H2O2 and tumor necrosis factor-alpha induce differential binding of the redox-responsive transcription factors AP-1 and NF-kappaB to the interleukin-8 promoter in endothelial and epithelial cells.

We previously demonstrated that tumor necrosis factor-alpha (TNFalpha) and H2O2 differentially regulate interleukin-8 (IL-8) and intercellular adhesion molecule (ICAM-1) gene expression in endothelial and epithelial cells. H2O2 induced IL-8 expression in the A549 and BEAS-2B epithelial cell lines, but not in the human microvessel endothelial cell line, HMEC-1 or human umbilical vein endothelial cells. In contrast, H2O2 induced ICAM-1 only in endothelial cells. Unlike H2O2, the proinflammatory cytokine TNFalpha induced IL-8 and ICAM-1 in both cell types. In this study, we examine the role of the redox-responsive transcription factors AP-1 and nuclear factor-kappaB (NF-kappaB) in the differential expression of IL-8. DNA binding studies using nuclear protein extracts from HMEC-1 and A549 cells stimulated with H2O2 or TNFalpha demonstrated differential activation and promoter binding of AP-1 and NF-kappaB. H2O2 activated AP-1 but not NF-kappaB in A549, whereas TNFalpha activated AP-1 as well as NF-kappaB. In HMEC-1, TNFalpha activated NF-kappaB but not AP-1, while H2O2 did not activate either transcription factor. The differential activation of the factors was also reflected in their differential binding to the IL-8 promoter. Moreover, the H2O2 concentration dependent increase in epithelial IL-8 mRNA expression directly corresponded to the H2O2 concentration dependent binding of AP-1 to the IL-8 promoter. Supershift analysis revealed H2O2 as well as TNFalpha induced AP-1 complexes containing c-Fos and JunD. TNFalpha induced NF-kappaB complexes containing Rel A (p65). Immunohistochemical staining of HMEC-1 and A549 cells revealed TNFalpha stimulated nuclear localization of Rel A, whereas no translocation of Rel A was detected in either cell type stimulated by H2O2. These data indicate that the cell type-specific induction of IL-8 gene expression by H2O2 and TNFalpha in HMEC-1 and A549 cells can be explained by the differential binding of AP-1 and NF-kappaB to the IL-8 promoter.

IL-8 is induced by oxidant stress (20,21), and antioxidants have been shown to inhibit IL-8 expression (22,23). H 2 O 2 increases IL-8 expression in epithelial cell lines, fibroblasts, and whole blood (20,21), and hypoxia followed by reoxygenation increases IL-8 expression in mononuclear and endothelial cells (24,25) and in the lung and myocardium in vivo (26,27). Nitric oxide, a reactive nitrogen species, has been shown to induce IL-8 through the activation of NF-B (28).
We recently reported that H 2 O 2 and TNF␣ differentially regulate IL-8 and ICAM-1 gene expression in epithelial and endothelial cells (20). IL-8 was induced by H 2 O 2 in epithelial cells but not in endothelial cells. In contrast, H 2 O 2 induced ICAM-1 in endothelial cells but not in epithelial cells. TNF␣ has also been shown to generate oxidant stress (29,30) and induce IL-8 and ICAM-1 expression in both cell types (20). In addition, TNF␣ and H 2 O 2 induce a differential pattern of CC chemokine expression in epithelial cells. While TNF␣ induced both RANTES and MCP-1, H 2 O 2 only induced MCP-1 in A549 cells (20). These studies suggest that oxidant stress constitute cell type-and gene type-specific activation signals in epithelial and endothelial cells that may critically influence the sitespecific recruitment of leukocyte subsets in inflammatory reactions.
The aim of the present study was to investigate the transcriptional mechanism by which H 2 O 2 and TNF␣ differentially regulate IL-8 gene expression in endothelial and epithelial cells. We demonstrate that the redox-sensitive transcription factors AP-1 and NF-B are differentially activated by H 2 O 2 and TNF␣ in HMEC-1 and A549 cells. Our findings suggest that the discordant binding of the redox-responsive transcription factors AP-1 and NF-B to the IL-8 gene promoter mediate the distinct pattern of IL-8 observed in epithelial and endothelial cells.
