Oxidative Stress Modulates DNA Methylation during Melanocyte Anchorage Blockade Associated with Malignant Transformation

Abstract

Both oxidative/nitrosative stress and alterations in DNA methylation are observed during carcinogenesis of different tumor types, but no clear correlation between these events has been demonstrated until now. Melanoma cell lines were previously established after submitting the nontumorigenicmelanocyte lineage, melan-a, to cycles of anchorage blockade. In this work, increased intracellular oxidative species and nitric oxide levels, as well as alterations in the DNA methylation, were observed after melan-a detachment, which were also associated with a decrease in intracellular homocysteine (Hcy), an element in the methionine (universal methyl donor) cycle. This alteration was accompanied by increase in glutathione (GSH) levels and methylated DNA content. Furthermore, a significant increase in dnmt1 and 3b expression was identified along melan-a anchorage blockade. l^G-Nitro-l-arginine methyl esther (l-NAME), known as a nitric oxide synthase (NOS) inhibitor, and N-acetyl-l-cysteine (NAC) prevented the increase in global DNA methylation, as well as the increase in dnmt1 and 3b expression, observed during melan-a detachment. Interestingly, both l-NAME and NAC did not inhibit nitric oxide (NO) production in these cells, but abrogated superoxide anion production during anchorage blockade. In conclusion, oxidative stress observed during melanocyte anchorage blockade seems to modulate DNA methylation levels and may directly contribute to the acquisition of an anoikis-resistant phenotype through an epigenetic mechanism.

Introduction

The interaction between cells and their surrounding extracellular matrix is a major determinant of cell behavior by modulating gene expression, cell growth, differentiation, migration, and overall tissue architecture. Cell survival also depends on anchorage to extracellular matrix and on cell-cell contacts. Apoptosis induced by loss of appropriate cell-matrix contact has been termed anoikis and plays an essential role in regulating homeostasis, as it maintains the correct cell number of high turnover tissues [1,2]. An important feature of transformed cells is loss of anchorage-dependent growth control, leading to increased survival time and facilitating an eventual reattachment. Thus, the ability of escaping anoikis regulation is a critical step in oncogenesis [3].

Reduced substrate adhesion or lack of growth factors represent adverse conditions that confer a dramatic rise in cellular stress, characterized by damage to various cellular constituents such as DNA, protein, and lipids that ultimately triggers death signals [4,5]. Low or subtoxic levels of reactive oxygen species (ROS) can act as potent second messengers in signal transduction pathways that regulate cell growth and transformation [6,7]. High levels of ROS have been considered direct DNA-damaging agents that increase the mutation rate and promote and maintain the oncogenic phenotype [8–10]. Neither superoxide (O₂·^-) nor hydrogen peroxide (H₂O₂) are particularly toxic, but cells can greatly increase the toxicity of superoxide by producing nitric oxide (NO) that may function as an antiapoptotic or proapoptotic agent depending on concentration and cell type [11]. Superoxide anion and NO react to produce peroxynitrite (ONOO^-) that can readily modify proteins and other molecules [12]. In addition to causing genetic changes, ROS may lead to epigenetic alterations that strongly affect gene expression without changing the DNA base sequence [13–15].

Epigenetic changes, including alterations in DNA methylation status, have been implicated with malignant transformation and progression of numerous tumor types. The addition of methyl (CH₃) radicals to the cytosine nucleotide is catalyzed in mammals mainly by three different DNA methyltransferases (Dnmt), namely DNMT 1, 3a, and 3b. DNMT3 group comprises two enzymes with catalytic activity (DNMT 3a and 3b) and DNMT 3L, which has a regulatory role [16,17].

Neoplastic cells present simultaneously regional DNA hypermethylation and global hypomethylation. Hypermethylation of promoter regions is associated with tumor suppressor gene silencing by interfering with the binding of transcription factors or by recruiting corepressor complexes containing, for example, histone deacetylases. Global DNA hypomethylation can result in oncogene expression, genomic instability, and loss of imprinting. Alterations in methylation patterns can thus determine the gene expression profile, which is crucial for comprehension of carcinogenesis [18–20].

Recently, an in vitro transformation model of murine melanocytes has been developed in our laboratory, after submitting a nontumorigenic melanocyte lineage, melan-a, to sequential cycles of anchorage blockade [21,22]. The aim of the present study was to determine whether anchorage blockade results in altered ROS and/or NO levels, and the resulting effect on DNA methylation, which may contribute to melanocyte malignant transformation.

