Differential Regulation of Interleukin-8 and Intercellular Adhesion Molecule-1 by H2O2 and Tumor Necrosis Factor-α in Endothelial and Epithelial Cells*

The reactive oxygen intermediate H2O2 can function as a signaling molecule to activate gene expression. In this study, we demonstrate that oxidant stress induced by tumor necrosis factor α (TNFα) or H2O2 differentially regulates intercellular adhesion molecule-1 (ICAM-1) and interleukin-8 (IL-8) gene expression in endothelial and epithelial cells. Northern blot analysis revealed that TNFα induced both ICAM-1 and IL-8 expression in either the A549 lung epithelial cell line or the human microvessel endothelial cell line (HMEC-1). In contrast, H2O2 selectively induced only ICAM-1 in HMEC-1 and only IL-8 in A549. This cell type-specific pattern of IL-8 expression was also observed in several other endothelial and epithelial cells. TNFα induced greater IL-8 gene expression as compared with H2O2, but the kinetics of induction were similar. The induction of epithelial IL-8 message was accompanied by a corresponding increase in functional IL-8 protein secretion as determined by a neutrophil motility assay. The increased neutrophil motility stimulated by conditioned media from H2O2- or TNFα-exposed A549 cells was completely inhibited by an anti-IL-8 antibody. TNFα and H2O2 also induced a differential pattern of CC chemokine expression in A549. While TNFα induced both RANTES and MCP-1, H2O2 induced only MCP-1. These data suggest that epithelial cells under oxidant stress contribute to the inflammatory cytokine network by selective production of IL-8, MCP-1, and RANTES, which may critically influence the site-specific recruitment of leukocyte subsets.


The reactive oxygen intermediate H 2 O 2 can function as a signaling molecule to activate gene expression. In this study, we demonstrate that oxidant stress induced by tumor necrosis factor ␣ (TNF␣) or H 2 O 2 differentially regulates intercellular adhesion molecule-1 (ICAM-1) and interleukin-8 (IL-8) gene expression in endothelial and epithelial cells. Northern blot analysis revealed that TNF␣ induced both ICAM-1 and IL-8 expression in either the A549 lung epithelial cell line or the human microvessel endothelial cell line (HMEC-1). In contrast, H 2 O 2 selectively induced only ICAM-1 in HMEC-1 and only IL-8 in A549. This cell type-specific pattern of IL-8 expression was also observed in several other endothelial and epithelial cells. TNF␣ induced greater IL-8 gene expression as compared with H 2 O 2 , but the kinetics of induction were similar. The induction of epithelial IL-8 message was accompanied by a corresponding increase
in functional IL-8 protein secretion as determined by a neutrophil motility assay. The increased neutrophil motility stimulated by conditioned media from H 2 O 2 -or TNF␣-exposed A549 cells was completely inhibited by an anti-IL-8 antibody. TNF␣ and H 2 O 2 also induced a differential pattern of CC chemokine expression in A549. While TNF␣ induced both RANTES and MCP-1, H 2 O 2 induced only MCP-1. These data suggest that epithelial cells under oxidant stress contribute to the inflammatory cytokine network by selective production of IL-8, MCP-1, and RANTES, which may critically influence the site-specific recruitment of leukocyte subsets.
The chemotactic factor, interleukin-8 (IL-8), 1 and the cell surface adhesion protein, intercellular adhesion molecule-1 (ICAM-1; CD54), are major inflammatory mediators implicated in neutrophil-mediated lung injury (1,2). IL-8, a member of the CXC chemokine family (3), is a potent activator and chemoattractant of neutrophils (4) and is secreted as a 72-or 77-amino acid protein by a wide variety of cell types including endothelial and epithelial cells (5,6). ICAM-1, a member of the immunoglobulin supergene family, is constitutively expressed at low levels on the cell surface of endothelial and epithelial cells, and its up-regulation is critical for the firm adhesion of neutrophils to the vascular endothelium prior to extravasation. Together, IL-8 and ICAM-1 coordinate the transcellular migration of neutrophils to sites of inflammation and injury (7). In this process, rolling neutrophils firmly adhere to activated endothelium expressing ICAM-1 and migrate across the endothelial barrier in response to a chemotactic gradient of IL-8 (8,9). The importance of IL-8 and ICAM-1 in immunopathology has been demonstrated by the ability of specific antibodies to reduce in vivo inflammatory reactions (10 -13).
