A promoter recruitment mechanism for tumor necrosis factor-alpha-induced interleukin-8 transcription in type II pulmonary epithelial cells. Dependence on nuclear abundance of Rel A, NF-kappaB1, and c-Rel transcription factors.

The alveolar macrophage-derived peptide tumor necrosis factor-alpha (TNFalpha) initiates pulmonary inflammation through its ability to stimulate interleukin-8 (IL-8) synthesis in alveolar epithelial cells through an incompletely described transcriptional mechanism. In this study, we use the technique of ligation-mediated polymerase chain reaction (LMPCR) to record changes in transcription factor occupancy of the IL-8 promoter after TNFalpha stimulation of A549 human alveolar cells. Using dimethylsulfate/LMPCR, no detectable proteins bind the TATA box in unstimulated cells. By contrast, TNFalpha rapidly induces protection of G residues at -79 and -80 coincident with endogenous IL-8 gene transcription. Using DNase I/LMPCR, we observe inducible protection of nucleotides -60 to -99 (the TNF response element) and nucleotides -3 to -32 (containing the TATA box). Surprisingly, extensive TATA box protection is only seen after TNFalpha stimulation. Using a two-step microaffinity isolation/Western immunoblot DNA binding assay, we observe that the NF-kappaB subunits Rel A, NF-kappaB1, and c-Rel inducibly bind the TNF response element; these proteins undergo rapid TNFalpha-inducible increases in nuclear abundance as a consequence of IkappaBalpha proteolysis. Furthermore, the peptide aldehyde N-acetyl-Leu-Leu-norleucinal, an agent that blocks both IkappaBalpha proteolysis and NF-kappaB subunit translocation, abrogates recombinant human TNFalpha-inducible IL-8 gene transcription. These studies demonstrate that IL-8 is activated by a promoter recruitment mechanism in alveolar epithelial cells, where NF-kappaB subunit translocation is required for (and coincident with) binding of the constitutively active TATA box-binding proteins.

The respiratory epithelium contributes to normal pulmonary function in its ability to clear inhaled particulates through mucociliary action, ensure alveolar patency through surfactant secretion, and facilitate bacterial opsonization through secretory immunoglobulin production. A large body of evidence now supports an additional role for the airway epithelial cell to amplify cytokine signals from pathogen-activated alveolar macrophages into secretion of chemokines, arachidonic acid metabolites, and phospholipid-molecules that recruit additional inflammatory cells into the airway mucosa (1,2). Of the potent alveolar macrophage-derived cytokines studied, the secretion of tumor necrosis factor ␣ (TNF␣), 1 in particular, has been implicated in the pathophysiology of neutrophilic-infiltrating disorders including acute lung injury from sepsis, silica-induced pulmonary fibrosis, allograph rejection, and acute respiratory tract infection (3)(4)(5)(6).
Although inducible IL-8 gene expression has been investigated intensively, controversy still exists as to mechanism responsible for TNF␣-inducible transcription in epithelial cells. In human gastric adenocarcinoma cells, TNF␣ activates IL-8 through a synergistic effect on two regions, one cis element located at nucleotides (nt) Ϫ126 to Ϫ120 that contains a putative AP-1 binding site, and a second element at nt Ϫ80 to Ϫ71 that contains a nuclear factor-B (NF-B) binding site (19). In human bronchial epithelium-derived HS-24 and BET-1a cell lines, an element encompassing nt Ϫ130 to Ϫ112 was shown to be essential for TNF-inducible IL-8 promoter activity (20). Finally, in HeLa epithelial cells, TNF␣ activates IL-8 through a mechanism involving cooperative binding of nuclear factor-IL6 (NF-IL6) and NF-B to a cis element encompassing nt Ϫ97 to Ϫ69 (21,22). Given the highly tissue-restricted pattern of NF-IL6 expression, the relevance of this latter study to alveolar epithelium is uncertain (23,24). Moreover, these studies have solely relied on transient transfection assays, a technique that does not always faithfully reproduce phenomenon of inducible gene expression within chromatin contexts (25); thus, whether these observations are relevant to endogenous IL-8 gene expression is problematic.
Therefore, in this study, we examine the mechanism for TNF␣-inducible gene transcription in human alveolar epithelial (A549) cells using the technique of ligation-mediated PCR (LMPCR) to record changes in binding to the endogenous IL-8 promoter. LMPCR in recombinant human TNF␣ (rhTNF␣)treated cells demonstrates inducible DMS protection of guanine (G) bases at Ϫ79 and Ϫ80 and a DNase I footprint over nt Ϫ60 to Ϫ99 of the endogenous IL-8 promoter, occurring simultaneously with the peak in its endogenous transcription (at 0.5-1 h). Surprisingly, although no stable TATA box binding could be observed in unstimulated cells, an additional extended footprint from nt Ϫ3 to Ϫ32, spanning the TATA box, was observed in rhTNF␣-stimulated cells. Transient transfection assays of 5Ј-deleted and site mutations of IL-8 promoter/luciferase plasmids demonstrate nt Ϫ60 to Ϫ99 is a TNF-response element (TNFRE). Inducible TNFRE DNA binding activity is composed of the NF-B subunits Rel A, NF-B1, and c-Rel. These subunits undergo rapid increases in nuclear abundance after rhTNF␣ treatment via a mechanism involving IB␣ proteolysis. Finally, the peptide aldehyde N-acetyl-Leu-Leu-norleucinal, an agent that blocks IB␣ proteolysis and inducible NF-B binding, completely blocks endogenous IL-8 expression. We conclude that the NF-B transcription factor complex, Rel A, c-Rel and NF-B1, activate endogenous IL-8 gene expression through a promoter recruitment mechanism.
