GKN1–miR-185–DNMT1 Axis Suppresses Gastric Carcinogenesis through Regulation of Epigenetic Alteration and Cell Cycle

Recently, gastrokine 1 (GKN1) was isolated from the gastric mucosa cells of several mammalian species including rats (1). GKN1 is a novel autocrine/paracrine protein that is specifically expressed in gastric mucosa (1, 2). Toback and colleagues reported that GKN1 protects the antral mucosa and promotes healing by facilitating restitution and proliferation after injury (3). Interestingly, GKN1 is downregulated in Helicobacter pylori–infected gastric epithelial cells, and the loss of GKN1 expression is detected in gastric cancer tissues and precancerous lesions such as intestinal metaplasia (4, 5). We also witnessed frequent loss of GKN1 expression in gastric cancers and confirmed tumor suppressor activity of GKN1 through a functional analysis (6). Moreover, GKN1 plays an important role in the epithelial–mesenchymal transition (EMT) and migration of gastric cancer cells (7). Here, we hypothesized that GKN1 plays an important role in cell-cycle progression. This hypothesis is supported by the recent finding that GKN1 inhibits cell growth by inducing G₂–M arrest in SGC-7901 cells (8). However, the molecular mechanism through which GKN1 inhibits the cell cycle is still unknown.

Thus, we focused on the effects of GKN1 on cell viability and proliferation, cell cycle, and epigenetic alteration of the cell-cycle–related proteins in AGS, MKN1 and MKN28 gastric cancer, and HFE-145–immortalized non-neoplastic gastric epithelial cells. Overall, we showed that GKN1 plays an important role in the cell-cycle regulation and the epigenetic alterations of tumor-associated genes in gastric cancers.

AGS, MKN1 and MKN28 gastric cancer cells, and HFE-145–immortalized non-neoplastic gastric epithelial cells were obtained from the American Type Culture Collection and Dr. Hassan Ashktorab (Howard University, Washington, DC). GKN1 cDNA and shGKN1 were cloned into the pcDNA3.1 expression vector and pSilencer 3.1 H1-neo (Invitrogen). AGS, MKN1, MKN28, and HFE-145 cells were transfected in 60-mm-diameter dishes with the expression plasmids (2 μg total DNA) using Lipofectamine Plus transfection reagent (Invitrogen) according to the manufacturer's recommendations.

For cell viability analysis, MTT assay was conducted at 24, 48, and 72 hours after transfection of each construct, as described previously (6). To further examine whether GKN1 contributes to the chemosensitivity of 5-fluorouracil (5-FU), MTT assay was conducted on AGS and HFE-145 cells at 24 and 48 hours after simultaneous treatment with GKN1, recombinant GKN1 protein and 5-FU.

For cell proliferation assay, a bromodeoxyuridine (BrdUrd) incorporation assay was conducted at 24, 48, and 72 hours after transient transfection of each construct using the BrdU Cell Proliferation Assay Kit (Millipore), according to the manufacturer's instructions.

To measure the proliferative ability of a single cell in vitro, a plate clonogenic assay was conducted. Briefly, AGS cells transfected with each construct were cultured in RPMI-1640 for 2 weeks to allow colony formation. Colonies were fixed in 1% formaldehyde, stained with 0.5% crystal violet solution, and counted using the colono-count program.

For cell-cycle analysis, AGS cells from each experimental group were collected and stained with propidium iodide (PI) for 45 minutes in the dark before analysis. The percentages of cells in different phases of the cell cycle were determined using a FACSCalibur Flow Cytometer with CellQuest 3.0 software (BD Biosciences).

Expression of G₀–G₁ phase proteins, including p53, p21, p16, p15, CDK4, cyclin D1, E2F, and G₂–M phase proteins, including cdc2, cyclin B, cdc25A, cdc25C, aurora A, and Plk1, was examined in AGS cells 24 and 48 hours after transfection with GKN1 and in HFE-145 cells after transfection with shGKN1, as described previously (6). After blocking, the membrane was subsequently probed with antibodies against G₀–G₁ phase proteins and G₂–M phase proteins. Protein bands were detected using enhanced chemiluminescence reagents (Amersham Pharmacia Biotech).

