Arsenite Exposure of Cultured Airway Epithelial Cells Activates κB-dependent Interleukin-8 Gene Expression in the Absence of Nuclear Factor-κB Nuclear Translocation*

Airway epithelial cells respond to certain environmental stresses by mounting a proinflammatory response, which is characterized by enhanced synthesis and release of the neutrophil chemotactic and activating factor interleukin-8 (IL-8). IL-8 expression is regulated at the transcriptional level in part by the transcription factor nuclear factor (NF)-κB. We compared intracellular signaling mediating IL-8 gene expression in bronchial epithelial cells culturedin vitro and exposed to two inducers of cellular stress, sodium arsenite (AsIII), and vanadyl sulfate (VIV). Unstimulated bronchial epithelial cells expressed IL-8, and exposure to both metal compounds significantly enhanced IL-8 expression. Overexpression of a dominant negative inhibitor of NF-κB depressed both basal and metal-induced IL-8 expression. Low levels of nuclear NF-κB were constitutively present in unstimulated cultures. These levels were augmented by exposure to VIV, but not AsIII. Accordingly, VIV induced IκBα breakdown and NF-κB nuclear translocation, whereas AsIIIdid not. However, both AsIII and VIV enhanced κB-dependent transcription. In addition, AsIII activation of an IL-8 promoter-reporter construct was partially κB-dependent. These data suggested that AsIII enhanced IL-8 gene transcription independently of IκB breakdown and nuclear translocation of NF-κB in part by enhancing transcription mediated by low levels of constitutive nuclear NF-κB.

Nuclear factor-B (NF-B) 1 was originally described as a constitutive nuclear transcription activator in mature B lymphocytes that bound a specific DNA sequence in the intronic enhancer of the immunoglobulin -light chain (Ig) gene and mediated constitutive Ig expression. (1). However, numerous subsequent studies have shown that NF-B is polymorphic. It is composed of homo-or heterodimers of at least five structurally related mammalian proteins that have a broad tissue distribution. Likewise, NF-B modulates the expression of a large number of genes whose products participate in immune, inflammatory, and environmental stress responses (2,3).
In many tissues NF-B mediates transient changes in gene expression in response to humoral and environmental stimuli. In this case, NF-B is held inactive in the cytoplasm by IBs, a family of inhibitor proteins that mask its nuclear translocation signal. The activation of NF-B is mediated in part by the inactivation of IBs through stimulus-specific posttranslational modifications of IBs. To date, the most commonly observed mechanism of IB inactivaton involves phosphorylation of two N-terminal serine residues by IB kinase, a large multimeric complex that receives input from a variety of signal transduction pathways (4,5). Phosphorylation of IBs by the IB kinase complex targets IBs for ubiquitination and proteolytic degradation. Upon its release from IB, NF-B translocates into the nucleus and binds to B response elements (RE) in the enhancer regions of target genes.
A number of pharmacological interventions that inhibit inducible B-dependent transcription, however, do not inhibit the translocation of cytoplasmic NF-B into the nucleus or its DNA binding activity (6 -10). So an increase in nuclear NF-B alone is not sufficient for the maximal activation of B-dependent transcription. Conversely, enhanced B-dependent transcription has been observed in the absence of an increase in nuclear NF-B in cells that have low levels of constitutive nuclear NF-B (11). This indicates that the mobilization of cytoplasmic NF-B is not invariably necessary for the activation of transcription. Thus B-dependent transcription is dependent upon both the abundance of nuclear NF-B and additional cooperative factors and regulatory processes that influence the transcription activating (transactivating) potential of NF-B.
Of the five known mammalian NF-B family members, p65 (RelA), RelB, c-Rel, p50/p105, and p52/p100, only three, p65, RelB, and c-Rel, are capable of transcriptional activation (2). The transactivation potential of the p65 subunit of NF-B has been shown to depend upon specific p65 protein domains. NF-B family members share a conserved 300-amino acid, N-terminal Rel homology domain that mediates dimerization, nuclear localization, and DNA binding. Phosphorylation of a cAMP-dependent protein kinase site in the Rel homology domain of p65 strongly increases B-dependent transcription and requires IB␣ degradation (6). In contrast, studies using constitutive nuclear chimeric transcription factors have suggested that the transactivation potential of p65 can also be regulated by nuclear processes that are independent of IB␣ degradation and nuclear translocation of p65. Transcriptional regulation by these processes requires one or both of two C-terminal transactivation domains of p65 (11,12).
The mitogen-activated protein (MAP) kinase signal transduction cascades (13)(14)(15) have been implicated as upstream regulatory pathways that mediate the activation of B-dependent transcription by processes that are independent of IB␣ degradation and NF-B nuclear translocation (7,9,11,12). We have recently demonstrated that two metal compounds, sodium arsenite and vanadyl sulfate, activate MAP kinases in airway epthelial cells in vitro (16). These metals also evoke a proinflammatory response as indicated by enhanced production of interleukin-8 (IL-8), an ␣-chemokine that is a neutrophil chemoattractant and stimulant (17,18). IL-8 gene transcription is induced by phorbol esters and the proinflammatory cytokines tumor necrosis factor-␣ and interleukin-1␤ (IL-1␤). This induction depends upon an enhancer region of the IL-8 gene located upstream of the transcription start site (base pairs Ϫ126 to Ϫ72), which includes activator protein-1, C/EBP, and NF-B response elements. All three of these cis-acting elements are necessary for maximal transcriptional activation, although there are tissue-specific differences in this dependence. The activator protein-1 and C/EBP elements are employed in a tissue-specific fashion, whereas the NF-B element is necessary in all tissues examined (19 -22).