Cell Culture and Treatments-The A549 human type II lung carcinoma cell line was grown and maintained in Dulbecco's modified Eagle's medium with 10% fetal calf serum, 1% penicillin/streptomycin, and 1% gentamycin to 90% confluence. HMEC-1 was cultured in MCDB-131 media (Life Technologies, Inc.) with 10% fetal calf serum, 1% penicillin/ streptomycin, 1% gentamycin, 1 g/ml hydrocortisone, and 0.01 g/ml epidermal growth factor to 90% confluence in 24-well dishes. Cells were washed twice with 1 ϫ PBS and covered with serum-free, phenol red-free and growth factor-free media for 24 h prior to agonist treatment. H 2 O 2 and TNF␣ were diluted to final concentration in serum and growth factor-free, phenol red-free media and applied to cells as indicated in the figure legends.
DNA Binding Studies-Electrophoretic mobility shift assays (EMSA) were performed essentially as described (13). Briefly, nuclear protein extracts (3-6 g of protein) prepared from A549 or HMEC-1 cells by the method of Osborn et al. (31) were incubated with 50,000 cpm (ϳ0.1 ng) of 32 P-end-labeled oligonucleotide probes listed in Table I for 20 -30 min at room temperature in 10 or 20 l reaction volumes containing 12% glycerol, 12 mM HEPES-NaOH (pH 7.9), 60 mM KCl, 5 mM MgCl 2, 4 mM Tris-Cl (pH 7.9), 0.6 mM EDTA (pH 7.9), 0.6 mM dithiothreitol, and 0.25 g of poly(dI-dC). To demonstrate binding specificity, 100-fold molar excess (10 ng) of a specific or nonspecific oligonucleotide as indicated in the figure legends was included in the binding reaction. For supershift analysis, antibodies (1-3 g) against AP-1 (c-Jun, c-Fos, JunD) or NF-B (Rel A, p50, c-Rel) subunits (Santa Cruz Biotechology, Santa Cruz CA) were included in the binding reaction. Protein-DNA and protein-DNA-antibody complexes were resolved in 5% polyacrylamide gels pre-electrophoresed for 30 min at room temperature in 0.25 ϫ TBE buffer (22.5 mM Tris borate and 0.5 mM EDTA, pH 8.3). Gels were dried and exposed to radiographic film with an intensifying screen at Ϫ70°C. Gel shifts were performed at least twice with using nuclear extract prepared from different batches of cells. Similar results were obtained and a representative gel is shown in the figures.
Immunostaining-A549 and HMEC-1 were grown on coverslips to 80% confluence. Cells were washed twice in PBS and fixed in 3.7% formaldehyde (diluted in PBS containing 0.2% Triton X-100) for 10 min. The coverslips were washed three times in PBS and covered with 0.2% bovine serum albumin diluted in PBS (blocking buffer) for 10 min. The blocking buffer was removed by draining and 4 -8 g of anti-human NF-B (Rel A/p65) antibody (rabbit polyclonal IgG purchased from Upstate Biotechnology, NY) diluted in 100 l of blocking buffer added to the cells for 60 min. The antibody was removed and the cells washed three times in PBS. The cells were then incubated with the secondary antibody fluorescein-tagged goat anti-rabbit IgG. After 60 min, the secondary antibody was removed and the cells washed three time in PBS. The coverslips were mounted to slides containing Citifluor. Immunofluorescent Rel A staining was detected using a fluorescent microscope.

H 2 O 2 and TNF␣ Induce Different
Patterns of AP-1 and NF-B Binding Activity in HMEC-1 and A549 -To investigate the role of the redox-sensitive transcription factors AP-1 and NF-B in the cell type-specific expression of IL-8 gene expression, we first performed a series of DNA binding studies using consensus binding sites for AP-1 and NF-B (Table I). Nuclear protein extracts were prepared from A549 epithelial cells and HMEC-1 endothelial cells exposed for 15, 30, or 60 min to concentrations of H 2 O 2 or TNF␣ shown previously to activate IL-8 gene expression. DNA binding activity was determined by the EMSA. As shown in Fig. 1, H 2 O 2 increased AP-1 binding activity in A549 (Fig. 1A) but not NF-B binding activity (Fig.  1B). AP-1 binding activity was detected as early as 15 min and was increased at 60 min (Fig. 1A). In contrast, TNF␣ increased both AP-1 and NF-B binding activity in A549 cells, although their kinetics of induction differed. AP-1 binding activity induced by TNF␣ peaked at 30 min (Fig. 1A), whereas NF-B binding activity continued to increase over the 60-min time course (Fig. 1B).