We also analyzed the expression levels of EZH2, DNMT1, HDAC1, and Tip60, which are involved in the epigenetic process, in AGS, MKN1, and MKN28 cells 24 and 48 hours after transfection with GKN1 and in HFE-145 cells transfected with shGKN1. Expression of DNMT1 and EZH2 was also examined in 80 frozen advanced gastric cancer and corresponding non-cancerous gastric tissues, HFE-145 cells, and 10 gastric cancer cell lines by Western blot analysis. After blocking, the membrane was subsequently probed with anti-DNMT1, anti-EZH2, anti-methyl histone H3, and anti-HDAC1 monoclonal antibody (BD bioscience). Approval was obtained from the Institutional Review Board of The Catholic University of Korea, College of Medicine (CUMC09U089).

Real-time reverse transcription (RT)-PCR was carried out using SYBR Green Q-PCR Master Mix (Stratagene) according to the manufacturer's instructions. DNMT1, EZH2, and CDKN2A mRNAs were quantified by SYBR Green quantitative PCR (qPCR) and normalized to mRNA of the housekeeping gene GAPDH. To investigate whether GKN1 directly regulates miR-185 expression, expression of primary and precursor miR-185 was also quantified by qPCR and normalized to human U6 snRNA. Data are reported as relative quantities according to an internal calibrator using the 2^−ΔΔCT method (9). Sequences of the primers are described in Supplementary Table S1.

The DNMT1 activity was analyzed using the DNMT1 Activity Assay Kit (Abcam) according to manufacturer's instructions. Briefly, the cell pellet was lysed in lysis buffer (10 mmol/L Tris-Hcl pH 7.5, 10 mmol/L NaCl, and 2 mmol/L MgCl₂) containing protease inhibitor mixture (Complete; Roche Molecular Biochemicals). Then, 6 μL of 20% NP-40 was added, and the mixture was incubated for 10 minutes at 4°C and centrifuged for 5 minutes. The supernatant was collected and the pellet containing the nuclei was resuspended in 50 μL of extraction buffer (20 mmol/L HEPES, pH 7.9, 420 mmol/L NaCl, 1.5 mmol/L Mgcl₂, 0.2 mmol/L EDTA, and 10% glycerol) followed by incubation for 30 minutes at 4°C and collection of the nuclear extract by centrifugation.

Methylation status of the promoter region of the p16 and E-cadherin genes in AGS cells was determined using sodium bisulfite treatment of the DNA followed by bisulfate genome sequencing (BGS), as described previously (6). Bisulfite-treated DNA was amplified using primers for BGS (Supplementary Table S1), and sequencing of the PCR products was carried out using the Cyclic Sequencing Kit (Perkin-Elmer), according to the manufacturer's recommendations.

The Student t test was used to analyze the effect of GKN1 on cell viability and proliferation. All experiments were carried out in triplicate to verify the reproducibility of the findings. Data are expressed as means ± SD from at least 3 independent experiments. Association between GKN1, miR-185, DNMT1, and EZH2 expression and the tumor histologic type was tested using χ²and Spearman correlation tests. P < 0.05 was considered the limit of statistical significance.

When we transfected AGS, MKN1, and MKN28 cells with GKN1, we witnessed a significant time-dependent inhibition in cell growth and proliferation (Fig. 1A and B and Supplementary Fig. S1). In contrast, when we silenced GKN1 with shGKN1 in HFE-145 cells with GKN1 expression, we found a time-dependent increase in cell growth and proliferation (Fig. 1A and B). In the colony formation assay, GKN1 dramatically reduced the number and size of surviving colonies in the AGS gastric cancer cell lines compared with the empty vector–transfected control cells (mock; Fig. 1C). In addition, treatment with 5-FU induced a decrease in time- and dose-dependent cell viability in both cell lines (Fig. 1D and E). AGS cells treated with GKN1 and recombinant GKN1 protein showed synergistic inhibition of cell viability with 5-FU in a dose- and time-dependent manner (Fig. 1F). In flow cytometric analysis, treatment with 5-FU alone had no effect on the cell cycle, whereas AGS cells treated with 5-FU and GKN1 or recombinant GKN1 protein showed increased numbers of sub-G₀ cells and indicated a modest effect on the cell-cycle G₀–G₁ and G₂–M phase progression (Fig. 1G). Furthermore, the exogenous GKN1 expression synergistically induced increased expression of p53 and p21 and decreased expression of cyclin B and D in AGS cells (Fig. 1H).

