HOTAIR Induces Methylation of PCDH10, a Tumor Suppressor Gene, by Regulating DNMT1 and Sponging with miR-148b in Gastric Adenocarcinoma

Patients and tissue samples

Forty-nine GC tissue and paired adjacent gastric tissue samples were collected from 49 patients who underwent surgical resection or endoscopic resection for GC at Kangdong Sacred Heart Hospital, Hallym University College of Medicine. All samples were frozen in liquid nitrogen immediately after resection and stored at −80℃ until use. The mean age of patients was 62.2±1.7 years, and the male/female ratio was 65.3%. This study was approved by the Ethics Committee of Kangdong Sacred Heart Hospital, and written informed consent was obtained from all patients (IRB no. 2017-11-007).

Cell lines and cell culture

A total of five GC cell lines (Kato III, MKN 28, MKN 45, MKN 74, AGS) and normal GES-1 cell line were used. Human GC cells were cultured in RPMI-1640 medium (Thermo Scientific, Rockford, IL, USA) supplemented with 10% fetal bovine serum (FBS), 1% penicillin, and streptomycin. The GC cells were incubated at an atmosphere of 5% CO₂ and 95% air at 37℃.

RNA extraction, reverse transcription, and quantitative real-time PCR

Total RNA was extracted from 49 GC tissues and cell lines using TRIzol reagent (Invitrogen, Carlsbad, CA, USA). For cDNA synthesis, 2.0 µg of total RNA was reverse transcribed according to the manufacturer's protocol. The level of HOTAIR was measured by real-time PCR using iQ SYBR Green Supermix (Applied Biosystems Inc., Carlsbad, CA, USA). Ct value of the sample was normalized to the U6 or GAPDH expression, and the 2^−ΔΔCt method was used to calculate the relative level. The primers used for this study are listed in Table 1.

Table 1
Primer Sequences Used for qRT-PCR

• Click for larger image
• Click for full table
• Download as Excel file

Small interfering RNA transfection

MKN 28 and MKN 74 cells (2×10⁵) were plated in 6-well culture plates and transfected after incubation for 24 h. Small interfering RNA (siRNA) and nonspecific control siRNA were synthesized (Invitrogen), and cells were transfected using Lipofectamine 2000 (Invitrogen). Target sequences for HOTAIR siRNAs were designed as follows: si-HOTAIR1, Sense: 50-GAACGGGAGUACAGAGAGAUU-30, Antisense: 50-AAUCUCUCUGUACUCCCGUUC-30; si-HOTAIR2, Sense: 50-CACAUGAACGCCCAGAGAUU-30, Antisense: 50-AAUCUCUGGGCGUUCAUGUGG-30.

HOTAIR-overexpressing plasmid construction

To overexpress HOTAIR in AGS cell, we used human HOTAIR cDNA (Addgene, Plasmid #26110, Cambridge, MA, USA).18 The HOTAIR cDNA was amplified using a PCR system, and the resulting product was inserted in the pcDNA3.1 vector (Addgene, Plasmid #47388) using TOPO cloning (Invitrogen) according to the manufacturer's protocol. The HOTAIR-expressing vector was sequenced and analyzed using Macrogen (Macrogeninc., Seoul, Korea). AGS cell was transfected with 1 ug pcDNA3.1-HOTAIR (pcDNA-HOTAIR) for 24 h using Lipofectamine 2000 (Invitrogen).

Cell proliferation analysis

MKN 28 and MKN 74 cells were transfected with 50 nM of si- HOTAIRs, and AGS cell was transfected with pcDNA-HOTAIR from 0 to 72 h. MTS assays (Promega, Madison, WI, USA) were used to measure cell proliferation in 96-well plates. The number of viable cells was determined by reaction with MTS reagent for 1 h in the dark, and the products of the reaction were measured by enzyme-linked immunosorbent assay (ELISA). Luminescence activity was measured using a Luminescence Microplate Reader (Molecular Devices Co., Sunnyvale, CA, USA).