In contrast to A549, HMEC-1 displayed a very different pattern of binding activity for the two transcription factors. As shown in Fig. 2, H 2 O 2 and TNF␣ did not increase AP-1 binding activity over the constitutive binding activity in HMEC-1 and in fact slightly reduced this constitutive binding activity at 15 and 30 min ( Fig. 2A). However, like A549, TNF␣ induced NF-B binding activity in HMEC-1 and the kinetics were similar in the two cell types (Fig. 2B). NF-B binding activity was detected at 15 min and continued to increase at 30 and 60 min. These data indicate that H 2 O 2 and TNF␣ differentially activate the redox-responsive transcription factors AP-1 and NF-B in A549 and HMEC-1 providing a potential mechanism for the cell type-specific expression of IL-8 and ICAM-1 in epithelial and endothelial cells we previously reported (13,20).

H 2 O 2 and TNF␣ Induce Different Patterns of AP-1 and NF-B Binding Activity to the IL-8 Promoter in A549 and
HMEC-1-To determine whether the differential activation of AP-1 and NF-B could mediate the cell type-specific expression of IL-8, we examined the AP-1 and NF-B binding activities on the IL-8 promoter. Oligonucleotide sequences corresponding to the AP-1 and NF-B-binding sites of the IL-8 promoter (Table  I) were assessed for binding activity by EMSA. As shown in Figs binding of AP-1 and NF-B to the consensus sites was observed on the IL-8 promoter sites. The correspondence between the consensus and IL-8-binding sites suggests that the distinct expression patterns of IL-8 induced by H 2 O 2 and TNF␣ in epithelial and endothelial cells is mediated by the differential activation of AP-1 and NF-B binding to the IL-8 promoter.
TNF␣ Induce Distinct NF-B Binding Complexes in HMEC-1 and A549 -To demonstrate that the IL-8 and consensus oligonucleotides were indeed binding similar AP-1 and NF-B proteins, we performed a series of competition experiments. As shown in Fig. 5, the AP-1 and NF-B binding complexes formed on the IL-8 promoter sites were competed by a 100-fold molar excess of the appropriate consensus oligonucleotide, indicating the IL-8 promoter sites also recognize and bind consensus AP-1 and NF-B binding complexes. Although the consensus NF-Bbinding sites competed with the IL-8 site, a comparison of the NF-B binding complexes in A549 and HMEC-1 revealed differences in mobility of the induced complexes (Fig. 5B). In A549 cells, the predominant NF-B binding activity migrated as a single complex between the non-induced, nonspecific complexes (NIC), whereas in HMEC-1 the predominant NF-B binding activity migrated as a cluster of complexes above the noninduced complex. Because NF-B is a family of proteins that bind as heterodimers, it is not unusual to detect multiple NF-B specific complexes. These data indicate that distinct NF-B binding complexes are selectively induced by TNF␣ in A549 and HMEC-1. In contrast to NF-B, the AP-1 binding complexes induced by H 2 O 2 or TNF␣ comigrated in the two cell types (Fig. 5A), suggesting similar AP-1 binding complexes are induced by H 2 O 2 and TNF␣.
Mutation of the IL-8 Promoter Binding Sites Abrogate NF-B and AP-1 Binding Activity-To determine if mutations of the AP-1 and NF-B-binding sites in the IL-8 promoter alter AP-1 and NF-B binding activity in A549 and HMEC-1, we introduced point mutations in the IL-8 promoter oligonucleotides ( Table I)  activity (18,32). As shown in Fig. 6A, mutation of the AP-1 or NF-B-binding sites disabled their ability to compete with their respective wild type binding sites. In addition, as shown in Fig. 6, B and C, no induced DNA binding activity was detected on the mutant oligonucleotide probes. These data demonstrate that mutation of the AP-1 and NF-B-binding sites in the IL-8 promoter that affect function also prevent H 2 O 2 and TNF␣ induced DNA binding activity.