Cell Culture and Treatment-A549 human alveolar type II-like epithelial cells (American Type Culture Collection) were grown in minimal essential medium containing 10% (v/v) fetal bovine serum, 10 mM glutamine, 100 IU penicillin/ml, and 100 g streptomycin/ml. rhTNF␣ was added to growth medium at a final concentration of 20 ng/ml. For in vitro modification of DNA, medium containing 1:1200 (v/v) DMS (diluted immediately prior to addition to cell culture) for 1 min at 37°C. Afterward, cells were washed in phosphate-buffered saline and genomic DNA was extracted using ion-exchange chromatography (A.S.A.P. kit, Boehringer Mannheim) for piperidine cleavage using previously published protocols (26,27). For in vivo treatment with DNase I, A549 cells in 100-mm dishes were first permeabilized for 1.5 min by 0.5 g/ml lysolecithin in buffer A (150 mM sucrose, 80 mM KCl, 35 mM HEPES (pH 7.4), 5 mM MgCl 2 , 5 mM CaCl 2 ). The permeabilized cells were washed in 4 ml of buffer A, and treated with DNase I either at 10 g/ml or 50 g/ml in solution B (buffer A with 80 mM NaCl substituted for KCl) for 2 min. The DNase I-treated cells were then washed with phosphatebuffered saline, and genomic DNA was isolated by ion exchange chromatography.
Ligation-mediated PCR (LMPCR)-hIL-8 gene-specific primers synthesized to amplify the noncoding strand using the downstream (d) primers were: d1, 5Ј-CACACAGTGAGAATGGTTCCT-3Ј; d2, 5Ј-CCG-GTGGTTTCTTCCTGGCTC-3Ј; and d3, 5Ј-CCGGTGGTTTCTTCCTG-GCTCTTGTCCTAGAAG-3Ј. Due to the presence of an adenine-thy-mine-rich tract upstream of nt Ϫ170, it was not possible to synthesize upstream primers with the appropriate annealing temperature for LMPCR (28) to footprint the noncoding strand. Unidirectional "linker primers" were synthesized as described (29). LMPCR was performed as reported, using the following modifications. For the composition of the first strand synthesis reaction, 2-4 g of modified genomic DNA was extended with 0.3 pmol primer d1 in a 30-l reaction volume with 2 units of Vent polymerase. Extension parameters were: denaturation, 5 min at 95°C; annealing, 30 min at 60°C; and extension, 10 min at 76°C. After ligation of undirectional linkers, DNA was ethanol-precipitated and amplified using 10 pmol of primer d2, 10 pmol of linker primer, and 2 units of Vent polymerase in a total volume of 100 l. PCR cycling parameters were: denaturation, 1 min at 95°C; annealing, 2 min at 66°C; and extension, 3 min at 76°C for 18 cycles (with 5 s added to extension step with each successive cycle). Samples were iced immediately and labeled using 2.3 pmol of [␥-32 P]ATP/T4 polynucleotide kinase end-labeled primer d3 by PCR (denaturation, 3.5 min at 95°C; annealing, 2 min at 67°C; extension, 10 min at 76°C for 2 cycles). Radiolabeled DNA was ethanol-precipitated, resuspended in formamide loading dye, and fractionated on a 6% polyacrylamide, 8 M urea sequencing gel. The gel was dried and exposed to Kodak XAR film using an intensifying screen at Ϫ70°C.
EMSAs and Microaffinity Purification-Nuclear extracts (NE) of A549 monolayers were prepared using our previously published protocol (18,24,34). Duplex oligonucleotides corresponding to Ϫ96 to Ϫ69 base pairs of hIL-8 promoter are listed below. Underlines indicate purine-to-noncomplementary-pyrimidine substitutions from the wild-type sequence. EMSAs included 15 g of total protein, 1 g of poly(dA-dT), and 20,000 cpm ␣-32 P-labeled double-stranded IL-8 probe. Competition and supershift assays were performed as described (18). Microaffinity purification of TNFRE-binding proteins was performed using chemically synthesized TNFRE wild-type (WT) oligonucleotides containing 5Ј biotin (Bt) on a flexible linker (Genosys, The Woodlands, TX). Twenty pmol of duplex Bt-TNFRE WT was incubated with 1 mg of control or rhTNF␣-treated A549 NE in the presence of 20 g of poly(dA-dT) in a 500-l volume of a buffer containing 8% (v/v) glycerol, 5 mM MgCl 2 , 1 mM dithiothreitol, 60 mM KCl, 1 mM EDTA, and 12 mM HEPES (pH 7.9) for 1 h at 4°C. Proteins bound to Bt-TNFRE WT were captured by addition of 50 l of a 50% (v/v) slurry of streptavidin-agarose beads (Pierce) and washed three times in binding buffer. TNFRE-binding proteins were then eluted by SDS-PAGE loading buffer for Western immunoblot analysis. For competition, a 5-fold excess of nonbiotinylated oligonucleotide was included in the initial binding reaction.
Western Immunoblot-Proteins were fractionated by SDS-PAGE and transferred onto polyvinylidene difluoride membranes as described (24,34). Affinity-purified rabbit polyclonal antibodies to Rel A, c-Rel, NF-B1, and Rel B were obtained commercially (Santa Cruz Biotechnology, Santa Cruz, CA) and used as primary antibody according to the supplier's recommendations. Secondary detection was using horseradish peroxidase-coupled donkey anti-rabbit antibody in the ECL enhanced chemiluminescence assay (Amersham Life Sciences) as described (18,24).
