INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

THE INDUCIBLE PLEIOTROPIC transcription factor nuclear factor-B (NF-B) is well recognized to play a pivotal role in the control of cellular inflammation. Now, a growing body of evidence points to an important role for NF-B in the growth of malignancies. Constitutive nuclear NF-B activation has been recently found important for proliferation of malignant melanoma (53), Hodgkin's disease (5), and carcinomas of the breast (56), ovary (12), colon (12), lung (7), head and neck (21), and pancreas (64). Repression of NF-B functionally interferes with normal and transformed cell proliferation (29, 32), and inhibition of NF-B by antisense strategies (28) or by overexpression of the NF-B inhibitor IB (5, 7, 21, 56) blocks tumor cell growth. Also, activation of NF-B has been linked with resistance of tumors to tumor necrosis factor- (TNF-)-induced apoptosis, anti-cancer chemotherapy, and radiation (9, 62, 63). Thus NF-B has emerged as a critically important regulator of transcription in neoplasms.

In unstimulated normal cells, NF-B resides in the cytoplasm as a dimeric protein complex bound to an inhibitor protein, designated IB (54). Agonist stimulation activates IB kinase, which phosphorylates serines-32 and -36 near the amino terminus of IB (3, 57), targeting the inhibitor for ubiquitination and proteolytic degradation by the 26S proteasome (46). The removal of IB unmasks the nuclear localization signal (10), allowing the NF-B complex to translocate to the nucleus, where it binds to its respective nucleotide sequence and transcriptionally regulates expression or repression of target genes. In most normal cell types, except for certain neurons (33) and mature B lymphocytes (39), nuclear NF-B activity is observed only transiently in response to inducers, including phorbol esters, viral transactivators, and cytokines, which might all act through a common mechanism involving the synthesis of reactive oxygen species as proximate messengers (52). In cultured melanoma cell lines, constitutive activation of NF-B has been causally related to enhanced degradation of IB (53) from elevated activity of IB kinase (19). Thus the constitutive activation of NF-B found in a number of tumors (5, 7, 12, 21, 53, 56, 64) may involve intermediate signaling events that are parallel to those occurring during activation of NF-B in normal cell lines (3, 57). Human tumor cells produce substantial amounts of reactive oxygen species spontaneously (18, 41, 60). Mitogenic signaling through both Ras (30) and Rac (31) is mediated by superoxide anion (O) production, and transfection with the superoxide-generating oxidase mox-1 transforms normal cells (58). Therefore, we determined whether reactive oxygen species might also mediate constitutive NF-B activation in malignant melanoma cell lines. Our findings suggest that endogenous redox stress is likely responsible for signaling NF-B activation in malignant melanoma cells and that scavengers of reactive oxygen species, especially of H₂O₂, are potent inhibitors of tumor cell proliferation. Moreover, the enzymatic source of redox stress promoting constitutive NF-B activation appears to be NAD(P)H:quinone oxidoreductase (NQO), which is overexpressed in many tumors (11) and can reduce membrane ubiquinone (35) for redox cycling with molecular oxygen to generate O at the plasma membrane, where stimulation of NF-B activation occurs.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Materials. Human malignant cell lines were obtained from American Type Culture Collection (Rockville, MD). RPMI medium 1640, Dulbecco's modified Eagle's medium, Leibovitz's L-15 medium, HEPES, antibiotic-antimycotic (10,000 units penicillin, 10,000 µg streptomycin, and 25 µg amphotericin B/ml), and trypsin-EDTA solution were purchased from the GIBCO BRL division of Life Technologies (Grand Island, NY). Fetal bovine serum (FBS) was purchased from HyClone Laboratories (Logan, UT). Rabbit polyclonal supershift antibodies for electrophoretic mobility shift assays (EMSAs) and for cyclin D1 were purchased from Santa Cruz Biotechnology (Santa Cruz, CA), and phosphospecific antibodies for immunoassay detection of the phosphorylated form of IB were purchased from New England Biolabs (Beverly, MA). Peroxidase-labeled donkey polyclonal anti-rabbit IgG was from Amersham Life Sciences (Buckinghamshire, England, UK). EMSA supplies, including DNA probes, were purchased from Promega (Madison, WI). Protease inhibitors were from Boehringer Mannheim (Indianapolis, IN). All other materials were purchased from Sigma Chemical (St. Louis, MO) unless specified.

Culture of malignant cell lines. Melanoma cell lines CRL 1585 and CRL 1619 were cultured in RPMI 1640 with 10% FBS and passed with nonenzymatic Cell Dissociation Solution (Sigma). LNCaP.FGC prostate adenocarcinoma cells were also cultured in RPMI 1640 with 10% FBS but passed with 0.05% trypsin and 0.53 mM EDTA. The adenosquamous lung carcinoma NCI-H596 cell line was grown in RPMI 1640 supplemented with 10% FBS, 10 mM HEPES, and 1.0 mM sodium pyruvate and passed with trypsin/EDTA. All of the above were grown in a 37°C humidified environment containing 5% CO₂-95% air. The breast carcinoma cell line MDA-MB-453 was grown in a 37°C humidified environment with free gas exchange with atmospheric air using Leibovitz's L-15 medium with 2 mM L-glutamine and 10% FBS and was passed with trypsin/EDTA.

Cell culture treatments. The effect of antioxidants on nuclear activation of NF-B was studied by incubating near-confluent (70%) cell cultures with antioxidant treatments for 1-48 h. Nuclear protein was harvested, and EMSAs were performed using DNA consensus-binding sequences. Nuclear translocation of the p65 component of NF-B was studied by immunoperoxidase staining, as outlined in Immunohistochemical localization of NF-B. The effect of antioxidants on expression of phosphorylated IB was studied by incubating near-confluent cell cultures with antioxidant treatments for 15 min to 48 h. Cells were lysed and protein expression levels were measured by immunoblot assay using a phosphospecific antibody for phosphorylated IB.

