β-Catenin gene promoter hypermethylation by reactive oxygen species correlates with the migratory and invasive potentials of colon cancer cells

Abstract

Purpose

Over half of the colon cancer patients suffer from cancer-related events, mainly metastasis. Loss of β-catenin activity has previously been found to facilitate cancer cell dissociation and migration. Here, we aimed to investigate whether epigenetic silencing of β-catenin induces human colon cancer cell migration and/or invasion.

Methods

HCT-116, Caco-2, HT-29 and SW620 cell migration and invasion capacities were assessed using scratch wound healing and Matrigel invasion assays, respectively. Confocal microscopy, qRT-PCR and Western blotting were performed to determine gene expression levels, whereas methylation-specific quantitative real-time PCR was used to assess the extent of β-catenin gene (CTNNB1) promoter methylation after treatment of the cells with TPA, hydrogen peroxide, 5-aza-2′-deoxycytidine and/or VAS2870.

Results

We found that treatment of HT-29 and Caco-2 cells (differentiated and low metastatic) with 12-O-tetradecanoyl phorbol-13-acetate (TPA; a tumor promoter) suppressed E-cadherin and β-catenin expression at both the mRNA and protein levels and, in addition, enhanced cell migration. Furthermore, we found that the CTNNB1 gene promoter methylation levels were higher in the more invasive HCT-116 and SW620 colon cancer cells than in HT-29 and CCD-841 (normal colon epithelial) cells. We also found that TPA or hydrogen peroxide induced CTNNB1 gene promoter methylation to a higher extent in HT-29 and CCD-841 cells than in HCT-116 and SW620 cells, and that the degree of CTNNB1 gene promoter methylation positively correlated with cell dissociation and migration. In addition, we found that co-treatment with 5-aza-2′-deoxycytidine (decitabine, a DNA methyl transferase inhibitor) and VAS2870 (a NADPH oxidase inhibitor) almost completely blocked the invasion of TPA-treated HT-29 and TPA-untreated HCT-116 and SW620 cells, and that these inhibitions surpassed those of the cells treated with decitabine or VAS2870 alone.

Conclusions

From our data we conclude that the extent of CTNNB1 gene promoter methylation by reactive oxygen species correlates with the migratory and invasive abilities of colon cancer cells. Our results suggest that epigenetic regulation of CTNNB1 may serve as a novel avenue to block colon cancer cell migration and invasion.

Introduction

Colon cancer is one of the commonest cancers, and more than half of those affected succumb to cancer-related events, mainly metastasis [1], which is caused by cells detaching from primary tumors and then migrating to surrounding tissues and/or distant sites [2]. These migratory mechanisms are classified as collective or single cell migration [1, 3–5]. Collective migration depends on integrin and cell adhesion molecules, whereas single cell migration, i.e., mesenchymal or amoeboid migration, does not depend on such molecules, but rather on breakage of adherens junctions due to adhesion molecule loss [6, 7].

E-cadherin plays an important role in the maintenance of epithelium integrity, and its loss has been associated with the acquisition of migratory and invasive abilities leading to cancer progression [8]. In most colorectal cancers the APC gene is mutated, and this leads to loss of binding affinity for β-catenin, axin and GSK-3β, which results in high cytoplasmic and nuclear levels of β-catenin and activation of gene transcription and, thus, to increased cell proliferation [9–11]. β-catenin is also associated with the structural integrity of the epithelial layer due to its binding to E-cadherin within adherens junctions [12] and, thus, loss of β-catenin facilitates the dissociation of cancer cells from the initial cell mass and their subsequent migration. β-catenin levels are regulated by the β-catenin destruction complex [13] and by gene silencing through promoter hypermethylation [14, 15], as has been shown by the close correlations found between β-catenin gene (CTNNB1) promoter hypermethylation and metastasis of non-small cell lung cancer and gastric cancer cells [11, 12] . DNA methylation is an epigenetic mechanism that affects gene expression. Hypermethylation of CpG regions of tumor suppressor gene promoters is frequently observed during tumorigenesis [16–19]. Moreover, it has been found that promoter hypermethylation-induced gene silencing may also regulate the expression of genes required for metastasis [20]. In addition, although heritable, epigenetic changes are reversible, which turns them into genuine therapeutic targets [21, 22]. For example, 5-aza-2′-deoxycytidine (decitabine; a DNA methylation inhibitor) has been shown to up-regulate the genes encoding E-cadherin (CDH1), fibronectin (FN1), N-cadherin (CDH2), and others during epithelial-mesenchymal transition (EMT) in non-small cell lung cancer [23]. In addition, low doses of DNA demethylating drugs have been approved for the treatment of myelodysplastic syndrome and leukemia [24].

