DNA demethylases of the TET family function as tumor suppressors in various human cancers, but their pathogenic contributions and mechanisms of action in gastric carcinogenesis and progression remain unclear. Here, we report that TET is transcriptionally upregulated in gastric cancer, where it correlates with poor prognosis. Mechanistic investigations revealed that TET facilitated gastric carcinogenesis through a noncoding function of the 3′UTR, which interacted with miR-26. This interaction resulted in sequestration of miR-26 from its target EZH2, which released the suppression on EZH2, and thereby led to EZH2 overexpression in gastric cancer. Our findings uncover a novel noncoding function for TET family proteins in facilitating gastric carcinogenesis. Cancer Res; 77(22); 6069–82. ©2017 AACR.

Gastric cancer is one of the most common malignancies and remains the second leading cause of cancer-related mortality worldwide (1). Despite the improvement in surgical techniques and targeted drug chemotherapy, the five-year overall survival rate for advanced gastric cancer remains unsatisfactory (2). Better understanding of the molecular mechanisms involved in this disease and developing of new biomarkers as well as more effective targeted therapeutic strategies are needed to improve patient outcomes.

The TET protein family including three members (TET1, TET2, and TET3) can catalyze the oxidation of 5-methyl-cytosine (5-mC) to 5-hydroxymethyl-cytosine (5-hmC) and promote DNA demethylation and has been shown to play a role in tumorigenesis (3). TET2-inactivating deletions and mutations frequently occur in hematopoietic malignancies and genetic inactivation of TET2 in the mouse hematopoietic system results in a CMML-like phenotype (4–6). Decreased expression of TET proteins and loss of 5-hmC have been reported in several solid tumors, with TET1 and TET2 exhibiting tumor-suppressive activities in breast, colon, and liver cancers (7–10). However, a recent report showed that TET1 plays a critical oncogenic role in MLL-rearranged leukemia through cooperation with MLL fusion proteins (11). To date, the role of TETs and their action model in gastric cancer remain elusive.

MicroRNA-26 (miR-26), a functional miRNA, has been found to function as a tumor suppressor in various cancers through targeting different oncogenes (12–16). Recently, we described downregulation of this miRNA in primary gastric cancer specimens (13). More importantly, functional in vitro and in vivo studies demonstrated that miR-26 suppressed gastric cancer cell proliferation, invasion, and tumor growth in mice (13). Intriguingly, TET genes have been discovered as direct targets of miR-26, with 4, 6, and 4 conserved binding sites in the 3′UTRs of TET1/2/3, respectively (17). Recently, multiple RNA transcripts have been validated in a number of studies to regulate other RNAs by competitively binding to shared miRNAs; thus, they act as endogenous decoys for miRNAs or competing endogenous RNAs (18–23). Both protein-coding mRNAs and noncoding RNAs such as pseudogenes, long noncoding RNAs (lncRNA), and circular RNAs have exhibited competing endogenous RNAs (ceRNA) activity in diverse species (24).

In the current study, we demonstrated that the three TET genes were overexpressed at the transcriptional level in gastric cancer and their overexpression correlated with a shortened survival in gastric cancer patients. Functional studies revealed that the TET family played an essential oncogenic role in gastric carcinogenesis and this effect was mediated by TETs as ceRNAs strongly enhancing EZH2 via competitively binding to miR-26.

Clinical samples

The gastric cancer tissue samples were collected from two separate cohorts. For cohort A, a total of 33 paired primary gastric cancer specimens and their adjacent tissues were collected at Cancer Hospital and Cancer Research Institute, Guangzhou Medical University. For cohort B, 76 pairs of gastric cancer and the matched normal tissues were obtained from patients at the affiliated Nanhua Hospital, University of South China and the Xiangya Hospital, Central South University, and follow-up information was also collected. All tissues were obtained immediately after surgery and stored at −80°C or paraffin-embedded. This study was approved by the Institutional Review Board of Guangzhou Medical University, and written informed consent was provided by all patients based on the Declaration of Helsinki. Clinical and histopathological characteristics of the patients are described in Supplementary Table S1.

Cell culture

HEK-293T cells were obtained in 2009 from the ATCC. The gastric cancer cell lines MGC-803, SGC-7901, BGC-823, AGS, MKN-45, and MKN-28 were obtained from the Chinese Academy of Medical Science and the gastric epithelial cell line GES-1 were from the Beijing Institute for Cancer Research in 2011. Cell lines involved in our experiments were reauthenticated by short tandem repeat analysis every 6 months after resuscitation in our laboratory. These cells were cultured in RPMI-1640 medium (Gibco) supplemented with 10% fetal bovine serum (Gibco), 100 U/mL penicillin sodium, and 100 mg/mL streptomycin sulfate at 37°C in 5% CO2.

RNA interference

Recombinant shRNA lentiviruses containing shTET1#1-3, shTET2#1-3, or shEZH2#1-2 were purchased from Sigma-Aldrich, while lentiviruses containing shTET3#1-3 were from FulenGen. The target sequences for TET1, TET2, TET3, and EZH2 were as follows: shTET1#1: 5′-TTGTGCCTCTGGAGGTTATAA-3′; shTET1#2: 5′-CCTCCAGTCTTAATAAGGTTA-3′; shTET3#3: 5′-GCAGCTAATGAAGGTCCAGAA-3′; shTET2 #1: 5′-CAGATGCACAGGCCAATTAAG-3′; shTET2#2: 5′-TTTCACGCCAAGTCGTTATTT-3′; shTET2#3: 5′-CCTCAAGCATAACCCACCAAT-3′; shTET3#1: 5′-GCGATTGCGTCGAACAAATAG-3′; shTET3#2: 5′-GGACAATCCCAAAGAGGAAGA-3′; shTET3#3: 5′-GCAGTTTGAGGCTGAATTTGG-3′; shEZH2#1: 5′-CCAACACAAGTCATCCCATTA-3′; shEZH2#2: 5′-TATGATGGTTAACGGTGATCA-3′.

Establishment of stable cell lines

SGC-7901 cells were infected with shRNA lentivirus targeting different regions of the TETs mRNA transcripts, and the cells stably expressing effective shRNAs (shTET1#2, shTET2#2, and shTET3#1) were generated by selection with 5 mg/L puromycin (Invivogen). To establish cell lines stably expressing the wild-type or the mutant type 3′UTR of TET members, MKN-28 cells were transfected with the expression plasmid pTET1-3U, p-TET3U, p-TET3-3U and the respective expression vectors with miR-26-binding site mutations in TET1/2/3 3′UTR, and were selected with 400 mg/L G418 (ThermoFisher Scientific). For generation of stable EZH2-depleted cell line, SGC-7901 cells were infected with recombinant shRNA lentiviruses containing shEZH2#2 or control shRNA and then incubated with 5 mg/L puromycin for selection.

Modulation of miR-26 Expression

LNA miR-26a mimics and LNA miR-26a/b inhibitors were obtained from Exiqon. miR-26a mimics were transfected at 20 nmol/L for upregulation of miR-26a in cultured cells. To knockdown miR-26, the cells were transfected with LNA miR-26 inhibitors at a final concentration of 50 nmol/L.

qRT-PCR

Total RNA was isolated using TRIzol reagent (Invitrogen). For miRNA analysis, reverse transcription and qPCR were performed using an all-in-one miRNA qRT-PCR Detection kit (GeneCopoeia). For mRNA detection, total RNA was reverse-transcribed using a Reverse Transcription System Kit (Promega) and qPCR was performed using a Power SYBR Green PCR Master Mix (ThermoFisher Scientific). Primer sequences are provided in Supplementary Table S2.

Antibodies

Antibodies against TET1 (Santa Cruz Biotechnology), TET2 (Abcam), TET3 (Santa Cruz Biotechnology), EZH2 (Cell Signaling Technology), 5-hmC (Active Motif), E-cadherin (Cell Signaling Technology), RUNX3 (Bioworld Technology), GFP antibody (Cell Signaling Technology), and β-actin (Santa Cruz Biotechnology) were used in this study.

Immunohistochemistry

Immunohistochemistry was performed under standard procedures. The positive percentage of TET1, TET2, TET3, or EZH2 was graded from 0 to 4 (0, <5% of positive cells; 1, 5%–24% positive cells; 2, = 25%–50% positive cells; 3, 51%–75% positive cells; 4, >75% positive cells). The slides were analyzed by two independent pathologists.