Ectopic expression of GKN1 resulted in increased numbers of sub-G₀ cells and showed a modest effect on the cell-cycle G₀–G₁ and G₂–M phase progression (Fig. 2A).

For G₀–G₁ arrest in AGS cells, GKN1 induced re-expression of p16 and downregulated expression of Cdk4, cyclin D, E2F1, and cyclin A (Fig. 2B). As CDKN2A is hypermethylated in gastric cancers (10), we investigated methylation status of the promoter region of the CDKN2A gene in AGS cells after treatment of 5-aza-dC. Expectedly, 5-aza-dC induced demethylation of the CDKN2A promoter region and re-expression of the p16 protein (Supplementary Fig. S2). In shGKN1-transfected HFE-145 cells, GKN1 silencing downregulated expression of p16 and p21, but upregulated expression of positive cell-cycle regulators including CDK4 and cyclin A (Fig. 2B). For G₂–M arrest, GKN1 downregulated expression of p-cdc2, PLK1, CDK2, cdc25a, cdc25c, and cyclin B in AGS cells, whereas GKN1 silencing in HFE-145 cells upregulated expression of p-cdc2, PLK1, and cdk2 (Fig. 2C).

As shown in Fig. 3A, GKN1 induced Tip60 expression, which acetylated DNMT1 (11), and reduced the expression of DNMT1, EZH2, methyl histone H3, and HDAC1 in AGS cells (Fig. 3A). In MKN1 and MKN28 cells, GKN1 also reduced expression of DNMT1 and EZH2 (Supplementary Fig. S3). However, GKN1 silencing in HFE-145 cells upregulated expression of these proteins. When we examined the DNMT1 activity in GKN1-transfected AGS cells and in shGKN1-transfected HFE-145 cells, decreased and increased DNMT1 activity was observed, respectively (Fig. 3B). Moreover, GKN1 inhibited DNMT1 and EZH2 mRNA expression and increased CDKN2A mRNA expression in AGS cells (Fig. 3C). In addition, DNMT1, EZH2, and HDAC1 silencing in AGS cells by shDNMT1, shEZH2, and shHDAC1 also significantly reduced cell growth (Supplementary Fig. S4A), increased CDKN2A mRNA expression, and induced G₀–G₁ and G₂–M cell-cycle arrest (Supplementary Fig. S4B and S4C).

In human gastric cancer tissues, GKN1-negative cancers showed overexpression of DNMT1 and EZH2, whereas GKN1-positive cancers showed underexpression of these proteins (Fig. 3D). Sixty-nine (86%) of 80 gastric cancers showed decreased expression of GKN1 and increased expression of DNMT1 (79%) and EZH2 (84%) proteins, compared with samples from the corresponding normal gastric mucosa (Fig. 3E). Histologically, reduced GKN1 expression and overexpression of DNMT1 and EZH2 were closely associated with diffuse type and poorly differentiated gastric cancer (P < 0.05, Supplementary Table S2). Moreover, HFE-145 cells showed strong GKN1 protein expression, but 10 gastric cancer cell lines did not express GKN1. HFE-145 cells did not express DNMT1 and EZH2, whereas there was strong expression of both proteins in 10 gastric cancer cell lines (Fig. 3F). To further confirm our results, we recapitulated GKN1, DNMT1, and EZH2 gene expression from the large cohorts of patients with gastric cancer that are available from the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) database (accession numbers GSE27342). GKN1 gene expression was significantly and consistently downregulated, whereas DNMT1 and EZH2 gene expression was significantly upregulated, in gastric cancer cohorts (Fig. 3G).