Because of their involvement in neutrophil migration, ICAM-1 and IL-8 are regulated by many of the same inflammatory mediators. Both molecules are induced by the proinflammatory cytokines TNF␣, IL-1, IL-6, and interferon ␥ (6, 14 -16), and inhibited by the anti-inflammatory cytokine IL-10 (17,18). ICAM-1 and IL-8 are regulated primarily at the level of gene transcription (19 -22) and their promoter regions contain functional binding sites for the transcription factors NF-B, C/EBP, and AP-1 (23,24). TNF␣ activates the IL-8 and ICAM-1 genes through a cooperative interaction between NF-B and C/EBP binding to a composite enhancer element within the proximal promoter (16,(25)(26)(27). In contrast to TNF␣, we recently demonstrated that H 2 O 2 activates ICAM-1 gene transcription through a distal enhancer containing binding site repeats for the transcription factors AP-1 and Ets (22). The AP-1/Ets composite elements also contain embedded within the repeat a so-called antioxidant response element (ARE), originally identified in the glutathione S-transferase gene (28). Recently, Munoz et al. (29) demonstrated that up-regulation of ICAM-1 transcription by antioxidants involved AP-1 binding to the ICAM-1 promoter.
Like ICAM-1, IL-8 is induced by oxidant stress (30), and antioxidants have been shown to inhibit IL-8 expression (31,32). H 2 O 2 increases IL-8 expression in epithelial cell lines, fibroblasts, and whole blood (30), and hypoxia followed by reoxygenation increases IL-8 expression in mononuclear and endothelial cells (33,34) and in the lung and myocardium in vivo (35,36). Nitric oxide, a reactive nitrogen species, has also been shown to induce IL-8 (37). As with ICAM-1, the IL-8 promoter contains a functional AP-1 site (38), and a potential ARE (39). However, the mechanism by which oxidants induce IL-8 expression is unknown and the regulation of oxidantinduced IL-8 and ICAM-1 in epithelial and endothelial cells has not been investigated.
TNF␣ has also been shown to generate oxidant stress (40,41). Therefore to determine whether IL-8 and ICAM-1 are induced by a common oxidant stress-mediated mechanism, we compared the effects of oxidant stress generated by H 2 O 2 and TNF␣ on the expression of IL-8 and ICAM-1 in either a human microvascular endothelial cell line (HMEC-1) or a human lung type II epithelial cell line (A549). The results herein demonstrate that, although TNF␣ and H 2 O 2 both generate oxidant stress, they differentially regulate the expression of the ICAM-1 and IL-8 genes in endothelial and epithelial cells. We further show that TNF␣ and H 2 O 2 can differentially induce other chemokine genes. We propose that oxidant stress constitutes cell type-and gene type-specific activation signals in epithelial and endothelial cells that may critically influence the site-specific recruitment of leukocyte subsets in inflammatory reactions. Cell Culture, Treatments, and IL-8 Secretion-The A549 human lung adenocarcinoma cell line representative of distal respiratory epithelium was grown in F-12 media with 10% fetal calf serum, 1% penicillin/ streptomycin, and 1% gentamycin to 90% confluency in 24-well culture dishes. HMEC-1, a simian virus (SV)-40 transformed human dermal microvessel endothelial cell line (42), was cultured in MCDB 131 media (Life Technologies, Inc.) with 10% fetal calf serum, 1% penicillin/streptomycin, 1% gentamicin, 1 g/ml hydrocortisone, and 0.01 g/ml epidermal growth factor to 90% confluency in 24-well dishes. HMEC-1 retain the morphologic, phenotypic, and functional characteristics of normal human microvascular endothelial cells (43,44). To eliminate effects of different factors in the growth media, 24 h prior to agonist treatment cells were washed twice with 1 ϫ PBS and placed under serum-starved conditions (i.e. media in which the supplement was diluted 20-fold to give 0.5% serum). Before agonist treatment, cells were washed twice with 1 ϫ PBS and covered with serum-free, phenol red-free media for 3 h. H 2 O 2 and TNF␣ were diluted to final concentration in serum-free, phenol red-free media. 200-l supernatant samples were taken and stored at Ϫ70°C until analysis for IL-8 production by ELISA (R & D Systems). Results are expressed as the mean of three separate experiments. Following the removal of supernatants, the cells were tested for viability using trypan blue exclusion and 3-(4,5-dimethyl thiazol-2-yl)-2,5-diphenyl tetrazolium bromide assay as described by DeForge et al. (30). Viability was greater than 95%.