Western blotting

Western blot analysis was performed using standard procedures. Briefly, total proteins were extracted and separated by 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto a polyvinylidene difluoride membrane. The membrane was then incubated with primary antibodies, followed by horseradish peroxidase-labeled secondary antibody (Santa Cruz Biotechnology) and detected by chemiluminescence.

Luciferase reporter assay

The reporter vector Luc-TET1-3UTR, Luc-TET2-3UTR, or Luc-TET3-3UTR was cotransfected with miR-26a, miR-26b, or control mimics into HEK-293T cells. Firefly and Renilla luciferase activities were detected with the Dual-Luciferase Reporter system (Promega) after 48 hours. On the other hand, the reporter plasmid Luc-EZH2-3UTR was cotransfected with the expression vectors pTET1-3U, pTET2-3U, pTET3-3U, or the respective mutation vectors with miR-26-binding site mutations in TET1/2/3 3′UTR into HEK-293T cells. Meanwhile, HEK-293T cells infected with TET1/2/3 shRNA lentiviruses were transfected with Luc-EZH2-3UTR plasmid. At 48 hours after transfection, firefly and Renilla luciferase activities were determined as above.

Chromatin immunoprecipitation

The chromatin immunoprecipitation assay was performed using a chromatin Immunoprecipitation Assay Kit (Millipore). After cells were crosslinked, lysed, and then sonicated, immunoprecipitation was performed using anti-EZH2, anti-H3K27me3, or IgG. The precipitated DNA was subjected to real-time PCR.

RNA immunoprecipitation assay

HEK-293T cells were cotransfected with MS2bs overexpressing vectors (MS2bs, MS2bs-TET1-3UTR, MS2bs-TET2-3UTR, MS2bs-TET3-3UTR, or MS2bp-EZH2-3UTR) and MS2bp-GFP overexpressing plasmid (Addgene). Meanwhile, HEK-293T cells expressing TET1/2/3 shRNA were transfected with the MS2bp-EZH2-3UTR. At 48 hours after transfection, cells were used to perform RNA immunoprecipitation (RIP) experiments using an anti-GFP antibody and the Magna RIP Immunoprecipitation Kit (Millipore) according to the manufacturer's instructions. The RNA was isolated with TRIzol reagent for further analysis.

Cell proliferation assay

Cells were seeded into 6-well plates, and the cell numbers were counting after 0 day, 1 day, 2 days, 3 days, 4 days, and 5 days of incubation using a Coulter Counter (Beckman Coulter) in triplicate.

Cell viability assay

Cells were seeded into 96-well plates and MTS assay was performed according to the manufacturer's instructions (Promega) after 0 days and 4 days of incubation.

Colony formation assay

Six-well plates were covered with a layer of 0.6% agar in medium supplemented with 20% fetal bovine serum. A total of 1,000 cells were prepared in 0.3% agar and cultured for 2 weeks at 37°C for two weeks. The numbers of colonies per well were counted.

Cell invasion assay

Cells in serum-free medium were seeded into the upper chamber with Matrigel in the insert of a 24-well culture plate (BD Biosciences) and 15% fetal bovine serum was added to the lower chamber as a chemoattractant. After 72 hours of incubation, invasive cells adhering to the lower membrane of the inserts were stained with crystal violet, counted, and imaged.

Cell cycle and apoptosis assay

For cell cycle analysis, cells were fixed in 70% ethanol and stained with 100 μg/mL propidium iodide. For apoptosis assays, cells were stained with propidium iodine/Annexin V-FITC staining (BD Biosciences). The cell cycle and apoptosis were then analyzed by flow cytometry FACS Calibur instrument (BD Biosciences).

Mouse xenograft model

All animal studies were approved by the Institutional Animal Care and Use Committee (IACUC) of Guangzhou Medical University. Standard animal care and laboratory guidelines were followed according to the IACUC protocol. Four-week-old female BALB/c nude mice were injected subcutaneously with 2 × 106 cancer cells (5 mice per group). Tumors were measured every 5 days and tumor volumes were calculated by the formula: volume = length × (width/2)2. Mice were sacrificed after 30 days of implantation and tumors were excised and weighed.

Statistical analysis

A Student t test or a χ2 test was used for two-sample comparisons. Differences among three or more groups were analyzed with a two-way analysis of variance. Overall survival curves were plotted using the Kaplan–Meier method, and survival differences were evaluated using a log-rank test. Pairwise expression correlation was analyzed by Pearson correlation tests. A P < 0.05 was considered as statistically significant.

Additional methods

Additional experimental methods can be found in the Supplementary Methods.

TET transcripts are significantly overexpressed in gastric cancer and correlate with prognosis of gastric cancer patients

To explore the potential role of TET genes in gastric cancer, we first detected their expression in 33 pairs of gastric cancer and the adjacent normal tissues (cohort A) using qRT-PCR and immunohistochemical staining. As shown in Fig. 1A, the mRNA levels of all three TET members were significantly elevated in gastric cancer compared with the corresponding normal tissues (Fig. 1A). Notably, the transcriptional levels of TETs were higher in higher-grade gastric cancer than that in lower-grade gastric cancer samples (Fig. 1B). However, we observed no conspicuous change in either TET proteins or 5-hmC levels in gastric cancer samples in comparison with adjacent normal tissues (Supplementary Fig. S1A). Thus, the transcriptional up-regulation of the TET family in gastric cancer may not result in increased protein levels.

Figure 1.

TET transcripts are upregulated in primary gastric cancer specimens and associated with poor clinical outcomes of gastric cancer patients. A, qRT-PCR analysis of the RNA levels of the TET family in 33 paired gastric cancer and adjacent normal tissues (cohort A). Using relative quantification methods, the mRNA expression was presented as 2−delCt. B, Assessment of TET1/2/3 RNA levels in 33 gastric cancer tissues according to their tumor stage. C, The transcriptional levels of TET1/2/3 in 76 pairs of gastric cancer and normal tissues (cohort B). The mRNA expression was presented as 2−delCt. D, Kaplan–Meier analysis of overall survival for gastric cancer patients between high and low TET expression groups based on each TET gene expression levels greater or less than the median.

Figure 1.

TET transcripts are upregulated in primary gastric cancer specimens and associated with poor clinical outcomes of gastric cancer patients. A, qRT-PCR analysis of the RNA levels of the TET family in 33 paired gastric cancer and adjacent normal tissues (cohort A). Using relative quantification methods, the mRNA expression was presented as 2−delCt. B, Assessment of TET1/2/3 RNA levels in 33 gastric cancer tissues according to their tumor stage. C, The transcriptional levels of TET1/2/3 in 76 pairs of gastric cancer and normal tissues (cohort B). The mRNA expression was presented as 2−delCt. D, Kaplan–Meier analysis of overall survival for gastric cancer patients between high and low TET expression groups based on each TET gene expression levels greater or less than the median.

Close modal

To investigate the clinicopathological implication of TET1/2/3 transcriptional levels in gastric cancer patients, we analyzed TET family expression in another cohort of 76 pairs of normal and gastric cancer tissues (cohort B). Consistent with our studies with cohort A specimens, the transcriptional levels of TETs were significantly upregulated in gastric cancer specimens relative to matched normal tissues (Fig. 1C). Moreover, the increased transcripts of TETs in gastric cancer were significantly associated with tumor invasion depth, clinical stage, lymph node metastasis, and poor overall survival of patients (Fig. 1D; Supplementary Table S3). Univariate and multivariate analyses revealed that clinical stage and TET1/3 were independent risk factors for overall survival (Table 1). In addition, there was a significant correlation between the expression of TET1/2 and TET2/3 expression in these gastric cancer specimens (Supplementary Fig. S1B).

Table 1.