EMSAs. Nuclear protein was isolated and DNA binding reactions were performed as previously described in detail (34). Monolayers were washed twice in cold Dulbecco's phosphate buffered saline (DPBS) and equilibrated 10 min on ice with 0.7 ml cold cytoplasmic extraction buffer (CEB), consisting of (in mM) 10 Tris, pH 7.9, 60 KCl, 1 EDTA, 1 dithiothreitol (DTT), with protease inhibitors (PI; 1 mM Pefabloc, 50 µg/ml antipain, 1 µg/ml leupeptin, 1 µg/ml pepstatin, 40 µg/ml bestatin, 3 µg/ml E-64, and 100 µg/ml chymostatin). The detergent Nonidet P-40 (NP-40) was added to a final concentration of 0.1%, and cells were dislodged with a cell scraper. Nuclei were pelleted by centrifugation and washed with CEB/PI. Nuclei were then incubated for 20 min on ice in nuclear extraction buffer (NEB; 20 mM Tris, pH 8.0, 400 mM NaCl, 1.5 mM MgCl₂, 1.5 mM EDTA, 1 mM DTT, and 25% glycerol) with PI, spun briefly to clear debris, and stored at 80°C until performance of EMSAs. EMSAs were performed using the consensus binding oligonucleotides 5'-AGTTGAGGGGACTTTCCCAG-GC-3' and 3'-TCAACTCCCCTGAAAGGGTCCG-5' for the p50 component of NF-B (Promega, Madison, WI), end-labeled by phosphorylation with [-³²P]ATP and T4 polynucleotide kinase. DNA-protein binding reactions were performed with 2 µg of nuclear protein (as determined by the Pierce method) and 50-100,000 cpm of ³²P end-labeled double-stranded DNA probe in 10 mM Tris · HCl, pH 7.5, 50 mM NaCl, 0.5 mM EDTA, 0.5 mM DTT, 1 mM MgCl₂, 50 µg/ml poly(dI-dC), and 4% glycerol. All components of the binding reaction, with the exception of labeled probe, were combined and incubated at room temperature for 10 min before addition of labeled probe and incubated for an additional 20 min. Competition experiments were performed with 10× unlabeled wild-type oligonucleotide sequences for NF-B added before labeled probe. Supershift assays were performed by adding 1.0 µg of supershift-specific antibodies for p65, p50, p52, Rel B, or c-Rel components of NF-B and incubated at room temperature for 30 min or overnight at 4°C before adding the probe. Samples were electrophoresed on a 5% nondenaturing polyacrylamide gel in Tris-glycine-EDTA (120 mM glycine and 1 mM EDTA in 25 mM Tris, pH 8.5) buffer. Gels were dried and analyzed by autoradiography at 80°C using an image intensifier screen. Densitometry of bands was performed using Kodak Digital Science 1D image analysis software (Eastman Kodak, Rochester, NY).

Immunohistochemical localization of NF-B. Cells grown on sterile coverslips and treated with catalase (3,000 U/ml), apocynin (150 µg/ml), dicumarol (250 µM), or DPBS or DMSO vehicles for 24 h were fixed for 20 min on ice with 4% paraformaldehyde in DPBS with protease inhibitors, PI [10 µl/ml of Sigma protease inhibitor cocktail, containing 4-(2-aminoethyl)benzensulfonyl fluoride, pepstatin A, trans-epoxysuccinyl-L-leucylamindo-(guanidino)butane (E-64), bestatin, leupeptin, and aprotinin]. Cells were permeabilized by treating for 2 min with 0.1% NP-40 in DPBS/PI, washed once with cold DPBS, and fixed as before for 10 min. Coverslips were incubated in 3% hydrogen peroxide for 30 min to suppress any remaining peroxidase and were washed three times in cold DPBS. The permeabilized and fixed cells were blocked for 2 h with 2% BSA in DPBS on ice and incubated overnight at 4°C with 1 µg/ml of anti-p65 antibody (Santa Cruz) diluted in 0.1% BSA/DPBS. Unbound anti-p65 was washed away with 2% BSA/DPBS, and bound antibody was stained by incubation with biotinylated goat anti-rabbit immunoglobulin diluted 1:50 in 0.1% BSA/DPBS for 45 min on ice. Excess secondary antibody was washed away by three washes with 2% BSA/DPBS on ice. After the cells were washed, they were incubated with a streptavidin-biotin-peroxidase complex at room temperature for 1 h, washed again, and incubated in 0.03% wt/vol 3-3'-diaminobenzidine with 0.003% vol/vol hydrogen peroxide until a brown reaction product could be seen. Cells were then counterstained with eosin and mounted on glass slides before viewing under light microscopy.

Immunoassay for phosphorylated IB and cyclin D1. Cells were lysed, and proteins were isolated and quantitated by immunoassay as previously detailed (13). Cells were placed on ice, washed twice with cold DPBS, scraped into 0.5 ml boiling buffer [10% (vol/vol) glycerol and 2% (wt/vol) SDS in 83 mM Tris, pH 6.8] to which 50 mM DTT had been added as a reducing agent, and sheared by four passages through a pipette. Aliquots were removed for protein determination, using the bicinchoninic acid (BCA) protein assay (Pierce). After 10% -mercaptoethanol and 0.05% bromphenol blue were added, lysates were boiled for 5 min and stored at 80°C until immunoblotting was performed. Proteins in defrosted samples were separated by SDS-PAGE on 12% polyacrylamide gels (15 µg protein/lane) and electrotransferred to 0.45 µm Hybond ECL nitrocellulose membranes (Amersham Life Sciences) using the wet transblot method in transfer buffer [0.025 M Tris, 0.192 M glycine, 2.6 mM SDS, and 20% (vol/vol) methanol; pH 8.8] at 100 V for 1 h. In samples assayed for phosphorylated IB, blots were blocked for 2 h at room temperature with blocking buffer [PBS with 0.1% Tween 20 and 5% fat-free milk powder (Carnation, Glendale, CA)]. After the blots were rinsed five times for 5 min each in PBS containing 0.1% Tween 20, they were incubated overnight at 4°C with primary rabbit polyclonal phosphospecific antibodies for the phosphorylated form of the NF-B inhibitor IB, diluted 1:1,000 in PBS with 0.1% Tween 20 and 5% BSA. After the blots were rinsed again as above, they were incubated for 1 h at room temperature with horseradish peroxidase (HRP)-conjugated secondary antibody diluted 1:5,000 in blocking buffer. Immunoblots were rinsed again as above and detected using an enhanced chemiluminescence method (ECL Western blotting detection system; Amersham Life Science) and autoradiography. Densitometry was performed as above.