In this study, we have investigated the role of NF-B mobilization in B-dependent gene transcription following treatment with As III and V IV . Our results suggest that in cultured airway epithelial cells both As III and V IV activated B-dependent transcription; however, V IV mobilized cytoplasmic NF-B, whereas As III did not.

EXPERIMENTAL PROCEDURES
Cell Culture and in Vitro Exposure-Primary normal human bronchial epithelial (NHBE) cells were obtained from healthy, nonsmoking adult volunteers. Epithelial specimens were obtained by cytologic brushing at bronchoscopy and subsequently expanded in culture as described previously (23). The human BEAS-2B bronchoepithelial cell line was cultured as described previously (24). Vanadyl sulfate or sodium arsenite (both from Sigma) were diluted in BEGM (NHBE) or KGM (BEAS-2B) before addition to the cell culture.
Analysis of IL-8 Expression by RT-PCR and Enzyme-linked Immunosorbent Assay-Extraction of RNA, first-strand cDNA synthesis, and DNA amplification were performed as described previously (23) using the following oligonucleotide primers: GAPDH, sense, CCATG-GAGAAGGCTGGGG, and antisense, CAAAGTTGTCATGGATGACC; IL-8, sense, TCTGCAGCTCTGTGTGAAGGTGCAGTT, and antisense, AACCCTCTGCACCCAGTTTTCCTT; and c-Jun, sense, CGAGCTGGA-GCGCCTGATAAT, and antisense, GCGTGTTCTGGCTGTGCAGTT. Following amplification, products were analyzed by alkaline gel electrophoresis through 2% agarose gels in 1ϫ Tris/borate/EDTA buffer. The gel was stained using 1 g/ml ethidium bromide and photographed under UV illumination with Polaroid type 55 P/N film (Polaroid, Cambridge, MA). The specific bands were quantified using the Kodak 1D Image Analysis Software (Eastman Kodak Company, Rochester, NY), and optical densities for IL-8 mRNA bands were normalized to GAPDH band intensities. IL-8 content in conditioned medium collected from NHBE cells treated with sodium arsenite or vanadyl sulfate was assayed using a commercial IL-8 enzyme-linked immunosorbent assay kit (R & D Systems).
Separation of Cytoplasmic and Nuclear Fractions-After washing NHBE cells with ice-cold PBS, 200 l of cold cytoplasmic extraction buffer, CEB (10 mM Tris-HCl, pH 7.9, 60 mM KCl, 1 mM EDTA, 1 mM dithiothreitol) with protease inhibitors (1 mM Pefabloc, 50 g/ml antipain, 1 g/ml leupeptin, 1 g/ml pepstatin, 40 g/ml bestatin, 3 g/ml E-64, 100 g/ml chymostatin; all purchased from Roche Molecular Biochemicals) was added to each well. Using a rubber policeman, cells were scraped up and transferred into a microcentrifuge tube. The cells were allowed to swell on ice for 15 min and then Nonidet P-40 (Sigma) was added to a final concentration of 0.1%, and the tube was vortexed for 10 s. Nuclei were pelleted by centrifugation at 15,000 ϫ g for 30 s. The supernatant containing the cytoplasmic fraction was mixed with [ 1 ⁄4] volume of 4ϫ loading buffer (62.5 mM Tris-HCl, pH 6.8, 10% glycerol, 2% SDS, 0.7M ␤-mercaptoethanol, 0.05% bromphenol blue), denatured at 95°C for 10 min, and stored at Ϫ70°C for immunoblot analysis. Protein content of a small aliquot of the cytoplasmic fraction was determined using the DC Bradford assay (Bio-Rad). The nuclei were washed with CEB and centrifuged again at 15,000 ϫ g for 30 s. The supernatant was aspirated, and the nuclei were incubated for 10 min on ice in nuclear extraction buffer (20 mM Tris-HCl, pH 8.0, 400 mM NaCl, 1.5 mM MgCl 2 , 1.5 mM EDTA, 1 mM dithiothreitol, 25% glycerol) with protease inhibitors. After brief centrifugation, the supernatants, containing the nuclear fraction, were either stored at Ϫ80°C until analysis by electrophoretic mobility shift assay or denatured and stored for immunoblot analysis as described above.
Promoter Reporter Constructs, Transfection, and Promoter-Reporter Assay-A region of the 5Ј flank of the IL-8 gene (Ϫ1370 to ϩ82) that included the transcription start site was synthesized by amplification of human genomic DNA (Promega, Madison, WI). The amplification products were subcloned into pCR2.1 (Invitrogen, San Diego, CA), and the insert of a clone with a suitable orientation was excised with KpnI and XhoI restriction enzymes (Promega) and inserted upstream of the coding region of the firefly luciferase gene in pGL2-basic (Promega), generating the construct p1.5IL8wt-luc. The NF-B and C/EBP response elements in p1.5IL8wt-luc were disrupted by site-directed mutagenesis using PCR and uracil-containing oligonucleotides as described (25,26). The NF-B response element was mutated from Ϫ82 GTGGAATTTCC Ϫ72 to Ϫ82 GaatAATTTCC Ϫ72 (27), generating p1.5IL8B Ϫ . The C/EBP response element was mutated from Ϫ92 GTTGCAAATC Ϫ83 to Ϫ92 GcTaC-gAgTC Ϫ83 (21), generating p1.5IL8C/EBP Ϫ . Mutations were confirmed by sequencing (University of North Carolina Automated DNA Sequencing Facility, Chapel Hill, NC).