Univariate and multivariate analyses of factors associated with overall survival

Univariate analysisMultivariate analysis
ViableHRP valueHRP value
Age (<60 vs. ≥60) 0.988 0.969 
Gender (male vs. female) 0.900 0.753 
Lauren (intestinal vs. nonintestinal) 1.584 0.168 
Histologic grade (well and moderate vs. poor) 1.402 0.348 
Invasion depth (bm/sm/mp vs.ss/se/si) 0.454 0.213 
TNM stage (I/II vs. III/IV) 4.480 0.037 2.342 0.007 
Lymph node metastasis (absent vs. present) 1.525 0.404 
TET1 expression 2.210 0.018 2.053 0.018 
TET2 expression 1.088 0.818 
TET3 expression 1.843 0.050 2.062 0.015 
Univariate analysisMultivariate analysis
ViableHRP valueHRP value
Age (<60 vs. ≥60) 0.988 0.969 
Gender (male vs. female) 0.900 0.753 
Lauren (intestinal vs. nonintestinal) 1.584 0.168 
Histologic grade (well and moderate vs. poor) 1.402 0.348 
Invasion depth (bm/sm/mp vs.ss/se/si) 0.454 0.213 
TNM stage (I/II vs. III/IV) 4.480 0.037 2.342 0.007 
Lymph node metastasis (absent vs. present) 1.525 0.404 
TET1 expression 2.210 0.018 2.053 0.018 
TET2 expression 1.088 0.818 
TET3 expression 1.843 0.050 2.062 0.015 

Abbreviations: bm, tumor invasion of mucosa; mp, muscularis propria; se, serosa penetration; si, invasion to adjacent structures; sm, submocosa; ss, subserosa.

Knockdown of TETs inhibits gastric cancer growth and cell proliferation

To investigate the effects of TET family members on gastric carcinogenesis, the SGC-7901 cell line (endogenous expression levels of TET family in gastric cell lines were shown in Supplementary Fig. S1C) was selected to stably deplete TET members with effective shRNAs (shTET1#2, shTET2#2, and shTET3#1) using a lentiviral system (Fig. 2A; Supplementary Fig. S1D). Because all three TET proteins can convert 5-mC to 5-hmC, we assessed global 5-hmC level in TET-depleted cells. Dot blotting analysis revealed a concomitant reduction in 5-hmC levels when TET family genes were knocked down (Supplementary Fig. S1E). These cells were then subcutaneously injected into nude mice. We found that knockdown of any TET member greatly decreased tumor growth as compared with the controls (Fig. 2B), indicating that TETs may facilitate gastric cancer tumor growth in vivo.

Figure 2.

Knockdown of TETs inhibits gastric cancer cell proliferation and growth in vivo and in vitro. A, Western blotting analysis was performed to assess the inhibition efficiency in SGC-7901 cells infected with control or TET1/2/3 shRNA lentiviruses. B, Knockdown of TETs inhibited gastric cancer tumor growth in a nude mouse model. SGC-7901 cells infected with control or TET1/2/3 shRNA lentiviruses were inoculated subcutaneously into nude mice and the mice were closely monitored for tumor growth. The mice were sacrificed and tumors were removed at 30 days after inoculation. Error bars, SD (n = 5 mice/group). C, Cell proliferation and viability assays in SGC-7901 cells infected with control or TET1/2/3 shRNA lentiviruses. Error bars, SD from three independent experiments performed in triplicate. D, Western blotting and cell proliferation analyses in AGS cells expressing control shRNA or TET1/2/3 shRNA. Error bars, SD from three independent experiments performed in triplicate. E, Soft-agar colony formation assay was performed in both SGC-7901 and AGS cells expressing TET1/2/3 shRNA or control shRNA. Error bars, SD from three independent experiments performed in triplicate. F, Western blotting analysis of protein expression in SGC-7901 cells stably expressing TET1/2/3 shRNA or control shRNA transfected with the corresponding TET-CDS (the synonymous mutant) plasmids or control vector. pcDNA represents empty vector pcDNA3.1. G, Cell viability assay and soft-agar colony formation assay with the indicated SGC-7901 cells transfected as noted. Error bars show SD from three independent experiments performed in triplicate. *, P < 0.05; **, P < 0.01; ***, P < 0.001. NS, not significant.

Figure 2.

Knockdown of TETs inhibits gastric cancer cell proliferation and growth in vivo and in vitro. A, Western blotting analysis was performed to assess the inhibition efficiency in SGC-7901 cells infected with control or TET1/2/3 shRNA lentiviruses. B, Knockdown of TETs inhibited gastric cancer tumor growth in a nude mouse model. SGC-7901 cells infected with control or TET1/2/3 shRNA lentiviruses were inoculated subcutaneously into nude mice and the mice were closely monitored for tumor growth. The mice were sacrificed and tumors were removed at 30 days after inoculation. Error bars, SD (n = 5 mice/group). C, Cell proliferation and viability assays in SGC-7901 cells infected with control or TET1/2/3 shRNA lentiviruses. Error bars, SD from three independent experiments performed in triplicate. D, Western blotting and cell proliferation analyses in AGS cells expressing control shRNA or TET1/2/3 shRNA. Error bars, SD from three independent experiments performed in triplicate. E, Soft-agar colony formation assay was performed in both SGC-7901 and AGS cells expressing TET1/2/3 shRNA or control shRNA. Error bars, SD from three independent experiments performed in triplicate. F, Western blotting analysis of protein expression in SGC-7901 cells stably expressing TET1/2/3 shRNA or control shRNA transfected with the corresponding TET-CDS (the synonymous mutant) plasmids or control vector. pcDNA represents empty vector pcDNA3.1. G, Cell viability assay and soft-agar colony formation assay with the indicated SGC-7901 cells transfected as noted. Error bars show SD from three independent experiments performed in triplicate. *, P < 0.05; **, P < 0.01; ***, P < 0.001. NS, not significant.

Close modal

The biological function of TETs in gastric cancer cells was also observed in vitro. Figure 2C and D showed that knockdown of TETs in SGC-7901 and AGS cells remarkably inhibited cell proliferation. Soft-agar colony formation assays revealed that knockdown of TETs markedly reduced colony number and size (Fig. 2E). Moreover, knockdown of TETs resulted in a significant increase of cells in G1 phase and a concomitantly notable decrease of cells in S phase and G2–M phase (Supplementary Fig. S2A). In contrast, the level of apoptosis was not significantly changed following TET1/2/3 knockdown (Supplementary Fig. S2B). In addition, matrigel invasion assay showed downregulation of TETs impaired gastric cancer cell invasion (Supplementary Fig. S2C). Subsequently, we constructed shRNA-resistant vectors (containing a synonymous mutant coding region of the corresponding TET gene) and then transfected these vectors into TET-depleted cells to restore the expression of TET proteins (Fig. 2F). Much to our surprise, far from rescuing the inhibition of cell growth by TET1 knockdown, ectopic expression of TET1 protein further aggravated this inhibition. Moreover, the inhibitory role of TET2/3 knockdown in gastric cancer growth was not affected by re-expression of the TET2/3 protein (Fig. 2G), implicating that the TET family might exert an essential oncogenic role independently of the protein in gastric cancer.

TET-induced GC growth relies on the 3′UTR and miR-26

To determine whether the TET family could operate independently of the protein in gastric cancer tumorigenesis, the 5′UTR, CDS and 3′UTR fragments of TET1, TET2, and TET3, as well as the full-length cDNA of TET genes were cloned into the expression vector (Supplementary Fig. S3), and these constructs were transfected into MKN-28 cells that expressed relatively low levels of the endogenous TET family (Supplementary Fig. S1C). Transfection of both the CDSs and the full-length cDNAs efficiently induced corresponding TET protein expression (compared with the CDS constructs, the full-length cDNA constructs more modestly express TET protein owing to the presence of miR-26–mediated suppression as mentioned later), while transfection of either 5′UTRs or 3′UTRs could not code for proteins (Supplementary Fig. S4A and S4B). TET2-CDS and TET3-CDS overexpression had no significant effect on cell growth while TET1-CDS resulted in a mild decrease. The growth ability of gastric cancer cells was also unaffected upon overexpression of the 5′UTRs. Nevertheless, overexpression of TET1/2/3 3′UTR or full-length cDNA markedly promoted cell growth (Fig. 3A and B). These data suggest that the oncogenic effect of the TET family on gastric cancer cells might be independent of the CDS (or protein) but instead relied upon the 3′UTR. Our studies also suggest that the CDSs and 3′UTRs of the TET genes may play different roles in gastric cancer.

Figure 3.