Materials and Methods

Cell Culture

Murine nontumorigenic melanocyte lineage melan-a [23], a kind gift from Dr. Michel Rabinovitch (Discipline of Parasitology, Universidade Federal de Säo Paulo - UNIFESP), was maintained in RPMI 1640 (Gibco, Carlsbad, CA), pH 6.9, containing 5% fetal bovine serum (Gibco), 40 mg/l gentamicin, and 200 nM phorbol 12-myristate 13-acetate (Sigma, St. Louis, MO) at 37°C in 5% CO₂. For anchorage blockade assays, adherent melan-a cells were removed from plates by trypsin treatment (D0) and plated (10⁵ cells/ml) on 1% agarose-coated dishes for 1 to 24 hours (D1h to D24h) in the conditions described above. In these conditions, melan-a cells do not adhere to the plastic and remain viable in suspension for at least 24 hours, as demonstrated by the Trypan blue exclusion assay (data not shown).

Measurement of Superoxide Anion (O₂·^-) by Flow Cytometry

Relative concentrations of intracellular O₂·^- were determined as described previously [24]. Control adherent cells (D0) and cells submitted to anchorage blockade for different periods of time (D1h, D3h, D5h, and D24h) were assayed for O₂·^- detection using dihydroethidium (DHE; Molecular Probes, Carlsbad, CA). Briefly, cells were washed in phosphate-buffered saline (PBS) and resuspended in PBS containing 10 µM DHE at 37°C for 30 minutes. After washing, cells were resuspended in PBS and analyzed (10,000 events per sample) by flow cytometry (FACScalibur; Becton Dickinson, San Juan, CA) (excitation wavelength = 480 nm; emission wavelength = 567 nm).

Glutathione Assay

Total glutathione (GSH) content was measured from cell lysates obtained during adhesion (D0) and deadhesion conditions (D1h, D5h, and D24h). Cells were mixed with the same volume of 2 M perchloric acid with 4 mM EDTA. After pH neutralization with K₃PO₄ and centrifugation, supernatants were collected for reaction with nicotinamide adenine dinucleotide phosphate (4 mg/ml in 0.5% NaHCO₃ buffer), 5,5′-dithiobis(2-nitro-benzoic acid) (1.5 mg/ml in 0.5% NaHCO₃ buffer), and GSH reductase (6 U/ml in 0.1 M KPO₄, pH 7, with 1 mM EDTA). Total GSH levels were obtained spectrophotometrically at 412 nm [25] and all determinations were normalized to the protein content [26]. Total GSH contents were expressed as nanomole per milligram of protein.

Nitric Oxide Detection

The fluorescent NO indicator, diaminofluorescein-2 diacetate (DAF-2DA), was used to measure intracellular [NO] [27] in adherent (D0) and suspended cells (D1h, D3h, D5h, and D24h). DAF-2DA readily enters the cells and it is hydrolyzed by cytosolic esterases to DAF-2, which is trapped inside cells. In the presence of NO and oxygen, the relatively nonfluorescent DAF-2 is converted into the highly green fluorescent triazole form DAF-2T. Thus, increases in DAF-2T fluorescence represent elevation of [NO]. Cells were incubated with DAF-2DA (10 µM) in 0.5 ml of PBS at room temperature for 30 minutes, rinsed with PBS, and analyzed by flow cytometry in a FACScan (Becton Dickinson) (excitation wavelength = 495 nm; emission wavelength = 515 nm). Alternatively, extracellular NO levels were also determined by a gas-phase chemiluminescence reaction of NO with ozone using a NO Analyzer (NOA 280; Sievers Instruments, Inc., Boulder, CO). NO-related species (nitrite, nitrate, S-nitrosothiols, and so on) are converted to NO in the purge vessel of the analyzer. The released NO is carried by an inert gas to the detector where it reacts with ozone to produce a chemiluminescence signal proportional to the concentration. A standard curve was established with a set of serial dilutions (0.1–100 µM) of sodium nitrate. The concentrations of NO metabolites in the samples were determined by comparing with the standard curve and expressed as micromole per liter. Data collection and analysis were performed using the NOAnalysis software (version 3.21; Sievers) [28].

Lipid Peroxidation Assay

Lipid peroxidation was measured by quantifying thiobarbituric acid-reactive substances, mainly malondialdehyde (MDA), formed during incubation using the thiobarbiturate-MDA adduct formation [29]. The product of the reaction between cell lysates and thiobarbituric acid was measured spectrophotometrically at 535 nm and values were expressed as nanomole per milligram of protein.