Nuclear Run-on Transcriptional Rate Assay-Nuclear suspensions from 10 7 control or rhTNF␣-treated A549 cells were used to incorporate [␣-32 P]UTP into nascent RNA at 30°C for 45 min (18). After purification of RNA by sequential trichloroacetic acid and ethanol precipitation, a constant 5 ϫ 10 6 cpm/ml from each treatment was used to hybridize immobilized single-stranded plasmid DNAs corresponding to hIL-8, ␤-actin, and cyclophilin cDNAs. Membranes were washed, and quantitated by exposure to a phosphorimager screen.

Kinetics of rhTNF␣-inducible IL-8 Transcription in A549
Cells-Human alveolar epithelial A549 cells were selected for study because they maintain features of type II pneumocytes through their morphological appearance, ability to secrete surfactant, and susceptibility to respiratory syncytial virus infection, all of which are features of human alveolar cells (2,18,35). Immunoreactive IL-8 was quantified by ELISA in A549 cell culture supernatants after stimulation with 20 ng/ml rhTNF␣ (Fig. 1A). A statistically significant 3-fold increase in IL-8 production was detected after 2 h (as compared with control supernatants); immunoreactive IL-8 continued to accumulate to a 53-fold increase at 24 h. Northern blot assays were performed to determine changes in steady state IL-8 mRNA abundance at times preceding the changes in protein secretion (0 -8 h; Fig.  1B). In the absence of rhTNF␣, IL-8 mRNA was barely detectable. However, accumulation of a single 1.8-kilobase IL-8 transcript was strongly induced by rhTNF␣ treatment, starting at 0.5 h and peaking at 2 h (a 110-fold increase). To determine the transcriptional component for induction of IL-8 mRNA, the kinetics of IL-8 gene transcription was measured by nuclear run-on assay for 0 -4 h after rhTNF␣ treatment (Fig. 1C). Slightly preceding the maximal changes in mRNA, IL-8 transcription rapidly peaked at 0.5-1 h (a 20-fold induction). Although IL-8 transcription declined thereafter, it still remained elevated compared with control levels at 2 h (7-fold) and 4 h (3-fold). The slight lag in mRNA accumulation is probably reflective of the time required for the longer-lived IL-8 mRNA levels to reach steady state within the cell.

FIG. 1. Kinetics of TNF␣-inducible IL-8 transcription in A549 cells.
A, changes in immunoreactive IL-8. Confluent monolayers of A549 cells (3 ϫ 10 6 cells/6-cm dish) were stimulated with 20 ng/ml rhTNF␣ for the indicated times prior to harvest of medium and assay for IL-8 using ELISA. Shown is the mean Ϯ S.D. of n ϭ 3 independent experiments. Relative to unstimulated controls, IL-8 protein concentration increased 3-fold at 2 h, 11-fold at 4 h, 23-fold at 8 h, 26-fold at 12 h, and 53-fold at 24 h. Untreated A549 produce Ͻ500 pg/ml IL-8 at all times tested. B, accumulation of IL-8 transcripts in response to rTNF␣. Total cellular RNA from confluent A549 monolayers was fractionated for Northern blot hybridization after rhTNF␣ stimulation for the indicated times. Top, autoradiogram after hybridization with hIL-8 cDNA probe. Bottom, exposure after hybridization with ␤-actin as internal control. Relative to unstimulated cells, normalized IL-8 RNA increases 2-fold at 15 min, 10-fold at 30 min, 43-fold at 1 h, 110-fold at 2 h, 69-fold at 4 h, and 35-fold at 8 h. C, kinetics of IL-8 transcription in response to rhTNF␣. Nuclei harvested from control or rhTNF␣-treated confluent A549 cell monolayers were subjected to nuclear run-on transcription assay. Equal amounts of incorporated label (in cpm) were hybridized to nitrocellulose strips containing immobilized IL-8, ␤-actin, cyclophilin (cyclo), or the negative control plasmid DNA (pGEM). Shown is an autoradiogram of simultaneously exposed strips taken from cells stimulated for 0, 0.5 h, 1 h, 2 h, and 4 h as indicated at top. A rapid and transient peak in IL-8 transcription occurred at 0.5-1 h. Relative to untreated cells, IL-8 transcription was increased by 20-fold at 0.5 and 1 h, 7-fold at 2 h, and 3-fold at 4 h.

