Introduction

It has been suggested that reactive oxygen species (ROS) participate in tumor progression by promoting DNA damage and/or altering cellular signaling pathways [1]. ROS are known to modulate growth signals and to regulate gene expression, leading to the sustained proliferation of cancer cells [2]. DNA methylation is the most frequent epigenetic alteration seen in mammalian genomes, and it often mediates transcriptional repression [3, 4]. Recently, evidence has emerged that both genetic and epigenetic changes underlie carcinogenesis [5] and, in particular, hypermethylation of CpG islands in promoter regions of tumor suppressor genes has been commonly observed in various cancers [6, 7]. Oxidative stress due to metabolic, dietary, and environmental factors leads to the excessive production of ROS. ROS production is associated with increased DNA damage including base modifications, deletions, strand breakage, and chromosomal rearrangements [8, 9] and chromosomal alteration with both hyper- and hypomethylation of DNA [10].

The runt-domain transcription factor 3 (RUNX3) is known to be a tumor suppressor involved in various cancers, including gastric cancer [11–14]. RUNX3 is necessary for gastric epithelial growth [15], neurogenesis of dorsal root ganglia [16, 17], and T-cell differentiation [18], and RUNX3 knockout mice exhibited gastric epithelial hyperplasia, reduced apoptosis, and sensitivity to transforming growth factor (TGF)-β [15]. Approximately 45% to 60% of human gastric cancers were reported to display loss of RUNX3 expression due to hemizygous deletion and promoter hypermethylation [15], while chronic gastritis and gastric adenomas showed RUNX3 hypermethylation, indicating a role for RUNX3 in gastric cancer progression [19]. Thus, an evaluation of the RUNX3 methylation status in normal gastric mucosa may provide an estimate of the risk for gastric carcinogenesis [20]. Besides hypermethylation, it was reported that RUNX3 inactivation also occurs in response to the mislocalization of the RUNX3 protein from the nucleus to the cytoplasm [21–25]. Ito et al. showed that when the RUNX3 protein was translocated to the nucleus in response to TGF-β, it acted as tumor suppressor; however, cytoplasmic RUNX3 protein did not elicit tumor suppressive activity, suggesting that it had been inactivated [21]. The gastrointestinal tract, especially the colon and rectum, is constantly exposed to ROS originating from endogenous and exogenous sources [26]. A role for ROS in colorectal cancer remains speculative, but numerous epidemiological studies suggest that oxidative stress may be important in cancer initiation and progression [14, 27–30].

In this study, we investigated whether oxidative stress can regulate the expression of the tumor suppressor gene, RUNX3, in colorectal cancer cells. The data showed that RUNX3 silencing by oxidative stress is mediated by epigenetic DNA methylation and suggested that it may play a role during the development of colorectal cancer.

Materials and methods

Cell culture

Human colorectal cancer cell lines, HT-29, SW-403, SNU-407, SNU-1033, and Caco-2, from the Korean Cell Line Bank (Seoul, Republic of Korea) and normal human colon FHC cells from the American Type Culture Collection (Rockville, MD, USA) were maintained at 37°C in an incubator with a humidified atmosphere of 5% CO2. HT-29, SW-403, SNU-407, and SNU-1033 cells were cultured in RPMI 1640 medium containing 10% fetal calf serum (FCS), streptomycin (100 μg/ml), and penicillin (100 U/ml). Caco-2 cells were cultured in minimum essential medium containing 10% FCS, streptomycin, and penicillin. FHC cells were cultured in a 1:1 mixture of Ham's F12 and Dulbecco's modified Eagle's medium containing HEPES (25 mM), cholera toxin (10 ng/ml), insulin (5 μg/ml), transferrin (5 μg/ml), hydrocortisone (100 ng/ml), and 10% FCS.