The TET-induced gastric cancer growth is 3′UTR- and miR-26–dependent. A, Cell viability and colony formation analyses of MKN-28 cells transfected with TET1/2/3 3′UTR, mutant 3′UTR, 5′UTR, CDS, and full-length cDNA constructs or empty vector. Error bars, SD from three independent experiments performed in triplicate. B, Cell cycle analysis of MKN-28 cells transfected with TET1/2/3 3′UTR and mutant 3′UTR constructs or control vector. C, Overexpression of miR-26a inhibited TET UTR-induced elevation of cell proliferation and soft-agar growth. Error bars, SD from three independent experiments performed in triplicate. D, Knockdown of miR-26 restored cell proliferation and soft-agar growth in TET-depleted SGC-7901 cells. Error bars, SD from three separate experiments performed in triplicate. E, Overexpression of wild-type but not mutant TET 3′UTRs promoted gastric cancer tumor growth in the xenograft nude mouse model. MKN-28 cells stably transfected with control, TET1/2/3 3′UTR, or mutant TET1/2/3 3′UTR plasmids were inoculated subcutaneously into nude mice and the mice were closely monitored for tumor growth. At 30 days after inoculation, the mice were sacrificed and the tumors were removed. Error bars, SD (n = 5 mice/group). *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 3.

The TET-induced gastric cancer growth is 3′UTR- and miR-26–dependent. A, Cell viability and colony formation analyses of MKN-28 cells transfected with TET1/2/3 3′UTR, mutant 3′UTR, 5′UTR, CDS, and full-length cDNA constructs or empty vector. Error bars, SD from three independent experiments performed in triplicate. B, Cell cycle analysis of MKN-28 cells transfected with TET1/2/3 3′UTR and mutant 3′UTR constructs or control vector. C, Overexpression of miR-26a inhibited TET UTR-induced elevation of cell proliferation and soft-agar growth. Error bars, SD from three independent experiments performed in triplicate. D, Knockdown of miR-26 restored cell proliferation and soft-agar growth in TET-depleted SGC-7901 cells. Error bars, SD from three separate experiments performed in triplicate. E, Overexpression of wild-type but not mutant TET 3′UTRs promoted gastric cancer tumor growth in the xenograft nude mouse model. MKN-28 cells stably transfected with control, TET1/2/3 3′UTR, or mutant TET1/2/3 3′UTR plasmids were inoculated subcutaneously into nude mice and the mice were closely monitored for tumor growth. At 30 days after inoculation, the mice were sacrificed and the tumors were removed. Error bars, SD (n = 5 mice/group). *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Close modal

Recently, some RNA transcripts have been reported to function as endogenous miRNA sponges, or ceRNAs, functionally regulating other transcripts by competitively binding to shared miRNAs (19, 21–23). TET family members have been previously reported as miR-26 targets with multiple conserved binding sites in the 3′UTR (the 3′UTR of TET1/2/3 contains 4, 6, and 4 conserved miR-26 binding sites, respectively; Supplementary Fig. S3; ref. 17). Of note, our group and others have previously shown that miR-26 functions as a tumor suppressor in gastric cancer and other numerous types of tumors (12–14, 25, 26), which is in disagreement with the oncogenic role of the TET family in gastric cancer discovered by the current study. We hypothesized that TETs might serve as ceRNAs for the miR-26 family to promote gastric carcinogenesis. To address this, we first verified whether TETs were direct targets of miR-26 in gastric cancer cells. Indeed, the expression levels of all three TET members were reduced in miR-26a/b-overexpressing cells, whereas miR-26 inhibition elevated the levels of this family (Supplementary Fig. S5A). Furthermore, luciferase reporter assays showed that miR-26a/b overexpression caused a significant decrease in luciferase reporter activities under the control of the 3′UTR of TET1/2/3, whereas the mutation of the miR-26 binding sites in TET1/2/3 3′UTR abolished the suppressive effects (Supplementary Fig. S5B). Dot blotting assays revealed that 5-hmC levels were diminished when miR-26a/b was overexpressed, and increased when miR-26 was inhibited (Supplementary Fig. S5C). These data strongly demonstrate that TETs are the bona fide targets of miR-26a/b.

To further investigate the miR-26 dependency of TET-induced tumorigenesis, we established the mutant TET1/2/3 3′UTR expression constructs in which the miR-26-binding sites were mutated (Supplementary Fig. S3), and transfected them into MKN-28 cells. We found that the mutation of the miR-26 binding sites offset the 3′UTR-induced neoplastic phenotypes, evidenced by unchanged cell proliferation, colony formation, and cell cycle progression (Fig. 3A and B). More importantly, ectopic expression of miR-26a attenuated the cell malignant phenotypes elicited by overexpression of the wild-type TET 3′UTRs (Fig. 3C), while inhibition of miR-26 rescued the phenotypes induced by the specific depletion of TETs (Fig. 3D).

To examine the effect of TET ceRNA activity on gastric cancer growth in vivo, MKN-28 cells stably expressing either wild-type or mutated TET 3′UTRs were subcutaneously transplanted into nude mice. As shown in Figure 3E, overexpression of the wild-type TET1/2/3 3′UTR caused a significant increase in tumor growth, whereas the mutation in miR-26–binding sites abolished this effect. Collectively, these results indicate that TET competing RNA activity dramatically promotes gastric cancer growth and that this effect is dependent on miR-26.

We next evaluated whether TETs might regulate miR-26 expression in gastric cancer cells. Our data showed that neither specific knockdown of TETs nor overexpression of TET 3′UTRs (wild-type or mutant) influenced miR-26a/b expression (Supplementary Fig. S5D).

TET ceRNAs elevate EZH2 expression by sequestering miR-26

To further explore the molecular basis of TETs acting as ceRNAs in gastric carcinogenesis, we examined which of the previously reported targets of miR-26 were altered in response to miR-26 depletion in GES-1 cells. As seen in Supplementary Fig. S6A, the expression of 10 genes, out of 31 targets, was increased more than 1.5-fold upon inhibition of miR-26. In parallel, we profiled the expression levels of these targets in GES-1 cells transfected with the TET2 3′UTR or empty vectors. The results showed that the expression of 6 genes was increased more than 1.5-fold (Supplementary Fig. S6B). Interestingly, only three genes (EZH2, NOS2, and SODD) were consistently upregulated in both miR-26–inhibiting cells and TET2 3′UTR-overexpressing cells. Given the established oncogenic role of EZH2 in multiple human malignancies (27–29), we selected EZH2 as a potential candidate for further studies.

We first detected the regulatory role of TETs in the expression of EZH2 mRNA and protein. As shown in Fig. 4A; Supplementary Fig. S6C–S6E, the wild-type, but not the mutant, 3′UTRs of TETs substantially increased EZH2 expression in MKN-28 and GES-1 cells, which was validated in the gastric cancer xenografts from animal experiments. Moreover, ectopic expression of miR-26a abrogated this increase (Fig. 4B). Conversely, specific knockdown of TETs decreased EZH2 expression in SGC-7901 cells, and inhibition of miR-26 overcame the decrease of EZH2 (Fig. 4C; Supplementary Fig. S6C). We then determined the effect of TET ceRNAs on the expression of known EZH2 targets such as CDH1 (encodes E-cadherin; ref. 30), RUNX3 (31), and miR-218 (32). As expected, we found a significant inhibition of target gene expression in cells expressing the wild-type 3′UTRs, and overexpression of miR-26a rescued this inhibition (Fig. 4A and B; Supplementary Fig. S6C and S6D). Western blot analysis of the xenograft tumor tissues confirmed reduced E-cadherin and RUNX3 proteins in TET1/2/3 3′UTR-overexpressing tumors, but not in the mutant 3′UTR-overexpressing tumors (Supplementary Fig. S6E). Consistent with these findings, specific knockdown of TETs in SGC-7901 cells upregulated expression of EZH2 targets, and this upregulation was abolished upon inhibition of miR-26 (Fig. 4C; Supplementary Fig. S6C). Additional studies showed that the expression levels of both EZH2 and the targets were changed by overexpression of TET1/2/3 full-length cDNA but not by transfection of TET1/2/3 CDS (Supplementary Fig. S6F).

Figure 4.