Composition of the AP-1 and NF-B Binding Complexes Induced by H 2 O 2 and TNF␣-To identify the AP-1 and NF-B components induced in A549 and HMEC-1, we used antibodies to AP-1 and NF-B subunits in supershift assays. As shown in Fig. 7, TNF␣ and H 2 O 2 induced the same AP-1 components in A549 cells. Antibodies to c-Fos and JunD produced supershifts of the AP-1 complex induced by H 2 O 2 or TNF␣ (Fig. 7A). In contrast, antibodies to c-Jun produced no supershifted com- Although TNF␣ induced multiple gel shift complexes, we were only able to detect Rel A (p65) as a component of the NF-B binding complexes in A549 and HMEC-1. As shown in Fig. 7B, Rel A (p65) antibodies supershifted NF-B complexes induced by TNF␣ in both cell types. In contrast, no supershifts were detected with p50 or cRel antibodies. In control experiments, the anti-p50 antibody did supershift NF-B binding complexes derived from monocytic cells, indicating that the p50 antibody was functional (data not shown).
To directly demonstrate that H 2 O 2 did not activate Rel A in HMEC-1 and A549 cells, we examined nuclear translocation of Rel A by immunohistochemistry. In resting cells, NF-B resides primarily in the cytoplasm complexed with an inhibitor protein I-B (33). Cellular activation leads to the phosphorylation of I-B, which target the inhibitor protein for ubiquination and then proteolysis by the proteasome (34). Proteolytic degradation of I-B permits NF-B to translocate to the nucleus and bind its recognition site in the promoter of IL-8 (35). To examine NF-B activation and nuclear translocation, HMEC-1 and A549 cells were treated with H 2 O 2 or TNF␣ for 60 min and stained with a rabbit anti-Rel A (p65) antibody. Localization of Rel A was detected using a fluorescein-tagged goat anti-rabbit secondary antibody. As shown in Fig. 8, TNF␣ dramatically increased the nuclear staining of the two cell types, whereas little or no increase in nuclear fluorescence was observed in either HMEC-1 or A549 cells stimulated with H 2 O 2 . These data support the DNA binding studies demonstrating that TNF␣ but not H 2 O 2 activates NF-B in HMEC-1 and A549 cells. Thiol Oxidation Inhibits H 2 O 2 Induction of AP-1 Activity and IL-8 Expression-AP-1 and NF-B binding activity is mediated by a conserved redox-sensitive cysteine residue within the DNA-binding domain (36,37). Thiol oxidation agents such as diamide can inhibit AP-1 and NF-B binding activity (38). To determine whether thiol oxidation of AP-1 or NF-B could modulate IL-8 gene expression, we examined the effect of diamide on H 2 O 2 induction of AP-1 activity and IL-8 protein secretion. Nuclear protein extracts from A549 cells pretreated for 60 min with increasing concentrations of diamide followed by stimulation with 800 M H 2 O 2 for 60 min were assessed for AP-1 binding activity. As shown in Fig. 10, diamide dose dependently inhibited H 2 O 2 induced AP-1 binding activity to the IL-8 promoter with complete inhibition at 800 M diamide. These data demonstrate that diamide-mediated thiol oxidation inhibits H 2 O 2 induction of AP-1 binding to the IL-8 promoter in A549 cells.