Identification of rhTNF␣-inducible Binding Sites on the Endogenous IL-8 Promoter-
To determine whether rhTNF␣ changes the composition of proteins interacting with the IL-8 gene promoter, two variations of the LMPCR (27) technique were used using IL-8 gene-specific primers. First, intact cells were treated with the membrane-soluble alkylating agent DMS, an agent that methylates exposed guanine (G) bases in the DNA major groove. The location of G methylation can be determined following extraction and piperidine cleavage. Fig.   2A shows the results of DMS/LMPCR by amplifying the DMSmodified coding strand from control or rhTNF␣-treated cells (at times corresponding to the peak in IL-8 transcription). As controls, deproteinized genomic DNA was DMS-treated and amplified in parallel. Surprisingly, in unstimulated cells, no stable protein-DNA interaction could be detected by DMS/ LMPCR (compare lanes 1 and 2). In contrast, rhTNF␣ treatment induces a rapid protection of G residues at nt Ϫ79 and Ϫ80 and the appearance of several hypersensitive sites, the FIG. 2. LMPCR of endogenous IL-8 gene promoter. A, LMPCR using DMS protection in vivo. Control and rhTNF␣treated A549 cells (for times indicated at top) were treated with 1:1200 (v/v) DMS for 1 min. Genomic DNA was then isolated, piperidine cleaved, and subjected to LMPCR (see "Experimental Procedures"). Shown is an autoradiogram of labeled PCR products after fractionation on a 6% polyacrylamide sequencing gel. N, "naked" DNA ladder prepared by DMS treatment of purified (deproteinized) A549 genomic DNA. A549 refers to DMS ladder produced in intact A549 cells for indicated times of rhTNF␣ treatment. At left is sequence of hIL-8 promoter corresponding to the region of interest with asterisks corresponding to protected G residues at nt Ϫ79 and Ϫ80, and arrow corresponding to the nt Ϫ82 hypersensitive site. B, LMPCR using DNase I modification in lysolecithin-permeabilized cells. Control or rhTNF␣-treated cells were lysolecithin-permeabilized, and various concentrations of DNase I were used to footprint the IL-8 gene in vivo using LMPCR. Shown is an autoradiogram after fractionation on a sequencing gel. DMS/LMPCR "G" ladder (lane 1); deproteinized genomic DNA using 5 g/ml DNase I (lane 2); lysolecithin-permeabilized unstimulated A549 cells using DNase I at 10 g/ml (lane 3) and 50 g/ml (lane 4); lysolecithin-permeabilized rhTNF␣-stimulated A549 cells using DNase I digestion at 10 g/ml (lane 5), and 50 g/ml (lane 6). In unstimulated cells, weak regions of protection between nt Ϫ55 to Ϫ68 and nt Ϫ99 to ϳϪ147 could be identified. By contrast, rhTNF␣-inducible regions correspond to nt Ϫ3 to 32 and nt Ϫ60 to Ϫ99 (marked on right). A rhTNF␣-inducible DNase I-hypersensitive site at nt ϳϪ66 (large arrow) and multiple hypersensitive sites (small arrows) are indicated. C, summary diagram of protein-DNA interactions of endogenous hIL-8 promoter in A549 cells. Sequence of human IL-8 promoter from ϩ1 to Ϫ147 is shown. Asterisks, G residues inducibly protected by rhTNF␣; small arrow corresponds to DMS-hypersensitive site. Brackets, areas of DNase I protection designated as FPo (for constitutive footprint) or FPi (inducible footprint). most prominent at nt Ϫ-82 of the coding strand (asterisks and arrow, respectively, in Fig. 2A). For additional information on genomic protein binding sites, LMPCR was used to assay IL-8 promoter occupancy after DNase I treatment in vivo. DNase I can be introduced into cultured cells after transient permeabilization of the cell membrane by lysolecithin without disruption of protein⅐DNA interaction (36). Fig. 2B demonstrates DNase I/LMPCR of deproteinized genomic DNA compared with control or rhTNF␣-treated A549 cells. As compared with the DNase I digestion pattern of deproteinized DNA, in unstimulated cells, a protected region between nt Ϫ55 to Ϫ68 and a larger extended protection could be detected from nt Ϫ99 to ϳϪ147 (the upper border was indistinct) with a hypersensitive site at nt ϳϪ115. Importantly, no stable protein-DNA interaction was detected over the proximal promoter from nt Ϫ3 to Ϫ32 (a region containing the TATA box). By contrast, after 1 h of rhTNF␣ treatment, two inducible regions could be identified: 1) nt Ϫ3 to Ϫ32 (corresponding to the TATA box), and 2) nt Ϫ60 to Ϫ99 (containing the G residues inducibly protected using DMS/LMPCR assay). Also in the rhTNF␣-treated ladder, the indistinct upstream border of the constitutive protected region became hypersensitive to DNase I cleavage, probably indicating a conformational change of the DNA. A summary of the DMS-and DNase I/LMPCR results is shown in Fig. 2C. That TATA box binding is rhTNF␣-inducible implies a promoterrecruitment model in IL-8 gene activation (see "Discussion").
The TNF Response Element Maps to the Inducible Contact Sites on the IL-8 Promoter-Both techniques of DNA modification indicated inducible protein binding to region contained in nt Ϫ60 to Ϫ99, and a constitutive site containing the putative AP-1 site nt Ϫ120 to Ϫ126. To assay for which cis elements are important in inducible activity, transient transfections of 5Јdeletions (Fig. 3A) or site mutations of the IL-8 promoter linked to luciferase reporter genes (Fig. 3B) were performed. Ϫ162/ ϩ44 hIL-8/LUC is 28 Ϯ 5-fold inducible by 20 ng/ml rhTNF␣ treatment (X Ϯ S.E., n ϭ 8 independent transfections). Fig. 3A shows the results of a representative transfection where -fold inducibility of Ϫ162/ϩ44, Ϫ132/ϩ44, Ϫ99/ϩ44 and Ϫ54/ϩ44 hIL-8/LUC reporters are compared. -Fold induction of 5Ј deletions to Ϫ99/ϩ44 and longer promoters are indistinguishable, but deletion to Ϫ54 is significantly less rhTNF␣-inducible, indicating the region nt Ϫ54 to Ϫ99, containing the rhTNF␣inducible LMPCR footprint, plays an indispensable role in promoter activity. Fig. 3B shows the kinetics of luciferase reporter induction for the Ϫ162/ϩ44 hIL-8/LUC wild-type (WT) plasmid in comparison to the same promoter containing a purine to noncomplementary pyrimidine mutation in the G residues Ϫ79, Ϫ80 and its mirror half-site on nt Ϫ72 and Ϫ73 on the noncoding strand (designated Ϫ162/ϩ44 hIL-8⌬/LUC). In the absence of rhTNF␣, the Ϫ162/ϩ44 hIL-8⌬/LUC reporter was consistently less active than the WT reporter, at 25 Ϯ 17% of the WT activity (n ϭ 4). Moreover, Ϫ162/ϩ44 hIL-8⌬/LUC was inert to rhTNF␣ stimulation (Fig. 3B), indicating an absolute requirement of these residues for rhTNF␣-inducible transcription.