Measurement of GRO- and IL-8 expression. GRO- and IL-8 were measured in culture medium from untreated and ebselen- or dicumarol-treated cells using commercial ELISA assays purchased from R&D Systems (Minneapolis, MN). This assay could not be used to detect GRO- production by catalase-treated cells because of interference in the peroxidase-based ELISA by catalase in the cell supernatant.

Measurement of proliferation in cell cultures. Proliferation of cultured cells was quantitated using a previously reported colorimetric method based on metabolic reduction of the soluble yellow tetrazolium dye MTT to its insoluble purple formazan by the action of mitochondrial succinyl dehydrogenase (13). This assay empirically distinguishes between dead and living cells. For proliferation studies, cells were seeded into 24-well uncoated plastic plates (Costar) at 15,000-50,000 cells per well and cultured with respective media and mitogens. After 24-96 h, medium was removed, cells were washed twice with 1 ml of sterile Dulbecco's modified PBS without Ca²⁺ or Mg²⁺ (DPBS), the medium was replaced with 1 ml/well fresh medium containing 100 µg/ml MTT, and plates were incubated an additional hour. MTT-containing medium was removed, 0.5 ml DMSO was added to each well, and the absorbance of the solubilized purple formazan dye was measured at 540 nm. A total of four to six wells was studied at each treatment condition. Preliminary studies were performed with 50-200 µg/ml MTT incubated for 15 min to 3 h to determine the optimum concentration and incubation time at which the rate of conversion was linear and proportional to the number of cells present. The absorbance of the MTT formazan reduction product (A₅₄₀) correlated with cell numbers counted by hemocytometer with R² = 0.99. In some experiments, the MTT assay and responses to mitogens and inhibitors were also confirmed by performing cell counts on 10 random fields/well of Giemsa-modified Wright's stained monolayers viewed at 40 power using a 0.01-cm² ocular grid.

Measurement of cytotoxicity and apoptosis. To assess for cytotoxicity, near-confluent cells cultured in 24-well plates were exposed to antioxidants or withdrawn from serum for 24 h. Medium was removed and replaced with DPBS containing 0.1% trypan blue. Cell death was assessed by counting the average number of trypan blue positive cells in five random fields counted of eight separate wells at 40 power using a 0.01-cm² ocular grid. To assess for apoptosis, cells were grown on glass slides in the presence of 10% FBS and treated with 3,000 U/ml catalase for 24 h. Apoptosis was studied by two methods. DNA cleavage was assessed by terminal deoxynucleotidyl transferase (TdT)-dependent 3'-OH fluorescein end-labeling of DNA fragments, using a Fluorescein-FragEL DNA fragmentation detection kit (Oncogene Research Products, Cambridge, MA). DNase-treated fixed cells were used as a positive control. Apoptosis was also evaluated by fluorescent-labeled annexin V staining of phosphatidylserine translocated to membrane surface. Externalized phosphatidylserine was stained using the Annexin-V-FLUOS staining kit (Roche Molecular Biochemicals, Indianapolis, IN) according to the manufacturer's instructions. Positive annexin V labeling was induced by treatment of cells for 24 h with 1 µM staurosporine.

Electron Microscopy. Cells were detached with trypsin/EDTA, fixed in 2.5% glutaraldehyde, postfixed in 1.0% osmium tetroxide, dehydrated sequentially in ethanol and propylene oxide, and embeded in spurr resin. Thin sections were stained sequentially with uranyl acetate and Sato's triple lead stain, and viewed using a Phillips CM10 transmission electron microscope.

DNA cell cycle measurements. To study the effect of antioxidant treatments on the DNA cell cycle, cells were grown to near confluence in 25-cm² plastic flasks and treated for 24 h. Cells were typsinized, washed twice in cold DPBS with 1 mM EDTA and 1% BSA, fixed 30 min in ice-cold 70% ethanol, and stained by incubation for 30 min at 37°C in a 10 µg/ml solution of propidium iodide in DPBS and 1 mg/ml RNase A. DNA cell cycle measurements were made using a FACStar^PLUS Flow Cytometer (Becton-Dickinson, San Jose, CA).

Measurement of reactive oxygen species. O generation was measured by the technique of superoxide dismutase (SOD)-inhibitable reduction of ferricytochrome c (24), employing a modification allowing absorbance reading with an automatic enzyme immunoassay reader (47). Confluent cells grown on 24-well plates were washed with DPBS and incubated in 5% CO₂-95% air at 37°C with 160 µM ferricytochrome c in a total volume of 550 µl of sodium bicarbonate-containing Krebs-Heinseleit buffer (44) or Hanks' balanced salt solution (HBSS), with and without copper-zinc SOD (1,000 U/ml). The absorbance of each well was measured at 550 nm initially and 3-24 h later using an ELx800 ultraviolet (UV) automated microplate reader (Biotek Instruments, Highland Park, VT). Monolayers were then washed with DPBS, and cell protein was measured using the BCA protein assay (Pierce). O generation, normalized to cell protein, was computed from the Beer-Lambert relationship (6) as the quotient of SOD-inhibitable increase absorbance over time divided by the difference between the molar extinction coefficients for ferricytochrome c and ferrocytochrome c (2.1 × 10⁴ M¹ · cm¹) (24). In some experiments, the following inhibitors of major oxidases were added to dissect potential sources of O generation: the quinone analog capsaicin (8-methyl-N-vanillyl-6-noneamide, 100 µM), the NQO inhibitor dicumarol (250 µM), the xanthine oxidase inhibitor allopurinol (1 mM), the cyclooxygenase inhibitor indomethacin (10 µg/ml), the cytochrome P-450 inhibitor cimetidine (300 µM), the nitric oxide synthase inhibitor N-nitro-L-arginine methyl ester (LNAME; 100 µM), and the mitochondrial respiratory chain inhibitor rotenone (2 µM).