In a previous study, we showed that 12-O-tetradecanoyl phorbol-13-acetate (TPA; a tumor promoter) can induce an invasive phenotype in HT-29 colorectal carcinoma cells through enhancing matrix metalloproteinase (MMP)-7 expression by increasing NADPH oxidase 2 (NOX2)-derived superoxide levels. Furthermore, in contrast to the up-regulation of MMPs by reactive oxygen species (ROS), E-cadherin expression has been found to be down-regulated by ROS-dependent induction of SNAIL and SLUG [25–28]. In addition to its regulatory effects on transcription factors, ROS may also function as catalysts for DNA methylation [29–31], as exemplified by ROS induced hypermethylation of the CDH1 gene promoter in hepatocellular carcinoma cells [32]. However, although hypermethylation of the CTNNB1 gene promoter results in gene silencing and increased invasion, the exact mechanism of CTNNB1 gene promoter hypermethylation has not been determined.

In the current study, we investigated the involvement of β-catenin and the mechanism underlying its expression during colon cancer cell migration and invasion, and compared these with those of E-cadherin, by using four different colon cancer-derived cell lines: (i) HT-29, which carries mutant (MT) APC and wild type (WT) CTNNB1 genes, and (ii) Caco-2, which carries MT APC and MT CTNNB1 genes; these cell lines are low invasive, and (iii) HCT-116 (WT APC, MT CTNNB1), and (iv) SW620 (MT APC, WT CTNNB1); these cell lines are high invasive [33, 34]. The expression of E-cadherin in the HT-29, Caco-2 and HCT-116 cells is higher than that in SW620 cells. The HT-29, Caco-2 and SW620 cell lines are microsatellite stable and chromosomal instable (CIN), whereas the HCT-116 cell line is microsatellite instable without CIN. The HT-29 and HCT-116 cell lines show a higher CpG island methylator phenotype (CIMP) than the Caco-2 and SW620 cell lines [35].

Materials and methods

Reagents and antibodies

All the reagents, unless otherwise specified, were purchased from Sigma-Aldrich (St. Louis, MO, USA). RPMI-1640 and Dulbecco’s modified Eagle’s medium (DMEM) were purchased from Hyclone (Logan, UT, USA), whereas fetal bovine serum (FBS), penicillin/streptomycin and Trizol reagent were purchased from Invitrogen Life Technologies (Carlsbad, CA, USA). The anti-β-Catenin rabbit monoclonal antibody was purchased from Sigma-Aldrich, and the anti-phospho-β-catenin (S33/37/T41) and anti-E-cadherin mouse monoclonal antibodies were purchased from Cell Signaling Technology (Boston, MA, USA). The anti-β-Actin antibody was purchased from Abcam (Cambridge, MA, USA), and trypsin/EDTA from Clonetics, Inc. (Walkersville, MD, USA).

Cell cultures

The human colorectal cancer-derived cell lines HT-29, HCT-116, Caco-2 and SW620, and the normal human colon cell line CCD-841 were obtained from the American Type Culture Collection (ATCC; Manassas, VA, USA). HT-29 and Caco-2 cells were grown in RPMI-1640 or DMEM respectively, and HCT-116, SW620 and CCD-841 cells were grown in High Glucose DMEM containing 10% FBS. All media contained 100 IU/ml penicillin and 100 μg/ml streptomycin. Cells were maintained in a humidified incubator at 37 °C in 5% CO₂, and when confluent, sub-cultured by splitting them at a ratio of 1:3.

Cell migration assay

Cell migration was assessed using a scratch wound healing assay and a IncuCyte Zoom Live-Cell Imaging System (Essen Bioscience; Ann Arbor, Michigan, USA) according to the manufacturer’s instructions. Briefly, HT-29 cells were seeded in 96-well plates at a density of 20,000 cells per well and incubated at 37 °C in 5% CO₂ for 18 h. Wounds were made in confluent attached cell layers using a wound maker provided by Essen Bioscience. After wounding, media were aspirated and cells were gently washed twice with culture medium. Next, media containing TPA at the indicated concentrations were added after which the plates were placed inside the imaging system and scanned repeatedly for 24 h.

Immunofluorescence microscopy

Cultured cells (1 × 10⁴ cells/dish) were treated with or without TPA (12 ng/ml) for 24 h and fixed with 4% paraformaldehyde in 1× PBS for 30 min. Next, the cells were washed with PBS and incubated with 2% BSA in 1× PBS for another 30 min at room temperature to block non-specific protein binding. After blocking, the cells were washed with PBS three times, incubated with Texas Red-X Phalloidin (Molecular Probes; Rockford, IL, USA) for 30 min and washed three times with 1× PBS. Finally, the cells were stained with DAPI for 1 min, washed with PBS, and images were captured using a fluorescence microscope at 400× magnification (Olympus Corporation; Tokyo, Japan).