We also examined whether GKN1 induces expression of miR-185, which directly targets DNMT1 (12, 13). A significantly increased level of miR-185 was detected in the AGS, MKN1, and MKN28 cells transfected with GKN1, but GKN1 silencing in HFE-145 cells significantly decreased miR-185 expression (Fig. 4A and Supplementary Fig. S5B). Interestingly, ectopic expression of c-Myc, which binds to a promoter of the miR-185 (http://genome.ucsc.edu/), inhibited primary, precursor, and mature miR-185 expression, but GKN1 upregulated expression of primary, precursor, and mature miR-185 by binding to and downregulating c-Myc (Supplementary Fig. S5C and S5D). In addition, miR-185 decreased cell viability in AGS, MKN1, and MKN28 cells but increased cell viability in HFE-145 cells (Fig. 4B and Supplementary Fig. S6A). miR-185 increased the sub-G₀ cell population and induced cell-cycle arrest, especially in G₂–M phase (Fig. 4C). In particular, miR-185 inhibited DNMT1 and EZH2 activities by binding to a plausible recognition site in the 3′-untranslated region (UTR) of the DNMT1 and EZH2 (Fig. 4D). Overexpression of miR-185 reduced the expression of DNMT1 and EZH2 in AGS, MKN1, and MKN28 cells, as well as the expression of the cell-cycle–related proteins including cyclin D, E, and B in AGS cells (Fig. 4E and Supplementary Fig. S6B). In HFE-145 cells, silencing of miR-185 induced upregulation of DNMT1, EZH2, and the positive cell-cycle regulators, but not HDAC1 and Tip60 (Fig. 4E). Moreover, overexpression of miR-185 in AGS cells resulted in decreased DNMT1 activity, DNMT1 and EZH2 mRNA expression, and increased CDKN2A mRNA expression (Fig. 4F and G). In MKN1 and MKN28 cells, we also found that miR-185 reduced DNMT1 and EZH2 mRNA expression (Supplementary Fig. S6C). However, HFE-145 cells with anti-miR-185 treatment showed expression or activity patterns exactly opposite to miR-185 in AGS cells (Fig. 4F and G). In human gastric cancer tissues, there was decreased expression of GKN1 and miR-185 in 69 (86.3%) of 80 gastric cancer tissues and increased DNMT1 and EZH2 expression in 69 (86%) cancer tissues. Statistically, we found a close relationship between GKN1 and miR-185 expression and an inverse relationship between miR-185 and DNMT1 and EZH2 in gastric cancer tissues (Fig. 4H). Furthermore, we recapitulated miR-185 expression from the large cohorts from the NCBI GEO database (accession numbers GSE23739) and found that miR-185 expression was significantly downregulated in gastric cancer cohorts (Fig. 4I). In addition, decreased expression of miR-185 and GKN1 and increased expression of DNMT1 and EZH2 were found in 10 gastric cancer cell lines (Fig. 4J).

Next, we investigated whether GKN1 activity is dependent on miR-185 expression. Treatment with anti-miR-185 in AGS, MKN1, and MKN28 cells transfected with GKN1 showed a moderate ablation of GKN1-induced growth-inhibitory activity (Fig. 5A and Supplementary Fig. S7) and cell-cycle arrest (Fig. 5B). Silencing of miR-185 with anti-miR-185 recovered expression of DNMT1, EZH2, and positive cell-cycle regulators and decreased expression of p16 (Fig. 5C). Also, miR-185 silencing in AGS cells transfected with GKN1 increased the number and size of surviving colonies, DNMT1 activity, and DNMT1 and EZH2 mRNA expression (Fig. 5D–F) and decreased CDKN2A mRNA expression (Fig. 6F).

To determine whether GKN1 functions as a hypomethylating agent, we sequenced 218- and 417-bp fragments, containing 17 and 34 CpG dinucleotides, respectively, in the promoter and around the translation start sites of the E-cadherin (Bis-E-cadherin) and in the promoter of CDKN2A (Bis-CDKN2A; Fig. 6A and C). For the E-cadherin gene, 7 methylated CpG sites were completely unmethylated in GKN1-transfected cells, but treatment of GKN1 with anti-miR-185 reverted this hypomethylation. For the CDKN2A gene, 30 (88.24%) of 34 CpG islands were methylated in AGS cells. Ectopic GKN1 expression turned 26 methylated CpG sites into unmethylated CpG sites, whereas anti-miR-185 treatment inhibited the hypomethylating activity of GKN1 (Fig. 6B and D and Supplementary Fig. S8A and S8B). Thus, re-expression of E-cadherin and CDKN2A was induced by demethylation due to the suppression of EZH2 and DNMT1 through upregulation of Tip60 and decreased HDAC1 activity following GKN1 expression (Fig. 6E).