Determination of oxidase activities and levels of oxidase components. Xanthine dehydrogenase/oxidase (XDH/XO) activity was measured using the spectrofluorometric assay described by Beckman et al. (8). Briefly, monolayers were washed twice in ice-cold DPBS, scraped, and frozen in liquid nitrogen. The cell pellet was sonicated in 1 ml of buffer containing 0.1 mM EDTA, 10 mM DTT, and 1% 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate in 50 mM phosphate buffer, pH 7.4. The cell lysates were centrifuged at 18,000 g for 30 min at 4°C. The supernatant was diluted to 2 ml with 50 mM phosphate buffer containing 0.1 mM EDTA, pH 7.4. Fluorescence was monitored at 390 nm with the excitation wavelength set at 345 nm. After achieving a stable baseline, 20 µl of 1 mM pterin were added, and the reaction was observed for 20 min to assay XO activity. Subsequently, 20 µl of 1 mM methylene blue were added as an electron accceptor to assay total XDH/XO activity, and the reaction was observed for 20 min.

Statistical analysis. Data are expressed as means ± SE for a minimum number of four observations, unless otherwise indicated. Differences between two groups were compared using the Student's t-test. Differences between multiple groups were compared using one-way analysis of variance. The post hoc test used was the Newman-Keuls multiple-comparison test. Two-tailed tests of significance were employed. Significance was assumed at P < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Antioxidants reduce constitutive activation of NF-B in malignant human cell lines. Constitutive activation of NF-B has been previously reported in malignant melanoma cell lines (19, 53). M1619 melanoma cells also consistently exhibited constitutive DNA binding activity for NF-B in nuclear protein (Fig. 1, A and B, lane 1). Several distinct bands were observed, all of which were eliminated by addition of excess specific unlabeled NF-B consensus oligonucleotides to the binding reaction (Fig. 1B, lane 2). Supershift experiments demonstrated that the second band (Fig. 1A, arrow) contained p65 (Fig. 1A, lane 2) and p50 (Fig. 1A, lane 3) NF-B components, but not p52, Rel-B, or c-Rel (Fig. 1A, lanes 4-6). Identity of proteins binding in the other bands is presently unclear. Constitutive nuclear translocation of NF-B was confirmed immunohistochemically by intense staining for p65 in M1619 nuclei (Fig. 2A). Treatment of cells for 24 h with the antioxidants catalase or N-acetylcysteine (NAC) or the glutathione peroxidase mimetic (55) ebselen substantially reduced constitutive NF-B DNA binding activity in nuclear extracts (Fig. 1, C and D). Furthermore, exposure of cells to catalase for 24 h also essentially eliminated immunohistochemical staining for p65 in cell nuclei (Fig. 2B). In addition, catalase treatment for 24 h also suppressed constitutive nuclear DNA binding activity for NF-B in M1585 melanoma cells. These results suggest that constitutive nuclear activation of NF-B in malignant melanoma cell lines may be the consequence of endogenous redox stress.

View larger version (98K): [in this window] [in a new window]   Fig. 1.   Antioxidant treatment reduces constitutive nuclear DNA binding activity for nuclear factor-B (NF-B) in malignant cell lines. Confluent cultures of M1619 cells were lysed, nuclear protein was extracted, and electrophorectic mobility shift assays (EMSAs) were performed as described in the text, using the 32P-labeled consensus oligonucleotide 5'-AGTTGAGGGGACTTTCCCAGGC-3' and 3'-TCAACTCCCCTGAAAGGGTCCG-5', specific for the p50 component of NF-B. A: M1619 cells demonstrated prominent consitutive nuclear DNA binding activity for NF-B (lane 1). Several distinct bands were observed. Supershift experiments demonstrated that the second band (lane 1, arrow) contained p65 (lane 2) and p50 (lane 3) NF-B components but not p52 (lane 4), Rel-B (lane 5), or c-Rel (lane 6). B: competition experiments demonstrating that all bands are eliminated by addition of excess specific unlabeled NF-B consensus oligonucleotide to the binding reaction. Lane 1, M1619 nuclear protein incubated with 32P-labeled NF-B consensus oligonucleotide; lane 2, addition to binding reaction of 10× unlabeled NF-B consensus oligonucleotide; lane 3, addition to binding reaction of 10× unlabeled consensus oligonucleotide specific for cAMP responsive element. C: antioxidant treatment for 24 h substantially reduces constitutive nuclear DNA binding activity for NF-B in M1619 cells. Lanes 1-3, nuclear protein from untreated positive control cells; other lanes represent nuclear protein from cells treated 24 h with 3,000 U/ml catalase (lanes 4-6), 20 mM N-acetylcysteine (lanes 7-9), or the glutathione peroxidase mimetic ebselen (25 µM; lanes 10-12). D: densitometry results of the p65/p50-containing bands from gels in C. *P < 0.001 vs. control cells treated with fetal bovine serum (FBS) and medium alone. E: catalase treatment (3,000 U/ml for 24 h) also decreased constitutive nuclear DNA binding activity for NF-B in other malignant melanoma cell lines. Lane 1, untreated M1585 melanoma cells; lane 2, M1585 cells treated with catalase. Serum deprivation does not eliminate constitutive nuclear DNA binding activity for NF-B in M1619 cells. Near-confluent cells were incubated in the presence (lane 3) or absence (lane 4) of 10% FBS. After 24 h, nuclear protein was isolated and EMSAs were performed. Serum deprivation slightly decreased, but did not eliminate, constitutive nuclear activation of NF-B, suggesting that the oxidant stress inducing NF-B activation is not induced by components of serum but is endogenous to the malignant cell.