Quantitative real-time PCR (qRT-PCR)

HT-29 cells, serum starved overnight, were incubated in the presence or absence of TPA, 5-hydroxytryptamine (5-HT) or hydrogen peroxide (H₂O₂) for up to 24 h. Next, extracted total RNAs were used as templates for cDNA synthesis using a GoScript Reverse Transcription system (#A5001, Promega Corporation; Madison, WI, USA). After this, PCR products were synthesized from cDNA in the presence of 0.5 U Taq DNA polymerase (Takara Bio Inc.; Shiga, Japan), after which mRNA levels were quantified using QuantiTect SYBR Green PCR kits (Qiagen; Hilden, Germany). The internal control GAPDH primers were supplied by Qiagen, whereas the other primers used were obtained from Bioneer (Daejeon, South Korea). The primers used were as follows; CTNNB1 (NCBI reference sequence NM_001098209.1) (sense 5’-GCTGGGACCTTGCATAACCTT-3′ and antisense 5’-ATTTTCACCAGGGCAGGAATG-3′), CDH1 (NCBI reference sequence NM_004360.4) (sense 5’-CCCACCACGTACAAGGGTC-3′ and antisense 5’-CTGGGGTATTGGGGGCATC-3′), vimentin (VIM) (NCBI reference sequence NM_003380.4) (sense 5’-AGCCGAAAACACCCTGCAAT and antisense 5’-CGTTCAAGGTCAAGACGTGC-3′) and CDH2 (NCBI reference sequence NM_001792.4) (sense 5’-CAACTTGCCAGAAAACTCCA and antisense 5’-ATGAAACCGGGCTATCTGCT-3′).

Western blot analysis

After treating the cells for the indicated time periods with TPA or other compounds, total proteins were extracted using radioimmunoprecipitation assay buffer containing a 1× protease inhibitor cocktail. The extracted protein levels were determined using a BCA protein assay kit (Thermo Scientific). Next, the proteins were separated by 10% or 12% SDS-PAGE and transferred to nitrocellulose membranes. To detect unphosphorylated proteins, the membranes were blocked with 5% skim milk in TBS-Tween 20 (TBST), and to detect phosphorylated proteins the membranes were blocked in 5% bovine serum albumin in TBST for 1 h. Next, the membranes were incubated overnight with specific primary antibodies at 4 °C, washed three times with TBST, incubated with a secondary antibody conjugated with horse-radish peroxidase at room temperature for 1 h, washed three times with TBST, and treated with enhanced chemiluminescent reagents (Thermo Scientific). Finally, the proteins were detected using a luminescent image analyzer, LAS-4000 mini (Fujifilm; Tokyo, Japan).

Confocal microscopy

HT-29 cells (5 × 10³ cells/dish) were seeded on confocal dishes, pretreated with or without inhibitor compounds for 1 h, and then treated with TPA or H₂O₂ for 24 h. Next, the cells were fixed with 100% methanol pre-chilled to −20 °C and incubated with 1% BSA (bovine serum albumin) in 1× PBST for 30 min to block non-specific binding. After blocking, the cells were washed three times for 5 min with PBS, incubated with diluted anti-β-catenin or anti-E-cadherin antibody in BSA containing 1% TBST overnight at 4 °C, stained in the dark for 1 h with an Alexa Fluor secondary antibody for 1 h at room temperature, washed, counterstained with 0.5 μg/ml DAPI, and rinsed with PBS. Images were captured using a K1-Fluo biological confocal fluorescence laser scanning microscope (Nanoscope Systems; Daejeon, South Korea).

siRNA-mediated silencing of NOX1, NOX2, SLUG and SNAIL

For gene silencing, HT-29 cells were seeded and transfected on the following day with siRNAs for 48 h (Bioneer Corporation) using a Dharmacon reagent (#T-2004-01, Thermo Scientific). Next, the transfected cells were observed under a confocal microscope and transfection efficiencies were confirmed by Western Blotting.