Homocysteine and Cysteine Quantification

Intracellular homocysteine (Hcy) was measured during adhesion (D0) and deadhesion conditions (D24h) by high-performance liquid chromatography (HPLC) with fluorimetric detection and isocratic elution. This methodology was adapted from Pfeiffer et al. [30] and it involves three steps, namely reduction of thiol groups using tris(carboxyethyl)-phosphine, protein precipitation with trichloroacetic acid, and derivatization with 7-fluorobenzene-2-oxy-1,3-diazolic-4-ammonium sulfate. The HPLC system used was a Shimadzu apparatus with a SIL-10ADvp automatic sample injector and a RF-10AXL fluorescence detector (Shimadzu, Tokyo, Japan). Chromatographic separation was performed using a C18 model Shim-pack CLC-ODS column (4.6 x 150 mm², with 5.0-µm microparticles; Shimadzu). The fluorescence was analyzed with a detector adjusted for excitation at 385 nm and emission at 515 nm. Total Hcy and cysteine (Cys) content, expressed as nanomole per milligram of protein, were calculated with a calibration curve using known Hcy, Cys, and cystamine concentrations as the internal standards.

5-Methylcytosine Content

Global DNA methylation was evaluated by staining cells with a specific monoclonal antibody against 5-methylcytosine (5-MeC; Oncogene, La Jolla, CA), as previously described [31]. Cells in adherent (D0) or nonadherent (D1h, D3h, and D24h) conditions were washed with PBS supplemented with 0.1% Tween 20 and 1% bovine serum albumin (PBS-TB), fixed with 0.25% paraformaldehyde at 37°C for 10 minutes, followed by 88% methanol at -20°C for at least 30 minutes. After washing, cells were treated with 2 N HCl at 37°C for 30 minutes, neutralized with 0.1 M sodium borate (pH 9.0), and blocked with 10% mouse serum for 20 minutes at 37°C. Then, cells were incubated with anti-5-MeC antibody at a final concentration of 1 µg/ml for 45 minutes at 37°C, followed by incubation with goat anti-mouse IgG conjugated with fluorescein (Kirkegaard & Perry Laboratories Inc. (KPL), Gaithersburg, MD) for 30 minutes at 37°C. Finally, cells were analyzed by flow cytometry in a FACScan. For nitric oxide synthase (NOS) inhibition and antioxidant assays, adhered melan-a cells were treated for 18 hours with 1 mM l^G-nitro-l-arginine methyl esther (l-NAME), 5 mM N-acetyl-l-cysteine (NAC), or 1% dimethylsulphoxide (DMSO), submitted to the anchorage blockade protocol as described above in the presence of inhibitors, and analyzed for 5-MeC content. The NOS inhibitor and antioxidants were used at noncytotoxic concentrations, estimated by methyl thiazol tetrazolium (MTT) assays.

Cytotoxicity Assays

MTT assay was performed to determine noncytotoxic concentrations of the NO inhibitor l-NAME and the antioxidants. Briefly, cells were seeded at a density of 2.5 x 10⁵ cells/ml in 96-well plates (in a final volume of 200 µl) in complete medium and replicates of three. After adhesion, cells were incubated with various concentrations of l-NAME (0.5, 1, 2, 4, and 8 mM), NAC (0.5, 1, 5, and 10 mM), catalase (5, 10, 50, and 100 U/ml), peroxidase (10, 50, 100, and 200 U/ml), and DMSO (0.5, 1, and 2%) in serum-free medium. After 24 hours of incubation, 20 µl of MTT (5 mg/ml; Sigma) was added and incubated for 1 hour at 37°C. Supernatant was then removed, and 100 µl of isopropanol was added. Culture plates were incubated for 15 minutes at room temperature to dissolve MTT crystals. Absorbance values were determined by an Multiskan MS (Labsystems, Vantaa, Finland) at a wavelength of 570 nm. Each experiment was repeated twice.

Western Blot Analysis

Protein extracts (50 µg) from adherent (D0) and suspended (D1h, D3h, D5h, and D24h) melan-a cells, treated or not with the nonspecific NOS inhibitor l-NAME (1 mM), were submitted to Western blot analysis using rabbit polyclonal antibodies against dnmt3b (Abcam, Cambridge, MA) and dnmt1 (Abcam), followed by incubation with peroxidase-conjugated anti-rabbit antibody (KPL) and developed using an enhanced chemiluminescence (ECL) detection reagent (GE Healthcare, Buckinghamshire, UK).