Apoptosis analysis

MKN 28 and MKN 74 cells were transfected with si-HOTAIRs or si-Control (siCT), and washed with 1X phosphate buffered saline (PBS; Thermo Scientific). After 48 h, the cells were pelleted and washed with PBS. After washing with 1X binding buffer (BD Bioscience, San Jose, CA, USA), the cells were stained with fluorescein isothiocyanate (FITC)-Annexin V and PI, using a FITC-Annexin V kit (BD Bioscience). The apoptotic ratio was measured using flow cytometry (BD Bioscience), and caspase 3/7 protein levels were calculated by ELISA after treatment with si-HOTAIRs.

Cell cycle analysis

MKN 28 and MKN 74 cells were transfected with 100 nM of si-HOTAIRs or siCT, washed with PBS, and then fixed with 75% ethanol overnight at -20℃. Cells were resuspended in PBS and treated with RNase for 30 min at room temperature. The cell nucleus was stained with propidium iodide (Sigma, Saint Louis, MO, USA), and incubated for 20 min in the dark. The proportion of cells within cell cycle phases were determined by flow cytometry (BD Biosciences), and analyzed with FlowJO7.6 program (Treestar, Ashland, OR, USA).

Invasion assay and migration assay

To perform invasion assay, the matrigel invasion assay was performed using BD biocoat trans-wells (BD Biosciences). MKN 28 and MKN 74 cells were transfected with si-HOTAIRs or si-CT. After 48 h, the transfected cells were re-plated in the upper chamber containing RPMI-1640 medium. The lower chamber was filled with RPMI-1640 medium containing 10% FBS. After 48 h, non-invading cells within the insert chamber were removed, and the upper layer of the trans-well was wiped with a cotton swab. The membrane of the bottom part of the upper chamber was fixed and stained with Diff-Quik solution (Dade Behring Inc., Newark, DE, USA). Invading cells were visualized using a virtual microscope (BX51; Olympus, Tokyo, Japan) in five random fields, counted, and averaged.

For migration analysis, GC cells were transfected with si-HOTAIRs and si-CT. The cells were grown for 48 h, and a wound was then generated using a P-20 tip. The width of scratched cells was measured at 0 and 48 h under a virtual microscope (BX51).

Western blot

The cells were lysed in 1_ RIPA buffer (Cell Signaling Technology, Danvers, MA, USA) containing protease inhibitor. Extracted proteins were separated by 8% to 10% SDS–polyacrylamide gel electrophoresis and transferred to a polyvinylidene difluoride membrane (GE Healthcare, Piscataway, NJ, USA). The membrane was blocked for 1 h at room temperature in tris–phosphate buffer containing 0.1% tween 20 with 5% skim-milk (BD Biosciences), and subsequently incubated at room temperature for 1–2 h with primary antibodies. The following primary antibodies were used for Western blot analysis:

PARP (Cell Signaling Technology, #9542), bcl-2 (Cell Signaling Technology, #2870), bcl-xl (Cell Signaling Technology, #2764), p53 (Santa Cruz Biotechnology, Santa Cruz, CA, USA, sc-126), cleaved caspase-9 (Cell Signaling Technology; 9504), SUZ12 (Abcam, Cambridge, MA, USA, ab12073), PCDH10 (Thermo Scientific, Cell Signaling Technology, #7237, PA5-31042), Vimentin (1:2000, Santa Cruz Biotechnology, sc-373717), Snail (1:1000, Cell Signaling Technology, 3879s), Slug (1:1000, Santa Cruz Biotechnology, sc-166476), Src (1:1000, Santa Cruz Biotechnology, sc-8056), β-actin (Bioworld Technology, Louis Park, MN, USA, AP0060), EZH2 (Abcam, ab186006), and DNMT1 (Abcam, ab13537).

Methylation-specific PCR

Genomic DNA was extracted from MKN 28 and MKN 74 cells using a DNeasy Blood & Tissue kit (Qiagen, Valencia, CA, USA). An EZ DNA methylation-gold kit (ZymoResearch, Irvine, CA, USA) was used for DNA bisulfate transformation. The following methylation-specific PCR (MS-PCR) primers for PCDH10 were used as described: methylation (forward: GTTAGGGAGGATGGATGTAAGTATC, reverse: GCG AAATAAAAACAATAAAACGAC), and un-methylation (forward: GTTAGGGAGGATGGATGTAAGTATT, reverse: CCCACA AAATAA AAACAATAAAA AA).