TET ceRNA activity regulates EZH2 expression in a miR-26–dependent manner. A, MKN-28 cells were transfected with empty vector or expression vectors containing the wild-type or mutant TET1/2/3 3′UTR. The protein levels of EZH2, E-cadherin, and RUNX3 were measured by Western blotting. B, Western blotting analysis of the protein expression in MKN-28 cells cotransfected with wild-type TET1/2/3 3′UTR vectors and miR-26a/control mimics. C, The protein levels of EZH2, E-cadherin, and RUNX3 were measured in TET-depleted SGC-7901 cells transfected with either miR-26 or negative control inhibitors. D and E, Chromatin immunoprecipitation followed by qRT-PCR showed that overexpression of wild-type TET1/2/3 3′UTR in MKN-28 cells increased, but knockdown of TETs in SGC-7901 cells decreased, both EZH2 (D) and H3K27me3 (E) enrichment at the promoters of the PRC2 target genes. The GAPDH gene was used as a negative control. Error bars, SD from three independent experiments. F, TET1/2/3 3′UTR coexpression in HEK-293T cells stimulated the luciferase activities of a luciferase reporter containing EZH2 3′UTR. Inhibition of TET genes, however, significantly reduced the luciferase activities. Error bars, SD from three independent experiments. G, Left, the experimental MS2-RIP model examined by miRNA qRT-PCR to detect miRNAs associated with the TET1/2/3 3′UTR and EZH2 3′UTR. Middle, MS2-RIP examined by qRT-PCR to identify both miR-26a and miR-26b endogenously associated with TET1/2/3 3′UTR and EZH2 3′UTR in HEK-293T cells. Right, MS2-RIP followed by qRT-PCR showed an increased association of miR-26 with EZH2 3′UTR in HEK-293T cells depleted for TET1, TET2, or TET3. Empty MS2bs vector and miR-191 were used as negative controls. Error bars, SD from three independent experiments. *, P < 0.05; **, P <0.01; ***, P < 0.001.

Figure 4.

TET ceRNA activity regulates EZH2 expression in a miR-26–dependent manner. A, MKN-28 cells were transfected with empty vector or expression vectors containing the wild-type or mutant TET1/2/3 3′UTR. The protein levels of EZH2, E-cadherin, and RUNX3 were measured by Western blotting. B, Western blotting analysis of the protein expression in MKN-28 cells cotransfected with wild-type TET1/2/3 3′UTR vectors and miR-26a/control mimics. C, The protein levels of EZH2, E-cadherin, and RUNX3 were measured in TET-depleted SGC-7901 cells transfected with either miR-26 or negative control inhibitors. D and E, Chromatin immunoprecipitation followed by qRT-PCR showed that overexpression of wild-type TET1/2/3 3′UTR in MKN-28 cells increased, but knockdown of TETs in SGC-7901 cells decreased, both EZH2 (D) and H3K27me3 (E) enrichment at the promoters of the PRC2 target genes. The GAPDH gene was used as a negative control. Error bars, SD from three independent experiments. F, TET1/2/3 3′UTR coexpression in HEK-293T cells stimulated the luciferase activities of a luciferase reporter containing EZH2 3′UTR. Inhibition of TET genes, however, significantly reduced the luciferase activities. Error bars, SD from three independent experiments. G, Left, the experimental MS2-RIP model examined by miRNA qRT-PCR to detect miRNAs associated with the TET1/2/3 3′UTR and EZH2 3′UTR. Middle, MS2-RIP examined by qRT-PCR to identify both miR-26a and miR-26b endogenously associated with TET1/2/3 3′UTR and EZH2 3′UTR in HEK-293T cells. Right, MS2-RIP followed by qRT-PCR showed an increased association of miR-26 with EZH2 3′UTR in HEK-293T cells depleted for TET1, TET2, or TET3. Empty MS2bs vector and miR-191 were used as negative controls. Error bars, SD from three independent experiments. *, P < 0.05; **, P <0.01; ***, P < 0.001.

Close modal

By chromatin immunoprecipitation (ChIP), we discovered that ectopic expression of wild-type, but not mutant, 3′UTRs of TETs facilitated an increased occupancy of EZH2 at the promoters of CDH1 and RUNX3, whereas specific knockdown of TETs significantly reduced the occupancy of EZH2 at the promoters (Fig. 4D). Because EZH2 was shown to regulate gene expression by trimethylating H3K27 (H3K27me3), we analyzed the level of promoter occupancy of the H3K27 histone mark. The occupancy of H3K27me3 at the promoters of CDH1 and RUNX3 was elevated when the wild-type 3′UTRs were ectopically expressed in MKN-28 cells, whereas inhibition of endogenous TETs showed the opposite effect (Fig. 4E).

Next, we determined the ability of TETs to decoy the miR-26 family from EZH2. The luciferase reporter vectors containing EZH2 3′UTR were cotransfected with the expression constructs, including either the wild-type or mutant 3′UTR of TETs into HEK293T cells. From luciferase activity analysis, we found that overexpression of the wild-type, but not the mutant, 3′UTRs of TETs significantly enhanced the luciferase activities of the EZH2 reporter vector (Fig. 4F, left). However, knockdown of TETs significantly reduced the luciferase activities (Fig. 4F, right). For further validation, the TET 3′UTRs or EZH2 3′UTR were cloned into a plasmid to transcribe RNA combined with MS2-binding sequences (MS2b) and then cotransfected with an MS2bp-GFP expression vector into HEK-293T cells. Subsequently, RIPs were performed with GFP antibodies (Fig. 4G, left). By qRT-PCR, the miR-26 family endogenously associated with TET 3′UTRs and EZH2 3′UTR, while the nontargeting microRNA (miR-191) did not with the 3′UTRs (Fig. 4G, middle). Importantly, knockdown of TETs increased the association of miR-26 with EZH2 3′UTR (Fig. 4G, right). Taken together, these data indicated that the TETs sequestered the miR-26 family to target EZH2.

To serve as a ceRNA, the abundance of TETs should be comparable with or higher than miR-26. We therefore used quantitative RT-PCR to quantify the exact copy numbers of TETs, miR-26 and EZH2. Absolute qPCR showed that there were roughly 150–250 copies of TET1/2/3 per SGC-7901 cell, whereas the copy numbers of miR-26s and EZH2 were about 50–150 and 300 per cell, respectively (Supplementary Fig. S6G). To determine whether the coexpression of TETs, miR-26 and EZH2 could be observed in gastric cancer cells, we performed RNA FISH assay to visualize the cellular distribution. Indeed, RNA FISH assay showed that TETs colocalize with miR-26 and EZH2 in the cytoplasm of gastric cancer cells (Supplementary Fig. S7). Based on our results, the TET family may be able to function as a ceRNA for miR-26.

EZH2 mediates TET-induced gastric carcinogenesis

The above results prompted us to examine whether EZH2 mediates TET-induced gastric carcinogenesis. To this end, we specifically inhibited EZH2 expression in SGC-7901 cells via two shRNAs. The knockdown effect by EZH2 shRNA was confirmed by Western blotting assay (Fig. 5A). As expected, downregulation of EZH2 significantly reduced EZH2 and H3K27me3 occupancies at the promoters of CDH1 and RUNX3 (Fig. 5B). In line with our findings, there was a clear upregulation of E-cadherin and RUNX3 in EZH2 knockdown cells (Fig. 5A). Moreover, knockdown of EZH2 led to a significant decrease in gastric cancer cell proliferation and anchorage-independent growth in vitro (Fig. 5C) and in tumor growth in vivo (Fig. 5D), which showed phenotypic overlap with TET depletion in gastric cancer cells. Increased E-cadherin and RUNX3 and reduced cell growth were also observed in another gastric cancer cell line AGS with EZH2 knockdown (Fig. 5E and F). For further confirmation, we introduced EZH2-CDS lentivirus into TET knockdown cells. Notably, overexpression of EZH2 protein abrogated the inhibitory effect of TET knockdown on gastric cancer growth (Fig. 5G and H). Collectively, these data indicated that EZH2 is a crucial player responsible for TET ceRNA function in gastric carcinoma.

Figure 5.

EZH2 mediates TET-induced gastric carcinogenesis. A, The expression levels of EZH2, E-cadherin, and RUNX3 were examined by Western blotting analysis in SGC-7901 cells infected with lentivirus containing either control shRNA or the shRNA targeting different regions of the EZH2 mRNA transcript. B, Chromatin immunoprecipitation followed by qRT-PCR showed that EZH2 knockdown in SGC-7901 cells reduced both EZH2 and H3K27me3 enrichment at the promoters of PRC2 target genes. The GAPDH gene was used as a negative control. Error bars, SD from three independent experiments. C, Cell proliferation, cell viability, and soft-agar colony formation assays were performed in SGC-7901 cells infected with two EZH2 shRNA lentiviruses or control shRNA lentivirus. Error bars, SD from three independent experiments performed in triplicate. D, EZH2 knockdown inhibited gastric cancer tumor growth in mice. SGC-7901 cells infected with EZH2#2 shRNA or control shRNA lentivirus were inoculated subcutaneously into nude mice, and the mice were closely monitored for tumor growth. The mice were sacrificed and tumors were removed at 30 days postinoculation. Error bars, SD (n = 5 mice/group). E and F, Protein expression analysis (E) and cell proliferation and soft-agar colony formation assays (F) in AGS cells infected with EZH2#1 and #2 shRNA or control shRNA lentiviruses. Error bars, SD from three independent experiments performed in triplicate. G and H, Protein expression analysis (G) and cell proliferation and soft-agar colony formation assays (H) were performed in SGC-7901 expressing TET1/2/3 shRNA infected with EZH2-CDS lentivirus or control lentivirus. Error bars, SD from three independent experiments performed in triplicate. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 5.