A B-dependent promoter-reporter construct, pNF-B-luc (Stratagene), was also used. It was composed of a 5ϫ tandem repeat of the NF-B RE of the mouse Ig gene intronic enhancer cloned upstream of a TATA box and a firefly luciferase cDNA. A constitutively active SV40 promoter-␤-galactosidase construct, pSV-␤-galactosidase (Promega) was used to adjust for well-to-well variation in cell number and transfection efficiency.
BEAS cells grown to 40 -80% confluence in 24-well tissue culture dishes were co-transfected with 236 pg of one of the IL-8 promoterluciferase vectors or pNF-B-luc and 14 pg of pSV-␤-galactosidase using 1.5 g of DOTAP transfection reagent (Roche Molecular Biochemicals). 48 h after transfection cultures were treated for 1 h with 50 M sodium arsenite or vanadyl sulfate and cultured for an additional 7 (arsenite) or 3 h (vanadium). Luciferase and ␤-galactosidase activity was determined using the Dual Light reporter gene assay system (Perkin-Elmer) and an AutoLumat LB953 luminometer (Berthold Analytical Instruments, Nashua, NH). Promoter activity was estimated as specific luciferase activity (luciferase counts/unit ␤-galactosidase counts).
Infection with Adenovirus-NHBE cells grown to about 80% confluence were infected with Ad5IB␣ (28) or a nonrecombinant control vector, Ad5CMV3, at a multiplicity of infection of 100 plaque-forming units/cell for 3-4 h. The infection mixture was aspirated, and the cells were incubated for another 24 h, before stimulation with sodium arsenite or vanadyl sulfate.
Immunoblot Analysis-Protein samples (50 g) were separated by SDS-polyacrylamide gel electrophoresis on 14% Tris-glycine gels, followed by immunoblotting using specific rabbit antibodies to IB␣ or p65 (both 1:1000, Santa Cruz Biotechnology, Santa Cruz, CA) for 1 h at room temperature. Antigen-antibody complexes were stained with horseradish peroxidase-conjugated goat anti-rabbit antibody (1:4000, Bio-Rad) and enhanced chemiluminescence (ECL) reagent and ECL film (both from Amersham Pharmacia Biotech). Immunoblot films were digitized, and the optical densities of specific antigen-antibody complexes were quantified as described above (see RT-PCR methods).
Indirect Immunofluorescent Localization of Hemaglutinin-tagged IB␣(S32A,S36A)-BEAS-2B cultures that had been infected with Ad5IB␣ (see above) or Ad5LacZ 24 h earlier were fixed for 5 min with 4% paraformaldehyde in CEB at room temprature, lysed for 2 min on ice with 0.2% Nonidet P-40 in CEB, washed once with CEB, fixed again for 20 min on ice, and finally blocked by incubation in 2% BSA/PBS on ice for 1 h. The hemagglutinin-tagged IB␣(S32A,S36A) was localized by incubation overnight in 1 g/ml mouse anti-hemagglutinin monoclonal antibody (Santa Cruz Biotechnology) diluted in 0.2% BSA/PBS followed by a 45-min incubation in a 1:1000 dilution of ALEXA 488 goat anti-mouse secondary antibody (Molecular Probes, Eugene, OR) diluted in 0.2% BSA/PBS. Samples were washed with 2% BSA/PBS and photographed on a Zeiss Axiovert 10 fluorescence microscope using a standard fluorescein excitation and emission filter set.

Exposure to Sodium Arsenite or Vanadyl Sulfate Enhanced IL-8 Gene Expression in NHBE Cells
-Exposure of primary human airway epithelial cells to noncytotoxic concentrations of sodium arsenite or vanadyl sulfate in vitro has been shown to enhance IL-8 expression (16,23). These observations were confirmed and extended by estimating the concentration thresholds for As III -and V IV -induced IL-8 expression. Levels of IL-8 protein in supernatants of NHBE cells cultured in the absence or presence of various concentrations of As III and V IV for 24 h are shown in Fig. 1. NHBE cultures constitutively expressed IL-8, and this expression was augmented in a dose-dependent fashion by challenge with the metals. The threshold concentration for metal-induced IL-8 production was lower for V IV (12.5 M) than for As III (25 M). Likewise, V IV induced greater increases in IL-8 production compared with As III when the metals were used at the same concentrations. These data indicated that V IV was a stronger stimulant than As III . The same doses of iron, nickel, and copper sulfate did not evoke IL-8 expression (not shown). Thus the response to As III and V IV was independent of colloidal properties of metal salts and dependent upon metal species-specific interactions with cellular constituents.
The As III -and V IV -induced IL-8 production by NHBE cultures was preceded by an increase in steady-state levels of IL-8 mRNA. As shown in Fig. 2A, both As III and V IV elevated IL-8 mRNA levels above the basal level within 2 h. Quantitative estimates of IL-8 mRNA abundance using GAPDH mRNA levels to normalize between samples showed that arsenite induced approximately a 2.4-fold increase and vanadium a 5-fold in-crease in steady-state IL-8 mRNA abundance (Fig. 2B). As in the case of IL-8 protein production, V IV showed greater potency than As III in inducing a response.