DNA extraction and quantitative methylation-specific real-time PCR

Genomic DNA was extracted from HT-29 cells treated with or without TPA (12 ng/ml) and H₂O₂ (5 μM) using a QIAamp DNA Mini Kit (Qiagen). The DNA obtained was treated with sodium bisulphite using an Epitect Bisulfite Kit (Qiagen). Next, methylation-specific real-time PCR assays were performed using the following primer pairs as described before [13]: CTNNB1 forward, 5′-GGAAAGGCGCGTCGAGT-3′ and reverse, 5′-TCCCCTATCCCAAACCCG-3′, with the TaqMan probe 5′-6FAM- CGCGCGTTTCCCGAACCG-TAMRA-3′; and β-actin (ACTB) forward, 5′-TGGTGATGGAGGAGGTTTAGTAAGT-3′ and reverse, 5′-AACCAATAAAACCTACTCCTCCCTTAA-3′, with the TaqMan probe 5′-6FAM-ACCACCACCCAACACACAATAACAAACACA-TAMRA-3′. The housekeeping gene ACTB was used to normalize the reaction data. For relative quantifications, the amounts of methylated DNA (percentage of the methylated reference) in the CTNNB1 gene promoter region were normalized to the methylation value of a universal methylated DNA control (Qiagen), which was used as a reference and set at 100%. The percentage methylation versus the methylated reference was defined using the following equation:

$$100\times 2^{\left(\text{sample ACTB}\left(\mathrm{ct}\right)–\text{sample}\mathit{CTNNB}1\left(\mathrm{ct}\right)\right)}/2^{\left(\text{calibrator ACTB}\left(\mathrm{ct}\right)–\text{calibrator}\mathit{CTNNB}1\left(\mathrm{ct}\right)\right)}.$$

Based on a previous study [13], cutoff values ≥ 4% were defined as “methylated” and < 4% as “unmethylated”.

DNA methyl transferase activity assay

DNA methyl transferase activities were assayed using an EpiQuick DNA methyl transferase Activity/Inhibition Assay Kit (EpiGentek; Farmingdale, NY, USA). Nuclear proteins were extracted using the EpiQuick Nuclear Extraction Kit I and protein concentrations in nuclear extracts were measured using a Bradford protein assay.

Scratch wound healing assay

Cell migration was assessed as previously described [36]. Briefly, cells were seeded in 6-well plates, after which confluent monolayers were treated with mitomycin (50 μg/ml) for 30 min and then scratched using a 1 mm wide tip. Next, the monolayers were rinsed with HBSS to remove cell debris and the remaining cells, treated with or without TPA (12 ng/ml), were allowed to migrate. Images were captured using an optical microscope at 100× magnification (Olympus Corporation, Japan).

Invasion assay

Matrigel invasion assays were performed as reported previously [26]. Briefly, HT-29, HCT-116 or SW620 cells treated with or without VAS2870 and decitabine were seeded in transwell inserts coated with Matrigel (BD Biosciences; Bedford, MA, USA). Next, the inserts were placed in 24-well plates containing media with or without TPA for 24 h, after which invading cells were H&E stained and photographed under a microscope at 200× magnification.

Statistical analysis

Student’s one-way ANOVA in Graph Pad Prism version 5.0 (San Diego, CA, USA) was used to determine the significance of intergroup differences. Three independent experiments were performed and results are expressed as mean ± SEM. P-values < 0.05 were considered statistically significant.

Results

TPA-induced decreases in β-catenin and E-cadherin levels correlate with TPA-induced HT-29 single cell migration

To determine the migratory capacity of colon cancer cells during invasion, HT-29 migration was analyzed after treating the cells with TPA, which is known to induce an invasive phenotype in these cells [26]. Using a scratch wound healing assay, we found that TPA induced HT-29 cell migration in a time- and concentration-dependent manner (Fig. 1a), and that this migration corresponded with increases in single cell populations in TPA-treated cells (Fig. 1b). The dissociation of TPA-treated cells from initial colonies and the morphological changes of migrating cells were confirmed using neuclear staining with DAPI and by filamentous actin (F-actin) staining with Texas Red phalloidin (Fig. 1c). In TPA-untreated cells F-actin was found to be present in the outer cells of the colonies, whereas TPA treatment induced cell scattering with F-actin being present in the individual cell membranes.

Fig. 1.

TPA facilitates the migration of HT-29 cells by enhancing cell detachment. a HT-29 cells were seeded in 96-well plates, and after treatment with mitomycin C for 30 min in the absence or presence of TPA, wound healing assays were performed using the IncuCyte Zoom Live-Cell Imaging System. b HT-29 cell morphology changes after TPA treatment captured by phase-contrast microscopy at 200×. c HT-29 cells were seeded in 6-well plates, treated with TPA for 24 h, and stained with DAPI and Texas Red-X phalloidin to detect F-actin and to assess cell morphology changes by fluorescence microscopy at 400×