RNA immunoprecipitation

For immunoprecipitation of endogenous RNA-protein complexes, cells were lysed with IP buffer (Thermo Fisher Scientific, Waltham, MD, USA), and resuspended in RNA immunoprecipitation (RIP) buffer (Abcam) with RNase inhibitor (GenDEPOT) and protease inhibitor (GenDEPOT, Barker, TX, USA). For shearing of chromatin, we used 20 cycles of shearing under cooling conditions, with 15 s on and 30 s off for each cycle (170–190 W) and water bath sonication. After sonication, the antibodies were added to the supernatant obtained by centrifugation, and then incubated overnight at 4℃ with a rotator. After incubation, 20 µL of MagnaChip protein magnetic beads (Millipore, Darmstadt, Germany) were added, and samples were reacted on a rotator at 4℃ for 1 h. After washing twice with RIP buffer, samples were dissolved with TRIzol reagent or RIPA.

Luciferase reporter assay

For the luciferase reporter assay, GC cells were co-transfected with miRNAs (miR-148b mimics or miR-148b inhibitors; Biomics Biotechnologies Co., Ltd., Nantong, China) and reporter vectors (pmirGLOWT or pmirGLOMUT) using Lipofectamine 2000. Luciferase activity was assayed 48 h after transfection using a DualLuciferase Reporter Assay system (Beyotime Institute of Biotechnology, Haimen, China). The values were normalized to those obtained for miRNA negative control transfection. All transfection experiments were performed in triplicate.

Statistical analysis

All of the analyzed data for continuous and categorical variables are presented as the mean±standard error and the number of lesions with the percentage. Statistical tests used to compare the measured results included the t-test, chi-square test, and Fisher's exact test. The expressions of HOTAIR in GC were categorized into low and high groups based on the median value of HOTAIR expression in GC tissue. A p-value<0.05 indicated a statistically significant difference for comparisons between groups. All of the statistical procedures were conducted using the statistical software SPSS for Windows (version 18.0; SPSS Inc., Chicago, IL, USA).

HOTAIR expression in GC cell lines and GC tissues

The relative RNA level of HOTAIR was higher in MKN 74, MKN 28, MKN 45, and KATO III cells compared to normal gastric cell line GES-1 (p<0.01); however, the level of HOTAIR in AGS cell was lower than that in GES-1 (p<0.05) (Fig. 1A). The expression of HOTAIR in GC tissues were significantly higher than that in non-tumor tissue (p<0.05) (Fig. 1B). The Cancer Genome Atlas data showed similar result (Fig. 1C). After transfection with si-HOTAIR, HOTAIR was significantly downregulated in MKN 28 and MKN 74 cells (Fig. 1D, left panel). In AGS cell, which showed low expression of HOTAIR compared to other cancer cell lines, HOTAIR was overexpressed by transfection with pc-DNA-HOTAIR (Fig. 1D, right panel).

Fig. 1
HOTAIR was overexpressed in GC cell lines and tissues, and knockdown of HOTAIR repressed GC cell proliferation. (A) The relative expression of HOTAIR by qRT-PCR in five GC cell lines. (B) The relative expression of HOTAIR by qRT-PCR in 49 GC tissues and paired non-tumor tissues. (C) The relative expression of HOTAIR in GC tissues and adjacent gastric tissues by TCGA data. (D) MKN 28 and MKN 74 cells were transfected with siHOTAIR1, si- HOTIAR2, and/or scrambled RNA; and AGS cell was transfected with pcDNA and pcDNA-HOTAIR. The relative HOTAIR expression by qRT-PCR assay is shown. (E) Cell viability was detected by MTS assay. Data are presented as the mean±standard deviation of three independent experiments. The asterisk (*) represents a statistically significant difference compared with scrambled control. ^*p≤0.05, ^**p≤0.01. HOTAIR, HOX transcript antisense intergenic RNA; GC, gastric cancer; TCGA, The Cancer Genome Atlas.