The Sodium Arsenite-and Vanadyl Sulfate-induced IL-8 Expression in Airway Epithelial Cells Was NF-B-dependent-
The enhanced levels of IL-8 mRNA induced by As III and V IV could be mediated by enhanced IL-8 gene transcription. Because extracellular stimulus-dependent IL-8 gene transcription has been shown to be regulated in part by the transcription factor NF-B (19 -22), the role of NF-B in the As III -and V IV -induced IL-8 expression was investigated. NF-B activity was suppressed in NHBE cultures by overexpression of a dominant negative IB␣ mutant (IB␣(S32A,S36A) in which serines 32 and 36 had been substituted with alanines. Overexpression of this mutant IB can sequester NF-B into IB␣(S32A,S36A)-NF-B complexes that are unresponsive to numerous stimuli that mobilize NF-B by activating IB kinases that specifically phosphorylate serines 32 and 36. NHBE cultures were infected with Ad5IB␣, an adenoviral expression vector encoding hemagglutinin-tagged IB␣(S32A,S36A) (28) or with a nonrecombinant control vector (Ad5CMV3). Analysis of IB␣ levels following infection by immunoblotting confirmed overexpression of IB␣ (data not shown). As expected, stimulation with As III or V IV up-regulated steady-state IL-8 mRNA levels in the control infected cultures (Fig. 3, Ad5-CMV3). In marked contrast, overexpression of the dominant negative IB␣ depressed both the As III -and V IV -induced increases in steady-state IL-8 mRNA abundance to levels below those observed in Ad5CMV3-infected, unstimulated cultures (Fig. 3, Ad5-IB␣). Basal IL-8 mRNA levels were also suppressed, suggesting that NF-B may regulate basal IL-8 expression in NHBE cultures. Arsenite also induced an increase in c-Jun mRNA levels, but this response was not affected (Fig. 3A, c-jun), indicating that IB␣(S32A,S36A) overexpression selec- tively inhibited signal transduction in the NHBE cultures.
Vanadyl Sulfate, but Not Sodium Arsenite, Induced IB␣ Breakdown and p65 Nuclear Translocation and Increased Nuclear NF-B DNA Binding Activity-To determine whether As III or V IV treatment mobilized cytoplasmic NF-B by inducing degradation of IBs, cytoplasmic fractions of As III -or V IVstimulated cells were subjected to immunoblotting analysis using IB␣-and IB␤-specific antibodies. Treatment with V IV induced a rapid (within 30 min) reduction in cytosolic levels of both IB␣ and IB␤ protein levels in NHBE cultures. In contrast, arsenite exposure had no effect on IB␣ or IB␤ levels ( Fig. 4), indicating that V IV , but not As III , induced IB degradation. To further test this inference, levels of the p65 subunit of NF-B in cytoplasmic and nuclear fractions were estimated by immunoblot analysis. Fig. 5 (A and B) shows that basal levels of nuclear p65 were detected in unchallenged cultures and that V IV , but not As III , induced an increase in ratio of nuclear to cytoplasmic p65 (n/c p65) compared with control ratios (Fig. 5C). Overexpression of IB␣(S32A,S36A) blocked the V IV -induced increase in n/c p65 but did not affect n/c p65 in controls or in As III -treated cultures (data not shown), suggesting that IB␣(S32A,S36A) did not alter the partitioning of NF-B between cytoplasm and nucleus in untreated and As IIItreated cultures but did prevent mobilization of NF-B.
As a final test of the apparent differential mobilization of cytoplasmic NF-B by As III and V IV , the influence of challenge with the metals on the levels of nuclear NF-B DNA binding activity were assessed. Electrophoretic mobility shift assays showed that the enhanced n/c p65 ratio observed in V IV -treated cultures coincided with enhanced NF-B DNA binding activity and that As III treatment did not induce a comparable effect (Fig. 5C, compare lane V with lane C). The p65 subunit of NF-B was a component of this DNA binding activity, because it could be supershifted with an anti-p65 antibody (data not shown). However, the steady-state levels of nuclear p65 in untreated and arsenite-treated cultures observed by immunoblotting ( Fig. 5A) were not detected by EMSA (Fig. 5C, lanes C and As), demonstrating that the detection of p65 by EMSA depended upon factors in addition to the mere presence of p65 in nuclear extracts. Given that basal levels of NF-B DNA binding activity were not detected, the EMSA did not rule out the possibility that arsenite mobilized some small quantity of cytoplasmic NF-B that was not detected. Because arsenite did not induce IB breakdown (Fig. 4) or an increase in the n/c p65 ratio (Fig. 5, A and B) and the EMSA was not contradictory, the data were consistent with the notion that V IV , but not As III , mobilized cytoplasmic NF-B.
Both Vanadyl Sulfate and Sodium Arsenite Enhanced B-dependent Transcription in Airway Epithelial Cells-The influence of As III and V IV on B-dependent transcription in airway epithelial cell cultures was investigated by transient transfection assays using a B-dependent promoter-reporter construct, pNF-B-luc. Because of the limited number of primary cells available, these assays were performed using the BEAS-2B human bronchoepithelial cell line (29). Similar to the primary cell lines, the BEAS-2B cells constitutively expressed low levels of IL-8 mRNA and protein in culture that were significantly augmented by treatment with As III or V IV and both basal and inducible expression were significantly reduced by IB␣(S32A,S36A) overexpression (data not shown). BEAS-2B cultures were transiently cotransfected with pNF-B-luc and pSV␤-galactosidase. The pSV␤-galactosidase construct directed ␤-galactosidase expression under the control of a constitutively active viral promoter that did not respond to either As III or V IV treatment (data not shown). Consequently, ␤-galactosidase activity could be used as a normalizing factor to adjust for well-to-well variation in transfection efficiency and cell number as well as an index of cell viability. 48 h after transfection, cultures were left untreated or were treated with 50 M metal for 1 h and then assayed for luciferase and ␤-galactosidase activity 7 h (As III ) or 3 h (V IV ) later. These conditions were based upon the kinetics of IL-8 protein expression, which showed that the response to V IV was rapid with IL-8 increases in the medium evident 4 h after exposure, whereas increased IL-8 protein was not apparent until 8 h after exposure to arsenite (data not shown). Unstimulated cultures supported transcription of pNF-B-luc as assessed by specific luciferase activity (Fig. 6A, Media Ϫ/Ϫ ), and brief exposure to both As III and V IV enhanced transcription above its basal level ( Inhibition of the V IV -induced B-dependent activity was expected, because IB␣(S32A,S36A), a cytoplasmic inhibitor, would be expected to prevent NF-B mobilization, and anti-p65 immunoblotting had shown that it prevented the V IV -induced increase in n/c p65 ratio (not shown). The inhibition of basal and arsenite-induced B-dependent transcription was unexpected, because they appeared to be independent of NF-B mobilization. However, indirect immunofluorescent localization of the hemagglutinin-tagged IB␣(S32A,S36A) transgene product demonstrated that it was present in both the nucleus and cytoplasm (Fig. 6B). The presence of nuclear IB␣(S32A,S36A) and inhibition of B-dependent transcription was in accordance with reports that IB␣ when uncharged with NF-B is imported into the nucleus where it can extract NF-B from transcription initiation complexes and inhibit B-dependent transcription (30 -33). Thus, IB␣(S32A,S36A) overexpres- Representative immunoblots are shown. B, densitometric analysis of optical densities of the anti-p65 immunoreactive bands from at least three independent experiments are shown. The data are expressed as the mean increase in p65 levels relative to unchallenged controls Ϯ S.E. C, a representative EMSA for NF-B DNA binding activity in NHBE cell nuclear extracts prepared from untreated cultures (C) and cultures treated with 50 M arsenite (As) or vanadyl sulfate (V) are shown. A radiolabeled double-stranded oligonucleotide corresponding to the NF-B RE of the MHC class II gene enhancer was used as probe (see Table I).