During single cell detachment from the initial colonies, TPA significantly suppressed the mRNA expression levels of the CDH1 and CTNNB1 genes in HT-29 cells, although the decrease in CTNNB1 expression was greater than that of CDH1 (Fig. 2a). In contrast, we found that TPA did not alter the mRNA expression levels of the mesenchymal marker genes VIM or CDH2. The TPA-induced changes in CTNNB1 and CDH1 mRNA levels were also observed in Caco-2 cells, another non-invasive human colon cancer cell line (Fig. 2b) and CCD-841 cells, a normal human colon epithelial cell line (Fig. 2c). Similar to the mRNA level changes, we found that the TPA-induced reductions in β-catenin protein levels were greater than those of E-cadherin in HT-29 (Fig. 2d), Caco-2 and CCD-841 cells (Fig. 2e). However, in SW620 cells (a metastatic colon cancer cell line) the basal expression levels of β-catenin and E-cadherin were significantly lower than in CCD-841, HT-29 or Caco-2 cells, and these expression levels were not affected by TPA treatment (Fig. 2e). Since E-cadherin levels are known to be repressed by the transcription factors SNAIL and SLUG, we also examined their expression levels in TPA-treated HT-29 cells. We found that the SNAIL and SLUG levels were increased by TPA treatment in a time-dependent manner (Fig. 2d), and that siRNA-mediated knock-down of SNAIL, but not of SLUG, blocked the TPA-induced decrease in E-cadherin expression (Fig. 2f). Although no phosphorylation of the β-catenin protein was detected (which marks β-catenin for proteosomal degradation), we found that TPA time-dependently reduced β-catenin protein levels in the same manner as it reduced its mRNA levels (Fig. 2d). We also found by confocal microscopy that the reductions in β-catenin levels were not associated with its nuclear translocation (Fig. 2g). To further confirm the observation that β-catenin levels decrease by TPA treatment, both cytosolic and nuclear β-catenin levels were examined using Western blot analyses. We found that in the untreated control condition the cytosolic and nuclear β-catenin levels in Caco-2 cells, which carry mutant APC and mutant CTNNB1 genes, was higher than that in HT-29 cells, which carry mutant APC and wild type CTNNB1 genes (Fig. 2h). In both cell lines, we found that treatment with Wnt3α induced nuclear translocation of β-catenin, whereas TPA treatment suppressed the β-catenin levels in both the cytosol and nucleus. We also found that co-treatment with Wnt3α and TPA blocked Wnt3α-induced nuclear translocation of β-catenin. Co-treatment also resulted in an increased expression of MMP-7, which is one of the Wnt targets and a critical molecule associated with colorectal cancer invasion and metastasis, to a larger extent than Wnt3α treatment alone (Fig. 2h).

Fig. 2.

TPA downregulates E-cadherin and β-catenin expression at the mRNA and protein levels in different colon cancer-derived cell lines. a–c Temporal mRNA expression levels in HT-29, Caco-2 and CCD-841 cells treated with TPA (12 ng/ml) analyzed by qRT-PCR. Bar graphs represent the mean ± SEM from three independent experiments. *P < 0.05 vs. untreated controls. d Temporal protein levels in HT-29 cells treated with TPA determined by Western blotting. *P < 0.05 versus untreated controls. e mRNA and protein expression levels of β-catenin and E-cadherin in CCD-841, Caco-2, HT-29 and SW620 cells treated with or without TPA for 12 or 24 h. *P < 0.05 vs. vehicle-treated controls. f HT-29 cells transfected with siRNAs targeting SNAIL and SLUG were treated with TPA for 24 h, and analyzed for β-catenin and E-cadherin expression. *P < 0.05 vs. vehicle-treated control cells (Mock). ^#P < 0.05 vs. NT siRNA-treated cells. g HT-29 cells treated with TPA for 24 h were incubated with anti-β-catenin and anti-E-cadherin antibodies, stained with DAPI and examined under a confocal microscope. h HT-29 and Caco-2 cells were treated with TPA, Wnt3α, or a combination of the two, after which the expression levels of β-catenin and MMP-7 were analyzed in cytosolic, nuclear and total cell extracts. *P < 0.05 vs. vehicle-treated control group of HT-29 cells. ^#P < 0.05 vs. Wnt3α alone-treated group