To relate the diamide effect on AP-1 binding activity with IL-8 expression, we measured culture supernatants at 24 h for IL-8 protein secretion. As shown in Fig. 11, diamide dose dependently inhibited the H 2 O 2 induction of IL-8 protein secretion, consistent with the close association between AP-1 binding activity and IL-8 gene expression in A549 cells. However, in contrast to the effect of diamide on AP-1 binding activity, IL-8 secretion was inhibited by much lower concentrations of diamide, suggesting that thiol oxidation can also effect IL-8 expression independently of AP-1 binding activity. Moreover, the H 2 O 2 induced IL-8 secretion was more sensitive to diamide than the spontaneous secretion of IL-8, suggesting that the constitutive and induced mechanisms of IL-8 expression may differ in A549 epithelial cells. Taken together, these data indicate that H 2 O 2 induction of IL-8 in A549 cells is mediated by  (13, 20). The up-regulation of ICAM-1 on the surface of endothelium is re-quired for the firm adhesion of rolling neutrophils and a chemotactic gradient of IL-8 is critical for the adherent neutrophils to migrate across the alveolar-capillary membrane during lung inflammation and injury (39). In this study, we investigated the mechanism by which IL-8 and ICAM-1 are differentially regulated by oxidant stress in epithelial and endothelial cells. As summarized in Table II, we demonstrate that H 2 O 2 differentially activates the redox-responsive transcription factors AP-1 and NF-B in epithelial and endothelial cells. H 2 O 2 selectively induced AP-1 in A549 cells, whereas TNF␣ induced both AP-1 and NF-B (Table II). Moreover, H 2 O 2 induction of AP-1 binding to the IL-8 promoter was closely associated with H 2 O 2 induction of IL-8 mRNA expression, suggesting that H 2 O 2 induces IL-8 in epithelial cells through the action of the transcription factor AP-1. In contrast, in HMEC-1, AP-1 binding activity on the IL-8 promoter was constitutive and H 2 O 2 did not stimulate an increase with this basal AP-1 binding activity. H 2 O 2 also did not stimulate NF-B binding activity in HMEC-1, whereas TNF␣ increased only NF-B binding activity (Table II). Consistent with the DNA binding studies, we recently showed that TNF␣ but not H 2 O 2 could induce IL-8 promoter activity in HMEC-1. 2 In contrast to H 2 O 2 , TNF␣ appears to induce IL-8 in epithelial and endothelial cells mostly through the activation of NF-B. As with AP-1, a redox mechanism appears to be involved since diamide, which is known to target a conserved cysteine residue in the DNA-binding domain of NF-B, inhibited TNF␣ induced IL-8 expression. 2 This conclusion is consistent with several studies demonstrating that TNF␣ activates IL-8 and ICAM-1 transcription through a cooperative interaction between NF-B and NF-IL-6 (C/EBP-␤) (17,19). Indeed, we found little or no binding activity on the IL-8 NF-IL-6 binding site, 2 consistent with the binding requirement of NF-B for NF-IL-6 to also bind to the IL-8 promoter (17). In agreement with our findings, several studies have shown that H 2 O 2 also does not activate NF-B in primary endothelial cells (13,40,41). However, H 2 O 2 has been reported to activate NF-B in porcine aortic endothelial cells and transformed endothelial cell lines (40,42,43), indicating H 2 O 2 activation of NF-B is complex and may depend on the endothelial cell type. In contrast to endothelial cells, in A549 epithelial cells, TNF␣ induced, in addition to NF-B, AP-1 binding activity on the IL-8 promoter, suggesting AP-1 may contribute to the TNF␣ induction of IL-8 in epithelial cells. Indeed, the duel binding of AP-1 and NF-B to the IL-8 promoter may account for the greater induction of IL-8 expression by TNF␣ than H 2 O 2 in A549 cells.
Oxidant stress has been reported to affect AP-1 and NF-B differently, suggesting distinct mechanisms of redox regulation of these transcription factors (44,45). Our studies suggest that H 2 O 2 and TNF␣ regulate IL-8 and ICAM-1 gene expression 2 V. Lakshminarayanan and K. A. Roebuck, unpublished data.   (46). They found that thiols induced NF-B but not AP-1, while oxidants induced AP-1 but not NF-B. Oxidants such as ozone have also been shown to activate IL-8 expression through AP-1 and NF-B in A549 cells (47). AP-1-like proteins have also been associated with the transcriptional regulation of ␥-glutamylcysteine synthetase-heavy subunit by oxidants in A549 cells (48). The differential activation of AP-1 and NF-B has also been shown in other cell systems (44, 45, 49 -53). We conclude that the redox regulation of AP-1 and NF-B contribute to the distinct patterns of gene expression induced in epithelial and endothelial cells by H 2 O 2 and TNF␣. A similar finding has been reported for the manganese superoxide dismutase gene in pulmonary epithelial cells (54 (62).