Part of the inducible DNase I/LMPCR footprint, nt Ϫ82 to Ϫ99, contains a putative NF-IL6 binding site, a protein that cooperatively binds and is transcriptionally synergistic with NF-B (21). Although we cannot detect NF-IL6 expression in A549 cells (24), we could not exclude a low level of NF-IL6 expression that participates in rhTNF␣ activation. For this, a mutation within the NF-IL6 binding site was produced (Ϫ162/ ϩ44 hIL-8⌬NF-IL6/LUC) corresponding to known mutations that disrupt NF-IL6 binding (see Ref. 21 and "Experimental Procedures"). In transient transfections, Ϫ162/ϩ44 hIL-8⌬NF-IL6/LUC showed similar basal reporter activity and was equally inducible as the wild-type promoter (81 Ϯ 23% of stimulated wild-type activity; n ϭ 4), indicating that NF-IL6 binding plays a minor, if any, role in IL-8 induction.
To determine whether the Ϫ60 to Ϫ99 nt region of IL-8 contained latent rhTNF␣-inducible enhancer activity, three Reporter activity directed by (TNFRE WT) 3 -rATLUC was 34fold inducible by rhTNF␣, whereas (TNFRE ⌬) 3 -rATLUC was not (1-fold; n ϭ 3). These data indicate that the complex specifically binding to nt Ϫ80 and Ϫ79 is required for constitutive and rhTNF␣-induced activation of the IL-8 gene promoter.
Characterization of rhTNF␣-inducible Proteins Binding the TNFRE-EMSAs were performed to identify the kinetics and specificity of proteins binding to the TNFRE. As shown in Fig.  4A, rhTNF␣ rapidly induced the binding of three complexes (Ci 1 -Ci 3 ) to the TNFRE with kinetics similar to those observed for the genomic footprinting ( Fig. 2A), and for the transcriptional activation (Fig. 1C) of endogenous IL-8. (Inducible complexes Ci 2 and Ci 3 are so closely comigrating as not to be separable except under long electrophoresis conditions (cf. Fig.  4C).) After 15 min, complex Ci 1 increased 13-fold, and complexes Ci 2/3 increased 12-fold, without changes in Co. To maximize the resolution of the individual complexes, EMSAs were fractionated longer (Fig. 4, B and C). Fig. 4B demonstrates binding specificity, where either unlabeled wild-type TNFRE, or NF-B mutated sequences (⌬) were used to compete Ci 1-3 in EMSA. TNFRE WT, but not TNFRE ⌬, is able to compete Ci 1-3 with high affinity. To again exclude the unlikely possibility that NF-IL6 was present within the C i complex, we also competed it with unlabeled NF-IL6 binding sites (37). No effect was seen when either the wild-type or mutant NF-IL6 binding site was used as a competitor (Fig. 4B). These data indicate that the rhTNF␣-inducible complexes bind with NF-B binding specificity.
NF-B is a family of homo-and heterodimeric proteins related by a conserved NH 2 -terminal ϳ300-amino acid Rel ho-mology domain; members of this family include the proteolytically processed NF-B1 (p50) and NF-B2 (p49) subunits, as well as the Rel A (p65), c-Rel, and Rel B subunits (reviewed in Ref. 38). Heterodimerization of NF-B subunits produces species with various intrinsic DNA-binding specificities (39), transactivation properties, and subcellular localization (38,40). Because of the distinctive nature of the NF-B complexes, and that we observed multiple inducible species of complexes, it was important to determine those that bound to the TNFRE. For this, we initially used supershift assays in EMSA (Fig. 4C); this assay indicated that anti-Rel A antibodies produced a strong supershift of Ci 1-3 complexes.

Recombinant Rel A Contacts the TNFRE in a Manner Indistinguishable from That of the Ci 1-3 Complex on the Endogenous IL-8 Promoter-DMS/LMPCR indicates that inducible contacts occur on
Ϫ79 and Ϫ80 nt of the IL-8 promoter. To establish that Rel A makes similar contacts, recombinant Rel A (p65(12-317); Ref. 18) was used in methylation interference assays. Recombinant Rel A contacts residues Ϫ79 and Ϫ80 on the coding strand and symmetrical residues Ϫ72 and Ϫ73 on the noncoding strand (Fig. 4D).

Two-step Microaffinity/Western Immunoblot Assay Demonstrates Additional NF-B Members Bind the TNFRE-Because
the supershift assay depends on the titer and affinity of the antibody used, we could not exclude the presence of other NF-B family members in the heterogeneous Ci 1-3 complex. To more rigorously identify other NF-B family members binding the TNFRE, we used a two-step microaffinity isolation/Western immunoblot assay. In this assay, Bt-TNFRE was used to bind control or rhTNF␣-stimulated A549 NE. TNFRE-binding proteins were captured by the addition of streptavidin agarose beads, washed, and the presence of bound proteins detected using Western immunoblot. Fig. 4E shows the result of this assay where the blot is probed for Rel A. Very little Rel A binding is detectable prior to rhTNF␣ stimulation, but its abundance increases dramatically (6-fold) after stimulation. Rel A is not detected when nonbiotinylated TNFRE WT is included as competitor in the initial binding step, but it is detectable when nonbiotinylated TNFRE ⌬ mutation is used, indicating sequence-specific binding. Using the same strategy, the presence of other NF-B subunits was assayed (Fig. 4F). Both NF-B1 and c-Rel inducibly bind the TNFRE (we were unable to detect the expression or binding of Rel B or NF-B2; data not shown). These data indicate that Rel A, c-Rel, and NF-B1 inducibly bind the TNFRE.