ROS reduces β-catenin and E-cadherin protein expression

Based on our previous report that TPA-induced NOX2 derived ROS is associated with gene expression changes [26], we next set out to investigate the effect of ROS on CTNNB1 and CDH1 gene expression. We found that in H₂O₂ treated HT-29 cells the mRNA expression patterns of CTNNB1, CDH1, CDH2 and VIM were similar to those observed in TPA treated HT-29 cells (Fig. 3a). After Western blot analysis (Fig. 3b) and confocal microscopy (Fig. 3c), we found that at non-toxic levels, H₂O₂ suppressed both the β-catenin and E-cadherin protein levels in HT-29 cells. We also found that H₂O₂ treatment increased the SNAIL and SLUG protein levels in a similar manner as TPA (Fig. 3b). Furthermore, we found that the TPA-induced decreases in β-catenin and E-cadherin levels were completely blocked by siRNA-mediated NOX2 silencing, but not by by siRNA-mediated NOX1 silencing (Fig. 3d). Subsequent confocal microscopy revealed that the dissociation of cells from cell clusters was also blocked by siRNA-mediated NOX2 silencing (Fig. 3e). In addition, we examined the effect of an endogenous factor, 5-hydroxytryptamine (5-HT), which has previously been shown to suppress E-cadherin expression by generating a short-term ROS flux [27, 37]. We found that 5-HT suppressed the mRNA (Fig. 3f) and protein (Fig. 3g) expression levels of both β-catenin and E-cadherin, but suppressed β-catenin to a lesser extent than E-cadherin.

Fig. 3.

NOX2-derived sustained ROS production reduces the mRNA and protein expression levels of β-catenin and E-cadherin. a–c HT-29 cells were treated with H₂O₂ (5 μM) for the indicated time periods after which qRT-PCR (a), Western blotting (b) and confocal microscopy (c) were performed to detect gene expression changes. *P < 0.05 vs. vehicle-treated controls. d, e HT-29 cells transfected with non-target siRNA (NT) or siRNA sequences specific for NOX1 or NOX2 for 24 h were treated with TPA for 24 h, after which they were subjected to Western blotting (d) and confocal microscopy (e) to assess the expression levels of β-catenin and E-cadherin. *P < 0.05 vs. vehicle-treated control cells (Mock). ^#P < 0.05 vs. NT siRNA-treated cells. f, g HT-29 cells treated with 5-HT (10 μM) were subjected to qRT-PCR (f) or Western blotting (g). *P < 0.05 vs. vehicle-treated control cells

Co-occurrence of CTNNB1 gene promoter methylation by NOX2-derived ROS and the migration of TPA-treated HT-29 cells

Next, we examined whether the TPA-induced down-regulation of β-catenin by TPA was due to epigenetic regulation by treating the cells with DNA methyl transferase (DNMT), histone deacetylase (HDAC) and proteasome inhibitors. We found that TPA-induced down-regulation of β-catenin and E-cadherin expression were inhibited by pre-treating HT-29 cells with decitabine, a DNMT inhibitor, to a similar magnitude as VAS2870 (a NOX2 inhibitor) and vitamin C (an antioxidant). However, pretreatment of HT-29 cells with MG-132 (a proteasome inhibitor) or trichostatin A (TRiA; a HDAC inhibitor) did not result in inhibition of TPA-induced reductions in the expression of β-catenin and E-cadherin (Fig. 4a). In addition, we found that TPA-induced dissociation (Fig. 4b) and migration (Fig. 4c) of HT-29 cells was prevented by decitabine, VAS2870 and vitamin C treatment. Similarly, we found that the DNMT activity in HT-29 cells was increased by TPA or H₂O₂ treatment, and that these increases were inhibited by VAS2870 or vitamin C (Fig. 4d), which suggests an involvement of ROS in DNMT activity in TPA-treated HT-29 cells. In contrast, we found that 5-HT did not enhance DNMT activity (Fig. 4e) in these cells. As expected from the observed relation between DNMT activity and gene promoter methylation, we found that TPA or H₂O₂ enhanced the methylation of the CTNNB1 gene promoter region in a time-dependent manner (Fig. 4e).

Fig. 4.

Hypermethylation of the CTNNB1 gene promoter facilitates the dissociation and migration of TPA-treated cells. a, b HT-29 cells were pretreated with decitabine (DNA methyltransferase inhibitor), MG-132 (proteasome inhibitor), TriA (histone deacetylase inhibitor), VAS2870 (NOX2/4 inhibitor) or Vitamin C (Vit C, an antioxidant) for 1 h and then with TPA for 24 h, after which Western blotting (a) and confocal microscopy (b) were performed. *P < 0.05 vs. vehicle-treated control cells. ^#P < 0.05 vs. TPA-treated cells. c Scratch wound healing assay performed in HT-29 cells treated with MG-132, VAS2870, TriA or decitabine for 1 h prior to TPA treatment for 24 h. d Equal amounts of protein extracts from HT-29 cells treated with 5-HT (10 μM), TPA (12 ng/ml) or H₂O₂ (5 μM) for 24 h in the absence or presence of Vit C, decitabine or VAS2870 were used to determine DNA methyltransferase activities. *P < 0.05 vs. vehicle-treated control cells. ^#P < 0.05 vs. TPA-treated cells. e HT-29 cells were treated with TPA or H₂O₂ for the indicated time periods, after which genomic DNAs were extracted and treated with bisulfite. Methylation-specific qPCR was performed on these DNAs to determine the degree of CTNNB1 gene promoter methylation. *P < 0.05 vs. vehicle-treated control cells