FIG. 6. Both vanadium and arsenite activated B-dependent transcription of a 5xNF-B-reporter construct in BEAS-2B cells.
BEAS-2B cultures were transiently cotransfected with pNF-B-luc and pSV␤-galactosidase (see "Experimental Procedures"). 24 h post-transfection, cultures were left uninfected (Ϫ/Ϫ) or were infected with Ad5CMV3 (ϩ/Ϫ) or Ad5IB␣ (Ϫ/ϩ) at an multiplicity of infection of 100 plaque-forming units/cell for 3 h. 48 h post-transfection, cultures were challenged with 50 M sodium arsenite or vanadyl sulfate for 1 h and harvested 7 or 3 h later, respectively. Specific luciferase activity in culture lysates was determined using ␤-galactosidase activity as a normalizing factor. The data are expressed as mean specific luciferase activity Ϯ S.E., n ϭ 5. A, both arsenite and vanadyl treatment enhanced B-dependent transcription. The inducible as well as the basal activity was inhibited by overexpression of IB␣(S32A,S36A). B, indirect immunofluorescent localization of the hemagglutinintagged IB␣(S32A,S36A) transgene product showed that it was present in both the cytoplasm and nucleus of Ad5IB␣-infected BEAS-2B cells. Nuclear IB␣(S32A,S36A) could explain the inhibition of the basal and arsenite-induced activity which were independent of NF-B mobilization. C, hemagglutinin immunoreactivity was not detected in cultures infected with Ad5LacZ, an expression vector encoding untagged ␤-galactosidase. Bar, 25 m. sion suggested that airway epithelial cell cultures supported a basal level of B-dependent transcription that was augmented by exposure to either As III or V IV .
Sodium Arsenite Induced B-dependent IL-8 Promoter-Reporter Activity in Airway Epithelium-The IB␣(S32A,S36A)mediated inhibition of IL-8 mRNA levels (Fig. 3B) suggested that arsenite may be stimulating B-dependent IL-8 gene transcription. To investigate this possibility, the influence of arsenite on the activity of an IL-8 promoter-luciferase construct was examined. The IL-8 promoter-reporter construct was active in unchallenged cultures (Fig. 7, Media Ϫ/Ϫ ), consistent with the observed basal expression of IL-8 mRNA and protein in cultures (Figs. 1-3). Moreover, basal transcriptional activity was suppressed by overexpression of IB␣(S32A,S36A) (Fig. 7, WT, compare Media Ϫ/ϩ to Media Ϫ/Ϫ ), consistent with the observed depression in basal IL-8 mRNA levels following infection with Ad5IB␣ (Fig. 3). Arsenite induced a significant increase in the transcriptional activity (Fig. 7, WT, compare 50 M As Ϫ/Ϫ to Media Ϫ/Ϫ ), whereas overexpression of IB␣(S32A,S36A) inhibited this response (Fig. 7, WT, compare 50 M As Ϫ/ϩ to 50 M As Ϫ/Ϫ ). There was, however, a residual difference in the activity of IL-8 promoter-reporter construct in unstimulated and arsenite-challenged cultures that had been infected with Ad5IB␣ (Fig. 7, WT, compare Media Ϫ/ϩ to 50 M As Ϫ/ϩ ). This suggested that only a portion of the arsenite-induced IL-8 promoter-reporter activity was B-dependent. Infection with the nonrecombinant adenovirus did not affect the basal or inducible transcriptional activity (Fig. 7, WT, compare Media Ϫ/Ϫ to Media ϩ/Ϫ and 50 M As Ϫ/Ϫ to 50 M As ϩ/Ϫ ). In addition, exposure to arsenite did not affect the activity of the promoterless parent luciferase vector of the IL-8 promoter-reporter construct or that of a constitutively active SV40 promoterluciferase construct (not shown). The specificity of the inhibition mediated by IB␣(S32A,S36A) overexpression was investigated by determining its effect on the activity of an IL-8 promoter reporter construct in which the NF-B response element had been inactivated by mutation. As expected, the dominant negative IB␣ did not affect the B-independent activity of the mutant IL-8 promoter (NF-B Ϫ 50 M As), indicating that IB␣(S32A,S36A) selectively inhibited B-dependent transcription. These data supported the notion that the arsenite-induced increase in IL-8 expression was partially dependent upon enhanced, B-dependent IL-8 gene transcription.