Rel A, NF-B1, and c-Rel Show rhTNF␣-inducible Changes in Nuclear Abundance-Western immunoblot was performed on cytoplasmic and sucrose cushion-purified NE taken from rhTNF␣-treated A549 cells (Fig. 5). 65-kDa Rel A was easily detected in the cytoplasm, but not the nucleus of unstimulated cells; by 30 min, a significant increase in nuclear Rel A abundance was detected. The increase in nuclear Rel A was transient and fell below the limits of detection after 4 h. 75-kDa c-Rel and 50-kDa NF-B1 underwent similar rapid increases in nuclear abundance. Like Rel A, c-Rel nuclear abundance de- promoter in vitro as the TNFRE-binding proteins do on the endogenous IL-8 promoter. E, two-step microaffinity isolation/Western immunoblot for Rel A. Control or rhTNF␣-treated A549 NE were affinity-purified for Bt-TNFRE binding in the absence or presence of nonbiotinylated TNFRE WT (WT) or TNFRE ⌬ (⌬) competitors. After capture on streptavidin agarose, complexes were eluted and assayed for Rel A by Western immunoblot using ECL (shown). Specific 65-kDa Rel A staining is faintly detectable in unstimulated NE and increases by 6-fold after rhTNF␣ stimulation. Binding is competed by WT and enhanced by ⌬ oligonucleotides. *, nonspecific cross-reacting material. Rel A binding is rhTNF␣-inducible and sequence-specific. F, NF-B1 and c-Rel inducibly bind TNFRE WT. Two-step microaffinity isolation/Western immunoblot was performed on control (Ϫ) and rhTNF␣-stimulated (ϩ) A549 NE, and assayed using preimmune or subunit-specific Rel A, NF-B1, or c-Rel antibodies in Western. Locations of 65-kDa Rel A, 50-kDa NF-B1, and 75-kDa c-Rel are indicated (arrows). *, nonspecific cross-reacting material. The small 33-kDa band in the Rel A lane is a proteolytic fragment of Rel A observed in some, but not all affinity isolations. creases after 4 h, whereas nuclear NF-B1 remains detectable. In data not shown, we have confirmed that the increase in nuclear Rel A, c-Rel, and NF-B1 occur in the absence of new protein synthesis, indicating that the mechanism is via cytoplasmic-nuclear translocation of preformed protein.
The Peptide Aldehyde, N-Acetyl-Leu-Leu-norleucinal (Nleu), Blocks Inducible TNFRE binding and Endogenous IL-8 Expression-NF-B is maintained in the cytoplasm through association with the IB family of inhibitory proteins, IB␣ being the most abundantly expressed (41,42). In response to signalinduced degradation of IB␣, NF-B undergoes nuclear translocation (43). Inhibitors of IB␣ proteolysis, then, would allow us to test the requirement of NF-B translocation on endogenous IL-8 transcription. The effect of the protease inhibitor Nleu, a peptide aldehyde with activity for the 26 S proteasome complex (primarily responsible for IB␣ proteolysis; Refs. 44 and 45), was used to determine the effect of rhTNF␣ on steady state IB␣ levels. Fig. 6A shows that rhTNF␣ induces a rapid and complete disappearance of IB␣ within 15 min, followed by its resynthesis after 1-4 h. By contrast, in the presence of Nleu, IB␣ proteolysis is significantly inhibited (compare 15 min time points). In addition, Nleu-pretreated A549 cells do not form the inducible Ci 1-3 complexes in response to rhTNF␣ (Fig.  6B). These data indicate that Nleu blocks inducible IB␣ pro-FIG. 6. Effect of the proteasome inhibitor Nleu on NF-B activation and IL-8 gene expression. A, IB␣ Western Immunoblot. Control or Nleu-pretreated A549 cells (100 M for 1 h) were stimulated with rhTNF␣ for the times indicated at top. Cytoplasmic extracts were prepared and abundance of IB␣ assayed using Western immunoblot using ECL (shown). In response to rhTNF␣, IB␣ is rapidly proteolyzed within 15 min and reaccumulates over 4 h. A ϳ3-kDa slower migrating form of IB␣ can be identified in rhTNF␣-treated extracts that probably represents the phosphorylated form of IBa ( P -IB␣). In Nleu-pretreated cells, IB␣ proteolysis is significantly inhibited. B, effect of Nleu on inducible Ci 1-3 binding by EMSA. Control or Nleu-pretreated cells were stimulated with rhTNF␣ for 1 h. NE were prepared assayed for Ci 1-3 complex formation by EMSA. A nonspecific complex was sometimes visualized in late-passage A549 cells (*). Nleu pretreatment impairs inducible TNFRE complex formation. C, effect of Nleu on rhTNF␣-inducible reporter activity. A549 cells transiently transfected with Ϫ162hIL-8/LUC and CMV-␤-gal reporters were stimulated for 4 h prior to harvest and reporter assay. Shown is a representative transfection of normalized luciferase reporter activity. Nleu blocks IL-8-driven luciferase reporter activity. D, effect of Nleu on inducible endogenous IL-8 expression. A549 cells were treated as in Fig. 7a. Total RNA was extracted and assayed for IL-8 mRNA by hybridization. An overexposed radiogram is shown to demonstrate the unstimulated levels. Nleu completely blocks endogenous IL-8 gene induction. teolysis and nuclear translocation of Rel A, c-Rel and NF-B1 (Ci 1-3 ) in pulmonary epithelial cells. Fig. 6C demonstrates that inducible expression of Ϫ162 hIL-8/LUC is completely blocked by pretreatment with Nleu. Finally, Northern blot assays were used to determine whether inhibition of Ci 1-3 resulted in inert IL-8 gene expression in rhTNF␣-stimulated A549 cells (Fig.  6D). Under these conditions, induction of IL-8 mRNA is completely abolished. DISCUSSION The alveolar macrophage-delivered cytokine, TNF␣, is a potent activator of a cytokine cascade in airway epithelial cells. TNF␣ secretion plays an import aspect of host-defense to respiratory pathogens and may also play a pathophysiological role in neutrophilic infiltrating and fibrotic pulmonary disease. In this study, we have examined the mechanism used by TNF␣ to activate expression of the IL-8 gene, encoding a product that accounts for the majority of neutrophilic chemotactic activity in pathogen-or cytokine-stimulated airway epithelial cell supernatants. Although IL-8 gene expression has been intensively studied in a variety of tissue types, the pleiotropic mechanisms identified for activation of its promoter warrant a systematic investigation in airway epithelial cells before rational design of therapeutic agents to block expression of this cytokine in lung can be undertaken. Our study demonstrates a rapid and transient burst in IL-8 expression after rhTNF␣ stimulation that largely accounts for changes in IL-8 mRNA.
