NADPH Oxidase NOX5-S Mediates Acid-induced Cyclooxygenase-2 Expression via Activation of NF-κBin Barrett's Esophageal Adenocarcinoma Cells*

We have shown that the NADPH oxidase NOX5-S may play an important role in the progression from Barrett's esophagus to esophageal adenocarcinoma (EA) by increasing cell proliferation and decreasing apoptosis. However, the mechanism of the acid-induced NOX5-S-mediated increase in cell proliferation is not known. We found that, in SEG1 EA cells, the acid-induced increase in prostaglandin E2 (PGE2) production was mediated by activation of cyclooxygenase-2 (COX2) but not by COX1. Acid treatment increased intracellular Ca2+, and a blockade of intracellular Ca2+ increase inhibited the acid-induced increase in COX2 expression and PGE2 production. Knockdown of NOX5-S or NF-κB1 p50 by their small interfering RNA significantly inhibited acid-induced COX2 expression and PGE2 production in SEG1 cells. Acid treatment significantly decreased IκBα and increased luciferase activity when SEG1 cells were transfected with an NF-κB in vivo activation reporter plasmid, pNF-κB-Luc. In a novel Barrett's cell line overexpressing NOX5-S, IκBα was significantly reduced, and luciferase activity increased when these Barrett's cells were transfected with pNF-κB-Luc. Overexpression of NOX5-S in Barrett's cells significantly increased H2O2 production, COX2 expression, PGE2 production, and thymidine incorporation. The increase in thymidine incorporation occurring in NOX5-S-overexpressing Barrett's cells or induced by acid treatment in SEG1 EA cells was significantly decreased by COX2 inhibitors or small interfering RNA. We conclude that acid-induced COX2 expression and PGE2 production depend on an increase in cytosolic Ca2+ and sequential activation of NOX5-S and NF-κB in SEG1 cells. COX2-derived PGE2 production may contribute to NOX5-S-mediated cell proliferation in SEG1 cells.

whereas NOX5-L is not detected in these cells. The expression of NOX5-S mRNA is significantly higher in these cells than in esophageal squamous epithelial cells. NOX5-S mRNA is also significantly higher in Barrett's tissues with high grade dysplasia than without dysplasia. We have also shown that acid-induced H 2 O 2 production is mediated by the NADPH oxidase NOX5-S and that acid-induced NOX5-S expression depends on an increase in intracellular calcium and activation of cyclic AMP response elementbinding protein (CREB) in SEG1 EA cells. Overproduction of ROS, derived from up-regulation of NOX5-S, increases cell proliferation and decreases apoptosis, possibly contributing to progression from intestinal metaplasia (Barrett's esophagus) to dysplasia and to adenocarcinoma.
COX2 (cyclooxygenase 2) may also play a role in the progression from BE to EA, since 1) COX2 overexpression has been demonstrated in both Barrett's metaplastic and adenocarcinoma cells (16); 2) COX2 expression increases significantly in ex vivo BE tissues pulsed with acid or bile salts, and this effect is attenuated by the selective COX2 inhibitor NS-398 (16); and 3) selective COX2 inhibitors significantly decrease the development of esophageal adenocarcinoma in a rat model of BE (17). The mechanisms mediating COX2-mediated tumorigenesis, however, are not fully understood. COX2 has been reported to play an important role in development of colorectal cancer (18), possibly by increasing cell migration and invasion (19) and decreasing apoptosis (20,21). In EA cells, selective COX2 inhibitors significantly decrease proliferation and increase apoptosis (22,23), suggesting that COX2-derived prostaglandin E2 may contribute to esophageal tumorigenesis, possibly by promoting cell proliferation and inhibiting apoptosis.
Whether acid-induced overexpression of NOX5-S increases cell proliferation through up-regulation of COX2 in Barrett's esophageal adenocarcinoma cells, however, is not known. In the present study, we show that acid-induced COX2 expression depends on an increase in intracellular calcium and sequential activation of NADPH oxidase NOX5-S and NF-B and that COX2 may contribute to a NOX5-S-mediated increase in cell proliferation in SEG1 cells.

EXPERIMENTAL PROCEDURES
Cell Culture and Acid Treatment-Human Barrett's adenocarcinoma cell lines SEG1 were derived from human esophageal Barrett's adenocarcinomas (24) and generously provided by Dr. David Beer. These cells are cultured in DMEM containing 10% fetal bovine serum and antibiotics.
Human Barrett's cell line was established and generously provided by Dr. S. J. Spechler and Dr. R. F. Souza. This cell line was derived from esophageal mucosal biopsies of patients with BE (intestinal metaplasia) and immortalized with telomerase as described previously (25). Cells were cultured in wells precoated with collagen IV (1 g/cm 2 ; BD Bioscience, Bedford, MA) and in Keratinocyte Medium-2 (Ca 2ϩ -free solution, Cambrex, Rockland, ME) supplemented with 1.8 mM CaCl 2 , 5% fetal bovine serum, 400 ng/ml hydrocortisone, 20 ng/ml epidermal growth factor, 0.1 nM cholera toxin, 20 g/ml adenine, 5 g/ml insulin, 70 g/ml bovine pituitary extract, and antibiotics. Fig.  1A shows Alcian blue staining-positive cells in cultured Bar-rett's cell line, indicating that this cell line contains mucussecreting cells (possibly Goblet cells). Mucin-2 and an intestine specific transcription factor CDX2 were also detectable ( Fig.  1B), confirming that these cells are intestinal metaplastic cells. Both cell lines were cultured at 37°C in a 5% CO 2 humidified atmosphere.