View larger version (121K): [in this window] [in a new window]   Fig. 2.   Constitutive nuclear translocation of NF-B is demonstrated in M619 cells by intense immunohistochemical staining for p65 in nuclei (A). In contrast, little p65 is present in nuclei from cells treated with the antioxidant catalase (B), the NAD(P)H oxidase inhibitor apocynin (C), or the NAD(P)H:quinone oxidoreductase (NQO) inhibitor dicumarol (D). Confluent cells were fixed in paraformaldehyde, permeabilized, stained using an antibody to the p65 component of NF-B and a streptavidin-biotin-immunoperoxidase based method outlined in the text, viewed under light microscopy using a green filter to enhance contrast, and photographed at ×980 magnification. Control untreated cells (A) show intense brown staining in nearly all nuclei, corresponding to the presence of anti-p65. In contrast, cells treated for 24 h with 3,000 U/ml catalase (B), 150 µg/ml apocynin (C), or 250 µM dicumarol (D) demonstrate little anti-p65 brown staining in nuclei. The nuclei from catalase, apocynin, and dicumarol-treated cells also display greater detail, with prominent nucleoli (B, C, and D) not seen in untreated cells shown (A). D: a 50 mM concentration of dicumarol was dissolved in water by drop-wise addition of 0.1 N NaOH. Addition of up to 5 µl of this solution per milliliter (to make 250 µM) did not change pH of medium.

View larger version (20K): [in this window] [in a new window]   Fig. 3.   Catalase decreases the amount of phosphorylated inhibitor of NF-B (IB). A: replicate immunoassays are shown of phosphorylated IB (IB-P) in untreated M1619 cells (lanes 1-4) and in cells treated for 24 h with 3,000 U/ml catalase (lanes 5-8), respectively. Catalase treatment virtually eliminates any detectable phosphorylated IB. B: mean ratios of the densitometrically determined sum intensities of experiments in A. No detectable phosphorylated IB exists in cells treated 24 h with catalase. *P < 0.01 compared with untreated cells.

View larger version (20K): [in this window] [in a new window]   Fig. 4.   Catalase inhibits protein expression of the cell cycle regulator cyclin D1. M1619 cells were grown to near confluence in 10% FBS and treated for 24 h with 3,000 U/ml catalase. Cells were lysed and the level of cyclin D1 was immunoassayed as described in MATERIALS AND METHODS. The band of interest is a doublet of ~36 kDa. The upper half of the doublet may represent phosphorylated protein, since cyclin D1 undergoes phosphorylation on threonine-286 (26). Both bands of the doublet were eliminated in all lanes when the primary antibody (sc-718, 5 µg in 20 ml; Santa Cruz) was incubated overnight at 4°C with a 4-fold excess of specific blocking peptide (sc-718-P, 20 µg in 20 ml; Santa Cruz) before this mixture was used in primary antibody staining of the Western blot (data not shown). Compared with cells incubated in 10% FBS and media alone (lanes 1-4), catalase treatment (lanes 5-8) transiently reduced cyclin D1 expression beginning after 4 h of treatment (lane 6 vs. lane 2), with indication of some recovery after 12 h (lane 8 vs. lane 4).

Antioxidants are antiproliferative against malignant human cell lines. Inhibiting NF-B activation in tumor cells has been shown to reduce cellular proliferation. We therefore studied whether the same antioxidants that decreased constitutive activation of NF-B also inhibited growth of malignant cell lines. At concentrations we have previously reported to inhibit growth of cultured human airway smooth muscle (13), NAC and catalase reduced proliferation of M1619 melanoma cells when added to culture medium (Fig. 5A). In contrast, copper-zinc SOD had no effect on cell growth. Growth inhibition from catalase was dose dependent (Fig. 5A), was shared by a variety of catalase preparations from different sources (data not shown), and was eliminated by protein inactivation (Fig. 5B). Catalase treatment did not appear to produce apoptosis. Catalase did not result in externalization of phosphatidylserine to the plasma membrane surface, studied by fluorescent-labeled annexin V staining (Fig. 6, A-C) and did not cause DNA fragmentation, studied by TdT-dependent 3'-OH fluorescein end-labeling of DNA fragments (data not shown). Also, catalase-treated M1619 cells demonstrated no electron microscopic evidence of membrane blebbing or condensation of chromatin (Fig. 6, E vs. D), but catalase did cause prominent vacuolization of the cell cytoplasm (Fig. 6, E vs. D) and significantly decreased trypan blue dye exclusion (1.4 ± 0.2 for vehicle-treated control cells vs. 10.1 ± 0.8 trypan blue positive cells/field for confluent cells treated 24 h with 3,000 U/ml catalase; P < 0.01). M1619 proliferation was also dramatically reduced by ebselen (Fig. 5C). Catalase and ebselen were antiproliferative even if added 24 h after melanoma cells were plated (Fig. 5D) and were effective against a wide range of cultured malignant cells, including carcinomas of the lung, prostate, and breast (Table 1). Together, these results indicate that reactive oxygen species may be important signaling molecules for growth of malignant cell lines and suggest that the proximate growth-signaling form of reactive oxygen may be H₂O₂.