The Basal and Arsenite-induced IL-8 Promoter-Reporter Activity Required the Compound C/EBP/NF-B Response Element of the IL-8 Promoter-
The B dependence of the basal and arsenite-induced activity of the IL-8 promoter-reporter construct suggested by the suppressive effects of IB␣(S32A,S36A) overexpression (Fig. 7) was confirmed by mutational analysis of the IL-8 promoter. Several studies have indicated that inducible B-dependent IL-8 gene transcription requires a compound C/EBP (nuclear factor-IL-6)/NF-B response element located upstream (base pairs Ϫ94 to Ϫ72) of the transcription start site in the IL-8 gene (20 -22, 34), although the C/EBP element may be dispensible in some instances (19). Consequently, the C/EBP and NF-B elements of the compound RE in the IL-8 promoterreporter construct were independently disrupted by site-directed mutagenesis, and the phenotype of these mutations was characterized by transient transfection of BEAS-2B cultures. The basal activity of the B Ϫ construct was reduced about 20-fold compared with the wild type construct (Fig. 8, compare lanes C at both WT and NF-B Ϫ ). Reversion of the B Ϫ construct to wild type restored basal activity to wild type levels (not shown), indicating that the reduction in basal activity was due solely to disruption of the NF-B response element. Disruption of the C/EBP RE also significantly reduced the basal IL-8 promoter-reporter activity (Fig. 8, compare lanes C at both WT and C/EBP Ϫ ), although to a lesser degree than mutation of the NF-B RE. Thus, the basal activity of the IL-8 promoterreporter construct was dependent upon both the NF-B and C/EBP elements of the IL-8 promoter. Moreover, the full basal activity of the wild type construct was approximately 2.7 times greater than the sum of the activities B Ϫ and C/EBP Ϫ constructs, suggesting that IL-8 promoter activity in unstimulated airway epithelium depends upon synergistic interactions between nuclear factors that bind to the C/EBP/NF-B compound RE of the IL-8 promoter.
Even though the overall activity of the B Ϫ construct was greatly reduced, it retained some arsenite responsiveness (Fig.  8, compare lane C at NF-B Ϫ to lane As at NF-B Ϫ ). Arsenite induced a 3.8 Ϯ 1-fold increase of the activity of the wild type construct, whereas it induced a significantly smaller 2.4 Ϯ Specific luciferase activity in culture lysates was determined using ␤-galactosidase activity as a normalizing factor. The data are expressed as the mean specific luciferase activity Ϯ S.E., n ϭ 5. IB␣(S32A,S36A) overexpression inhibited arsenite induced wild type IL-8 promoterreporter activity, whereas the activity of the NF-B Ϫ construct was unaffected, suggesting that the IB␣ transgene product specifically inhibited the function of NF-B RE of the IL-8 promoter .   FIG. 8. Both basal and arsenite-induced IL-8 promoter-reporter activity was dependent upon the compound C/EBP/ NF-B response element of the IL-8 gene. BEAS-2B cultures were transiently cotransfected with pSV␤-galactosidase and wild type (wt) or mutant IL-8 promoter-reporter constructs in which either the NF-B RE (NF-B Ϫ ) or the C/EBP RE (C/EBP Ϫ ) had been disrupted (see "Experimental Procedures"). 48 h post-transfection, cultures were left untreated (Media) or challenged with 50 M sodium arsenite for 1 h (As) and harvested 7 h later. Specific luciferase activity in culture lysates was determined using ␤-galactosidase activity as a normalizing factor. The data are expressed as the mean specific luciferase activity Ϯ S.E., n ϭ 9. 0.5-fold increase in the activity of the B Ϫ construct. This suggested that transcription factors in addition to NF-B dominated responsiveness of the IL-8 promoter to arsenite exposure. Mutation of the C/EBP RE suppressed the arsenite inducibility of the IL-8 promoter-reporter construct (Fig. 8, compare lane C at C/EBP to lane As at C/EBP Ϫ ). Thus the C/EBP RE had a greater influence on the arsenite-inducible activity than the NF-B RE had. As in the case of basal activity, the activity of the wild type construct was approximately 6.3 times greater than the sum of activities of the B Ϫ and C/EBP Ϫ constructs, suggesting synergistic interactions between transcription factors.
Because the basal and arsenite induced IL-8 promoter-reporter activity was dependent upon the compound C/EBP/ NF-B response element of the IL-8 promoter (Fig. 8), nuclear extracts of NHBE cultures were analyzed by EMSA for DNA binding activities specific for this sequence. There were detectable levels of a single DNA binding activity in the nuclei of unchallenged cultures that were enhanced following treatment with arsenite for 1 h (Fig. 9A, arrow). The increases were transient, returning to control levels after 4 h of exposure. The activity was specific for the sequence of the compound C/EBP/ NF-B RE, because competition with 100-fold molar excess of unlabeled probe inhibited radiolabeled complex formation (Fig.  9B). Mutation of either half of the response element resulted in a significant reduction in DNA binding (Fig. 9C, compare wt with mNF-B and mC/EBP). This basal and enhanced DNA binding activity for the C/EBP/NF-B compound response element and its sensitivity to mutation correlated with the observed basal and arsenite-induced activation of the IL-8 promoter-reporter construct (Figs. 7 and 8) and its inhibition by disruption of the compound C/EBP/NF-B response element (Fig. 8).