For acid treatment, SEG1 cells were exposed to acidic DMEM (pH 4.0), acidic DMEM (calcium-free and 1 mM EGTA) (pH 4.0), acidic medium plus NS-398 (10 Ϫ6 M), acidic medium plus valeryl salicylate (10 Ϫ5 M), or normal DMEM (control) for 1 h, washed, and cultured in fresh medium (pH 7.2, without phenol red) for an additional 24 h. For the NS-398 and valeryl salicylate group, NS-398 or valeryl salicylate was added to the culture medium in this additional 24-h culture. Finally, the culture medium and cells are collected for measurements. Acidic DMEM (pH 4.0, 250 l) was added to each well in a 12-well plate, and the final pH was about 4.9 after a 1-h incubation.
NOX5-S-overexpressing Stable Barrett's Cell Line-Barrett's cells were transfected with NOX5-S plasmid or pCMV-Tag5A plasmid by using Lipofectamine 2000. From the second day after the transfection, NOX5-S or empty vector-transfected cells were selected with 200 g/ml G418 for 4 weeks. These Barrett's cells were cultured as described above.
Mucosal Organ Culture-Endoscopic mucosal biopsies were obtained from patients with documented BE undergoing endoscopy for cancer surveillance. As clinically recommended, one biopsy was taken from each quadrant every 2 cm in the entire length of the Barrett's esophagus. All mucosal samples were divided in half using an aseptic technique. One half was used for histology and examined by a pathologist; the other was placed immediately in ice-cold culture medium and transported to the laboratory. BE mucosa confirmed to be intestinal metaplasia by pathological examination were used for the studies. The experimental protocols were approved by the Human Research Institutional Review Committee at Rhode Island Hospital.
Biopsies were cultured as described previously (16,26,27). Briefly, BE mucosal biopsy specimens were randomly assigned to acid, acid plus calcium-free, or control groups. The biopsy specimens were placed on a sterilized stainless wire mesh (Flynn & Enslow, Inc., San Francisco, CA) within a Falcon center-well organ culture dish (BD Biosciences) so that culture medium (0.9 ml) just covered the surface of the biopsy. Organ Magnification is ϫ40. B, Western blot analysis showed that mucin-2 and an intestinespecific transcription factor CDX2 were detectable in Barrett's cell line. These data confirmed that these cultured cells are intestinal metaplastic cells.
culture dishes were then placed on racks in the Modular Incubator Chamber (Billups-Rothenberg, Inc., Del Mar, CA), perfused with 95% oxygen and 5% carbon dioxide, and then cultured at 37°C. Organ culture was performed in RPMI 1640 supplemented with 10% fetal bovine serum, 5 g/ml insulin, CaCl 2 (1.377 mM), glutamine (2 mM), glucose (3.66 mg/ml), 500 units/ml streptomycin, and 250 units/ml penicillin. The final concentration of calcium in the medium was 1.8 mM. BE mucosal biopsy tissues were first equilibrated in culture for 2 h and then exposed to acidic medium (pH 4.0), acidic medium without calcium plus 1 mM EGTA, or control medium (pH 7.2) for 1 h. After washing twice, BE mucosa biopsies were cultured in fresh medium without phenol red (pH 7.2) for an additional 24 h. Finally the culture medium was collected for measurement of H 2 O 2 , and the levels of H 2 O 2 were normalized for protein content.
Small Interfering RNA (siRNA) Transfection-24 h before transfection at 40 -50% confluence, cells were trypsinized and diluted 1:5 with fresh medium without antibiotics (1-3 ϫ 10 5 cells/ml) and transferred to 12-well plates (1 ml/well). Transfection of siRNAs was carried out with Lipofectamine 2000 (Invitrogen) according to the manufacturer's instruction. Per well, 75 pmol of siRNA duplex of NOX5, COX2, NF-B1 p50, or scrambled siRNA formulated into liposomes were applied; the final volume was 1.2 ml/well. After a 4-h transfection, the transfection medium was replaced with regular medium. 12 h (NOX5 siRNA) or 48 h (p50 siRNA) later, cells were exposed to acidic medium (pH 4, 1 h), washed, and cultured in fresh medium (pH 7.2, without phenol red) for an additional 24 h. For COX2 siRNA, cells were cultured without acid treatment for 48 h after transfection. Finally, the culture medium and cells were collected for measurements. Transfection efficiencies were determined by fluorescence microscopy after transfection of Block-it fluorescent oligonucleotide (Invitrogen) and were about 90% at 48 h.
Reverse Transcription-PCR-Total RNA was extracted by TRIzol reagent (Invitrogen) for the cultured cells and extracted by the RNAqueous kit (Ambion Inc., Austin, TX) for the biopsy tissues according to the protocols of the manufacturers. 1.5 g of total RNAs from cultured cells or 0.5 g of total RNA from tissues was reversely transcribed by using a SUPERSCRIPT TM kit first strand synthesis system for reverse transcription-PCR (Invitrogen) or a Sensiscript RT kit (Qiagen, Valencia, CA), respectively.