Reverse Transcription-Polymerase Chain Reaction

Total RNA was extracted from adherent and deadherent melan-a cells with TRIzol reagent (Invitrogen, Carlsbad, CA), according to the manufacturer's instructions. One microgram of RNA was reverse-transcribed to cDNA with Superscript III (Invitrogen). The resulting single-strand cDNAs were amplified by polymerase chain reaction (PCR) as follows: initial 5 minutes at 94°C, followed by 42 cycles of denaturing at 94°C for 30 seconds, with combined annealing at 56.5°C for 30 seconds, and extension at 72°C for 60 seconds. The β-actin mRNA was used to normalize the amounts of RNA in the cell samples, and the intensities of the resulting PCR bands were used to calculate the ratio of endothelial NOS (eNOS)/β-actin intensities. PCR fragment amplification was confirmed by agarose gel staining with ethidium bromide. PCR detection of eNOS and actin mRNA was carried out by using the primers: eNOS F 5′ CATGGAAATGTCAGGCCCG 3′; eNOS R 5′ TTCCACAGAGAGGATTGTAGC 3′; actin F 5′ CGAGGCCCAGAGCAAGAGAG 3′; actin R 5′ AGGAAGAGGATGCGGCAGTGG 3′.

Statistical Analysis

GraphPad Prism 3.03 software (San Diego, CA) was used for statistical analysis. One-way analysis of variance test was initially performed to analyze sample groups and unpaired t tests were used to compare suspended cells samples to adherent cells controls. All experiments were performed at least three times, each one in triplicates, and one representative experiment is shown.

Results

Anchorage Blockade Induces Oxidative and Nitrosative Stress

Several types of environmental stress induce ROS and reactive nitrogen species production, but few studies address the relationship of forced anchorage impediment and free radicals levels in melanocytes. To investigate the effect of anchorage blockade on ROS production, adherent (D0) and deadherent melan-a cells (D1h, D3h, D5h, and D24h) were assayed for intracellular superoxide anion (O₂·^-). Intracellular O₂-^- levels, which have been shown to be increased in human metastatic melanomas compared to normal melanocytes [32], become elevated progressively along melan-a anchorage blockade. In adherent conditions (D0), 1% of cells are positive for the O₂·^- marker, whereas 30% of cells maintained in suspension for 5 (D5h) to 24 hours (D24h) showed positivity (Figure 1).

Figure 1.

Anchorage blockade of nontumorigenic melanocytes (melan-a) induces superoxide anion (O₂·^-) production. Intracellular O₂·^- concentration were analyzed by flow cytometry in suspended (D1h, D3h, D5h, and D24h) and adherent (DO) melan-a cells using the nonfluorescent cell-permeant probe DHE, which, in contact with O₂·^-, becomes highly fluorescent. Light gray lines represent nonstained and black lines represent stained melan-a cells. Numbers inside graphs represent percentage of positive cells relatively to nonstained controls.

Oxidized and reduced GSH levels are commonly used as a measurement of intracellular redox state [33], and total GSH levels were investigated during exposure of melan-a cells to anchorage blockade. The ratio of intracellular GSH relative to total protein content was not significantly increased during the first hours in suspension (D1h, P = .064 and D5h, P = .159), when H₂O₂ levels were maximum (data not shown), but became elevated in a statistically significant manner only after 24 hours in detached cells (D24h, P = .020; Figure 2A), when H₂O₂ levels returned to concentrations found in adherent cells (data not shown), suggesting that increased glutathione levels could be acting in the removal of excessive H₂O₂ through reduction.

Figure 2.

Stress response to anchorage blockade induces nitric oxide production and alterations in homocysteine levels. Total cellular glutathione levels (measured by colorimetry), indicating an antioxidative response, become significantly elevated after 24 hours of anchorage blockade (A). Intracellular nitric oxide concentrations, analyzed by the NO fluorescent indicator, DAF (B), as well as one of the main lipid peroxidation products, malondialdehyde (MDA; C), also become significantly elevated along anchorage blockade. Levels of the glutathione precursor homocysteine (Hcy), but not cysteine (Cys), determined by HPLC showed a drastic reduction when maintained in suspension for 24 hours, compared to adhered control cells (D and E, respectively). Endothelial NO synthase (eNOS) mRNA expression was strongly induced after melan-a anchorage blockade (F). The murine endothelial cell line L229 was used as an eNOS expression positive control. Actin mRNA expression was used as an internal control and numbers below this figure show eNOS/actin ratio. Numbers inside flow cytrometry graphs (B) represent percentage of stained positive cells (black lines) relatively to nonstained controls (light gray lines). Asterisks indicate statistical significance levels, determined by unpaired t tests between each suspended cell sample (D1h, D3h, D5h, D8h, and D24h) compared to adhered control cells (D0). *P < .05, **P < .01, ***P < .0005.

ROS production has been associated also with that of NO, whose derivatives (N₂O₃, ${\text{NO}}_2^{\text{-}}$, peroxynitrite, and others) possess a strong oxidative capacity [34]. We estimated intracellular NO concentration in adherent (D0) and deadherent (D1h, D8h, and D24h) melan-a (Figures 2B and 4C); NO levels are higher in samples collected from suspended cell cultures than from adherent melan-a cells (D0), peaking around 5 hours in suspension and decreasing after 24 hours in deadherent conditions (Figure 2B). In addition, exposure of melan-a melanocytes to anchorage blockade resulted in a significant expression of eNOS (type III), different from adherent cells (D0) that did not express this enzyme (Figure 2F). Neither adherent nor suspended melan-a cells expressed neuronal NOS (nNOS) (type I) (data not shown).