Nuclear extracts were also examined using a radiolabeled probe corresponding to the solitary C/EBP response element of the IL-6 gene (Table I). A DNA binding activity was observed in unstimulated cultures, and this activity was enhanced following exposure to arsenite (Fig. 9D, arrow, lanes C and As). These data demonstrated the presence of a constitutive nuclear factor in airway epithelium that binds the C/EBP response element and whose activity was increased by arsenite exposure. DISCUSSION In this study we investigated As III -and V IV -induced NF-B activation pathways, which culminate in IL-8 gene expression in airway epithelial cells. Both the As III -and V IV -induced IL-8 expression were NF-B-dependent; however, V IV induced IB␣ degradation and NF-B translocation, whereas exposure to As III failed to do so. Thus, despite the B dependence of arsenite-induced gene expression, there was no detectable mobilization of cytoplasmic NF-B, suggesting that the response to arsenite was mediated by low levels of constitutive nuclear NF-B that were detected in the airway epithelial cell cultures.
The presence of low levels of constitutive nuclear NF-B was suggested by several pieces of evidence: (i) Nuclear p65 was detected by immunoblotting of nuclear extracts of unchallenged cultures (Fig. 5); (ii) EMSA of nuclear extracts using a recognized functional compound C/EBP/NF-B response element of the IL-8 gene revealed basal nuclear levels of a B-dependent DNA binding activity (Fig. 9B); (iii) unstimulated cultures supported B-dependent transcription from both 5xNF-B-reporter (Fig. 6) and IL-8 promoter-reporter constructs (Figs. 7 and 8); and (iv) basal expression of IL-8 mRNA was B-dependent (Fig. 3).
The B dependence of basal IL-8 mRNA expression and basal activities of the promoter-reporter constructs was suggested by their suppression following global inhibition of NF-B function by overexpression of a dominant negative IB␣ mutant (Figs. 3, 6, and 7). The mutant IB␣ was present not only in the cytoplasm but also in the nucleus (Fig. 6B). This is in accordance with observations that IB␣ is imported into the nucleus when uncharged with NF-B (30), a likely situation when IB␣ is overexpressed. Nuclear IB␣ inhibits B-dependent transcription (31)(32)(33), which was also observed here. The FIG. 9. Arsenite enhanced the levels of a nuclear DNA binding activity for the C/EBP/NF-B RE of the IL-8 promoter in NHBE cells. A, nuclear extracts isolated from NHBE cultures were analyzed for DNA binding activities by EMSA using a radiolabeled probe corresponding the compound C/EBP/NF-B RE of the IL-8 promoter (see Table I). A nuclear DNA binding activity for the compound RE in unstimulated cultures (lane C, arrow) was transiently enhanced by a 1-h exposure to 50 M sodium arsenite but returned to basal levels after 4 h of stimulation. B, competition with 100-fold molar excess of wild type probe inhibited radiolabeled complex formation with nuclear factors isolated from unstimulated cultures (lane C) or cultures challenged with 50 M arsenite for 1 h (lane As). C, nuclear extracts were examined by EMSA for their affinity for a radiolabeled wild type compound RE (wt) or mutant compound RE in which the NF-B site (mNF-B) or the C/EBP site (mC/EBP) had been disrupted (see Table I). Nuclear factors from both unstimulated cultures (lane C) and cultures challenged with 50 M sodium arsenite (lane As) had substantially reduced affinity for the mutated compound RE (arrow). D, EMSA of nuclear extracts from untreated cultures (lane C) or cultures treated with 50 M sodium arsenite for 1 h using a radiolabeled probe corresponding to the C/EBP response element of the IL-6 gene (see Table I) is shown. Arsenite exposure enhanced the basal activity of a nuclear factor that bound to the C/EBP RE (arrow). specificity of the inhibition for B-dependent processes was suggested by a number of observations. Overexpression of IB␣(S32A,S36A) did not inhibit the As III -induced increase in c-Jun message (Fig. 3) or the B-independent activity of the IL-8 promoter (Fig. 7). Additional studies indicate that overexpression of the mutant IB␣ does not inhibit basal or phorbol myristate acetate-induced activator protein-1-dependent transcription but does inhibit phorbol myristate acetate-induced Bdependent transcription. 2 Thus, it is clear that IB␣(S32A,S36A) overexpression did not inhibit transcription in a nonspecific fashion. The data consistently supported the notion that there were low levels of constitutive nuclear NF-B and basal IL-8 expression in airway epithelial cell cultures.
The origin of the low levels of constitutive nuclear NF-B and IL-8 expression is not clear. Environmental stresses because of artificial cell culture conditions have been shown to elicit IL-8 from cultured peripheral blood mononuclear cells, whereas freshly isolated (naïve) peripheral blood mononuclear cells do not express IL-8 (35). Thus, it is possible that stresses because of artificial culture conditions in addition to As III are acting on the primary cell lines and that these stresses establish the low levels of constitutive nuclear NF-B and IL-8 expression that are a prerequisite for the As III -induced B-dependent transcription. Alternatively, the constitutive nuclear NF-B and IL-8 expression may be a tissue characteristic of airway epithelium in vivo and in vitro. Recent clinical studies using RT-PCR have shown that IL-8 mRNA is expressed in naïve (uncultured) biopsies of airway epithelium (36). In addition, low levels of IL-8 are invariably found in the airway lining fluid of normal healthy individuals (36 -40). Although there are other cell types in the airway that can produce IL-8, epithelial cells are by far the most abundant. The expression of IL-8 by airway epithelium in vivo may be related to the role the epithelium plays in host defense. The airway is in an unusual physiological situation. It is constantly exposed to respirable environmental pathogens and toxicants, protected only by a thin layer of fluid containing mucus and proteins. Many of the pathogens and toxicants to which airway epithelial cells are constantly exposed are capable of inducing translocation of NF-B into the nucleus in vitro (41)(42)(43)(44)(45). Thus, the IL-8 expression detected in normal airway epithelium may be a consequence of chronic low levels of stress because of environmental pathogens and toxicants. Alternatively, airway epithelium may constitutively express low levels of IL-8 that mediate heightened immune surveillance of the airways. Even though these possibilites cannot be distinguish at the moment, the primary cell lines appear to be a reasonable model of airway epithelium in vivo.