RNA isolation and reverse transcription-polymerase chain reaction

Total RNA was isolated from cells using Trizol reagent (GibcoBRL, Grand Island, NY, USA), and cDNA was amplified using 1 ml of the reverse-transcription reaction buffer, primers, dNTPs, and 0.5 U of Taq DNA polymerase in a final volume of 25 ml as described previously [31]. The polymerase chain reaction (PCR) conditions consisted of 5 min at 94°C for the initial denaturation, followed by 35 cycles of 1 min at 94°C, 1 min at 55°C, 1 min at 72°C, and a final elongation step of 7 min at 72°C. The following primers were used to amplify the RUNX3 cDNA [32]: sense, 5′-GGCAATGACGAGAACTAC-3′ (located in exon 2); antisense, 5′-GGAGAATGGGTTCAGTTC-3′ (located in exon 5). The amplified products were resolved by 1% agarose gel electrophoresis, stained with ethidium bromide, and photographed under UV light using Image Quant™ TL analysis software (Amersham Bioscience, Sweden).

Western blot analysis

Nuclear extracts were prepared using a nuclear protein extraction kit (Cayman Chemical, Ann Arbor, MI, USA), and nuclear extracts were lysed on ice with 1 ml of lysis buffer (10 mM Tris–HCl, pH 7.9, 10 mM NaCl, 3 mM MgCl2, and 1% NP-40) for 4 min. After centrifugation for 10 min at 3,000 × g, the pellets were re-suspended in 50μl of extraction buffer (20 mM HEPES, pH 7.9, 20% glycerol, 1.5 mM MgCl2, 0.2 mM EDTA, 1 mM DTT, and 1 mM PMSF), incubated on ice for 30 min, and then centrifuged at 13,000 × g for 5 min. The supernatants were stored at −70°C after determination of the protein concentration. Aliquots of the lysates (40 μg of protein) were boiled for 5 min and were electrophoresed on 10% SDS-polyacrylamide gel. Blots of the gels were transferred to nitrocellulose membranes, which were subsequently incubated with primary antibodies, and then with secondary immunoglobulin-G-horseradish peroxidase conjugates (Pierce, Rockford, IL, USA). Protein bands were detected using an enhanced chemiluminescence Western blotting detection kit (Amersham, Little Chalfont, Buckinghamshire, UK). The protein bands were visualized using a luminescent image analyzer.

Quantitative real-time PCR

cDNA synthesis was performed with 2 μg RNA using a Superscript kit (Invitrogen, Carlsbad, CA, USA). Quantitative real-time PCR was performed with 2× SYBR Green Mastermix (Invitrogen, Carlsbad, CA, USA) and 900 nM primers for RUNX3 and 18 s rRNA. The thermal cycling parameters were as follows: an initial cycle of Taq activation for 1 min at 94°C, 1 min at 55°C, and 1 min at 72°C, optimized annealing temperature 5 s at 58–60°C, 30 s at 72°C, and a detection step of 8 s at 80°C. Reactions were performed in duplicate and the specificity was monitored using melting curve analysis after cycling. The following primers were designed by Bioneer Co. Ltd. (Seoul, Republic of Korea) (5′–3′) 18 s rRNA, 5′-CAGCCACCCCAGATTGAGCA-3′ and 5′-TAGTAGCGACGGGCGGTGTG-3′; (5′–3′) RUNX3, 5′-GACAGCCCCAACTTCCTCT-3′ and 5′-CACAGTCACCACCGTACCAT-3′. All data were tested first for normality and data with a non-normal distribution were subjected to square root transformation prior to statistical analyses. The comparative cycle threshold (C t) method was used to calculate the relative changes in gene expression in the iQ5 real-time PCR system (Bio-Rad Laboratories, Hercules, CA, USA). The 2-delta-delta C t value was calculated after 18 s RNA normalization [33].

Methylation-specific PCR

DNA bisulfate modification was carried out with the Methylamp™ DNA modification kit (Epigentek, Pittsburgh, PA, USA) according to the manufacturer’s instructions. To analyze RUNX3 DNA methylation, methylation-specific (MS)-PCR was performed using an Epitect MSP kit (Qiagen, Valencia, CA, USA). The methylated and unmethylated RUNX3 primer sets were as follows [15]: unmethylated RUNX3, 5′-TTATGAGGGGTGGTTGTATGTGGG-3′ and 5′-AAAACAACCAACACAAACACCTCC-3′; methylated RUNX3, 5′-TTACGAGGGGCGGTCGTACGCGGG-3′ and 5′-AAAACGACCGACGCGAACGCCTCC-3′. The PCR products were separated on 6% non-denaturing polyacrylamide gels, stained with ethidium bromide, and visualized under UV light.