Materials
mRNA Expression Studies-Northern blotting was performed as described previously (17,22). All solutions used for the isolation of RNA were treated with diethyl pyrocarbonate (0.1%) and sterilized or prepared with diethyl pyrocarbonate-treated water. All glassware was autoclaved at 240°C to remove any traces of RNase. Cells were washed twice with ice-cold PBS. Total RNA was isolated with the Trizol (Life Technologies, Inc.) one step total RNA isolation method according to the manufacturer. Lysate was run through a 22-gauge needle and extracted with phenol/chloroform. Samples were mixed vigorously and incubated at room temperature for 2-3 min then spun at approximately 12,000 ϫ g for 15 min, and the aqueous phase collected. The RNA was precipitated with equal volumes of isopropyl alcohol at Ϫ70°C for 1 h. Following the precipitation, 20 g of RNA per treatment was resolved in a 1.25% formaldehyde agarose gel. All gels were analyzed for the amount and integrity of RNA by ethidium bromide staining to detect 18 S and 28 S ribosomal RNA bands. RNA was transferred to Duralon UV© (Stratagene) using Posiblot© (Stratagene), and RNA was UV crosslinked to the membrane by Stratalinker© (Stratagene). Blots were prehybridized with preheated (65°C) QuickHyb© Solution (Stratagene) and hybridized with 330 l of QuickHyb© solution/cm 2 of membrane containing radiolabeled cDNA probe overnight. The blot was washed twice with 2 ϫ SSC containing 0.1% SDS for 15 min at room temperature followed by washing twice with 0.1 ϫ SSC containing 1% SDS at 55°C for 30 min. The blot was covered in plastic wrap and exposed to radiographic film overnight. The blot was stripped by incubation at 95°C with 2 ϫ SSC containing 0.1% SDS five times for 10 min. Once stripped, the blot was rehybridized with GAPDH (glyceraldehyde-3phosphate dehydrogenase) probe. GAPDH cDNA, ICAM-1 cDNA, and IL-8 cDNA were labeled using Prime It II© (Stratagene) random priming kit with [␣-32 P]dCTP (3,000 Ci/mmol).
RNase protection assays were performed using an RNase protection assay kit purchased from Pharmagen (San Diego, CA). In brief, total RNA was isolated from stimulated A549 cells with Trizol (Life Technologies, Inc.). Multiprobe, hCK-5, which contains templates for the chemokines RANTES, MCP-1, and IL-8, and the housekeeping genes L-32 and GAPDH, was labeled with [␣-32 P]UTP using T7 RNA polymerase. 3 ϫ 10 6 cpm of labeled probe was hybridized to 5 g of total RNA for 16 h at 56°C. mRNA probe hybrids were treated with RNase mixture and phenol-choroform extracted. Protected hybrids were resolved on a 6% denaturing polyacrylamide sequencing gel and exposed to radiographic film overnight. Laser densitometry was performed using a Personal Densitometer SI (Molecular Dynamics, Inc. Sunnyvale, CA).
Neutrophil Motility Studies-The neutrophil motility assay was adapted from Marks et al. (45). In brief, whole blood from normal human subjects was drawn in sodium/heparin-coated vacutainer. Polymorphonuclear lymphocytes (PMN) were isolated using one-step Lymphocyte Separation Media, a Ficoll-hypaque solution with a density of 1.077 g/ml (Biowhittaker, Walkersville, MD) as per the manufacturer's instructions. Erythrocytes were removed from the PMN pellet with a 20-s hypotonic lysis treatment. The cell pellet was washed in PBS, resuspended in 2 M media (150 mM NaCl, 5 mM KCl, 1 mM MgCl, 1 mM CaCl 2 , 20 mM HEPES, 10 mM glucose, pH 7.4) and counted with a Coulter counter. Final cell preparation contained greater than 97% PMN. Microscope chamber slides were precoated with 10 g/ml vitronectin for 60 min. 40 ϫ 10 3 PMN were added to the vitronectincoated slides and allowed to adhere for 5 min at 37°C. Nonadherent PMN were removed by washing and 200 l of culture conditioned medium from H 2 O 2 -or TNF␣-stimulated epithelial or endothelial cells was added. A random visual field was selected on the microscope and the PMN (ϳ12-15) within this field were monitored for motility. Images were captured and saved for later analysis. Percent motility was determined as the percent of PMN in the field that moved one cell length within 3 min (45). In this assay, 50% of the PMN move in response to stimulation with the chemoattractant fMLP within this time frame (46).