Figure 4.

l-NAME and NAC do not inhibit NO production, but abrogate O₂·^- generation during melan-a anchorage blockade. Melan-a cells submitted to anchorage blockade (D3h) produced significant amounts of NO, estimated by NO analyzer, compared to adherent melan-a cells (D0). NO generation was not affected by either l-NAME (LN), NAC, or catalase (Cat) treatment (A). Intracellular O₂·^- levels were determined by flow cytometry (FACScan; Becton Dickinson) using DHE staining. l-NAME and NAC, but not catalase (Cat), treatment strongly inhibited O₂·^- generation by deadhered melan-a cells (D3h) (B). C: nontreated melan-a cells. Numbers inside graphs represent percentage of positive stained cells (black lines) relatively to control cells incubated in the absence of DHE (light gray lines). ***P < .0005.

Lipid hydroperoxides are the initial products of unsaturated fatty acid oxidation caused by peroxynitrite, the reaction product of NO with O₂·^- [35]. One of the major aldehyde products of lipid peroxidation isMDA and, as shown in Figure 2C, its levels increase progressively, soon after melanocyte anchorage blockade. These results clearly show that anchorage impediment in melanocytes results initially in increased levels of ROS and NO, later followed by accumulation of total glutathione.

Anchorage Blockade Induces Hcy Depletion

Approximately half of the intracellular GSH pool in cells is derived from Hcy through the transsulfuration pathway [36]. Hcy is a metabolite in the methionine cycle and has two major fates: 1) transsulfuration catalyzed by cystathionine β-synthase leading to cystathionine and 2) remethylation catalyzed by methionine synthase or by betaine Hcy methyltransferase. In the former pathway, cystathionine is subsequently converted to Cys, a precursor of GSH [37]. Cells can make the necessary Cys from methionine, or they can take it up from the surrounding fluids. Because the maintenance of intracellular GSH pool is regulated by Cys concentrations, which in vivo main source is the Hcy-dependent transsulfuration pathway, we compared Hcy concentrations in adherent (D0) and suspended (D24h) melan-a cells. Melan-a cells maintained in suspension for 24 hours show a dramatic decrease in intracellular Hcy concentrations compared to adherent melan-a cells (Figure 2D), whereas Cys levels remain stable in this condition (Figure 2E).

Anchorage Blockade Alters 5-MeC Content

In the second metabolic pathway, Hcy is remethylated to methionine, which is further converted to S-adenosylmethionine (SAM) by methionine adenosyltransferase. SAM, the universal methyl donor, is demethylated by several enzymes, such as DNA methyltransferases [38]. Alterations in Hcy levels in deadherent melan-a cells led us to investigate whether global DNA methylation was altered during melana anchorage blockade. Figure 3A (CTRL) shows that global DNA methylation is augmented in suspended melan-a cells (D1h, D3h, and D24h) when compared to adherent ones (D0, 47.3% of 5-MeC-positive cells). 5-Methylcytosine levels are elevated as soon as 1 hour (70.2% positive cells) and become progressively higher up to 24 hours after adhesion blockade (95.9% positive cells).

Figure 3.

Total DNA methylation levels and dnmt1 and 3b expression become elevated during forced anchorage blockade, and this response is abrogated by the l-NAME and NAC. 5-MeC content, analyzed by flow cytometry using a specific antibody (Oncogene), augments along melan-a adhesion blockade (D1 - 24h) compared to the adherent counterpart (D0) and this increase is abolished when these cells are treated with 1 mM ^l-NAME or with the antioxidant NAC, but not with DMSO (A). Numbers inside graphs represent percentage of positively stained cells (black lines) relative to control cells incubated in the absence of primary antibody (light gray lines). Suspended melan-a cells (D1 - 24h) present elevated expression of dnmt1 (D) and 3b (B and C) along time compared to adherent control cells (D0), as analyzed by Western blot using specific antibodies against dnmt1 and 3b (Abcam), and l-NAME treatment inhibits both dnmt1 and 3b expression in suspension condition [D3h and D24h for dnmt3b (C); D5h for dnmt1 (D)]. D0: adhered melan-a; D1h, D3h, D24h: suspended melan-a cells. -: nontreated cells, +: treated cells. Numbers below blots indicate dnmt/actin ratio.