• Click for larger image
• Download as PowerPoint slide

The effect of HOTAIR on proliferation and apoptosis of GC cells

We have already demonstrated that HOTAIR inhibits GC proliferation and apoptosis in the previous study14; however, in order to identify the reproducibility in a new experimental environment, we examined MTS assay. From 48 hr after treatment with si-HOTAIR1 and si-HOTAIR2, the proliferation of MKN 28 and MKN 74 were significantly reduced, respectively, compared to the si-CT (Fig. 1E). In the contrary, overexpression of HOTAIR upregulated the proliferation in AGS cell (Fig. 1E).

Next, we analyzed the effect of HOTAIR on apoptosis of MKN 28 and MKN 74 cells using PI/Annexin-V. Si-HOTAIRs were significantly increased in the early-to-late apoptotic ratio compared to the control in both cell lines (Fig. 2A). Protein levels of two anti-apoptotic markers, such as Bcl-2 and Bcl-xL, were decreased by si-HOTAIRs, while an induction of apoptotic factors such as p53, Bax, cleaved caspase-9 and cleaved PARP were increased compared to the control cells (Fig. 2B). Si-HOTAIRs significantly increased the caspase 3/7 activity in MKN 28 and 74 cells (Fig. 2C). Flow cytometry analysis revealed a significant increase in the proportion of cells in the sub-G1 phase in si-HOTAIRs compared to siCT (Fig. 2D). Si-HOTAIRs subsequently reduced the proportion of cells in S and G2 compared to siCT (Fig. 2D). These data suggested that HOTAIR consistently inhibits apoptosis in GC cells.

Fig. 2
Knockdown of HOTAIR induced cell apoptosis and inhibited invasiveness and migratory capacity. (A) PI/annexin-V staining on flow cytometry was performed in transfected MKN 28 and MKN 74 cells (A; left panel), and the apoptotic ratio was measured by flow cytometry (A; right panel). (B) Western blot analysis of apoptotic markers after treatment with si-HOTAIRs. (C) Caspase Glo 3 and 7 assays after treatment with si-HOTAIRs. (D) Cell cycle distributions was analyzed by flow cytometry in GC. (E) Matrigel invasion assay after transfection with si-HOTAIRs. (F) Wound healing assay after transfection with si-HOTAIRs. The width of the scratch wound was observed by microscopy at 0 and 48 h. (G) Western blot analysis of epithelial-mesenchymal transition markers after treatment with si-HOTAIRs. The asterisk (*) represents a statistically significant difference compared with scrambled control. ^*p≤0.05, ^**p≤0.01. Data are presented as the mean±standard deviation of three independent experiments. HOTAIR, HOX transcript antisense intergenic RNA; GC, gastric cancer.

• Click for larger image
• Download as PowerPoint slide

The impact of HOTAIR on invasion and migration of GC cells

We have already shown that HOTAIR promotes the invasion and migration of GC cells14; however, in order to identify the reproducibility in a new experimental environment, we measured the migration and invasion of GC cells. We found that the knockdown of HOTAIR markedly inhibited cell invasive ability (Fig. 2E). Next, the effects of HOTAIR silencing on migratory capacity was confirmed by scratch wound healing assay. Compared to si-CT, wound closure was repressed by si- HOTAIRs (Fig. 2F). We measured the epithelial mesenchymal transition related proteins after the treatment of si-HOTAIR. When compared with siCT, si-HOTAIR down regulated Vimentin, Src, Slug, and Snail (Fig. 2G). This result showed that knockdown of HOTAIR suppressed GC cell migration.

The relation between HOTAIR expression and clinicopathologic characteristics of GC

Clinicopathologic features of GC were analyzed according to the median level of HOTAIR in GC tissues (Table 2). Diffuse and mixed type cancers by Lauren classification were more prevalent in the high HOTAIR group (high vs. low; diffuse and mixed type, 64.7% vs. 37.5%, p=0.04). Moreover, tumors in the high HOTAIR group showed a more advanced Tumor-Node-Metastasis (TNM) stage than those in the low HOTAIR group (high vs. low; stage II, III, 82.4% vs. 56.3%, p=0.04). The depth of tumor invasion showed more advanced T stage in the high HOTAIR group. Age, sex, proportion of H. pylori infection, carcinoembryonic antigen level, and histologic differentiation were not related to HOTAIR expression. In addition, lymphovascular invasion and lymph node metastasis were not statistically different according to the HOTAIR level.