Luciferase Assay-24 h before transfection, NOX5-S-overexpressing Barrett's cells, control Barrett's cells transfected with pCMV-Tag5A, or SEG1 cells were seeded in 24-well plates. Each well of cells was transfected with Lipofectamine 2000 (Invitrogen) and 0.2 g of either pGL3-Basic (as no promoter control) or an NF-B in vivo activation reporter plasmid pNF-B-Luc, which contains five repeats of NF-B binding element GGGGACTTTCC in the enhancer element of the plasmid. For acid treatment, SEG1 cells were treated with acidic medium (pH 4, 1 h) 12 h after transfection and then cultured for an additional 24 h.
Luciferase activity was assayed 24 h (Barrett's cells) or 36 h (SEG1 cells) after transfection. Cell extracts were prepared by lysing the cells with lysis buffer (Roche Applied Science). The lysate was centrifuged at 13,000 rpm for 10 min to pellet the cell debris. The protein concentration in the supernatants was determined. The luciferase activities in the cell lysates were measured using Luciferase assay substrate (Roche Applied Science) and normalized to protein content.
All reactions were performed in triplicate in a 25-l total volume containing a 1ϫ concentration of Brilliant SYBR Green QPCR Master Mix (Stratagene), a 100 nM concentration of each sense and antisense primer, 1 l of cDNA, and 30 nM reference dye. Reactions were carried out in a Stratagene Mx4000 multiplex quantitative PCR system for one cycle at 94°C for 5 min; 40 cycles at 94°C for 30 s, 59°C for 30 s, and 72°C for 30 s; one cycle at 94°C for 1 min; and one cycle at 55°C for 30 s. Fluorescence values of SYBR Green I dye, representing the amount of product amplified at that point in the reaction, were recorded in real time at both the annealing step and the extension step of each cycle. The Ct, defined as the point at which the fluorescence signal was statistically significant above background, was calculated for each amplicon in each experimental sample using Stratagene Mx4000 software. This value was then used to determine the relative amount of amplification in each sample by interpolating from the standard curve. The transcript level of each specific gene was normalized to GAPDH amplification.
Western Blot Analysis-Cells was lysed in Triton X lysis buffer containing 50 mM Tris⅐HCl (pH 7.5), 100 mM NaCl, 50 mM NaF, 5 mM EDTA, 1% (v/v) Triton X-100, 40 mM ␤-glycerol phosphate, 40 mM p-nitrophenylphosphate, 200 M sodium orthovanadate, 100 M phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, 1 g/ml pepstatin A, and 1 g/ml aprotinin. The suspension was centrifuged at 15,000 ϫ g for 5 min, and the protein concentration in the supernatant was determined. Western blot was done as described previously (28). Briefly, after these supernatants were subjected to SDS-PAGE, the separated proteins were electrophoretically transferred to a nitrocellulose membrane at 30 V overnight. The nitrocellulose membranes were blocked in 5% nonfat dry milk and then incubated with appropriate primary antibodies followed by a 60-min incubation in horseradish peroxidase-conjugated secondary antibody (Amersham Biosciences). Detection was achieved with an enhanced chemiluminescence agent (Amersham Biosciences).
Cytosolic Calcium Measurements-SEG1 cells were loaded with 1.25 M Fura-2/AM for 40 min and placed in a 5-ml chamber mounted on the stage of an inverted microscope (Carl Zeiss). The cells were allowed to settle onto a coverslip at the bottom of the chamber. The bathing solution is the HEPES-buffered solution (pH 7.4) containing 112.5 mM NaCl, 3.1 mM KCl, 2.0 mM KH 2 PO 4 , 10.8 mM glucose, 24.0 mM HEPES (sodium salt), 1.9 mM CaCl 2 , 0.6 mM MgCl 2 , 0.3 mg/ml basal medium Eagle amino acid supplement, and 0.08 mg/ml soybean trypsin inhibitor. The Ca 2ϩfree medium is the HEPES-buffered solution without CaCl 2 but with 200 M 1,2-bis(2-aminophenox)ethane-N,N,NЈ,NЈ-tetraacetic acid (BAPTA), which completely blocked KCl-induced Ca 2ϩ influx (30). When the Ca 2ϩ -free medium was used, the bathing solution was changed twice with Ca 2ϩ -free medium after the cells had settled to the bottom of the chamber. When thapsigargin or cyclopiazonic acid was used, cells were preincubated with thapsigargin (3 M) or cyclopiazonic acid (1 M) for 1 h and then utilized for acid treatment.
Ca 2ϩ measurements were obtained using a modified dual excitation wavelength imaging system (Ion-Optix Corp. Milton, MA) as described previously (31). Ratiometric images were masked in the region outside the borders of the cell, since low photon counts give unreliable ratios near the edges. We developed a method for generating an adaptive mask that follows the borders of the cell as Ca 2ϩ changes. A pseudoisosbestic image (i.e. an image insensitive to Ca 2ϩ changes) was formed in computer memory from a weighted sum of the images generated by 340-nm excitation and 380-nm excitation. This image was then thresholded (i.e. values below a selected level were considered to be outside the cell and assigned a value of 0). For each ratiometric image, the outline of the cell was determined, and the generated mask was applied to the ratiometric image. This method allows the imaging of the changes in Ca 2ϩ . Our algorithm has been incorporated into the IonOptix software. This algorithm calculates the conversion of the ratios of fluorescence elicited by 340-nm excitation to 380-nm excitation to Ca 2ϩ concentrations using techniques previously described in detail by Grynkiewicz et al. (32).