H 2 O 2 Increases IL-8 but Not ICAM-1 mRNA Expression in
Epithelial Cells-To investigate the effects of oxidant stress on epithelial and endothelial cells, we utilized the model cell lines A549 and HMEC-1, respectively. A549 is a human type II epithelial cell line derived from a lung alveolar cell adenocarcinoma (47). HMEC-1 is a human dermal microvessel endothelial cell line immortalized by transfection of the SV-40 large T antigen (42). H 2 O 2 increases IL-8 release from A549 cells (30). We recently reported that H 2 O 2 stimulated ICAM-1 expression in human umbilical vein endothelial cells (HUVEC) (22). Thus, both endothelial and epithelial cells can mediate oxidant stress activation signals. To determine whether H 2 O 2 induction of IL-8 and ICAM-1 was mediated by a common mechanism, we compared the induction of the IL-8 and ICAM-1 genes in the two cell types. We examined first the effect of H 2 O 2 on epithelial IL-8 and ICAM-1 expression by Northern analysis. Total RNA was isolated from A549 cells exposed for 3 h to H 2 O 2 (100 -800 M). As shown in Fig. 1 1 with lanes 4 and 5). In contrast, steady-state ICAM-1 mRNA, although induced by TNF␣ (see Fig. 2), was only weakly induced by 400 M H 2 O 2 . This is considerably less than the 2.5-3-fold induction we reported for ICAM-1 in human umbilical vein endothelial cells (22). We previously reported that actinomycin D, a potent transcriptional inhibitor, prevented H 2 O 2 induction of endothelial ICAM-1 gene expression (22). Similarly, actinomycin D prevented the H 2 O 2 induction of IL-8 (Fig. 1, lane 6 1, 4, and 7). Total RNA was isolated and analyzed by Northern blot. A, autoradiogram; B, bar graph of densitometry of bands from IL-8 and ICAM-1 normalized to GAPDH signal and expressed as fold increase over spontaneous mRNA expression (lanes 3, 6, and 9). Data are representative of three separate experiments.
were treated with H 2 O 2 (800 M) or TNF␣ (100 units/ml) and the medium was monitored over time for IL-8 accumulation by ELISA. As shown in Fig. 3, H 2 O 2 and TNF␣ markedly increased IL-8 secretion over that spontaneously released from unstimulated control cells. The kinetics of H 2 O 2 (Fig. 3A) and TNF␣ (Fig. 3B) (Fig. 4, A and C). As in A549 cells, TNF␣ induced both ICAM-1 and IL-8 in HMEC-1 (Fig. 4, B Fig. 5, H 2 O 2 increased IL-8 secretion by less than 50% over that spontaneously secreted by HMEC-1 even at the highest concentrations of H 2 O 2 examined (400 M). In contrast, TNF␣ increased IL-8 secretion more than 3-fold. These data demonstrate a close association between IL-8 gene induction and IL-8 protein secretion in HMEC-1.

Cells-To determine the effect of H 2 O 2 and TNF␣ on endothelial IL-8 protein secretion, we monitored IL-8 release into the medium 24 h after treatment with increasing concentrations of H 2 O 2 . As shown in
Because A549 and HMEC-1 are grown in very different media, it is conceivable that the different components in their media might influence the pattern of gene expression in the two cell types. For example, in contrast to A549 cells, HMEC-1 are grown in the presence of hydrocortisone and epidermal growth factor. To determine whether these agents affect the induction of IL-8, A549 cells were cultured on the same growth media as the HMEC-1 cells for 72 h prior to stimulation under serum-free conditions. Analysis of the medium 24 h after stimulation by TNF␣, showed in three independent experiments no statistical difference in IL-8 induction (data not shown).