DNA methyltransferase activity

DNA methyltransferase (DNMT) activity in nuclear extract was detected using a colorimetric assay with the EpiQuik DNMT activity assay kit (Epigentek). In this assay, the cytosine-rich DNA substrate is stably coated on the strip wells, and DNMT transfers a methyl group from S-denosylmethionine to a cytosine in the DNA substrate. The methylated DNA can be recognized using a 5-methylcytosine antibody. The levels of methylated DNA, which are proportional to the enzymatic activity, were then colorimetrically quantified. The results were expressed in absorbance units at 450 nm and were represented as a percentage of the activity.

Histone deacetylase (HDAC) activity

Histone deacetylase (HDAC) activity in nuclear extract was measured using the EpiQuik HDAC activity assay kit (Epigentek). The nuclear extracts were incubated with a specific substrate for 1 h at 37°C, followed by the capture antibody for 60 min, and then the detection antibody for 30 min at room temperature. Absorbance was determined using a microplate spectrophotometer at 450 nm and was represented as a percentage of the activity.

Immunoprecipitation and Western blot for detection of HDAC1

DNMT1 was immunoprecipitated from the nuclear extracts using a DNMT1 antibody. Immunoprecipitated complex was collected with protein G beads and washed with immunoprecipitation buffer. Equal amounts of precipitates were run on SDS–polyacrylamide gel and Western blotting was performed with an HDAC1 antibody.

Chromatin immunoprecipitation assay

Chromatin immunoprecipitation (ChIP) assay was performed using the SimpleChIP™ enzymatic chromatin IP kit (Cell Signaling Technology, Danvers, MA, USA) according to the manufacturer’s protocol with slight modifications. Cells were pretreated with 1 mM NAC for 1 h, with H2O2 treatment for 48 h and cross-linked by addition of 1% formaldehyde. Chromatin was prepared and digested with nuclease for 12 min at 37°C. ChIP assay was performed with a rabbit DNMT1 polyclonal antibody (Abcam, Cambridge, MA, USA) and normal rabbit IgG. The antibodies were added to the chromatin digests and incubated with constant rotation overnight at 4°C. ChIP-grade protein G magnetic beads were added to capture the immunoprecipitated complexes. The beads were washed and the immunoprecipitates eluted with ChIP elution buffer. The cross-links were reversed by incubation at 65°C for 30 min. Proteinase K was added and incubated at 65°C for 2 h. The immunoprecipitated DNA fragments were then purified using spin columns, and the DNA recovered from the immune-precipitated complex was subjected to PCR with 35 cycles. The promoter region of RUNX3 gene in ChIP assay exists between −1618 and −1307 of RUNX3 gene sequence. The primers for the RUNX3 gene promoter were: forward, 5′-GGTTGCAGAAGTCACAGG-3′; reverse, 5′-AATTTGCTTAGAACGTCCG-3′ [34, 35]. PCR products were separated on 2% agarose gels and DNA bands were visualized using the Image program (NIH, Bethesda, MD, USA).

Immunocytochemistry

Cells plated on coverslips were fixed in 4% paraformaldehyde for 30 min and permeabilized with 0.1% Triton X-100 in PBS for 2.5 min. Cells were treated with blocking medium (3% bovine serum albumin in PBS) for 1 h and incubated with RUNX3 antibody diluted in blocking medium for 2 h. The primary RUNX3 antibody was detected by a 1:500 dilution of the FITC-conjugated secondary antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA) for 1 h. After washing with PBS, the stained cells were mounted on microscope slides in the mounting medium with DAPI (Vector, Burlingame, CA, USA). Images were collected using the LSM 510 program on a Zeiss confocal microscope.

Colony formation assay

Cells were seeded into 60-mm dishes at a density to produce approximately 200 colonies per dish. After treatment of H2O2 at 100 μM, the cultures were further incubated for 10 days. The resultant colonies were fixed with 75% ethanol and 25% acetic acid and stained with trypan blue. The colonies containing 50 or more cells were considered viable.