[ 3 H]Thymidine Incorporation-Barrett's cells without (control) or with NOX5-S overexpression were cultured in the pres-FIGURE 2. Role of NOX5-S in COX2 expression in SEG1 EA cells. A typical example of Western blot analysis (A) and summarized data (B) showed that pulsed acid treatment significantly increased COX2 expression. This increase was significantly decreased by knockdown of NOX5-S with NOX5 siRNA. Knockdown of NOX5-S also significantly reduced the basal levels of COX2 protein (n ϭ 3). Transfection of siRNAs was carried out with Lipofectamine 2000. Per well, 75 pmol of siRNA duplex of NOX5 or scrambled siRNA formulated into liposomes were applied. After a 4-h transfection, the transfection medium was replaced with regular medium. 12 h later, cells were exposed to acidic medium (pH 4, 1 h), washed, and cultured in fresh medium (pH 7.2) for an additional 24 h. Finally, the culture medium and cells were collected for measurements. A typical example of Western blot analysis (C) and summarized data (D) showed that knockdown of NOX5-S had no effect on COX1 expression (n ϭ 3). E, pulsed acid treatment significantly increased PGE 2 production. Knockdown of NOX5-S remarkably reduced PGE 2 production at the basal condition as well as in response to acid treatment in SEG1 cells (n ϭ 5). The data suggest that NOX5-S may contribute to acid-induced COX2 expression and PGE 2 production in SEG1 EA cells. F, In SEG1 esophageal adenocarcinoma cells, the acid-induced increase in PGE 2 production was inhibited by the COX2 inhibitor NS-398 (10 Ϫ6 M) but not by the COX1 inhibitor valeryl salicylate (10 Ϫ5 M) (n ϭ 3), suggesting that acidinduced PGE 2 production is mediated by activation of COX2 but not of COX1. Differences between different groups were tested using ANOVA. #, p Ͻ 0.05; **, p Ͻ 0.01; ****, p Ͻ 0.0001, compared with pH 7.2 or pH 7.2 ϩ scrambled siRNA group. *, p Ͻ 0.05; ***, p Ͻ 0.001; ##, p Ͻ 0.0001, compared with pH 4 or pH 4 ϩ scrambled siRNA group. For acid treatment, SEG1 cells were cultured as described above and treated with acidic medium (pH 4.0) for 1 h at the confluence of 40 -50%. Having been washed three times, cells were cultured at pH 7.2 for 24 h in the presence or absence of NS-398 or valeryl salicylate. A part of culture medium was collected for PGE 2 measurement, and then cells were incubated with methyl-[ 3 H]thymidine (0.05 Ci/ml) for 4 h.

NOX5-S and COX2 in Esophageal Adenocarcinoma
After cells were washed three times with PBS to remove unincorporated radioactivity, cells were collected and homogenized with a lysis buffer containing (pH 7.4): 50 mM HEPES, 50 mM NaCl, 1% Triton X-100, 1% Nonidet P-40, 0.1 mM phenylmethylsulfonyl fluoride, and 1 mM dithiothreitol. Methyl-[ 3 H]thymidine uptake was measured in a scintillation counter, and the level of protein in the homogenates was also determined. The level of methyl-[ 3 H]thymidine incorporation was normalized to protein content. PGE 2 Measurement-PGE 2 in culture medium was quantified by using a PGE 2 enzyme immunoassay kit (Cayman Chemical Co., Ann Arbor, MI).
Amplex Red Hydrogen Peroxide Fluorescent Assay-Levels of H 2 O 2 in culture medium were measured by using the Amplex Red H 2 O 2 assay kit (Molecular Probes, Inc., Eugene, OR). This assay uses the Amplex Red reagent (10-acetyl-3,7dihydroxyphenoxazine) to detect H 2 O 2 . In the presence of peroxidase, the Amplex Red reagent reacts with H 2 O 2 in a 1:1 stoichiometry to produce the red fluorescent oxidation product, resorufin. Fluorescence is then measured with a fluorescence microplate reader using excitation at 540 nm and emission detection at 590 nm.
Protein Measurement-The amount of protein was determined by colorimetric analysis (Bio-Rad) according to the method of Bradford (33).
Statistical Analysis-Data are expressed as mean Ϯ S.E. Statistical differences between two groups were determined by Student's t test. Differences between multiple groups were tested using analysis of variance (ANOVA) and checked for significance using Fisher's protected least significant difference test.

Role of NOX5-S in Acid-induced COX2 Expression and PGE 2
Production-In SEG1 esophageal adenocarcinoma cells, acid treatment significantly increased COX2 (Fig. 2, A and B) but not COX1 expression (Fig. 2, C and D). In addition, acid treatment significantly increased PGE 2 production. This PGE 2 increase was inhibited by the COX2 inhibitor NS-398 but not by the COX1 inhibitor valeryl salicylate (Fig. 2F). The data suggest that acid-induced PGE 2 production is mediated by activation of COX2 but not by COX1.
We have previously shown that NOX5 siRNA effectively knock down NOX5 protein in SEG1 cells (27). Fig. 2 shows that knockdown of NOX5-S by transfection of SEG1 EA cells with NOX5 siRNA significantly decreased COX2 expression (Fig. 2,  A and B) and PGE 2 production (Fig. 2E) at basal condition as well as in response to acid treatment. Knockdown of NOX5-S, however, had no effect on COX1 expression (Fig. 2, C and D). The data suggest that NOX5-S may contribute to acid-induced COX2 expression and PGE 2 production in EA cells.