The redox regulation of AP-1 and NF-B may also be involved in the differential expression of CC chemokines. We reported previously that H 2 O 2 induces MCP-1 but not RAN-TES in A549 cells, whereas TNF␣ induced both chemokines (20). MCP-1 has been reported to be regulated by redox mechanisms (63)(64)(65), involving NF-B and AP-1 (66 -70), and RAN-TES has recently been shown to be regulated by NF-B (71). We propose that the differential activation of redox-responsive transcription factors like AP-1 and NF-B set up distinct patterns of gene expression in epithelial and endothelial cells that may critically influence the site-specific recruitment of leukocyte subsets during inflammatory responses.
The differential activation of AP-1 and NF-B by H 2 O 2 and TNF␣ are consistent with our previous results demonstrating that H 2 O 2 and TNF␣ activate ICAM-1 gene expression through distinct cis-acting elements in the ICAM-1 promoter (13). H 2 O 2 activated ICAM-1 transcription through an element of the promoter that contain antioxidant responsive elements (ARE), whereas TNF␣ targeted the proximal promoter containing a composite binding site for NF-B and NF-IL-6 (13). It was previously demonstrated that TNF␣ activates ICAM-1 and IL-8 through the cooperative interaction of the NF-B and NF-IL-6-binding sites (7,(17)(18)(19). H 2 O 2 induces ICAM-1 expression in HMEC-1 (20), even though H 2 O 2 does not induce AP-1 or NF-B in these endothelial cells. In contrast, H 2 O 2 does not induce ICAM-1 in A549 cells even though AP-1 is activated. These data suggest that H 2 O 2 induction of ICAM-1 in endothelial cells is not mediated by AP-1 or NF-B, but more likely by a novel redox-responsive transcription factor.
In EAhy926, an epithelial/endothelial hybrid cell line generated from the fusion of A549 and human umbilical vein endothelial cells, H 2 O 2 induces both ICAM-1 and IL-8, indicating that fusion of endothelial and epithelial cells abrogates their ability to differentially regulate the two genes (13,20). The dominance of the H 2 O 2 gene induction suggests the involvement of positive trans-acting factors in the discordant oxidant regulation of ICAM-1 and IL-8 in epithelial and endothelial cells. ICAM-1 expression was induced through a H 2 O 2 responsive region of the promoter containing tandem 16-base pair AP-1/Ets composite sites (13). These AP-1/Ets repeats also have homology to known antioxidant responsive elements (ARE) first identified as redox-responsive elements in GST Ya subunit gene (72)(73)(74)(75)(76). AP-1 can physically interact with Ets at AP-1/Ets composite sites to functionally activate gene transcription in response to H 2 O 2 (77)(78)(79). H 2 O 2 increased binding activity on the ICAM-1 ARE sites and mutation of either the AP-1 or Ets motif abrogated the H 2 O 2 induced binding activity (13). However, in A549 cells, ICAM-1 expression is not stimulated by H 2 O 2 , whereas H 2 O 2 induced ICAM-1 in HMEC-1 (20). Since H 2 O 2 did not induce AP-1 or NF-B in HMEC-1, it appears that H 2 O 2 induction of ICAM-1 in endothelial cells is mediated by novel redox responsive transcription factors. AREbinding proteins capable of binding the ICAM-1 AP-1/Ets sites have recently been identified providing a potential mechanism by which H 2 O 2 could activate ICAM-1 in endothelial cells independently of AP-1 or NF-B (80 -82). Indeed, oxidant stress can differentially induce other transcription factors in epithelial and endothelial cells. For example, we previously demonstrated that H 2 O 2 differentially induces the winged-helix transcription factor, HFH-11 in A549 and HMEC-1 (83). H 2 O 2 increased HFH-11 expression in HMEC-1 but not in A549 cells (20).
In conclusion, we have demonstrated that H 2 O 2 and TNF␣ can differentially activate the redox-responsive transcription factors AP-1 and NF-B in epithelial and endothelial cells. We propose this differential regulation of AP-1 and NF-B leads to different patterns of gene expression in epithelial and endothelial cells that may be critical for their function during oxidant stress.