l-NAME and NAC Abrogate DNA Hypermethylation and the Concomitant Increase in dnmt1 and 3b Expression during Anchorage Blockade

Considering that the increase in reactive species production and DNA methylation have a temporal association in this model, we tested whether these phenomena share a causal relationship. Figure 3A shows that l-NAME, a known NOS inhibitor, and the cell-permeant antioxidant NAC impair global DNA hypermethylation observed during melan-a anchorage blockade. l-NAME completely reversed 5-MeC levels increase after 3 (D3h, 10.4% positive cells) and 24 hours (D24h, 7.3% positive cells) in suspension, compared to nontreated control cells (CTRL) at the same time points (D3h, 84.6% positive cells; D24h, 95.9% positive cells). NAC was also able to exert a significant inhibitory effect over 5-MeC content elevation, at an earlier time point (D1h, 24.9% positive cells) than l-NAME and in 24 hours after anchorage blockade (D24h, 41.8% positive cells).

Other antioxidants, such as catalase (Sigma), DMSO (Sigma), and peroxidase (type II horseradish peroxidase, Sigma), were unable to exert the inhibitory effect over 5-MeC levels observed with l-NAME and NAC (Figure 3A, for DMSO; and data not shown, for catalase and peroxidase). Intriguingly, l-NAME-treated adherent cells (D0) showed a marked elevation in 5-MeC content, not detected after treatment with other compounds (NAC: Figure 3; DMSO, catalase, and peroxidase: data not shown). This observation is being further investigated in our laboratory, and may be related to Ras activation status [39].

The DNA methyltransferases implicated in establishing de novo methylation patterns are dnmt3a and 3b, whereas maintenance of DNA methylation pattern is performed by dnmt1 during DNA replication process. We analyzed protein expression by Western blot at different time points (D0, D1h, D3h, D5h, and D24h), and Figure 3B shows a marked increase in dnmt3b expression along anchorage blockade. A significant increase in dnmt1 expression was also observed after anchorage impediment (D5h), as shown in Figure 3D. Corroborating the hypothesis that oxidative stress regulates DNA methylation, l-NAME was capable of abolishing the increase both in dnmt1 and 3b expression, observed during anchorage blockade of melan-a cells (D3h and D24h, Figure 3C; D5h, Figure 3D). In addition, l-NAME-treated adherent cells (D0), which present higher levels of 5-MeC as shown above, also present a much higher expression of dnmt3b protein, but not dnmt1. Taken together, these data reinforce the proposition that oxidative stress regulates DNA methylation through DNA methyltransferase expression modifications.

l-NAME and NAC Do Not Inhibit NO Levels in Melan-a Melanocytes, But Totally Impair O₂·^- Production

Despite its known capacity of inhibiting NOS, l-NAME failed to reduce intracellular NO levels in nonadherent melan-a cells (Figure 4A), which, although unexpected, is in accordance with previous data showing that l-NAME was unable to inhibit NO production in different human melanoma cell lines [40], and with results indicating that l-NAME can be a source of nonenzymatically produced NO [41]. In addition, the NAC and catalase antioxidants did not modify NO levels in melan-a cells maintained in suspension (Figure 4A). Furthermore, l-NAME and NAC, but not catalase, were able to completely abrogate O₂·^- production in this condition (Figure 4B).

Discussion

Normal melanocytes depend on signals provided by both keratinocytes (in the epidermal melanin unit) and extracellular matrix to maintain their normal homeostasis, and their ectopic localization, as in dysplastic nevi and melanomas, is associated with morphologic and functional alterations [42,43].

Our group has recently established a melanocyte transformation model where several melanoma lineages were derived from immortalized, but nontumorigenic, melanocytes (melan-a cells) submitted to rounds of anchorage blockade followed by reattachment during normal culture conditions [21,22]. This model allows the identification of biochemical, genetic, and epigenetic alterations involved in homeostatic control loss and the carcinogenic process.

A significant increase in oxidative stress, as shown directly by elevated intracellular levels of O₂·^- (Figure 1) and indirectly by GSH (Figure 2A) and MDA concentrations (Figure 2C), is observed in melan-a cells submitted to anchorage blockade, beginning as soon as 1 hour. The increase in GSH levels could also indicate a response to oxidative stress [33,36] (Figure 5) because we observed a significant increase in GSH levels only 24 hours after adhesion blockade (Figure 2A), when H₂O₂ production returns to basal levels (data not shown). Increased ROS intracellular levels induced by anchorage blockade have been implicated as mediators of anoikis of human endothelial cells [44] and colorectal carcinoma cells [45]. However, elevated levels of O₂·^- have also been associated with protection from apoptotic death induced by cytotoxic agents or CD95 activation [46,47]. Nair et al. [48] showed that the decision to commit to programmed cell death during oxidative stress is, at least for neuronal murine cells, determined stochastically by each cell. This mutually exclusive decision involves either extracellular-regulated kinase or p53 activation pathways. It is probable that such life or death choice is also taken by melan-a cells maintained in forced deadherent culture conditions, and preliminary results from our laboratory demonstrate that p53 expression is increased in premalignant cell lines, but is lost in cell lines with full malignant phenotype (manuscript in preparation).