EZH2 mediates TET-induced gastric carcinogenesis. A, The expression levels of EZH2, E-cadherin, and RUNX3 were examined by Western blotting analysis in SGC-7901 cells infected with lentivirus containing either control shRNA or the shRNA targeting different regions of the EZH2 mRNA transcript. B, Chromatin immunoprecipitation followed by qRT-PCR showed that EZH2 knockdown in SGC-7901 cells reduced both EZH2 and H3K27me3 enrichment at the promoters of PRC2 target genes. The GAPDH gene was used as a negative control. Error bars, SD from three independent experiments. C, Cell proliferation, cell viability, and soft-agar colony formation assays were performed in SGC-7901 cells infected with two EZH2 shRNA lentiviruses or control shRNA lentivirus. Error bars, SD from three independent experiments performed in triplicate. D, EZH2 knockdown inhibited gastric cancer tumor growth in mice. SGC-7901 cells infected with EZH2#2 shRNA or control shRNA lentivirus were inoculated subcutaneously into nude mice, and the mice were closely monitored for tumor growth. The mice were sacrificed and tumors were removed at 30 days postinoculation. Error bars, SD (n = 5 mice/group). E and F, Protein expression analysis (E) and cell proliferation and soft-agar colony formation assays (F) in AGS cells infected with EZH2#1 and #2 shRNA or control shRNA lentiviruses. Error bars, SD from three independent experiments performed in triplicate. G and H, Protein expression analysis (G) and cell proliferation and soft-agar colony formation assays (H) were performed in SGC-7901 expressing TET1/2/3 shRNA infected with EZH2-CDS lentivirus or control lentivirus. Error bars, SD from three independent experiments performed in triplicate. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Close modal

EZH2 overexpression correlates with high expression of TET transcripts in gastric cancer specimens

Finally, we sought to verify whether our findings could be extended to gastric cancer patients. EZH2 RNA expression was detected in the collection of human gastric cancer specimens and paired normal tissues as shown above. As shown in Fig. 6A, the EZH2 transcripts were significantly overexpressed in gastric cancer specimens of both cohorts A and B. Kaplan–Meier survival curves showed that EZH2 overexpression correlated with a shortened overall survival (Fig. 6B). We also examined EZH2 protein levels by immunohistochemistry. Out of 109 pairs of gastric cancer patients, 81% (88/109) of cases had higher EZH2 expression in gastric cancer vs. normal tissues (Fig. 6C). We further assessed the correction between TETs and EZH2 in gastric cancer samples and found that the EZH2 transcripts were significantly correlated with both TET2 and TET3 transcripts in both patient cohorts (Fig. 6D). The EZH2 transcripts had no significant association with TET1 RNA expression, although we see a trend (data not shown). Next, we analyzed the gastric cancer expression data from two publicly published datasets (GSE51105, ref. 33; GSE35809, refs. 34 and 35) in Gene Expression Omnibus to confirm this correction. This correction was true in both gastric cancer sample cohorts. EZH2 transcript in gastric cancer tissues had a significant correlation with both TET2 and TET3 expression but not with TET1 expression (Fig. 6E and data not shown). These data further support our findings that the TET family functions as a ceRNA for EZH2 in gastric cancer.

Figure 6.

EZH2 is overexpressed in human gastric cancer tissues and correlates with TET transcripts. A, The RNA levels of EZH2 were detected by qRT-PCR in gastric cancer specimens and paired normal tissues of both cohorts A and B. B, Kaplan–Meier analysis of overall survival based on EZH2 expression in 76 gastric cancer samples. C, Immunohistochemistry analysis of EZH2 protein levels in 109 pairs of gastric cancer and normal gastric tissues. Quantitative analysis of EZH2 staining showed significantly higher staining intensity in gastric tumor samples compared with normal tissues. Error bars, SEM. D, EZH2 RNA expression correlated with the RNA levels of TET2 and TET3 in gastric cancer specimens of both cohorts A and B. Each data point represents an individual gastric cancer sample, and the Pearson score shows a positive and significant correlation. E, mRNA expression data from Gene Expression Omnibus datasets (GSE51105 and GSE35809) were used to show EZH2 and TET2/3 were correlated in gastric cancer tissue samples. F, Model for TET-regulating EZH2 by acting as a ceRNA. In normal gastric mucosa tissues, TET family expression is steady, leading to binding of miR-26 molecules to EZH2, thus suppressing EZH2 expression. In gastric cancer, overexpression of the TET family sequesters the miR-26 family to target EZH2, thereby resulting in EZH2 overexpression, which in turn promotes gastric cancer progression.

Figure 6.

EZH2 is overexpressed in human gastric cancer tissues and correlates with TET transcripts. A, The RNA levels of EZH2 were detected by qRT-PCR in gastric cancer specimens and paired normal tissues of both cohorts A and B. B, Kaplan–Meier analysis of overall survival based on EZH2 expression in 76 gastric cancer samples. C, Immunohistochemistry analysis of EZH2 protein levels in 109 pairs of gastric cancer and normal gastric tissues. Quantitative analysis of EZH2 staining showed significantly higher staining intensity in gastric tumor samples compared with normal tissues. Error bars, SEM. D, EZH2 RNA expression correlated with the RNA levels of TET2 and TET3 in gastric cancer specimens of both cohorts A and B. Each data point represents an individual gastric cancer sample, and the Pearson score shows a positive and significant correlation. E, mRNA expression data from Gene Expression Omnibus datasets (GSE51105 and GSE35809) were used to show EZH2 and TET2/3 were correlated in gastric cancer tissue samples. F, Model for TET-regulating EZH2 by acting as a ceRNA. In normal gastric mucosa tissues, TET family expression is steady, leading to binding of miR-26 molecules to EZH2, thus suppressing EZH2 expression. In gastric cancer, overexpression of the TET family sequesters the miR-26 family to target EZH2, thereby resulting in EZH2 overexpression, which in turn promotes gastric cancer progression.

Close modal

Taken together, our results suggest a new model previously unidentified: elevated expression of TET RNAs in gastric cancer sequesters the miR-26 family to target EZH2, and thereby leads to EZH2 overexpression, which in turn promotes gastric carcinogenesis (Fig. 6F).

Decreased expression or activities of TET proteins and loss of 5-hmC have been reported in a wide spectrum of human cancers, including breast, colorectal, lung cancers, and hematopoietic malignancies (6–10). Moreover, the role for TET family members as tumor suppressors in several types of cancers has been described (5, 6, 36–38). For instance, homozygous or heterozygous loss of TET2 in mouse models enhances hematopoietic stem cell activity, resulting in a CMML-like phenotype (4, 5, 39, 40). Downregulation of TET1 in colon tumors causes constitutive activation of the WNT pathway due to DNA hydroxymethylation, resulting in uncontrolled tumor growth, according to Neri and colleagues (9). However, contrary to the previous studies, a recent report uncovered an oncogenic function of the TET gene in MLL-rearranged leukemia (11). TET1 is significantly upregulated in MLL-rearranged leukemia and plays a critical oncogenic role through its cooperation with MLL fusion proteins in regulating a set of important oncogenic cotargets (11). In the present study, we found that all three TET genes were strongly up-regulated in gastric cancer at the transcriptional level and the upregulation of TET transcripts was an unfavorable prognostic factor in gastric cancer patients. Functional studies demonstrated that the TET family played an essential oncogenic role in gastric cancer. To our knowledge, ours is the first study to show the oncogenic potential of the TET family in solid tumors.