H 2 O 2 and TNF␣ Differentially Induce IL-8 Secretion in Endothelial and Epithelial
Cells-To further demonstrate that the cell type-specific induction of IL-8 is not the result of factors in the growth media but a general property of epithelial and endothelial cells, we examined the induction of IL-8 in several other epithelial and endothelial cell types. As shown in Table I, H 2 O 2 induced IL-8 secretion in BEAS-2B, a bronchial epithelial cell line, but not in HUVEC or HPAEC (human pulmonary artery endothelial cell), two primary endothelial cell types. In contrast, TNF␣ induced IL-8 secretion in both the epithelial and endothelial cell types. Thus, the differential pattern of IL-8 expression appears to be a general feature of these cell types.
We previously demonstrated that the ICAM-1 promoter is induced by H 2 O 2 in EAhy926 cells (22), a epithelial/endothelial hybrid cell line generated from fusion of the A549 cell line with HUVEC (48). As shown in Table I

IL-8 Secretion Induced by H 2 O 2 or TNF␣ Increases Neutrophil Motility-
To determine whether the differential pattern of IL-8 secretion induced by H 2 O 2 or TNF␣ was functionally significant, we compared the effects of conditioned media of H 2 O 2or TNF␣-stimulated cells on neutrophil motility using a microvideo imaging system. HMEC-1 and A549 cells were stimulated with H 2 O 2 or TNF␣ for 24 h. Supernatants from the stimulated cells were applied to adherent human neutrophils for 3 min and the effects recorded. As shown in Fig. 6, conditioned media from A549 cells exposed to H 2 O 2 for 24 h induced distinct shape changes in the neutrophils, suggesting that a soluble factor secreted from the A549 cells can activate the neutrophils.
To determine whether this soluble factor secreted from A549 cells could increase neutrophil motility, the number of neutrophils displaced one cell width within 3 min was determined in two separate visual fields in three separate experiments. As shown in Fig. 7, conditioned media from H 2 O 2 -treated A549 cells (Fig. 7A) stimulated neutrophil motility approximately 3-fold. In contrast, conditioned media from H 2 O 2 -treated HMEC-1 was unable to increase neutrophil motility over control supernatants from unstimulated cells (Fig. 7B). Conditioned media from TNF␣-stimulated HMEC-1 and A549 cells both stimulated neutrophil motility. These data suggest that H 2 O 2 selectively induces the release of a chemotactic factor from A549 cells that activates neutrophils and stimulates their motility.
To determine whether the chemotactic factor released by A549 cells was IL-8, the conditioned media was depleted of IL-8 using immobilized anti-IL-8 antibody. As shown in Fig. 7A, depletion of IL-8 from A549 cell-conditioned media completely abrogated the neutrophil migration induced by H 2 O 2 or TNF␣. These data demonstrate that the increased neutrophil motility was functionally mediated by secreted epithelial IL-8.

TNF␣ and H 2 O 2 Induce Different Patterns of Chemokine Expression in A549 Cells-Epithelial cells secrete both CC and
CXC chemokines in response to proinflammatory cytokines. To determine whether TNF␣ and H 2 O 2 differentially induce other chemokines in A549 cells, we examined the gene induction of the CC chemokines RANTES and MCP-1. RANTES is a chemoattractant for eosinophils and MCP-1 is a chemoattractant for monocytes and memory T-cells (49). Total RNA was isolated from A549 cells stimulated for 3 h with TNF␣ (100 units/ml) or H 2 O 2 (800 M) and analyzed by RNase protection. As shown in Fig. 8

DISCUSSION
Chemokines and cell adhesion molecules are critical protein factors in the recruitment of leukocytes to sites of inflammation and injury, and oxidant stress is an important regulator of their expression (30,50). In this study, we demonstrate that ICAM-1 and IL-8 orchestrate the transendothelial migration of neutrophils to sites of inflammation and injury (7). The up-regulation of ICAM-1 on the surface of the endothelium is required for the firm adhesion of rolling neutrophils (1) 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 (8,9). H 2 O 2 produced at sites of inflammation could constitute cell type-and gene type-specific activation signals capable of inducing different patterns of ICAM-1 and IL-8 expression in pulmonary cells of the alveolarcapillary membrane. Bradley et al. (51) reported that H 2 O 2 can also interfere with cytokine signaling through the impairment of their cell surface receptors. The oxidant regulation of the ICAM-1 and IL-8 genes may have evolved to promote their discordant expression during the inflammatory response.