In this study, we addressed the molecular mechanism underlying the tumor suppressor activity of GKN1 in gastric cancer. GKN1 significantly reduced the cell viability, proliferation, and colony formation of AGS cells (Fig. 1). Treatment of GKN1 and 5-FU synergistically inhibited cell viability and proliferation by upregulating p53 and p21 (Fig. 1). The activation of p53 and p21 by 5-FU was previously reported in mouse L-TK and AGS cells (14, 15). The fact that GKN1 suppressed cell proliferation implies that GKN1 can modulate cell-cycle- regulating components. Negative cell-cycle regulators such as p15 and p21 are the key modulators that suppress cyclin D1/CDK4, 6 or cyclin E/CDK2 complex in the G₁–S transition (16). Here, we found a concomitant increase of G₀–G₁ phase and G₂–M phase in GKN1-transfected cells (Fig. 2A). GKN1 selectively induced p16 re-expression and elicited concomitant suppression of CDK4, cyclin D1, and E2F expression in the G₁–S transition (Fig. 2B). GKN1 also inactivated G₂–M -phase progression in AGS cells by regulating the expression of cdc25, PLK1, and cyclin B (Fig. 2C). These findings are consistent with previous reports (2, 8) and strongly suggest that GKN1 functions as a gastric tumor suppressor by regulating a set of proteins involved in cell-cycle control and has synergic effects with 5-FU.

Because GKN1 induced p16 re-expression in AGS cells (Fig. 2), we set out to determine whether GKN1 inhibits the expression of epigenetic regulators (17). Interestingly, we observed a near-complete inactivation of DNMT1, EZH2, methyl histone H3, and HDAC1 in GKN1-transfected cells (Fig. 3). In addition, DNMT1, EZH2, and HDAC1 silencing in AGS cells suppressed cell growth by inducing G₀–G₁ and G₂–M cell-cycle arrest (Supplementary Fig. S4). Furthermore, GKN1 expression was inversely correlated with DNMT1 and EZH2 expression in gastric cancer tissues and cell lines (Fig. 3D–F). DNMT1 is primarily involved in the maintenance of methylation during DNA replication phase, and the overexpression of DNMT1 has been reported in gastric cancer (18). EZH2, a histone methyltransferase, is involved in epigenetic silencing of a large number of genes involved in differentiation and proliferation (19). EZH2 expression is also regulated at transcriptional, posttranscriptional, and posttranslational levels in human cancer (20), and its overexpression is considered to be an important positive regulator of cancer cell growth in multiple human malignancies (21). Moreover, EZH2 promotes EMT by interacting with Snail and suppressing expression of E-cadherin (22). Snail induces DNA methylation of E-cadherin promoter by recruiting HDAC1 and DNMT1 (23). In particular, EZH2 containing polycomb repressor complex 2 transcriptionally represses cell-cycle suppressor INK-ARF to drive cell-cycle progression (19) and acts as a major enzyme that methylates lysine-27 of histone H3 (H3-K27; ref. 21). Thus, these results suggest that GKN1 inhibits the development and progression of gastric cancer by regulating the expression of epigenetic regulatory components and EMT-related proteins. Interestingly, a previous study that involved a group of patients with gastric cancer treated with cisplatin/5-FU–based neoadjuvant chemotherapy revealed that the concordant methylation of multiple genes suggested an association with worse response to therapy (24). As GKN1 synergistically enhanced the effect of cell growth inhibition induced by 5-FU (Fig. 1), all of these data suggest that GKN1 functions as a tumor suppressor by regulating abnormal epigenetic modification associated with gastric tumorigenesis, such as DNA methylation, DNA acetylation, and histone modification.