Constitutive nuclear NF-B has been observed previously in mature B cells (46), activated monocytes and macrophages (47), neurons (48), vascular endothelial cells (49), and fibroblasts (11). These studies suggest that the proportion of NF-B that is constitutively nuclear and that the subunit composition of nuclear NF-B varies with tissue type as well as with cellular differentiation and activation state. How constitutive nuclear levels of NF-B are set remains unclear. In neurons it may be determined by an autocrine activation loop (48). B lymphocyte maturation is associated with decreased stability of IB␣ as well as de novo expression of nuclear RelB-p52/p100 heterodimers, which are not efficiently inhibited by IB␣ (50,51). In certain tumor cell lines, elevated constitutive nuclear NF-B is associated with decreased stability of IB␣ (52,53), whereas dysfunctional IB␣ mutants have been observed in other tumor cells (54). Generalizing these observations leads to the suggestion that levels of constitutive nuclear NF-B may also vary as a consequence of genetic polymorphisms in NF-B or in the macromolecules that regulate NF-B expression or activity. The findings presented here may provide a framework for anticipating variation in the effects of arsenite exposure depending on tissue type, cellular differentiation or activation state, and possibly individual susceptibility.
Arsenite activated B-dependent transcriptional activity of a 5xNF-B-reporter and IL-8 promoter-reporter construct in airway epithelium without increasing the low levels of constitutive nuclear NF-B. This finding is consistent with a recent study in which enhanced B-dependent transcription induced by Ras or Raf transformation of NIH 3T3 fibroblasts was mediated by constitutive nuclear NF-B (11). In this case, transformation was shown to activate a chimeric transcription factor composed of the DNA-binding domain of the yeast Gal4 transcription factor and the C-terminal TA1 transactivation domain of p65. This suggested that transcriptional activation was due to functional activation of p65. Using the same strategy, the transcription promoting activity of the TA1 and TA2 transactivation domains of p65 have also been shown to be responsive to tumor necrosis factor-␣ (12). Both of these studies suggested that activation does not depend upon translocation of the hybrid transcription factor into the nucleus but rather is a consequence of nuclear processes that are dependent upon activation of p38 and/or ERK1/2 MAP kinases. Arsenite has been shown to activate ERK1/2, JNK/SAPK, and p38 MAP kinases in a variety of tissues (55-60) including airway epithelial cells (16). Taken together, these studies present a plausible mechanism by which arsenite may activate B-dependent gene transcription in the presence of low levels of constitutive nuclear NF-B. Currently, we are addressing the role of MAP kinase activation in arsenite-induced B-dependent transcription in airway epithelium.
The arsenite-induced activation of the IL-8 promoter-reporter construct was shown to depend upon the compound C/EBP/NF-B response element of the IL-8 gene (Fig. 8). Previously, this compound RE was shown to be essential for the activation of IL-8 promoter-reporter constructs by phorbol esters, tumor necrosis factor-␣, and IL-1␤ (20 -22). Both p65 and the ␤ isoform of C/EBP (C/EBP␤) family of transcription factors (61) have been shown to bind this RE in vitro (21,27) and synergistically activate transcription dependent upon the C/EBP/NF-B RE (21,34). However, it is likely that other combinations of C/EBP and NF-B isoforms may cooperate at this site, because numerous synergistic combinations of the six known isoforms of C/EBP with NF-B isoforms have been suggested (61). In this study, analysis of nuclear extracts by EMSA 2 W. Reed, unpublished observations. GGCTGGGGATTCCCCATCT IL-6 C/EBP TGCAGATTGCGCAATCTGCA indicated the presence of a DNA binding activity for the compound RE in unstimulated cultures whose activity was transiently enhanced by arsenite exposure (Fig. 9A). This activity correlated with the basal and arsenite-enhanced activity of the IL-8 promoter-reporter construct (Fig. 8). Additional analysis of nuclear extracts by EMSA indicated that a nuclear factor that binds a solitary C/EBP RE is present in unstimulated cultures and its activity is enhanced by arsenite exposure (Fig.  9D). C/EBP isoforms can be phosphorylated on independent regulatory sites by cAMP-dependent protein kinase, calcium/ calmodulin-dependent protein kinase, protein kinase C, and MAP kinase, and these modifications influence nuclear translocation and DNA binding activity (61,62). Based upon our data, we would predict that a C/EBP-like factor complexed with constitutive nuclear p65, both in functionally activated states, explain the arsenite-induced DNA binding activity for the C/EBP/NF-B RE, as well as the dependence of arsenite-induced IL-8 promoter-reporter activity upon the compound RE. Further studies are necessary to identify the factor or factors binding to the C/EBP/NF-B RE of the IL-8 promoter as well as their individual transactivation potentials in unstimulated and arsenite-treated airway epithelium.
In conclusion, the study presented here describes a distinct mechanism of enhanced B-dependent gene expression in airway epithelium. In the absence of IB␣ breakdown and mobilization of cytoplasmic NF-B, exposure to arsenite increased B-dependent transcription and gene expression, implying a potential role for basal levels of nuclear NF-B in inducible gene transcription.