Table 2
Clinicopathological Analysis according to Median Level of HOTAIR in Gastric Cancer

• Click for larger image
• Click for full table
• Download as Excel file

The effect of HOTAIR on tumor suppressor gene, PCDH10, in GC

We noticed that HOTAIR knockdown reduced PCDH10 in breast cancer cell in microarray7; therefore, we need to validate PCDH10 as a new target of HOTAIR in GC. We performed qRT-PCR to analyze the mRNA level of PCDH10 in 49 GC tissues, and the mRNA of PCDH10 was significantly higher in adjacent non-tumor tissue compared to GC tissues (p<0.05) (Fig. 3A). We evaluated the protein level of PCDH10 by Western blot in four GC tissues. The expression of PCDH10 was downregulated in GC tissue; and oppositely, the expression of HOTAIR was upregulated in GC tissue (Fig. 3B).

Fig. 3
HOTAIR promoted the methylation of PCDH10 by regulating EZH2 and DNMT1. (A) The relative expression of PCDH10 by qRT-PCR in GC tissues and paired adjacent gastric tissues. (B) Western blot analysis of PCDH10 and RT-PCR analysis of HOTIAR in four GC tissues and paired adjacent gastric tissues. (C) The relative expression of PCDH10 by qRT-PCR after transfection with si-HOTAIRs in MKN 28 and MKN 74 cells. (D) Western blot analysis of PCDH10 after treatment with si-HOTAIRs (upper panel), and MS-PCR analysis of PCDH10 after treatment with si-HOTAIRs. (E) The interaction between DNMT1, EZH2, and HOTAIR was confirmed by RIP analysis. The bars present relative enrichment. (F) Western blot analysis of DNMT1 and EZH2 after treatment with si-HOTAIRs in MKN 28 and MKN 74 cell lines and pcDNA-HOTAIR in AGS cell line. (G) The relative expression of p53 after transfection with pcDNA-HOTAIR in GES-1 cell by qRT-PCR. (H) Western blot analysis of p53 after transfection with pcDNA-HOTAIR in GES-1 cell. (I) The relative expression of PCDH10 after transfection with pcDNA-p53 and pcDNA-HOTAIR in AGS cell by qRT-PCR. Data are presented as the mean±standard deviation of three independent experiments. The asterisk (*) represents a statistically significant difference compared with scrambled control. ^*p≤0.05, ^**p≤0.01. HOTAIR, HOX transcript antisense intergenic RNA; PCDH10, protocadherin 10; GC, gastric cancer; MS-PCR, methylation-specific PCR; RIP, RNA immunoprecipitation.

• Click for larger image
• Download as PowerPoint slide

We investigated the PCDH10 level in GC cells after knockdown of HOTAIR. Both si-HOTAIR1 and 2 significantly upregulated the mRNA of PCDH10 (Fig. 3C) and protein (Fig. 3D, upper panel) in MKN 28 and MKN 74. In MS-PCR, si-HOTAIR1 and 2 significantly reduced the methylation of PCDH10 compared to the control in in MKN 28 and MKN 74 (Fig. 3D, lower panel). To investigate the related pathway of HOTAIR to control methylation of PCDH10, we performed RIP to identify the interaction between DNMT1, EZH2, SUZ12, and HOTAIR. HOTAIR significantly interacted with DNMT1 and EZH2, but not with SUZ12 (Fig. 3E). DNMT1 and EZH2 were downregulated by si-HOTAIRs in MKN 28 and MKN 74 cells, while overexpression of HOTAIR induced upregulation of DNMT1 and EZH2 in AGS cell (Fig. 3F).

Since PCDH10 is known as a transcriptional target of p53,15 we investigated the expression of PCDH10 and p53 after overexpression of HOTAIR. The overexpression of HOTAIR significantly reduced the translation and transcription of p53 in normal gastric cell line GES-1 (Fig. 3G and H). This result suggested that p53 reduction by HOTAIR may be a major cause of PCDH10 transcription reduction. Next, we conducted an experiment to compare the effects of p53 and HOTAIR on PCDH10 transcription. In AGS cell, p53 significantly increased PCDH10 transcription; however, co-infection of HOTAIR, which is expected to induce the promoter methylation of PCDH10 and reduce PCDH10 transcription, reduced only a part of PCDH10 transcription by p53 (Fig. 3I). This result disproved the fact that although HOTAIR is involved in the transcription of PCDH10, the effect of p53 is greater than HOTAIR.