The degree of CTNNB1 gene promoter methylation and the migratory and invasive abilities of colon cancer cells are related

Since cancer cells are known to have different abilities to migrate and invade surrounding tissues, we set out to examine the effect of CTNNB1 gene promoter methylation on the migratory abilities of different colon cancer-derived cells. We found that in the absence of TPA, CCD-841 and HT-29 cells tended to form aggregated clusters, whereas HCT-116 and SW620 cells were loosely aggregated (Fig. 5a). The levels of CTNNB1 gene promoter methylation in normal CCD-841 and HT-29 cancer cells were found to be significantly lower than those in SW620 and HCT-116 cells (Fig. 5b). After treatment with TPA for 24 h, we observed increases in cell dissociation and migration, and these increases were more obvious in CCD-841 and HT-29 cells than in SW620 and HCT-116 cells (Fig. 5a and c). Furthermore, we found that TPA significantly induced CTNNB1 gene promoter methylation in all four cell lines, but more so in CCD-841 and HT-29 cells (Fig. 5b). In addition, we found that the TPA-induced increases in HT-29 cell invasion were significantly inhibited by decitabine and VAS2870, and that co-treatment with decitabine and VAS2870 almost completely blocked TPA-induced HT-29 cell invasion (Fig. 5d). The invasive behavior of HCT-116 and SW620 cells was found to be similarly inhibited by decitabine or VAS2870, and to be synergistically inhibited by decitabine plus VAS2870 (Fig. 5e).

Fig. 5.

The degree of CTNNB1 gene promoter hypermethylation correlates with colon cancer cell dissociation, migration and invasion. a Morphologies of CCD-841, HT-29, SW620 and HCT-116 cells before and after treatment with TPA for 24 h. b Methylation-specific quantitative real-time PCR performed on CCD-841, HT-29, HCT-116 and W620 cells treated with or without TPA for 24 h. *P < 0.05 vs. untreated CCD-841 cells. c Scratches were made in confluent monolayers of CCD-841, HT-29, HCT-116 and SW620 cells, after which the cells were treated with mitomycin C for 30 min and cell migration with or without TPA was measured. Images were captured at 100× using a phase contrast microscope (Olympus Corporation, Japan). d HT-29 cells seeded in matrigel coated inserts were treated with decitabine, VAS2870 or both in the absence or presence of TPA (12 ng/ml) for 24 h. Invading cells were stained with hematoxylin and eosin and photographed at 200×. *P < 0.05 vs. vehicle control cells. ^#P < 0.05 vs. TPA-treated cells, ^$P < 0.05 vs. decitabine-treated cells. ^&P < 0.05 vs. VAS2870-treated cells. e Matrigel invasion assay of HCT-116 and SW620 cells treated with decitabine, VAS2870 or both. After 24 h, invading cells were stained and counted. *P < 0.05 vs. untreated control cells, ^#P < 0.05 vs. decitabine-treated cells. ^$P < 0.05 vs. VAS2870-treated cells

Discussion

Mutations of the CTNNB1 and/or APC genes in colon cancer result in β-catenin accumulation and nuclear translocation which, in turn, enhances the transcription activity of β-catenin-T-cell factor (Tcf) and the expressions of its target genes, which govern cell growth, invasion and metastasis [9, 38, 39]. Next to this transcription regulatory role, β-catenin also constitutes an integral part of adherens junctions together with E-cadherin and, as such, it plays a crucial role in the maintenance of epithelial integrity. Down-regulation or functional loss of the E-cadherin/β-catenin complex negatively impacts intercellular adhesion which, as a consequence, may result in cell migration and tumor progression [40]. Here we show that similar to E-cadherin, β-catenin down-regulation is closely associated with colon cancer migration and invasion. During epithelial to mesenchymal transition (EMT) mesenchymal markers (such as N-cadherin and vimentin) are upregulated and epithelial markers (such as E-cadherin) are downregulated [41]. In contrast, we found that TPA-induced colon cancer cell migration and invasion led to expression inhibition of β-catenin and E-cadherin, whereas that of N-cadherin and vimentin remained unchanged. Based on these molecular changes and the rounded cell morphology and cortical localization of F-actin observed, we conclude that the migratory pattern of TPA-treated HT-29 colon cancer cells is amoeboid rather than mesenchymal.