Here, we observed that the RNA levels of TETs were significantly elevated in gastric cancer, but their protein levels remained unchangeable. We supposed that the possible reasons might lie in the posttranscriptional regulation of TETs. Concordantly with the tumor-suppressive role in most previous studies (9, 41–43), here the TET1 protein inhibited gastric cancer growth. However, either TET2 or TET3 protein did not show any obvious effects on gastric carcinogenesis. In contrast, overexpression of TET1/2/3 3′UTRs is sufficient to cause an increase in gastric cancer growth in vitro and in vivo, suggesting that at least part of the function of TETs in gastric carcinogenesis is coding independent. We further showed that the TET RNA functioned as a ceRNA to sequester the miR-26 family, leading to EZH2 overexpression in gastric cancer. As a methyltransferase component of the polycomb repressive complex 2 (PRC2), EZH2 can catalyze the trimethylation of histone H3 at lysine 27 (H3K27me3), which silences specific gene transcription. Accumulating evidence has suggested that EZH2 overexpression or an activating mutation could be an important driver of tumor development and progression, and that inactivation of EZH2 may be therapeutically effective in many cancers (44). In this study, we demonstrated that EZH2 at both the mRNA level and protein level is overexpressed in gastric cancer samples and knockdown of EZH2 can attenuate gastric cancer cell growth, consistent with the previous reports (44, 45). More importantly, TET knockdown-elicited inhibition of gastric cancer growth can be rescued by enforced expression of EZH2, indicating that EZH2 serves as a major target in mediating the effect of TET ceRNA in gastric cancer tumorigenesis.

We also investigated the correction of EZH2 and TET members in clinical gastric cancer samples and found that the EZH2 transcripts were significantly correlated with both TET2 and TET3 transcripts in gastric cancer patient cohorts. These data further support the TET family functioning as a ceRNA for EZH2 in gastric cancer. In addition, our qRT-PCR results and GEO dataset analysis showed that there was no significant association between RNA expression of EZH2 and TET1 in gastric cancer samples. However, our in vitro and in vivo data suggest that the 3′UTRs of TET1-, TET2-, and TET3 exhibit comparable roles in gastric cancer tumorigenesis (Figure 3). It is conceivable to hypothesize that the EZH2 regulation by multiple factors in addition to TETs may account for the absence of significant correlation between expression of EZH2 and TET1 in gastric cancer.

Although much of this ceRNA activity is driven by the EZH2 pathway, it is possible that there might be additional TET ceRNA targets in gastric cancer progression, such as NOS2 and SODD, as they were consistently induced upon both miR-26 knockdown and TET2 3′UTR overexpression in our study. These and other genes regulated by TET ceRNA deserve further investigation in gastric cancer.

In summary, our findings provide evidence of a previously unrecognized noncoding function for the TET family in gastric cancer. Apart from the insights into gastric cancer pathogenesis, our data suggest that the TET/miR-26/EZH2 signaling axis represents potential targets for future prevention and treatment of human gastric cancer. Moreover, our studies identify that the TET family may be prognostic biomarkers predictive for the outcomes of gastric cancer patients.

No potential conflicts of interest were disclosed.

Conception and design: M. Deng, R. Zhang, Z. He

Development of methodology: M. Deng, R. Zhang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Deng, R. Zhang, Z. He, Q. Qiu, X. Lu, J. Yin, H. Liu

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Deng, R. Zhang

Writing, review, and/or revision of the manuscript: M. Deng, Z. He

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): X. Jia

Study supervision: Z. He

We thank Jian Ma for help in the design of experiments. We also thank Hailin Tang for technical assistance.

This work was supported by the National Natural Science Foundation of China (81472625, 81101526, and 81672452), the Science and Technology Planning Project of Guangdong Province (2014A020212741), the Scientific Research Project of Guangzhou Municipal University (1201410235), and the "Guangzhou Scholar" Research Project of Guangzhou Municipal University (1201561588).

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.

