell network in pulmonary tuberculosis. The effects of this proinflammatory cytokine on CCL5 secretion are shown to be specific for lower airway alveolar epithelial cells. Identifying the alveolar epithelium as a significant cellular source of CCL5 and defining the role of IL-1 may help in the design of local immunotherapeutic interventions to regulate T cell and monocyte recruitment to diseased lung infected with tuberculosis, particularly in patients with organisms resistant to conventional therapies.