Next, to identify a potential molecular pathway involved in the regulation of DNMT1 and EZH2 expression by GKN1, we analyzed miR-185, which is known to directly target DNMT1 and regulate DNA methylation (12, 13). Interestingly, GKN1 upregulated miR-185 expression in gastric epithelial cells (Fig. 4A) and miR-185 inhibited cell growth by inducing cell-cycle arrest, mainly at G₂–M phase, with inactivation of DNMT1 and EZH2 and increase of CDKN2A mRNA expression (Fig. 4). Unexpectedly, GKN1 downregulated HDAC1 and Tip60 in an miR-185–independent manner (Fig. 4E). Before these experiments, we confirmed miR-185 expression levels after miR-185 and anti-miR-185 transfection in AGS and HFE-145 cells, respectively (Supplementary Fig. S5). Previously, the binding site of the miR-185 in the 3′-UTR of the DNMT1 has been reported (12). Here, we found that miR-185 inhibits EZH2 activity by binding to a plausible recognition site in the 3′-UTR of the EZH2 (Fig. 4D). In gastric cancer tissues and cell lines, the GKN1 expression level was positively associated with miR-185 expression and inversely correlated with DNMT1 and EZH2 expression (Fig. 4H and J). Furthermore, miR-185 silencing suppressed the inhibitory effects of GKN1 on cell proliferation, cell-cycle, tumorigenicity, expression of DNMT1 and EZH2, and DNMT1 activity (Fig. 5). All of these data indicate that GKN1 suppresses gastric cancer cell growth by downregulating epigenetic regulators and positive cell-cycle components in an miR-185–dependent and -independent manners.

To define how GKN1 regulates the expression of miR-185, we analyzed expression of primary and precursor miR-185 and c-Myc in AGS and HFE-145 cells and found that GKN1 upregulated expression of both primary and precursor miR-185 by binding to and downregulating c-Myc (Supplementary Fig. S5C and S5D). However, further studies are strongly necessary to clarify the molecular mechanism of c-Myc downregulation by the GKN1.

As DNMT1 is primarily involved in the maintenance of methylation during DNA replication (25), we examined the methylation status of the CDKN2A and E-cadherin genes and found that GKN1-induced miR-185 converted hypermethylated CDKN2A and E-cadherin to the unmethylated form. Detailed methylation analysis of 32 and 17 CpG sites at the CDKN2A and E-cadherin CpG islands, respectively, confirmed the demethylated status (Fig. 6). Thus, these results strongly suggest that GKN1 functions as a hypomethylating agent and inhibit gastric carcinogenesis by regulating epigenetic alterations in cell-cycle regulatory components.

On the basis of our data, we propose the following model describing the role of GKN1 in cell cycle and epigenetic alterations. GKN1-induced miR-185 inhibits EZH2 and DNMT1 activity. Also, GKN1 downregulates DNMT1 by inhibition of HDAC1 and induction of Tip60 in an miR-185–independent manner. Therefore, GKN1 leads to demethylation of E-cadherin and the CDKN2A promoter region and re-expression of E-cadherin and CDKN2A. Thus, the data here suggest that GKN1 has an miR-185–dependent and -independent mechanism for chromatic and DNA epigenetic modification and cell-cycle regulation (Fig. 6E). It is plausible that modulating GKN1-induced miR-185 activity and stimulating its anti-cancerous effect could significantly impact the development of novel cancer treatments, which will ultimately achieve the goal of gastric cancer prevention and remission.

No potential conflicts of interest were disclosed.

Conception and design: J.H. Yoon, S.W. Nam, J.Y. Lee, W.S. Park

Development of methodology: J.H. Yoon, S.W. Nam, W.S. Park

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J.H. Yoon, Y.J. Choi, W.S. Choi, D. Smoot, W.S. Park

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J.H. Yoon, Y.J. Choi, W.S. Choi, W.S. Park

Writing, review, and/or revision of the manuscript: J.H. Yoon, W.S. Park

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): H. Ashktorab, W.S. Park

Study supervision: J.H. Yoon

This work was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and Technology (2012R1A2A2A01002531).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.