To confirm the role of NOX5-S in COX2 expression, we established a NOX5-S-overexpressing stable cell line in telomerase-immortalized Barrett's cells. As expected, transfection of NOX5-S plasmid significantly increased NOX5-S mRNA (Fig.  3C) and protein expression (Fig. 3, A and B) in these cells. In addition, overexpression of NOX5-S significantly increased H 2 O 2 production in Barrett's cells (Fig. 3D), suggesting that in these cells NOX5-S may be constitutively active or activated by mediators present in culture medium. In addition, in Barrett's cells, overexpression of NOX5-S significantly increased COX2 expression (Fig. 4, A and B) but did not affect COX1 expression (Fig. 4, C and D). Overexpression of NOX5-S significantly increased PGE 2 production, an increase that was blocked by NS-398 but not by valeryl salicylate (Fig. 4E). The data suggest that overexpression of NOX5-S induces COX2 expression and PGE 2 production.
Role of NF-B in Acid-induced COX2 Expression-Since NF-Bdependent COX2 expression has been indicated in other preparations (34 -36), we examined whether NF-B mediates acid-induced COX2 expression. Transfection of SEG1 EA cells with NF-B1 p50 siRNA significantly reduced the p50 protein expression (Fig. 5, A and  B) and significantly decreased COX2 expression and prostaglandin E 2 production at the basal condition as well as in response to acid treatment (Fig. 5, C-E), suggesting that acid-induced COX2 expression may depend upon activation of NF-B.
Pulsed acid treatment significantly decreased the expression of IB␣ (Fig. 6, A and B), supporting acid-induced activation of NF-B. To confirm this conclusion, we transfected SEG1 cells with NF-B cis-reporter plasmid pNF-B-Luc, which contains five repeats of NF-B binding element GGG-GACTTTCC in the enhancer element of the plasmid. Acid treatment significantly increased the luciferase activity (Fig. 6C), indicating that acid treatment induces activation of NF-B.
Since acid up-regulates the expression of NOX5-S, we examined whether NOX5-S activates NF-B. In the NOX5-S-overexpressing Barrett's cell line, the expression of IB␣ was significantly decreased (Fig. 7, A and B). When normal Barrett's cells were co-transfected with NOX5-S and pNF-B-Luc plasmids, the luciferase activity increased significantly (Fig. 7C), suggesting that NOX5-S activates NF-B.
Role of Ca 2ϩ in Acid-induced COX2 Expression and PGE 2 Production-We have shown that acid-induced NOX5-S expression depends on increased intracellular calcium (27). Since NOX5-S contributes to acid-induced COX2 expression, we examined whether an intracellular calcium increase mediates acid-induced COX2 expression.
As previously shown, acid treatment significantly increased intracellular Ca 2ϩ concentration in Fura-2/AM-loaded SEG1 cells, an increase that was blocked by removal of extracellular Ca 2ϩ (Fig. 8, A and B). Acid treatment increased intracellular Ca 2ϩ in a time-dependent manner (Fig. 8C), and depletion of intracellular Ca 2ϩ stores by the Ca 2ϩ -ATPase inhibitors thapsigargin (37) and cyclopiazonic acid (38) did not affect the acidinduced Ca 2ϩ increase (Fig. 8, C and D), suggesting that the showed that overexpression of NOX5-S significantly increased COX2 expression. A typical example of Western blot analysis (C) and summarized data (D) showed that overexpression of NOX5-S did not affect COX1 expression. E, overexpression of NOX5-S significantly increased PGE 2 production, an increase that was blocked by the COX2 inhibitor NS-398 but not by the COX1 inhibitor valeryl salicylate. The data suggest that overexpression of NOX5-S may induce COX2 expression and PGE 2 production. #, p Ͻ 0.02, paired t test, n ϭ 3. The differences between different groups in panel E were tested using ANOVA. *, p Ͻ 0.01, compared with control group where cells were transfected with pCMV-Tag5A plasmid; **, p Ͻ 0.01, compared with the NOX5-Soverexpressing group (NOX5S), n ϭ 3.
acid-induced Ca 2ϩ increase depends on Ca 2ϩ influx. Fig. 8, A and B, have been published (27) and are included here for readers' convenience.
In SEG1 cells, acid-induced COX2 expression and PGE 2 production were significantly decreased by removal of extracellular calcium (Fig. 8, E and F), suggesting that acid-induced COX2 expression and PGE 2 production may depend on a cytosolic Ca 2ϩ increase.
To test whether acid up-regulates COX2 expression in Barrett's esophageal mucosa, we cultured human BE mucosal biopsies in an oxygen-enriched environment and exposed these biopsy tissues to acid (pH 4). Similarly, acid treatment significantly increased COX2 expression in BE mucosal biopsies. Acid-induced COX2 expression was significantly decreased by the removal of extracellular calcium (Fig. 9A), suggesting that calcium mediates acid-induced COX2 expression in BE mucosa. Similarly, PGE 2 levels in the culture medium significantly increased after acid exposure, when compared with control (Fig. 9B). This PGE 2 increase was blocked by the removal of extracellular calcium (Fig. 9B), suggesting that calcium is involved in acid-induced PGE 2 production in BE mucosa.