E-cadherin loss can be achieved through trans-repression via transcription factors and/or though gene promoter hypermethylation. It has previously been reported that induction of the transcription factors SNAIL and SLUG, which are linked to EMT, may lead to suppression of E-cadherin expression [8, 42]. In the present study, we consistently observed SNAIL induction together with decreases in E-cadherin expression in TPA-induced migrating HT-29 cells. We also found that siRNA-mediated knock-down of SLUG did not result in inhibition of E-cadherin down-regulation in TPA-treated HT-29 cells, which indicates that SNAIL rather than SLUG acts as the trans-repressor of CDH1 expression in these cells. Previously, CDH1 gene promoter hypermethylation has been noted during breast and prostate cancer metastasis [43]. Here, we found that decitabine prevented TPA-induced E-cadherin down-regulation in colon cancer cells. Since promoter hypermethylation rather than transcription factors like SNAIL or SLUG were found to be involved, the down-regulation of the CTNNB1 gene differed somewhat from that of the CDH1 gene. Of note, we found that the CTNNB1 gene silencing status closely correlated with the migratory and invasive potentials of the colon cancer cells tested. In particular, we found that the basal level of CTNNB1 promoter methylation was much higher in the more invasive HCT-116 and SW620 cells than in the less invasive HT-29 and normal colonic CCD-841 cells. Treatment with TPA resulted in CTNNB1 gene promoter methylation in CCD-841 and HT-29 cells to a larger extent than in HCT-116 and SW620 cells, and to a concomitant induction of higher levels of single cell dissociation and migration of HT-29 cells compared to those of HCT-116 and SW620 cells. These observations suggest that higher levels of CTNNB1 gene promoter methylation may result in the dissociation and migration of colon cancer cells. We also found that decitabine may significantly inhibit colon cancer cell invasion, which is in concordance with the suppression of lung cancer cell invasion after preventing CTNNB1 gene promoter methylation [15]. We also found that the inhibitory effect of decitabine and VAS2870 co-treatment on colon cancer cell invasion was notably enhanced compared to treatment with either compound alone.

In analogy of previous reports on ROS-dependent hypermethylation of the CDH1 gene promoter [44], we found that CTNNB1 gene promoter hypermethylation was also mediated through ROS. Although a similar phenomenon has been reported before [45], we found that the mechanism by which H₂O₂ or TPA-generated superoxide downregulates CTNNB1 gene expression involves promoter hypermethylation. This notion was strengthened by our finding that the increase in DNMT activity induced by H₂O₂ (5 μM) was similar to that observed after TPA treatment, and that this increase in activity could be blocked by decitabine, VAS2870 or vitamin C. In addition, we found that siRNA-mediated knock-down of NOX2 prevented TPA-induced decreases in β-catenin expression. On the other hand, we found that 5-HT did not induce DNMT activity. Although TPA and 5-HT both produce ROS, they differ in their duration of ROS production induction, i.e., 5-HT induces transient ROS production [27, 37] whereas TPA induces sustained ROS production through NOX2 induction in HT-29 cells [26]. In agreement with these observations, we noted increases in methylation of the β-catenin gene promoter induced by TPA or H₂O₂ in a time-dependent manner, which suggests that sustained ROS generation is required for inducing β-catenin gene promoter methylation.

In the present study, four different colon cancer-derived cell lines (HT-29, Caco-2, HCT-116, and SW620) with different molecular and phenotypic features were used. Our current study showed that regardless these molecular/phenotypic differences, the degree of CTNNB1 gene promoter methylation by ROS was positively associated with cell dissociation and migration. In HT-29 and Caco-2 cells, TPA treatment led to a suppression of the β-catenin levels in the cytosol and nucleus, whereas Wnt3α led to nuclear translocation of β-catenin in both cell lines. Co-treatment with Wnt3α and TPA led to a blockade of Wnt3α-induced nuclear translocation of β-catenin, confirming that down-regulation of β-catenin may be associated with a reduction in transcriptional activity as a co-activator. However, MMP-7, which is one of the Wnt3α targets and a critical molecule associated with colorectal cancer invasion and metastasis, was found to be increased through the co-treatment, which may be explained by our previous results indicating that high level induction of MMP-7 by TPA may be mediated by NF-κB and AP-1 via NOX2-dependent ROS [26]. Our results further emphasize that decreases in β-catenin levels may be associated with cancer cell dissociation rather than transcription activity loss within the Wnt/β-catenin signaling cascade.

Taken together, we show that sustained ROS production may induce hypermethylation of the CTNNB1 gene promoter and that the resulting gene silencing may facilitate the migration and invasion of colon cancer cells. Our results also suggest that epigenetic CTNNB1 gene regulation may serve as a therapeutic tool in colon cancer patients.

Compliance with ethical standards

Conflict of interest

The authors have no conflict of interest to declare.