Figure 5.

O₂·^- production and its possible relation to methionine cycle. Lower levels of Hcy may result in lower levels of ADMA and higher NOS activity and/or expression. The concomitant local generation of NO and O₂·^- can result in peroxynitrite production, which, in turn, may lead to eNOS uncoupling and production of higher O₂·^- levels. Activated Ras, in the presence of high O₂·^- concentrations, may induce the dnmt's expression and, consequently, increase global DNA methylation. Met, methionine; SAM, S-adenosylmethionine; Cys, cysteine; GSH, glutathione; DDAH, dimethylarginine dimethylaminohydrolase; ADMA, asymmetric dimethylarginine; eNOS, endothelial NO synthase; CH₃, methyl radical; NO, nitric oxide; O₂·^-, superoxide anion; dnmt, DNA methyltransferases. Adapted from Laurent et al. [47] and Nair et al. [48].

Oxidative stress is also commonly associated with relatively high levels of reactive nitrosative species and reactive oxygen nitrogen species [49]. NO is synthesized by a family of enzymes termed NOS. Two of them, the so-called endothelial (eNOS) and neuronal (nNOS) isoforms, are expressed constitutively and generate NO for cell signaling purposes. The inducible isoform releases NO in large amounts during inflammatory or immunologic reactions and is involved in host tissue damage responses. NO production (Figures 2B and 4A) and MDA levels (Figure 2C) are clearly amplified in deadherent melan-a cells (D1–D24h), compared to their adherent counterparts (D0). In addition, anchorage blockade induced eNOS expression by melan-a cells (Figure 2F), but not nNOS expression (data not shown).

Intracellular Hcy concentration was shown to fall drastically in melan-a submitted to anchorage blockade for 24 hours (Figure 2D). Interestingly, the concentration of Cys, a product of Hcy metabolism and a GSH precursor, did not change after anchorage blockade (Figure 2E). The observed low levels of intracellular Hcy suggest that both metabolic pathways in which it is involved (transsulfuration and remethylation) could be overly active. DNA molecules are one of the possible targets of methylation reactions [50] and elevated plasma levels of Hcy are associated with DNA hypomethylation [51]. Our results show that global DNA methylation, estimated by 5-MeC content, is clearly augmented in the first hours after melan-a detachment (Figure 3A; CTRL) when the Hcy level is significantly decreased (Figure 2D). The expression of both dnmt1 and 3b were also increased along melan-a anchorage blockade (Figure 3, B–D), providing a functional explanation for the observed rise in 5-MeC content.

Some groups have shown that both ROS and NO can affect DNA methylation status [15,52–54]. In our work, treatment of melan-a cells with l-NAME or NAC, but not with other antioxidants (DMSO, catalase, and peroxidase), resulted in inhibition of 5-MeC content (Figure 3A), as well as in dnmt1 and 3b expression increase (Figure 3C) along melan-a anchorage blockade. This effect does not seem to evolve NO production because both l-NAME and NAC were unable to inhibit NO synthesis (Figure 4A), but rather seems related to O₂·^- levels, whose production was abrogated by these inhibitors (Figure 4B). Absence of nitric oxide production inhibition by a fairly known NOS inhibitor l-NAME has already been described previously in melanoma cells [40], as well as in other cell types [41]. As an l-arginine analog, l-NAME itself seems to be a source of nonenzymatically produced NO [41], which could explain the observed maintenance of NO levels after l-NAME addition to culture media. Hydrogen peroxide scavenger antioxidants, such as DMSO (Figure 3A), catalase, and peroxidase (data not shown), did not alter either 5-MeC, dnmt1 and 3b expression or O₂·^- production. Conversely, NAC can specifically induce the expression [55] and the activity [56] of manganese superoxide dismutase (MnSOD), apart from its role as a precursor of GSH, which could explain its effects on reducing O₂·^- levels in melan-a cells submitted to anchorage blockade.