COX2 Mediates the Acid-induced Increase in Cell Proliferation-In SEG1 cells, pulsed acid treatment significantly increased thymidine incorporation. This increase in thymidine incorporation was significantly reduced by NS-398 but not by valeryl salicylate (Fig. 10A). In these experiments, NS-398 effectively blocked the acid-induced increase in PGE 2 production (Fig. 2F). The data suggest that COX2 may mediate the acid-induced increase in cell proliferation.
We have shown that pulsed acid treatment increases cell proliferation in SEG1 EA cells via activation of NADPH oxidase NOX5-S (27). Therefore, we examined whether COX2 contributes to NOX5-S-mediated increase in cell proliferation.
In Barrett's cells, overexpression of NOX5-S significantly increased the thymidine incorporation. This increase in thymidine incorporation was significantly decreased by NS-398 but not by valeryl salicylate (Fig. 10B). In the culture medium collected in these experiments, NS-398 has been demonstrated to abolish the increase of PGE 2 production induced by overexpression of NOX5-S (Fig. 4E). In addition, a very selective COX2 inhibitor CAY10404 (39,40) also significantly diminished the basal production of PGE 2 in the culture medium of Barrett's cells and almost abolished the increase of PGE 2 production induced by overexpression of NOX5-S (Fig. 11E). In these same cells as in Fig. 11E, CAY10404 also significantly reduced the thymidine incorporation at the basal condition and partially inhibited the increase of thymidine incorporation caused by the overexpression of NOX5-S (Fig. 11F). Similarly, knockdown of COX2 by COX2 siRNA, which effectively knocked down COX2 (Fig. 11, A and B), significantly reduced PGE 2 levels in the culture medium of Barrett's cells at the basal condition and almost blocked the  (B) showed that transfection of cells with NF-B1 p50 siRNA significantly reduced the p50 protein expression, indicating that p50 siRNA effectively knocks down p50 protein (n ϭ 3). A typical example of Western blot analysis (C) and summarized data (D) showed that knockdown of NF-B1 p50 significantly decreased COX2 expression at basal condition as well as in response to acid treatment (n ϭ 3). E, knockdown of NF-B1 p50 significantly decreased PGE 2 production at basal condition as well as in response to acid treatment (n ϭ 3). The data suggest that acid-induced COX2 expression may depend on activation of NF-B. Transfection of siRNAs was carried out with Lipofectamine 2000. Per well, 75 pmol of siRNA duplex of NF-B1 p50 or scrambled siRNA formulated into liposomes, were applied. After a 4-h transfection, the transfection medium was replaced with regular medium. 48 h later, cells were exposed to acidic medium (pH 4, 1 h), washed, and cultured in fresh medium (pH 7.2) for an additional 24 h. Finally, the culture medium and cells were collected for measurements. *, p Ͻ 0.05, t test n ϭ 3; The differences between multiple groups were tested using ANOVA. **, p Ͻ 0.02; ***, p Ͻ 0.0001, compared with pH 7.2 ϩ scrambled siRNA group; #, p Ͻ 0.0001, compared with pH 4 ϩ scrambled siRNA group.
increase of PGE 2 production in response to overexpression of NOX5-S (Fig. 11C). In these cells used in Fig. 11C, COX2 siRNA significantly lessened the thymidine incorporation at the basal condition as well as in response to the overexpression of NOX5-S (Fig. 11D). These data suggest that the NOX5-S-induced increase in cell proliferation at least in part depends on activation of COX2.

DISCUSSION
We have previously shown that NOX1 and NOX5-S are the major isoforms of NADPH oxidase in SEG1-esophageal adenocarcinoma cells. The expression of NOX5-S mRNA is significantly higher in these cells than in esophageal squamous epithelial cells. NOX5 mRNA is also significantly higher in Barrett's tissues with high grade dysplasia than without dysplasia, suggesting that NOX5-S may play an important role in the progression from BE to esophageal adenocarcinoma. Pulsed acid treatment significantly increases H 2 O 2 production in both SEG1-EA cells and BE mucosa. In SEG1 cells, acid treatment increases mRNA expression of NOX5-S but not NOX1, and knockdown of NOX5 by NOX5 siRNA abolishes acid-induced H 2 O 2 production, suggesting that NOX5-S mediates acid-induced H 2 O 2 production. In addition, acid treatment increases intracellular calcium and phosphorylation of CREB. Acid-induced NOX5-S expression and H 2 O 2 production are significantly inhibited by removal of extracellular calcium and by knockdown of CREB using CREB siRNA, suggesting that acidinduced NOX5-S expression depends on intracellular Ca 2ϩ increase and activation of CREB. Overproduction of ROS derived from up-regulation of NOX5-S increases cell proliferation and decreases apoptosis. In this study, we examined whether NADPH oxidase NOX5-S up-regulates COX2 expression, thereby causing increase in cell proliferation in SEG1 EA cells.
Role of NADPH Oxidase NOX5-S in Acid-induced COX2 Expression and PGE 2 Production-We have shown that acid treatment increases NOX5-S expression and H 2 O 2 production in SEG1 cells (27). H 2 O 2 is known to induce COX2 expression in other preparations (41,42). Therefore, we examined whether NOX5-S contributes to acid-induced COX2 expression.
In SEG1 EA cells, acid-induced PGE 2 production may depend on activation of COX2 but not of COX1, since 1) acidinduced PGE 2 production was inhibited by the COX2 inhibitor NS-398 (39, 43) but not by the COX1 inhibitor valeryl salicylate (44) (Fig. 2F), and 2) acid treatment significantly increased COX2 expression (Fig. 2, A and B) but not COX1 expression (Fig. 2, C and D). Our data are consistent with a previous report (45).