Statistical analysis

All measurements were performed in triplicate and all values are represented as the mean ± standard error of the mean. The results were subjected to analysis of variance using Tukey’s test to analyze differences. P < 0.05 was considered significant.

Results

RUNX3 expression in colorectal cancer cell lines

RUNX3 mRNA and protein expression levels were analyzed by RT-PCR and Western blotting in five colorectal cancer cell lines: HT-29, SW-403, SNU-407, SNU-1033, and Caco-2. High levels of RUNX3 expression were observed in all cell lines except for HT-29 (Fig. 1a, b). SNU-407, which showed the highest level of RUNX3 expression, was selected for further study.

Fig. 1
figure 1

Expression of RUNX3 in colorectal cancer cell lines. a RUNX3 mRNA level was analyzed by RT-PCR in the colorectal cancer cell lines, HT-29, SW-403, SNU-407, SNU-1033 and CaCo-2. b Nuclear extracts were electrophoresed and the cell lysates were immunoblotted using a RUNX3 antibody

Full size image

ROS down-regulated RUNX3 expression in colorectal cancer cell lines

Cells were treated with 100 μM of H2O2 during the 48-h incubation period and H2O2 (100 μM), for 48 h, did not show a cytotoxic effect to the SNU-407 cells. However, H2O2 treatment produced a significant decrease in RUNX3 mRNA and protein levels as demonstrated by RT-PCR, real-time PCR, and Western blot analysis (Fig. 2a–c). NAC pretreatment in SNU-407 cells prevented the decreased RUNX3 expression by H2O2 treatment (Fig. 2d, e), suggesting that ROS are involved in regulating the tumor suppressor gene, RUNX3. In order to determine whether ROS-down-regulated RUNX3 expression is specific to SNU-407 cells, the mRNA and protein expression level of RUNX3 in SNU-1033 cells were also detected by RT-PCR, real-time PCR, and Western blot analysis. A similar pattern was observed in SNU-1033 cells, indicating that ROS-down-regulated RUNX3 expression is not unique to SNU-407 cells (Fig. 3).

Fig. 2
figure 2

Oxidative stress regulates RUNX3 expression in SNU-407 cells. Cells were treated with 100 μM H2O2 during incubation for 48 h. Expression of RUNX3 following H2O2 treatment was assessed using a RT-PCR, b real-time PCR, and c Western blotting. Asterisk indicates significant difference from control (p < 0.05). Cells were pretreated with 1 mM NAC and 1 h later 100 μM of H2O2 was added and cells were incubated for 48 h. RUNX3 mRNA and protein expression were assessed using d RT-PCR and e Western blotting

Full size image
Fig. 3
figure 3

ROS-down-regulated RUNX3 expression is not unique to SNU-407 cells. SNU-1033 cells were treated with 100 μM H2O2 during incubation for 48 h. Expression of RUNX3 following H2O2 treatment was assessed using a RT-PCR, b real-time PCR, and c Western blotting. Asterisk indicates significant difference from control (p < 0.05)

Full size image

ROS-induced methylation of the RUNX3 promoter

H2O2-treated cells showed methylation of the RUNX3 promoter as indicated by MS-PCR analysis, while pretreatment with the antioxidant, NAC, blocked H2O2-induced methylation (Fig. 4a). Pretreatment with the DNA methyltransferase inhibitor, 5-Aza-dC, decreased the level of RUNX3 promoter methylation induced by H2O2 treatment (Fig. 4b).

Fig. 4
figure 4

Oxidative stress induces methylation of the RUNX3 Promoter. Cells were pretreated with a 1 mM NAC or b 10 μM 5-Aza-dC and, 1 h later, exposed to H2O2 for 48 h. MS-PCR analysis of RUNX3 was performed on genomic DNA isolated from SNU-407 cells. Methylation was detected by the presence of a PCR product amplified by methylation-specific primers. Demethylation was detected by the presence of a PCR product amplified with unmethylation-specific primers. M methylated, U unmethylated DNA

Full size image

ROS-induced methylation-related proteins

H2O2-treated cells induced the expression and increased the activity of the methylation-related proteins, HDAC1 and DNMT1, in a time-dependent manner (Fig. 5a–c). In addition, H2O2 treatment increased the binding of HDAC1 to DNMT1 (Fig. 5d). The ChIP assay data demonstrated that H2O2 treatment increased the recruitment of DNMT1 to the RUNX3 promoter (Fig. 5e).