View larger version (29K): [in this window] [in a new window]   Fig. 5.   Antioxidants inhibit growth of cultured malignant melanoma cells. A: bovine liver catalase and N-acetylcysteine (NAC), but not copper-zinc superoxide dismutase (SOD), are antiproliferative against M1619 cells. Cells stimulated with 10% FBS were plated at a density of 50,000 cells/well, and antioxidants were added to wells at the indicated concentrations (mM or U/ml). After 48 h, proliferation was quantitated by assessing the cell number-dependent reduction of the soluble yellow tetrazolium dye 3-[4,5-dimethylthiazol]-2yl-2,5-diphenyl tetrazolium bromide (MTT) to its insoluble formazan, measured as the absorbance at 540 nm (A540) (13). *P < 0.001 vs. FBS alone. B: antiproliferative activity of 3,000 U/ml catalase is abolished by boiling the protein. *P < 0.001 vs. FBS alone; +P < 0.001 vs. active catalase. C: proliferation of M1619 cells was also inhibited in a dose-dependent manner by the glutathione peroxidase mime ebselen. Cells were cultured for 72 h and proliferation was measured as above. DMSO (5 µl) served as a vehicle control for ebselen-treated cells. *P < 0.001 vs. FBS alone. D: catalase and ebselen reduce M1619 cell proliferation even when added 24 h after cells are plated. Cells stimulated with 10% FBS were plated at a density of 50,000 cells/well and grown for 24 h before antioxidants were added to wells at the indicated concentrations (U/ml or µM). After an additional 48 h, proliferation was quantitated as above. DMSO (5 µl) served as a vehicle control for ebselen-treated cells. *P < 0.001 vs. FBS; +P < 0.001 vs. DMSO vehicle-treated cells.

View larger version (104K): [in this window] [in a new window]   Fig. 6.   compared with untreated control cells cells (A) or cells treated 24 h with 1 µM staurosporine (B), treatment with 3,000 U/ml catalase for 24 h (C) did not induce apoptosis in cultured M1619 cells. The top half of A, B, and C photographed under visible light corresponds to the fluorescent micrograph below each. Apoptosis was studied by staining for externalization of phosphatidylserine to the plasma membrane surface using fluorescent-labeled annexin V (Annexin-V-FLUOS staining kit; Roche Molecular Biochemicals, Indianapolis, IN, used according to manufacturer's instructions). However, compared with vehicle-treated controls (D), catalase (3,000 U/ml for 24 h) did induce prominent vacuole formation in M1619 cell cytoplasm (E) but did not cause nuclear or plasma membrane blebbing or chromatin condensation. Electron micrographs are shown at ×8,900 magnification for D and ×6,610 for E.

Antioxidants reduce polyploidy and increase S-phase fraction in M1619 cells. Antioxidant treatments that reduced constitutive activation of NF-B also had profound effects on the cell cycle. Results of DNA cell cycle analysis of M1619 cells are shown in Fig. 7. Untreated M1619 melanoma cells are a rapidly proliferating, desynchronized malignant line composed of both diploid and tetraploid cells. Over 30% of cells are tetraploid (Fig. 7A). Treatment with catalase for 24 h (Fig. 7B) substantially reduces the fraction of tetraploid cells (12.8%) and increases the total fraction of cells in S-phase from 41.2 (untreated) to 56.7% (catalase treated). Similar changes were seen after treatment with NAC (data not shown). While the exact signaling relationships responsible for these events is not clear, this suggests the possibility that antioxidants impair progression into G₂-M.

View larger version (32K): [in this window] [in a new window]   Fig. 7.   Catalase decreases polyploidy and increases the fraction of cells in S-phase. Near-confluent monolayers of M1619 cells were incubated in RPMI 1640 and 10% FBS in the presence or absence of 3,000 U/ml catalase. After 24 h, cells were harvested, ethanol fixed, permeabilized with proteinase K, stained with propidium iodide, and subjected to DNA cell cycle analysis. Whereas a large fraction (30.2%) of untreated cells (A) were tetraploid, only 12.8% of catalase-treated cells (B) were tetraploid. Antioxidant treatment also increased the total percentage of cells in S-phase from 41.2% in untreated controls (A) to 56.7% in cells treated with catalase (B), suggesting a slowing of progression into the G2-M phase of the cell cycle. Red, diploid G0-G1; yellow, tetraploid G0-G1; hatched, S-phase; green, cell aggregates; purple, cell debris, which may also represent necrotic and apoptotic cells. G2-M was hidden by tetraploid G0-G1 or by debris and could not be analyzed.

Melanoma cells release endogenously generated reactive oxygen species. Inhibition of constitutive NF-B activation in malignant cells by antioxidants suggests that neoplastic cells may be exposed to constant redox stress that in turn activates the IB kinase complex. One potential source of oxidant stress might be reactive oxygen species produced endogenously by the tumor cell. M1619 cells progressively reduced ferricytochrome c placed in the buffer medium. Approximately half the ferricytochrome c reduction was inhibitable by SOD, indicating that O produced much of the observed reduction (A₅₅₀ after 3 h = 0.012 ± 0.001 with ferricytochrome c alone vs. 0.005 ± 0.001 for ferricytochrome c + SOD; A₅₅₀ after 12 h = 0.021 ± 0.001 with ferricytochrome c alone vs. 0.007 ± 0.001 for ferricytochrome c + SOD; n = 12 each and P < 0.001 for both time points). When they were incubated in buffer alone without serum stimulation, melanoma cells produced 4.69 ± 0.16 µmol O/mg cell protein over 24 h. These results suggest that the source of oxidant stress driving constitutive activation of NF-B in these cells is endogenously generated in an autocrine fashion and is not the consequence of stimulation by growth factors present in serum. Endogenous sources of redox stress are supported by the finding that M1619 cells incubated in serum-free medium for 24 h still exhibited substantial constitutive NF-B DNA binding activity in nuclear protein (Fig. 1E, lanes 3 and 4).

Reactive oxygen species in melanoma cells are generated by an NAD(P)H-dependent enzymatic activity similar to NQO. M1619 cells had no measureable xanthine oxidase activity, and no evidence was detected of mRNA specific for the p22 and gp91^phox components of neutrophil NADPH oxidase or for the newly described mox-1 (or NOH-1L) oxidase. Neither cellular reduction of ferricytochrome c nor proliferation were reduced by the xanthine oxidase inhibitor allopurinol, the cycloxygenase inhibitor indomethacin, the cytochrome P-450 inhibitor cimetidine, the nitric oxide synthase inhibitor LNAME, or the mitochrondrial respiratory chain inhibitor rotenone. However, ferricytochrome c reduction was significantly decreased by the quinone analog capsaicin (Fig. 8A). Capsaicin, as well as the NAD(P)H oxidase inhibitors diphenylene iodonium chloride and apocynin (4'-hydroxy-3'-methoxy-acetophenone), significantly reduced proliferation of M1619 melanoma cells at 48 h (Fig. 8B). Also, apocynin treatment of cells for 24 h reduced constitutive nuclear translocation of NF-B as assessed by immunohistochemistry (Fig. 2, C vs. A). Taken together, these results suggest that the source of endogenous O generation stimulating NF-B activation and influencing cellular proliferation in this cell line is an NAD(P)H oxidoreductase activity distinct from the gp91^phox neutrophil NADPH oxidase, mox-1 or the NOH-1L oxidase.