A very interesting feature of the NOS enzymes is that they not only generate NO but also produce O₂·^- themselves [57–59]. This phenomenon is referred to as the uncoupled state of NOS and has been associated with risk factors for some pathologies and considered as an abnormality of NOS function [60,61]. As mentioned above, eNOS expression was significantly induced during melan-a anchorage blockade (Figure 2F), as well as superoxide anion production (Figure 1). The exposure of eNOS to oxidants, including peroxynitrite, may cause increased enzymatic uncoupling and generation of superoxide anion [62] (Figure 5). Exogenous NOS inhibitor l-NAME can impair the transfer of electrons to molecular oxygen, inhibiting O₂·^- production [58], which could explain the abrogation of O₂·^- production in our model (Figure 5). In a similar way, asymmetric dimethylarginine (ADMA), an endogenous NOS inhibitor, can regulate the balance of NO and O₂·^- production from NOS. ADMA is indirectly stimulated by Hcy [63] (Figure 5) and elevated plasma levels of the latter are associated with DNA hypomethylation [51]. In this context, lower Hcy concentrations could be indirectly related to NOS uncoupling and increased O₂·^- production, which, in turn, might directly contribute to DNA methylation regulation through modulation of dnmt's expression (Figure 5).

In our model, Ras activation is strongly induced few hours after melan-a anchorage blockade (Machado, J. Jr., personal communication). Aberrant activation of this oncogene has been implicated in many aspects of the malignant phenotype, including uncontrolled proliferation and functional and morphologic alterations [64,65]. MacLeod et al. [66] showed that DNA methylation may be regulated by the Ras signaling pathway by activating the Jun transcription factor, which in turn transactivates the dnmt promoter by interacting with AP-1 sites. The expression of either dnmt1, 3a, or 3b has already been shown to be upregulated by activated Ras in different experimental conditions [67–69] (Figure 5). Interestingly, Sephashvili et al. [70] demonstrated that l-NAME increases the production of S-adenosylhomocysteine (SAH), leading to the reduction in SAM/SAH ratio and in methylation reactions, only in cells presenting mutated oncogenic RasH, possibly through enhancement of superoxides. Otherwise, these authors showed that l-NAME-treated wild-type Ras-expressing cells present increased SAM/SAH ratio, indicative of augmented methylation reactions. In our work, increased 5-MeC content and dnmt1 and 3b expression were abrogated by l-NAME during melan-a anchorage blockade, when Ras is activated, suggesting that this dnmt's upregulated expression may be related to Ras activation status. Curiously, l-NAME treatment of adherent melan-a cells, which present nonactivated Ras, leads to increases both in global DNA methylation and in dnmt3b, but not dnmt1 expression. This fact may be related to the experimental conditions, because dnmt1, and not dnmt3b expression, is cell cycle-dependent. In most studies, the expression of dnmts induced by Ras has been found to depend on the cell type and context [67]. The impact of Ras activation status and oxidative stress in DNA methylation is under intensive investigation in our laboratory.

Anchorage-independent growth (anoikis resistance) has traditionally been described as an acquired in vitro characteristic, which is well correlated with in vivo tumorigenic capacity [71,72] but the acquisition of this phenotype is not considered a causative event in carcinogenesis. Nevertheless, using basically the same deadhesion protocol used in our work, Rak et al. [73] obtained tumorigenic variants after enforced anchorage-independent growth of a nontumorigenic immortalized epithelial cell line. Using a similar approach, Zhu et al. [74] obtained a melanoma cell line with greater metastatic potential compared to the parental cell line. More interestingly, Seftor et al. [75] showed that adhesion of normal human melanocytes to a modified extracellular matrix could induce the expression of several genes associated with a malignant phenotype. This cancer-associated expression profile was reverted after several days in physiological adherent conditions, and the authors suggest that melanocyte phenotype is controlled by epigenetic mechanisms that include DNA methylation.

In our model of melanocyte transformation, cell-substrate adhesion blockade is the only induced modification in culture conditions, which resulted in full malignant transformed cell lines after several anchorage blockade cycles [22]. The first hours of this process, as described in this work, present a significant increase in oxidative/nitrosative stress, and this metabolic imbalance may have a causal role in modifications of DNA methylation status and dnmt1 and 3b expression, as shown by l-NAME and NAC inhibition assays. These results begin to unveil important clues about the initial events in the carcinogenic process related to microenvironmental changes.

Abbreviations

ADMA

asymmetric dimethylarginine

DAF-2DA

diaminofluorescein-2 diacetate

DHE

dihydroethidium

Dnmt

DNA methyltransferase

GSH

glutathione

Hcy

homocysteine

l-NAME

L^G-nitro-l-arginine methyl esther

MTT

methyl thiazol tetrazolium

NAC

N-acetyl-l-cysteine

NO

nitric oxide

NOS

nitric oxide synthase

PCR

polymerase chain reaction

ROS

reactive oxygen species

SAM

S-adenosylmethionine

SAH

S-adenosylhomocysteine