Fig. 5
figure 5

Oxidative stress induces methylation of the RUNX3 promoter by activation of methylation-related proteins. a Western blotting analysis was performed with HDAC1 and DNMT1 antibodies. b DNMT activity was detected using a DNMT activity assay kit. Asterisk indicates significant difference from control (p < 0.05). c HDAC activity was measured using the HDAC activity assay kit. Asterisk indicates significant difference from control (p < 0.05). d Cell lysates were immunoprecipitated with a DNMT1 antibody, and precipitates were Western-blotted with a HDAC1 antibody. e ChIP analysis was performed with a DNMT1 antibody and primers amplifying the RUNX3 promoter region; the band shows the level of DNMT1 in the RUNX3 promoter region. Input represents amplification of total DNA from whole cell lysate. DNMT1 level in the RUNX3 promoter region was quantified and presented as an index of the intensity. Asterisk indicates significant difference from control (p < 0.05) and two asterisks indicate significant difference from H2O2 treatment (p < 0.05)

Full size image

Cytosine methylation inhibitor prevented the reduction in RUNX3 expression due to ROS

5-Aza-dC treatment prevented the levels of RUNX3 mRNA and protein reduced by H2O2 treatment (Fig. 6a, b). These results suggested that oxidative stress decreased the expression of the tumor suppressor gene, RUNX3, via DNA methylation, and that this epigenetic change is associated with RUNX3 inactivation.

Fig. 6
figure 6

Effect of a DNA methylation inhibitor on RUNX3 expression under oxidative stress. Cells were pretreated with 10 μM 5-Aza-dC and, 1 h later, exposed to H2O2 for 72 h. a The RUNX3 mRNA expression was assessed using RT-PCR and b RUNX3 protein expression was analyzed by Western blotting

Full size image

ROS-induced inactivation of RUNX3 by mislocalization

The mislocalization of the nuclear RUNX3 protein to the cytoplasm was observed initially in gastric cancers [21], breast cancers [22, 23], colorectal polyps [24], and colorectal cancers [25, 36]. As shown in Fig. 7a, b, H2O2-treated cells exhibited increased levels of RUNX3 protein in the cytosol and decreased levels in the nucleus, while NAC pretreatment reduced the loss of RUNX3 from nucleus to cytosol in H2O2-treated cells, indicating that H2O2 treatment resulted in the mislocalization of RUNX3 from the nucleus to cytosol.

Fig. 7
figure 7

Oxidative stress induces mislocalization of RUNX3. Cells were pretreated with 1 mM NAC and, 1 h later, exposed to H2O2 for 48 h. a Confocal images of RUNX3 antibody and FITC-conjugated secondary antibody staining indicate the location of RUNX3 protein (green); DAPI staining indicates the nucleus (blue). The merged images of H2O2-treated cells reveal the cytosolic location of RUNX3 protein. b Nuclear and cytoplasmic fractions were extracted and quantified. Extracts were electrophoresed and immunoblotted using a RUNX3 antibody

Full size image

The influence of down-regulation of RUNX3 expression by ROS

To investigate whether the decrease in RUNX3 expression by ROS could influence cell phenotype, cell proliferation rate was examined by colony formation. As shown in Fig. 8a, b, H2O2 significantly down-regulated RUNX3 expression and increased cell proliferation. To further investigate whether the decrease in RUNX3 expression by ROS is reversible, cells were treated with 100 μM H2O2 for 24 h, and then the culture medium was changed. After 48 h, RUNX3 expression was detected by Western blotting. H2O2 eventually suppressed RUNX3 expression, indicating that H2O2-induced down-regulation of RUNX3 is irreversible (Fig. 8c).