In addition to the differential regulation of the CXC chemokine IL-8, we found that TNF␣ and H 2 O 2 can also selectively regulate CC chemokine expression. In A549 epithelial cells, TNF␣ induced both RANTES and MCP-1, whereas H 2 O 2 selectively induced only MCP-1. In contrast to IL-8, H 2 O 2 induced MCP-1 in HUVEC (52). The differential induction of RANTES, MCP-1, and IL-8 in epithelial and endothelial cells by H 2 O 2 suggests that oxidant stress can regulate the expression of both the CC and CXC chemokines. Although we were unable to detect any significant induction of IL-8 expression in any of the endothelial cells we examined, it has been reported that oxidant stress generated by hypoxia followed by reoxygenation can induce IL-8 expression in human endothelial cells (34) and  With regard to IL-8 functional activity, the cell type-specific IL-8 mRNA expression and protein secretion was associated with increased neutrophil motility, demonstrating that oxidant activation of the IL-8 gene leads to the secretion of biologically active IL-8. These data suggest oxidant-exposed epithelial cells can contribute to the inflammatory cytokine network through the selective production of CC and CXC chemokines such as IL-8, RANTES, and MCP-1, which could critically influence the recruitment of leukocyte subsets to sites of inflammation and injury. Indeed, the differential expression of chemokines in epithelial and endothelial cells may contribute to the immuno-  pathophysiology of several inflammatory diseases. Differential expression of GRO␣, ENA-78, and IL-8 has been reported in a model of psoriasis (54). TNF␣ has also been shown to differentially regulate RANTES and IL-8 in rheumatoid synovial fibroblasts (55). Respiratory syncytial virus also selectively induces RANTES and IL-8 in upper airway epithelial cells (56,57).
In addition to their cell type-and gene type-specific regulation, chemokines are also regulated in a stimulus-dependent manner. Transforming growth factor ␤ 1 induces IL-8 expression in epithelial cells (58), but inhibits IL-8 expression in endothelium (59). TNF␣ induces RANTES and MCP-1 in human corneal keratocytes but not in corneal epithelial cells (60). TNF␣ induces IL-8 and MCP-1, but not RANTES in HMEC-1 (61). In contrast, we found TNF␣ induces all three chemokines in A549 cells, while H 2 O 2 induces IL-8 and MCP-1, but not RANTES. This differential expression of chemokines suggests TNF␣ and H 2 O 2 activate distinct signaling pathways in epithelial and endothelial cells. TNF␣ and H 2 O 2 also activate T-cells through distinct signaling pathways that are thought to converge to activate NF-B (62). Consistent with H 2 O 2 and TNF␣ activation signals stimulating different second messenger pathways, we found that the anti-inflammatory cytokine IL-10 can differentially inhibit H 2 O 2 and TNF␣ induction of IL-8. 2 Thus, in A549 cells IL-10 inhibited H 2 O 2 but not TNF␣ induction of IL-8.