Inducible Promoter Recruitment as a Mechanism for IL-8 Gene Transcription-Inducible gene transcription is due to interactions between activator proteins and general transcription factors (GTFs) through one of several experimentally separable mechanisms: 1) promoter recruitment, where activators recruit the GTFs TFIID/TFIIA and TFIIB to the promoter, or 2) conformational change, where activator binding alters the conformation of a preexisting preinitiation complex (reviewed in Ref. 46). To distinguish these mechanisms, we have applied the technique of LMPCR, a powerful tool used for the identification of protein-DNA interactions on gene promoters within their native chromatin context. This study, to our knowledge the first to demonstrate constitutive and rhTNF␣-inducible binding of the native hIL-8 promoter, is surprising because of the paucity of stable protein-DNA interactions in the proximal IL-8 promoter. LMPCR is, in part, a population assay that relies on a homogeneous population of target promoters for its interpretation. Although it is possible that some proteins engage the TATA box of some, but not all, of the unstimulated IL-8 promoters in our cultured A549 cells the presence of which does not result in either a DMS or DNase I footprint, it is clear that additional protein-TATA interactions occur coincidentally with NF-B binding. We interpret these data to mean that IL-8 gene activation occurs via promoter recruitment, rather than a conformational change of a preassembled preinitiation complex. In our data not shown, and in the hands of others (47), TATA binding in vitro is constitutive and not hormone-regulated.
How might rhTNF␣-dependent TATA binding occur within the context of endogenous gene? One explanation may be the role of chromatin-mediated repression, such as condensation with histone H1 or nucleoprotein structures that prevent binding of the GTFs to the TATA box (48,49). A well characterized example of this phenomenon is the mouse mammary tumor virus, a retrovirus containing a glucocorticoid receptor-regulated long terminal repeat. In the absence of glucocorticoid, the mouse mammary tumor virus long terminal repeat is actively repressed as the consequence of a phased array of nucleosomes for which binding prevents the constitutive proteins NF1 and TFIID from interacting with their respective CCAAT and TATA sites (reviewed in Ref. 50). In response to ligand-acti-vated glucocorticoid receptor, a shift in the nucleosome array occurs, which opens up the previously occluded CCAAT and TATA sites for binding by their cognate ("constitutive") proteins (47,50). In this way, inducible transcription factors effect gene expression through a two-step mechanism: 1) to "derepress" promoters through nucleosome rearrangement and 2) to "activate" their expression (48). Our studies on IL-8 have not systematically examined changes in nucleosomal array; however, Nakamura et al. have reported the appearance of TNF␣inducible DNase I-hypersensitive site ϳ120 nt 5Ј to exon 1 (20), a characteristic of nucleosomal rearrangement. The role of chromatin structure, if any, in controlling constitutive or inducible IL-8 gene expression will require additional investigation.
In yeast expressing an inducible form of a mutant TATA box-binding protein (TBP) that selectively recognizes a unique TATA box sequence, it was demonstrated previously that an important role of the activation domain of an inducible transactivator is to functionally increase TBP recruitment to a promoter, a process that was rate-limiting for transcriptional activation (51). Our data provide a direct demonstration of GTF recruitment in the rapid and highly inducible transcription of IL-8 in a system directly relevant to pulmonary inflammation. With regard to promoter activation, it is significant that Rel A and c-Rel subunits, but not NF-B1, have been observed to make stable direct protein-protein interactions with the TBP within the TFIID complex (52), thus providing a potential mechanism for activation of hIL-8. It is notable that the transcriptional profile of IL-8 exactly coincides with the nuclear abundance of Rel A and c-Rel (cf. Fig. 1C with Fig. 5). Whether NF-B binding alone is sufficient for GTF recruitment to the IL-8 promoter, or whether other factors or binding sites additionally participate, will require systematic evaluation in a chromatin-reconstituted system in vitro.