View larger version (26K): [in this window] [in a new window]   Fig. 8.   Ferricytochrome c reduction and cellular proliferation of melanoma cells are reduced by quinone analogs and NAD(P)H oxidase inhibitors. A: capsaicin inhibits ferricytochrome c (Cyto c) reduction. M1619 cells grown on 24-well plates were washed with DPBS and incubated in 5% CO2-95% air at 37°C with 160 µM ferricytochrome c in total volume of 550 µl of Hanks' balanced salt solution with and without the quinone analog capsaicin (100 µM final concentration) added in 5 µl/ml of ethanol. The absorbance of each well was measured at 550 nm initially and 3 h later using an ELx800 ultraviolet automated microplate reader (Biotek Instruments, Highland Park, VT). *P < 0.001 compared with vehicle control, n = 12 for each treatment. B: capsaicin and NAD(P)H oxidase inhibitors decrease melanoma cell proliferation. Cells stimulated with 10% FBS were plated at a density of 50,000 cells per well, and inhibitors were added to wells in the following final concentrations and vehicles: diphenylene iondonium chloride (DPI), 25 µM in Dulbecco's PBS; capsaicin, 100 µM in 5 µl/ml of ethanol; and apocynin, 150 µg/ml in 5 µl/ml of DMSO. After 48 h, proliferation was quantitated as in Fig. 4. *P < 0.001 compared with respective vehicle control. Independent experiments confirmed dose-response reduction of proliferation with each inhibitor.

View larger version (94K): [in this window] [in a new window]   Fig. 9.   The common dioxin-inducible form of NAD(P)H:quinone oxidoreductase (NQO1) is expressed by tumor cell lines. RNA was harvested and PCR was performed for NQO1 using the primers 5'-CAGCGCCCCGGACTGCACCAGAGCC and 3'-GGGAAGCCTGGAAAGA-TACCCAGA (25). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as an internal control. Lane 1, M1619 melanoma cells; lane 2, H520 squamous lung carcinoma cells; lane 3, H596 adenosquamous lung carcinoma cells; lane 4, M1585 melanoma cells; and lane 5, LNCaP prostate carcinoma cells.

View larger version (40K): [in this window] [in a new window]   Fig. 10.   Ferricytochrome c reduction, NF-B activation, and cellular proliferation of melanoma cells are reduced by dicumarol, an inhibitor of NQO. A: the NQO inhibitor dicumarol inhibits ferricytochrome c (Cyto c) reduction by M1619 cells. Dicumarol (250 µM) was added to confluent M1619 cell cultures in complete medium before each experiment. After 60 min, dicumarol-containing medium was removed, cells were washed with Dulbecco's PBS, and ferricytochrome c reduction was studied as in Fig. 7A. A 50 mM concentration of dicumarol was dissolved in water by drop-wise addition of 0.1 N NaOH. Addition of up to 2.5 µl of this solution per ml (250 µM final concentration) did not change the pH of complete medium. *P < 0.01 compared with untreated control cells. B: dicumarol reduces constitutive NF-B activation in M1619 cells. Near-confluent cultures of M1619 cells incubated with complete medium alone or medium containing 250 µM dicumarol for 24 h. Cells were then lysed, nuclear protein was isolated, and EMSAs were performed as described in Fig. 1. Constitutive NF-B DNA binding was greatly reduced in dicumarol-treated cells (lanes 4-6) compared with cells incubated in growth medium alone (lanes 1-3). C: densitometry results of the p65/p50-containing bands from gels in B. *P < 0.001 vs. control cells treated with FBS and medium alone. D: dicumarol inhibits melanoma cell production of the autocrine growth factor GRO-. Near-confluent M1619 cells were incubated with or without dicumarol at the concentrations indicated. After 24 h, GRO- concentration was measured in media. *P < 0.001 compared with no dicumarol. E: dicumarol inhibits melanoma cell production of the autocrine growth factor interleukin-8 (IL-8). Near-confluent M1619 cells were incubated with or without dicumarol at the concentration indicated. After 24 h, IL-8 concentration was measured in media. *P < 0.001 compared with no dicumarol. F: dicumarol inhibits proliferation of M1619 cells. Cells stimulated with 10% FBS were plated at a density of 50,000 cells/well, and dicumarol was added to medium in the concentrations shown. After 48 h, proliferation was quantitated as in Fig. 4. *P < 0.001 compared with respective vehicle control. G: vitamin K does not prevent growth inhibition from dicumarol. M1619 cells stimulated with 10% FBS were plated at a density of 50,000 cells/well, and dicumarol or dicumarol plus an equimolar concentration of vitamin K2 (Vit K) were added to medium in the concentrations shown. The vehicle for vitamin K2 (5 µl/ml of DMSO) was added to all wells. *P < 0.01 compared with FBS alone; +P < 0.001 compared with DMSO vehicle control. Dicumarol alone vs. dicumarol + vitamin K2 were not different at either concentration.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The transcription factor NF-B is increasingly recognized to play an important role in development and progression of a variety of malignancies. While the exact signaling relationships linking activation of NF-B to neoplastic cell proliferation are incompletely understood, NF-B is constitutively activated in a number of malignancies (5, 7, 9, 12, 21, 28, 53, 56, 62-64), and inhibition of NF-B activation both reduces tumor cell growth (5, 7, 28, 56) and sensitizes malignant cells to TNF--mediated apoptosis (9, 62, 63). In addition, NF-B activation by anti-tumor chemotherapy may play a role in mediating resistance of tumor cell lines to currently available chemotherapy treatments by upregulating anti-apoptotic gene products (9, 62, 63). Thus unraveling the mechanism by which NF-B is constitutively activated in neoplasms would represent an important advance in tumor cell biophysiology. In normal cells, agonist stimulation activates IB kinase, leading to IB phosphorylation (3, 57), ubiquitination, and proteolytic degradation (46). Uncoupled from its inhibitor, the NF-B complex translocates to the nucleus to transcriptionally regulate expression or repression of target genes (10). In cultured melanoma cells, constitutive activation of NF-B is the result of elevated constitutive activity of the IB kinase complex (19), resulting in enhanced degradation of IB (53). Thus constitutive activation of NF-B in tumors may follow intermediate signaling events that are identical to those occurring during activation of NF-B in normal cell lines (3, 57).