Fig. 8
figure 8

The influence of down-regulation of RUNX3 expression by ROS on cell proliferation. a Cells were treated with 100 μM H2O2 for the indicated time and RUNX3 expression was assessed by Western blotting. b After treatment with H2O2, cells were allowed to grow for 10 days. The resulting colonies were stained with trypan blue and counted. Asterisk indicates significant difference from control (p < 0.05). c Cells were treated with 100 μM H2O2 for 24 h, and then the culture medium was changed. After 48 h, RUNX3 expression was detected by Western blotting

Full size image

Discussion

An imbalance in the production/removal of ROS can be either directly or indirectly involved in the initiation, promotion, and progression phases of carcinogenesis [37]. ROS may cause damage to DNA and chromosomes, induce epigenetic alterations such as hypermethylation of DNA or histone modification, interact with oncogenes or tumor suppressor genes, and impart changes to immunological mechanisms [10, 38, 39]. The tumor suppressor, RUNX3, is frequently inactivated in gastric cancer tissues [15, 21] and its aberrant activity is closely related to metastatic outcome [40]. It has been postulated that the mechanism for RUNX3 inactivation in cancer cells and tissues involves DNA hypermethylation in the promoter region [15, 41]. Abnormal de novo methylation of CpG islands occurs frequently in various cancers, including colorectal cancers; moreover, hypermethylation of CpG islands is associated with transcriptional inactivation [42–50]. The RUNX3 gene is located on human chromosome 1p36 [15], a region in which many genes play roles in the maintenance of chromosome stability, suppression of tumorigenesis, control of apoptosis, and DNA methylation [51, 52]. Deletions in the 1p36 region are common in colorectal cancers [51–55], suggesting that gene loss in this region might be implicated in chromosome instability [52].

In our study, all of the human colorectal cancer cell lines SW-403, SNU-407, SNU-1033, and Caco-2, except for HT-29, showed expression of RUNX3 mRNA and protein. A previous report confirmed that the RUNX3 promoter was methylated and RUNX3 was not expressed in the cell line HT-29 [32]. Our data showed that RUNX3 expression was down-regulated in response to H2O2 treatment in SNU-407 cells and that this down-regulation was abolished with the ROS scavenger, NAC. MS-PCR data revealed that oxidative stress affected the methylation of the RUNX3 promoter and that treatment with the DNA methylation inhibitor, 5-Aza-dC, resulted in the re-expression of RUNX3, suggesting that an epigenetic regulatory mechanism may be involved in RUNX3 silencing by ROS-induced methylation. DNA methylation catalyzed by DNMTs and histone deacetylation catalyzed by HDACs play an important role in the regulation of gene expression during carcinogenesis. Considerable evidence has accumulated in the elucidation of the molecular mechanisms in which DNA methylation is involved in tumor suppressor gene silencing [56].

The methylation of CpG dinucleotides is catalyzed by DNMTs, which transfer the methyl moiety from the methyl donor, S-adenosylmethionine, to the 5′ position of the cytosine ring. DNMT1 has a preference for methylating hemimethylated DNA and is referred to as the primary maintenance DNMT [57, 58]. HDAC and DNMT1 cooperate to initiate and sustain epigenetic gene silencing [59]. In vivo studies have shown that DNMT1 associates with HDAC1 to deacetylate chromatin and silence gene transcription [60, 61]. In support of these data, our results showed that H2O2 treatment resulted in DNMT1 and HDAC1 up-regulation with increased expression, increased activity, increased binding of DNMT1 to HADC1, and increased DNMT1 binding to the RUNX3 promoter. In addition, 5-Aza-dC treatment prevented the decrease in RUNX3 mRNA and protein levels by H2O2 treatment. The mislocalization of active nuclear RUNX3 protein to the cytoplasm has been observed in various cancers [21–26]. In the present study, H2O2 treatment inhibited the nuclear localization and expression of RUNX3, and this inhibition was abolished by NAC pretreatment.

Taken together, the data presented in this study suggest that oxidative stress may play an important role in inhibiting the activation of the tumor suppressor, RUNX3, in human colorectal cancer cells, via promoter methylation, and that this epigenetic regulation may be associated with the progression of colorectal cancer.