The TNFRE Is an Essential Cis Element for Inducible IL-8 expression-Our data in alveolar epithelial cells are not consistent with previous transient transfection work mapping Ϫ130 to Ϫ112 as a major site for rhTNF␣-inducibility in bronchial cells (20). In type II A549 cells using both DMS and DNase I protection assays in LMPCR, we observe a rapid and stable binding of NF-B-like proteins to a region between nt Ϫ60 to Ϫ99 that makes specific G contacts at Ϫ79 and Ϫ80 indistinguishably from those produced by recombinant Rel A. Moreover, this NF-B binding is required for functional activity of the transiently transfected IL-8 promoter, establishing a direct correlation between genomic binding and transcriptional activation in the same system. In data not shown, we have demonstrated that rhTNF␣-inducible Ci 1-3 are also detected in normal human bronchial epithelial cells, indicating that NF-B activation is a general phenomenon in pulmonary epithelium.
The indispensable role of NF-B binding for inducible IL-8 gene expression is shown by the complete blockade of inducible promoter activity when NF-B contact sites were mutated in transient transfections (Fig. 3A), and the effect of Nleu on NF-B activation and endogenous IL-8 gene expression. Although Rel A is the predominant NF-B family member binding to the IL-8 TNFRE by supershift assay, our two-step microaffinity isolation/Western immunoblot assay clearly demonstrates the existence of other NF-B members in the Ci 1-3 complex. There has been controversy as to which NF-B members bind to IL-8; although one group was only able to identify Rel A binding to the IL-8 NF-B site (and not NF-B1 or c-Rel; Ref. 53), others have detected Rel A, NF-B1, and c-Rel binding (21). Our data are consistent with the latter studies; the presence of Rel A, NF-B1, and c-Rel binding is sufficient to explain the heterogeneous migration pattern of Ci 1-3 in EMSA. Our failure to demonstrate NF-B1 and c-Rel in the supershift assay may be either due to epitope masking within the complex, or the use of low affinity antibodies. Moreover, in data not shown, we have not been successful in UV cross-linking these complexes because Rel A and NF-B1 do not cross-link equivalently. The use of the microaffinity isolation/Western immunoblot assay avoids all of these potential artifacts. Characterization of the NF-B subunits within the binding complex is important because various subunits produce complexes with different transactivation properties (54). Presently, we cannot say which NF-B combination is the most potent activator of IL-8; in future experiments, it will be of interest to compare transactivation potencies of various NF-B combinations. The identification of multiple NF-B subunits that bind to IL-8 is important information in formulating a strategy for blocking IL-8 gene expression because strategies such as antisense oligonucleotides to specific, individual NF-B members may not be effective.
NF-IL6 was reported to play a role in cooperative binding of the NF-B complex to IL-8 in rhTNF␣-stimulated HeLa cells (21). We are unable to implicate a role for NF-IL6 in either the binding or regulation of IL-8 in alveolar epithelial cells because: 1) NF-IL6 site mutations have no detectable effect on reporter activity, 2) NF-IL6 binding cannot be documented in gel shift assay (Fig. 4B), 3) NF-IL6 is not detectable either in supershift EMSA or in the two-step microaffinity/Western immunoblot assay (data not shown), and 4) constitutive NF-IL6 expression cannot be detected in A549 cells (24).
IB␣ Degradation and Resynthesis in Controlling IL-8 Expression-IB␣ directly associates with Rel A and serves as a primary regulator of NF-B activation through its ability to inhibit its DNA-binding and nuclear translocation (34,41). Our Western immunoblots show that inducible IB␣ proteolysis occurs in response to rhTNF␣ in type II pulmonary cells, as has been demonstrated in other cell lines (34,43,55). The role of IB␣ degradation in controlling IL-8 activation is inferred by the inducible rapid proteolysis that exactly coincides with: 1) Rel A, c-Rel, and NF-B1 nuclear translocation; 2) TNFRE binding of the native IL-8 gene; and 3) activation of endogenous IL-8 transcription. This association is further underscored by the effect of the proteasome inhibitor Nleu. Nleu pretreatment inhibits IB␣ proteolysis, NF-B DNA binding, and IL-8 gene transcription.
Conversely, IB␣ resynthesis appears to play a central role in termination of IL-8 gene expression. Our nuclear run-on assays demonstrate that endogenous IL-8 gene transcription is transient, peaking by 0.5-1 h and declining thereafter, even though the stimulus, rhTNF␣, is still present in the culture medium. Resynthesis of cytoplasmic IB␣ can be detected after 4 h; this phenomenon also coincides with diminished nuclear Rel A and c-Rel by Western immunoblot (Fig. 5), and IL-8 transcription termination (Fig. 1C). We note that, in fibroblasts from ib␣-deficient mice, persistent NF-B activation is seen upon rhTNF␣ stimulation, underscoring the role of IB␣ in terminating the NF-B effect (56). Thus, IB␣ degradation and resynthesis is required for endogenous IL-8 transcriptional activation and termination (respectively) in pulmonary epithelial cells.
In summary, we demonstrate the alveolar-derived cytokine TNF␣ activates a rapid and transient burst in IL-8 transcription. Using LMPCR, we implicate a model for inducible protein binding to the endogenous IL-8 gene a region from nt Ϫ60 to Ϫ99 (the TNFRE) with contact points at G Ϫ79 and Ϫ80 coincident with TATA box recruitment. The TNFRE is activated through inducible binding of Rel A, NF-B1, and c-Rel NF-B subunits that undergo IB␣ proteolysis-dependent nu-clear translocation. Therapeutic targets aimed at this signaling pathway may be efficacious to augment or prevent neutrophilic chemotaxis in specific clinical conditions.