In SEG1 EA cells, knockdown of NOX5-S by NOX5 siRNA significantly decreased acid-induced COX2 expression and PGE 2 production (Fig. 2, A, B, and E). However, knockdown of NOX5-S did not affect COX1 expression (Fig. 2, C and D). The data suggest that NOX5-S may mediate acid-induced COX2 expression and PGE 2 production in SEG1 EA cells. This result was further supported by our findings that overexpression of NOX5-S by transfection with NOX5-S plasmid significantly increased COX2 expression as well as PGE 2 production in Barrett's cells (Fig. 4, A, B, and E) but did not affect COX1 expression (Fig. 4, C and D) and that PGE 2 production induced by overexpression of NOX5-S was blocked by COX2 inhibitors NS-398 (Fig. 4E), CAY10404 (Fig. 11E), and COX2 siRNA (Fig.  11C) but not by COX1 inhibitor valeryl salicylate (Fig. 4E). It is known that low doses of H 2 O 2 are mitogenic and promote cell proliferation, whereas intermediate doses result in growth arrest, such as replicative senescence, and high doses cause cell   JUNE 1, 2007 • VOLUME 282 • NUMBER 22 death via either apoptotic or necrotic mechanisms (46). We have shown that NOX5-S is overexpressed in SEG1 cells (27). To avoid the possibility that transfection with NOX5-S plasmid would produce too much H 2 O 2 , which might cause cell apoptosis in SEG1 cells, we transfected Barrett's cells with NOX5-S plasmid.

NOX5-S and COX2 in Esophageal Adenocarcinoma
Overexpression of NOX5-S significantly increased H 2 O 2 production in Barrett's cells (Fig. 3D), suggesting that NOX5-S may be constitutively active or activated by mediators present in culture medium. However, functional regulation of NOX5-S by its subunits is not known and needs to be further explored.
Role of NF-B in Acid-induced COX2 Expression-In SEG1 EA cells, transcription factors responsible for acid-induced COX2 expression are not known. It has been reported that NF-B mediates COX2 expression induced by interleukin-1␤ and tumor necrosis factor (34 -36). In addition, NF-B binding sites have been identified in the promoter region of the COX2 gene (47). Therefore, we examined the role of NF-B in acid-induced COX2 expression.
NF-B is thought to be a family of Rel domain-containing proteins, including Rel A (also called p65), Rel B, c-Rel, NF-B1 (p105/p50), and NF-B2 (p100/p52). p105 and p100 are larger precursor proteins containing IB (an inhibitor of B)-like ankyrin repeat sequences in their carboxyl termini. Because of their IB-like ankyrin repeat sequences, these precursors are retained in the cytoplasm and require proteolytic processing to generate their mature DNA-binding proteins, p50 and p52, respectively (48). In the cytoplasm NF-B is in an inactive state, and its activity is regulated by at least two pathways. In the first pathway, a heterotrimer composed of p50, p65, and IB is degraded in a ubiquitin-dependent reaction, leading to the translocation of the p65-p50 dimers to the nucleus (48). In the second pathway, the dimers consisting of p100 and Rel B undergo proteolytic removal of the IB-like COOH-terminal domain of p100, allowing Rel B-p52 dimers to translocate to nucleus, where NF-B activates gene transcription (48).
We found that in SEG1 cells, knockdown of NF-B1 p50 by p50 siRNA significantly decreased COX2 expression and PGE 2 production at basal condition as well as in response to acid treatment (Fig. 5, C-E), suggesting that NF-B1 may be responsible for COX2 expression in SEG1 cells. This result is further supported by our findings indicating that NF-B is activated by acid treatment, since in SEG1 esophageal adenocarcinoma cells, pulsed acid treatment significantly decreased the expression of IB␣ (Fig. 6, A and B) and increased the luciferase activity (Fig. 6C) in SEG1 cells transfected with NF-B cis-reporter plasmid pNF-B-Luc, which contains five repeats of NF-B binding element GGGGACTTTCC in the enhancer element of the plasmid.  (27) and are included here for the reader's convenience. C, acid treatment increased intracellular Ca 2ϩ in a time-dependent manner, and that depletion of intracellular Ca 2ϩ stores by Ca 2ϩ -ATPase inhibitor thapsigargin (37) had no effect on acid-induced Ca 2ϩ increase (control n ϭ 20, thapsigargin n ϭ 17). D, depletion of intracellular Ca 2ϩ stores by Ca 2ϩ -ATPase inhibitor cyclopiazonic acid (CPA) (38) had no effect on acid-induced Ca 2ϩ increase (control n ϭ 20, cyclopiazonic acid n ϭ 10). The data suggest that Ca 2ϩ influx may mediate acid-induced Ca 2ϩ increase. E, in SEG1 cells, acid-induced COX2 mRNA expression was significantly decreased by removal of extracellular calcium (n ϭ 5). F, acidinduced PGE 2 production was significantly reduced by removal of extracellular calcium (n ϭ 5), suggesting that acid-induced COX2 expression and PGE 2 production may depend on intracellular Ca 2ϩ increase. Cultured cells were exposed to acidic medium (pH 4.0) for 1 h and then cultured at pH 7.2 for an additional 24 h. Finally, the culture medium and cells were collected for measurement. COX2 mRNA was measured by real time PCR. The differences between different groups were tested using ANOVA. *, p Ͻ 0.01, compared with the pH 7.4 or 7.2 group. **, p Ͻ 0.01; # p Ͻ 0.02; ##, p Ͻ 0.05, compared with the pH 4 group.