With regard to the transcriptional mechanism, oxidant stress has been shown to modulate the DNA binding activities of the transcription factors AP-1 and NF-B (63)(64)(65). Both the IL-8 and ICAM-1 promoters contain binding sites for AP-1 and NF-B. Moreover, these transcription factors have been demonstrated to be involved in IL-8 and ICAM-1 expression (29, 66 -69). NF-B is critical for the TNF␣ response of both ICAM-1 and IL-8. The TNF␣ response is mediated by the cooperative binding of NF-B and C/EBP to adjacent binding sites in the proximal regions of the IL-8 and ICAM-1 promoters (16,(25)(26)(27). This cooperative binding mechanism appears to function in both epithelial and endothelial cells, since TNF␣ activated the expression of IL-8 and ICAM-1 in both cell types, although the TNF␣ induction of IL-8 was substantially greater in A549 cells than in any of the endothelial cells examined. We have recently shown that the antioxidant pyrrolidine dithiocarbamate, a potent inhibitor of NF-B, can abrogate the H 2 O 2 induction of IL-8 in A549 cells. 2 In contrast, pyrrolidine dithiocarbamate itself induces ICAM-1 expression (29) and does not inhibit H 2 O 2 induction of ICAM-1 in endothelial cells (22), suggesting a role for NF-B in the cell type-specific induction of these genes. Das et al. (70) have demonstrated that AP-1 and NF-B are differentially regulated by oxidant and antioxidants in A549 cells, providing a potential mechanism by which genes containing AP-1 and NF-B-binding sites could be selectively activated by oxidant stress. They found that thiols induced NF-B but not AP-1 in A549 cells, while oxidants like H 2 O 2 induced AP-1 but not NF-B. Differential activation of AP-1 and NF-B has also been shown in other systems (63,71,72). The differential activation of these transcription factors has been shown to be regulated by the intracellular levels of glutathione disulfide (GSSG) (73). In this regard it is interesting to note that the cellular glutathione (GSH) levels are markedly higher in A549 epithelial cells than in fibroblasts (74). Indeed, high levels of GSH could account for the higher concentrations of H 2 O 2 (100 -800 M) required to activate IL-8 gene expression in A549 cells compared with the low concentrations of H 2 O 2 (10 -400 M) needed to activate ICAM-1 in endothelial cells. Thus, different GSSG/GSH ratios in epithelial and endothelial cells could lead to cell type-specific gene expression via differential activation of redox-sensitive transcription factors like AP-1 and NF-B.
With regard to the oxidant regulation of ICAM-1 expression, we showed that the NF-B-and C/EBP-binding sites are not sufficient to mediate the H 2 O 2 response (22). This is in agreement with the present study suggesting TNF␣ and H 2 O 2 activate gene expression through distinct mechanisms. The H 2 O 2 response is mediated by a region of the ICAM-1 promoter containing two 16-base pair direct repeats, binding sites for the transcription factors AP-1 and Ets (22). AP-1/Ets composite elements have been found in promoters of the macrophage scavenger receptor gene (75) as well as the mouse glutathione S-transferase Ya subunit gene (76). The macrophage scavenger receptor AP-1/Ets composite element is nearly identical to the ICAM-1 AP-1/Ets elements and can mediate H 2 O 2 activation signals (22). The AP-1/Ets element from the glutathione Stransferase promoter can also mediate an H 2 O 2 response (76). The AP-1/Ets element is also similar to the ARE found in several anti-oxidant response genes (65,(77)(78)(79). Indeed, the ARE binds AP-1 and has been shown to mediate H 2 O 2 transcriptional responses (80,81). In contrast to ICAM-1, the IL-8 promoter does not appear to contain an AP-1/Ets composite element. The IL-8 promoter, however, does possess separate consensus AP-1 and ARE sites. Although the role these elements play in mediating oxidant stress induction of the IL-8 gene is presently not known, we have recently shown that binding activity on the IL-8 ARE is induced by H 2 O 2 . 2 Thus, sequence and positional differences between the ICAM-1 and IL-8 ARE sites may contribute to the cell type-specific induction of these genes.
In addition to AP-1 and NF-B, H 2 O 2 may induce cell typespecific transcription factors, which in turn could induce the different patterns of IL-8 and ICAM-1 expression in epithelial and endothelial cells. Recently, we demonstrated that oxidant stress induced by H 2 O 2 also activates HFH-11, a winged helix transcription factor that is expressed in embryonic epithelial cells and whose expression is reactivated in adult cells by proliferative signals (82). As with ICAM-1, HFH-11 is induced by H 2 O 2 in HMEC-1 (82), but not A549 cells. 3 However, since the kinetics of H 2 O 2 induction of HFH-11 was identical to that of ICAM-1, we do not believe HFH-11 mediates the H 2 O 2 induction of ICAM-1.
In summary, H 2 O 2 and TNF␣ can induce different patterns of chemokine gene expression in endothelial and epithelial cells. We believe this differential expression induced by oxidant stress could critically influence the site-specific recruitment of leukocyte subsets during inflammatory reactions when epithelial and endothelial cells are under oxidant stress.