The initial upstream response to agonist activation of IB kinase and subsequent degradation of IB in normal cells has been proposed to be production of O followed by generation of H₂O₂ (51). The compelling argument for this sequence of events is the observation that many of the known inducers of NF-B also induce generation of reactive oxygen species (2, 3, 51, 52). Human tumor cells produce substantial amounts of reactive oxygen species spontaneously (18, 41, 60). Mitogenic signaling through both Ras (30) and Rac (31) is mediated by O production, and transfection with the O-generating oxidase mox-1 transforms normal cells (58). We therefore proposed that oxidant stress from generation of reactive oxygen species might play a similar role in the constitutive activation of NF-B in malignant melanoma cell lines. Cultured M1619 melanoma cells displayed prominent constitutive nuclear activation of NF-B, demonstrated by conspicuous nuclear NF-B DNA binding activity in EMSAs (Fig. 1A, lane 1) and abundant immunohistochemical staining for the p65 component of NF-B in cell nuclei (Fig. 2A). Constitutive activation of NF-B was greatly decreased by antioxidant treatment of melanoma cells with the H₂O₂ scavenger catalase (Fig. 1C, lanes 4-6, and Fig. 2B), the sulfhydryl donor N-acetylcysteine (Fig. 1C, lanes 7-9), or the glutathione peroxidase mimetic ebselen (Fig. 1C, lanes 10-12). Reduction in the amount of phosphorylated IB (Fig. 3) by catalase suggests that antioxidant treatment prevents NF-B activation by influencing early signaling pathways regulating activity of the IB kinase complex. The interruption of NF-B functionality by antioxidant treatment was confirmed by the observation that catalase treatment of M1619 cells decreases expression of the NF-B-regulated (29, 32) cell cycle protein cyclin D1 (Fig. 4) and that ebselen causes a 40% reduction in protein expression of melanoma growth stimulatory activity/GRO-, an NF-B-regulated autocrine growth factor important in melanoma cell proliferation (65, 66). The effectiveness of catalase in reducing constitutive NF-B activation in M1585 melanoma cells (Fig. 1E, lane 2) suggests that oxidant stress may commonly underlie constitutive NF-B activation in other malignant melanoma cell lines. Addition of catalase to growth medium did not induce prominent apoptosis (Fig. 6, C and E) or necrosis, suggesting that antioxidant treatment decreased NF-B activation by selective interruption of oxidant-mediated signaling events, and not by indiscriminate cellular toxicity. Even in serum-free medium, melanoma cells progressively reduced ferricytochrome c in an SOD-inhibitable manner and continued to prominently display nuclear NF-B DNA binding activity in EMSAs (Fig. 1E, lane 4). This implies that the source of oxidant stress stimulating constitutive NF-B activation in malignant melanoma cells is endogenous in origin and autocrine in behavior.

Antioxidant treatment with catalase or ebselen also dramatically reduced growth of M1619 melanoma cells in a dose-dependent fashion, whether added at the time of initial cell culture (Fig. 5, A and C) or 24 h later, when cells had firmly attached and already begun rapid proliferation (Fig. 5D). M1619 cells treated with catalase demonstrated an increase in the fraction of cells in S-phase and a decrease in the percentage of polyploid cells (Fig. 7), consistent with impairment of progression into the G₂-M phase of the cell cycle. Paralleling their activity for inhibiting constitutive NF-B activation, catalase or ebselen also retarded growth of a variety of malignant cell lines, including M1585 melanoma, adenosquamous lung cancer, and carcinomas of the prostate and breast (Table 1). The importance of oxidant signaling in growth regulation is firmly established in normal tissues (13, 27, 30, 31, 50, 61). Our findings add to evidence presented by others (16, 18, 40, 41, 58, 60) that reactive oxygen species are also important growth signals for development and progression of malignancies. Because NF-B is the prototypical oxidant-activated transcription factor, it is tempting to speculate that oxidant stress selectively promotes malignant cell growth through activation of NF-B. However, other transcription factors such as activator protein-1 are redox regulated (48), and H₂O₂ can activate both phosphatidylinositol 3-kinase (49) and the p44 and p42 extracellular signal-regulated (ERK1 and ERK2) kinases of the mitogen-activated protein (MAP) kinase superfamily (1), which are thought to play key roles in the transduction of mitogenic signals to the cell nucleus. Also, we have recently reported posttranslational redox regulation for levels of the early response gene product c-Fos (13). Thus endogenous autocrine oxidant generation might redundantly impact on malignant cellular proliferation at multiple levels of regulation.

The initial form of reactive oxygen produced as a signaling intermediate by melanoma cells appears to be O. However, the effectiveness of catalase, but not SOD, at interrupting NF-B activation (Fig. 1) and malignant cellular proliferation (Fig. 5) suggests that H₂O₂ formed by dismutation of O either spontaneously or by extracellular SOD is the proximately important oxidant signaling intermediate. The importance of H₂O₂ as the relevant signaling intermediate for oxidant regulation of cell growth has been previously observed (13, 30, 31, 60, 61). The reactive oxygen species driving NF-B activat