HOTAIR regulates methylation of PCDH10 as a ceRNA by sponging miR-148b

In a previous study, several potential miRNA targets of HOTAIR were identified in GC by computer-aided algorithm.19, 20, 21, 22 In this study, we further explored the underlying mechanism of HOTAIR as ceRNA, and we examined miR-148b as target of HOTAIR, as it was down-regulated in GC in previous study.20

The expression level of miR-148b was significantly lower in four GC cells, including MKN 28 and MKN 74, compared to GES-1 cell (Fig. 4A). In RIP, miR-148b interacted with HOTAIR in both MKN 28 and MKN 74 cells (Fig. 4B). Both si-HOTAIR1 and 2 increased the miR-148b expression (Fig. 4C), and the treatment of miR-148b mimic reduced the HOTAIR expression (Fig. 4D), indicating an inverse correlation between HOTAIR and miR-148b in GC cells.

Fig. 4
Reciprocal negative regulation of miR-148b and HOTAIR on PCDH10 expression. (A) The relative miR-148b expression by qRT-PCR in five GC cell lines. (B) RIP with anti-Ago2 was performed in MKN 28 and MKN 74 cells transfected with HOTAIR and miR-148b. HOTAIR and miR-148b expression level were detected using qRT-PCR. (C) The relative miR-148b level after transfection with si-HOTAIRs by qRT-PCR. (D) The relative HOTAIR level after transfection with miR-148 mimic by qRT-PCR. (E) The relative luciferase activity of HOTAIR after treatment with miR-148b mimic/inhibitor in MKN 28 and 74 cells transfected with wild type/mutant type HOTAIR. (F) Western blot analysis of DNMT1 and PCDH10 after treatment with si-HOTAIRs and miR-148b mimic. (G) Western blot analysis of DNMT1 after treatment with si-HOTAIRs and miR-148b inhibitor. Data are presented as the mean±standard deviation of three independent experiments. The asterisk (*) represents a statistically significant difference compared with scrambled control. ^*p≤0.05, ^**p≤0.01. HOTAIR, HOX transcript antisense intergenic RNA; PCDH10, protocadherin 10; GC, gastric cancer; RIP, RNA immunoprecipitation.

• Click for larger image
• Download as PowerPoint slide

To verify whether miR-148b regulates HOTAIR as a direct target, a predicted binding site for miR-148b was identified using an online software program,23 and wild-type or mutant miR-148b target binding sequences were cloned into the pmirGLO luciferase vector. Following co-transfection with pmirGLO luciferase construct (pmirGLO-mutant or pmirGLO-wild-type) and miR-148b mimic or inhibitor, a dualluciferase assay was performed to determine the luciferase activity.

Our data showed that MKN 28 cell co-transfected with the constructs containing pmirGLO-Mut and miR-148b mimic had significantly higher luciferase activity compared to those transfected with pmirGLO-WT and miR-148b mimic (p<0.05) (Fig. 4E). In MKN 74 cell, the result was similar with MKN 28. In addition, the relative luciferase activity was significantly lower in MKN 28 and MKN 74 cells co-transfected with the constructs containing pmirGLO-Mut and inhibitor of miR-148b compared to those transfected with pmirGLO-WT and inhibitor of miR-148b (p<0.01) (Fig. 4E).

In Western blot analysis, treatment with si-HOTAIRs and miR-148b mimic reduced DNMT1 expression and increased PCDH10 expression compared to the control in MKN 28 and MKN 74 cells (Fig. 4F). HOTAIR overexpression increased DNMT1 expression and reduced PCDH10 expression in AGS cell; however, treatment with miR-148b mimic showed the opposite result (Fig. 4F). Finally, we found that DNMT1 was downregulated after si-HOTAIRs, and it was restored after treatment with miR-148b inhibitor in MKN 28 and MKN 74 cells (Fig. 4G).