Our data show that COX2 expression and PGE 2 production in SEG1 cells were significantly decreased by knockdown of either NOX5-S or NF-B1 p50. It is known that NF-B may be activated by ROS (49). However, it is not known whether NOX5-S activates NF-B. We found that in NOX5-S-overexpressing Barrett's cell line, overexpression of NOX5-S significantly decreased the expression of IB␣ (Fig. 7, A and B) and increased the luciferase activity in Barrett's cells transfected with both NOX5-S and pNF-B-Luc plasmids (Fig. 7C), suggesting that NOX5-S may activate NF-B. This result is consistent with the literature showing that the p47 phox subunit of NADPH oxidases participates in the activation of Rel A in endothelial cells (50). Therefore, it is possible that acid-induced COX2 expression and PGE 2 production may depend on sequential activation of NOX5-S and NF-B.
Role of Ca 2ϩ in Acid-induced COX2 Expression and PGE 2 Production-We have shown that acid treatment significantly increased intracellular Ca 2ϩ concentration in Fura-2/ AM-loaded SEG1 cells, an increase that was blocked by removal of extracellular Ca 2ϩ (Fig.  8, A and B), suggesting that intracellular Ca 2ϩ increase may be due to Ca 2ϩ influx. This was further consolidated by the findings that depletion of intracellular Ca 2ϩ stores by Ca 2ϩ -ATPase inhibitors thapsigargin (37) and cyclopiazonic acid (38) had no effect on acid-induced Ca 2ϩ increase (Fig.  8, C and D). The mechanisms of acid-induced Ca 2ϩ influx are not known and need to be further explored. Since acid increases intracellular Ca 2ϩ , causing up-regulation of NADPH oxidase NOX5-S, we examined the role of calcium in acidinduced COX2 expression.
In SEG1 cells, acid-induced COX2 expression and PGE 2 production were significantly decreased by removal of extracellular calcium (Fig. 8, E and F), suggesting that acid-induced COX2 expression and PGE 2 production may depend on intracellular Ca 2ϩ increase. Similarly, acid treatment significantly increased COX2 mRNA expression and PGE2 production in cultured human BE mucosal biopsies, an increase that was blocked by removal of extracellular calcium (Fig. 9, A and B), indicating that SEG1 cells might be a suitable in vitro model to study acid-induced changes. These data further support the possibility that NADPH oxidase NOX5-S mediates acid-induced COX2 expression.
COX2 Mediates the Acid-induced Increase in Cell Proliferation-We have shown that NOX5-S mediates the acid-induced increase in cell proliferation and decrease in cell apoptosis in SEG1 cells (27). However, the mechanisms of NOX5-S-mediated increase in cell proliferation are not known.
In SEG1 cells, pulsed acid treatment significantly increased thymidine incorporation. This increase in thymidine incorporation was significantly reduced by NS-398 but not by valeryl salicylate (Fig. 10A). In these experiments, NS-398 effectively blocked the acid-induced increase in PGE 2 production (Fig. 2F). The data suggest that COX2 mediates the acid-induced increase in cell proliferation. This is consistent with the literature showing that selective COX2 inhibitors significantly decrease cell proliferation and increase apoptosis in EA cell lines in vitro (22,23) and that COX2 mediates the acid-induced increase in cell proliferation in SEG1 EA cells (45). In addition, we found that NOX5-S contributes to acid-induced COX2 expression (Fig.  2) via activation of NF-B (Fig. 5). Therefore, we examined A B FIGURE 9. A, in organ-cultured esophageal mucosal biopsies from patients with Barrett's esophagus, pulsed acid treatment significantly increased COX2 mRNA expression, an increase that was significantly reduced by removal of extracellular calcium, suggesting that acid-induced COX2 expression is mediated by calcium influx. B, in the culture medium of Barrett's mucosa, acid-induced PGE 2 production was blocked by removal of extracellular calcium, suggesting that calcium influx is required for production of PGE2. Cultured mucosal tissues were exposed to acidic medium (pH 4.0) for 1 h and then cultured at pH 7.2 for an additional 24 h. Finally, the culture medium and tissues were collected for measurement. ANOVA was used for statistical analysis. *, p Ͻ 0.01, compared with pH 7.2; **, p Ͻ 0.01, compared with the pH 4 group; n ϭ 3.
FIGURE 10. COX2 mediates the acid-induced increase in cell proliferation. A, in SEG1 cells, pulsed acid treatment significantly increased thymidine incorporation. This increase in thymidine incorporation was significantly reduced by NS-398 but not by valeryl salicylate, suggesting that COX2 may mediate the acid-induced increase in cell proliferation (n ϭ 3). B, in Barrett's cells, overexpression of NOX5-S significantly increased the thymidine incorporation. This increase in thymidine incorporation was significantly decreased by NS-398 but not by valeryl salicylate. The data suggest that the NOX5-S-induced increase in cell proliferation at least in part depends on activation of COX2 (n ϭ 5). ANOVA was used for statistical analysis. ***, p Ͻ 0.0001, compared with the control or pCMV-tag5 group; **, p Ͻ 0.0001, compared with the acid or NOX5S group.