1.
Siegel
R
,
Ma
J
,
Zou
Z
,
Jemal
A
. 
Cancer statistics, 2014
.
CA Cancer J Clin
2014
;
64
:
9
29
.
2.
Camargo
MC
,
Kim
WH
,
Chiaravalli
AM
,
Kim
KM
,
Corvalan
AH
,
Matsuo
K
, et al
Improved survival of gastric cancer with tumour Epstein–Barr virus positivity: an international pooled analysis
.
Gut
2014
;
63
:
236
43
.
3.
Rasmussen
KD
,
Helin
K
. 
Role of TET enzymes in DNA methylation, development, and cancer
.
Genes Dev
2016
;
30
:
733
50
.
4.
Quivoron
C
,
Couronne
L
,
Della Valle
V
,
Lopez
CK
,
Plo
I
,
Wagner-Ballon
O
, et al
TET2 inactivation results in pleiotropic hematopoietic abnormalities in mouse and is a recurrent event during human lymphomagenesis
.
Cancer Cell
2011
;
20
:
25
38
.
5.
Li
Z
,
Cai
X
,
Cai
CL
,
Wang
J
,
Zhang
W
,
Petersen
BE
, et al
Deletion of Tet2 in mice leads to dysregulated hematopoietic stem cells and subsequent development of myeloid malignancies
.
Blood
2011
;
118
:
4509
4518
.
6.
Delhommeau
F
,
Dupont
S
,
Della Valle
V
,
James
C
,
Trannoy
S
,
Masse
A
, et al
Mutation in TET2 in myeloid cancers
.
N Engl J Med
2009
;
360
:
2289
301
.
7.
Yang
H
,
Liu
Y
,
Bai
F
,
Zhang
JY
,
Ma
SH
,
Liu
J
, et al
Tumor development is associated with decrease of TET gene expression and 5-methylcytosine hydroxylation
.
Oncogene
2013
;
32
:
663
9
.
8.
Lian
CG
,
Xu
Y
,
Ceol
C
,
Wu
F
,
Larson
A
,
Dresser
K
, et al
Loss of 5-hydroxymethylcytosine is an epigenetic hallmark of melanoma
.
Cell
2012
;
150
:
1135
46
.
9.
Neri
F
,
Dettori
D
,
Incarnato
D
,
Krepelova
A
,
Rapelli
S
,
Maldotti
M
, et al
TET1 is a tumour suppressor that inhibits colon cancer growth by derepressing inhibitors of the WNT pathway
.
Oncogene
2015
;
34
:
4168
76
.
10.
Sun
M
,
Song
CX
,
Huang
H
,
Frankenberger
CA
,
Sankarasharma
D
,
Gomes
S
, et al
HMGA2/TET1/HOXA9 signaling pathway regulates breast cancer growth and metastasis
.
Proc Natl Acad Sci U S A
2013
;
110
:
9920
5
.
11.
Huang
H
,
Jiang
X
,
Li
Z
,
Li
Y
,
Song
CX
,
He
C
, et al
TET1 plays an essential oncogenic role in MLL-rearranged leukemia
.
Proc Natl Acad Sci U S A
2013
;
110
:
11994
9
.
12.
Yang
X
,
Zhang
XF
,
Lu
X
,
Jia
HL
,
Liang
L
,
Dong
QZ
, et al
MicroRNA-26a suppresses angiogenesis in human hepatocellular carcinoma by targeting hepatocyte growth factor-cMet pathway
.
Hepatology
2014
;
59
:
1874
85
.
13.
Deng
M
,
Tang
HL
,
Lu
XH
,
Liu
MY
,
Lu
XM
,
Gu
YX
, et al
miR-26a suppresses tumor growth and metastasis by targeting FGF9 in gastric cancer
.
PLoS One
2013
;
8
:
e72662
.
14.
Lu
J
,
He
ML
,
Wang
L
,
Chen
Y
,
Liu
X
,
Dong
Q
, et al
MiR-26a inhibits cell growth and tumorigenesis of nasopharyngeal carcinoma through repression of EZH2
.
Cancer Res
2011
;
71
:
225
33
.
15.
Ji
J
,
Shi
J
,
Budhu
A
,
Yu
Z
,
Forgues
M
,
Roessler
S
, et al
MicroRNA expression, survival, and response to interferon in liver cancer
.
N Engl J Med
2009
;
361
:
1437
47
.
16.
Zeitels
LR
,
Acharya
A
,
Shi
G
,
Chivukula
D
,
Chivukula
RR
,
Anandam
JL
, et al
Tumor suppression by miR-26 overrides potential oncogenic activity in intestinal tumorigenesis
.
Genes Dev
2014
;
28
:
2585
90
.
17.
Fu
X
,
Jin
L
,
Wang
X
,
Luo
A
,
Hu
J
,
Zheng
X
, et al
MicroRNA-26a targets ten eleven translocation enzymes and is regulated during pancreatic cell differentiation
.
Proc Natl Acad Sci U S A
2013
;
110
:
17892
7
.
18.
Karreth
FA
,
Reschke
M
,
Ruocco
A
,
Ng
C
,
Chapuy
B
,
Leopold
V
, et al
The BRAF pseudogene functions as a competitive endogenous RNA and induces lymphoma in vivo
.
Cell
2015
;
161
:
319
32
.
19.
Tay
Y
,
Kats
L
,
Salmena
L
,
Weiss
D
,
Tan
SM
,
Ala
U
, et al
Coding-independent regulation of the tumor suppressor PTEN by competing endogenous mRNAs
.
Cell
2011
;
147
:
344
57
.
20.
Yuan
JH
,
Yang
F
,
Wang
F
,
Ma
JZ
,
Guo
YJ
,
Tao
QF
, et al
A long noncoding RNA activated by TGF-beta promotes the invasion-metastasis cascade in hepatocellular carcinoma
.
Cancer Cell
2014
;
25
:
666
81
.
21.
Kumar
MS
,
Armenteros-Monterroso
E
,
East
P
,
Chakravorty
P
,
Matthews
N
,
Winslow
MM
, et al
HMGA2 functions as a competing endogenous RNA to promote lung cancer progression
.
Nature
2014
;
505
:
212
7
.
22.
Poliseno
L
,
Salmena
L
,
Zhang
J
,
Carver
B
,
Haveman
WJ
,
Pandolfi
PP
. 
A coding-independent function of gene and pseudogene mRNAs regulates tumour biology
.
Nature
2010
;
465
:
1033
8
.
23.
Wang
Y
,
Xu
Z
,
Jiang
J
,
Xu
C
,
Kang
J
,
Xiao
L
, et al
Endogenous miRNA sponge lincRNA-RoR regulates Oct4, Nanog, and Sox2 in human embryonic stem cell self-renewal
.
Dev Cell
2013
;
25
:
69
80
.
24.
Tay
Y
,
Rinn
J
,
Pandolfi
PP
. 
The multilayered complexity of ceRNA crosstalk and competition
.
Nature
2014
;
505
:
344
52
.
25.
Liu
P
,
Tang
H
,
Chen
B
,
He
Z
,
Deng
M
,
Wu
M
, et al
miR-26a suppresses tumour proliferation and metastasis by targeting metadherin in triple negative breast cancer
.
Cancer Lett
2015
;
357
:
384
92
.
26.
Gao
J
,
Liu
QG
. 
The role of miR-26 in tumors and normal tissues (Review)
.
Oncol Lett
2011
;
2
:
1019
1023
.
27.
Ning
X
,
Shi
Z
,
Liu
X
,
Zhang
A
,
Han
L
,
Jiang
K
, et al
DNMT1 and EZH2 mediated methylation silences the microRNA-200b/a/429 gene and promotes tumor progression
.
Cancer Lett
2015
;
359
:
198
205
.
28.
Manning
CS
,
Hooper
S
,
Sahai
EA
. 
Intravital imaging of SRF and Notch signalling identifies a key role for EZH2 in invasive melanoma cells
.
Oncogene
2015
;
34
:
4320
32
.
29.
Campbell
RM
,
Tummino
PJ
. 
Cancer epigenetics drug discovery and development: the challenge of hitting the mark
.
J Clin Invest
2014
;
124
:
64
9
.
30.
Cao
Q
,
Yu
J
,
Dhanasekaran
SM
,
Kim
JH
,
Mani
RS
,
Tomlins
SA
, et al
Repression of E-cadherin by the polycomb group protein EZH2 in cancer
.
Oncogene
2008
;
27
:
7274
84
.
31.
Wang
C
,
Liu
Z
,
Woo
CW
,
Li
Z
,
Wang
L
,
Wei
JS
, et al
EZH2 Mediates epigenetic silencing of neuroblastoma suppressor genes CASZ1, CLU, RUNX3, and NGFR
.
Cancer Res
2012
;
72
:
315
24
.
32.
Li
CH
,
To
KF
,
Tong
JH
,
Xiao
Z
,
Xia
T
,
Lai
PB
, et al
Enhancer of zeste homolog 2 silences microRNA-218 in human pancreatic ductal adenocarcinoma cells by inducing formation of heterochromatin
.
Gastroenterology
2013
;
144
:
1086
1097
e9
.
33.
Busuttil
RA
,
George
J
,
Tothill
RW
,
Ioculano
K
,
Kowalczyk
A
,
Mitchell
C
, et al
A signature predicting poor prognosis in gastric and ovarian cancer represents a coordinated macrophage and stromal response
.
Clin Cancer Res
2014
;
20
:
2761
72
.
34.
Wu
Y
,
Grabsch
H
,
Ivanova
T
,
Tan
IB
,
Murray
J
,
Ooi
CH
, et al
Comprehensive genomic meta-analysis identifies intra-tumoural stroma as a predictor of survival in patients with gastric cancer
.
Gut
2013
;
62
:
1100
11
.
35.
Chia
NY
,
Deng
N
,
Das
K
,
Huang
D
,
Hu
L
,
Zhu
Y
, et al
Regulatory crosstalk between lineage-survival oncogenes KLF5, GATA4 and GATA6 cooperatively promotes gastric cancer development
.
Gut
2015
;
64
:
707
19
.
36.
Tefferi
A
,
Pardanani
A
,
Lim
KH
,
Abdel-Wahab
O
,
Lasho
TL
,
Patel
J
, et al
TET2 mutations and their clinical correlates in polycythemia vera, essential thrombocythemia and myelofibrosis
.
Leukemia
2009
;
23
:
905
11
.
37.
Abdel-Wahab
O
,
Mullally
A
,
Hedvat
C
,
Garcia-Manero
G
,
Patel
J
,
Wadleigh
M
, et al
Genetic characterization of TET1, TET2, and TET3 alterations in myeloid malignancies
.
Blood
2009
;
114
:
144
7
.
38.
Yamazaki
J
,
Taby
R
,
Vasanthakumar
A
,
Macrae
T
,
Ostler
KR
,
Shen
L
, et al
Effects of TET2 mutations on DNA methylation in chronic myelomonocytic leukemia
.
Epigenetics
2012
;
7
:
201
7
.
39.
Shide
K
,
Kameda
T
,
Shimoda
H
,
Yamaji
T
,
Abe
H
,
Kamiunten
A
, et al
TET2 is essential for survival and hematopoietic stem cell homeostasis
.
Leukemia
2012
;
26
:
2216
23
.
40.
Moran-Crusio
K
,
Reavie
L
,
Shih
A
,
Abdel-Wahab
O
,
Ndiaye-Lobry
D
,
Lobry
C
, et al
Tet2 loss leads to increased hematopoietic stem cell self-renewal and myeloid transformation
.
Cancer Cell
2011
;
20
:
11
24
.
41.
Fu
HL
,
Ma
Y
,
Lu
LG
,
Hou
P
,
Li
BJ
,
Jin
WL
, et al
TET1 exerts its tumor suppressor function by interacting with p53-EZH2 pathway in gastric cancer
.
J Biomed Nanotechnol
2014
;
10
:
1217
30
.
42.
Lin
LL
,
Wang
W
,
Hu
Z
,
Wang
LW
,
Chang
J
,
Qian
H
. 
Negative feedback of miR-29 family TET1 involves in hepatocellular cancer
.
Med Oncol
2014
;
31
:
291
.
43.
Pei
YF
,
Tao
R
,
Li
JF
,
Su
LP
,
Yu
BQ
,
Wu
XY
, et al
TET1 inhibits gastric cancer growth and metastasis by PTEN demethylation and re-expression
.
Oncotarget
2016
;
7
:
31322
35
.
44.
Yamaguchi
H
,
Hung
MC
. 
Regulation and Role of EZH2 in Cancer
.
Cancer Res Treat
2014
;
46
:
209
22
.
45.
Tsang
DP
,
Cheng
AS
. 
Epigenetic regulation of signaling pathways in cancer: role of the histone methyltransferase EZH2
.
J Gastroenterol Hepatol
2011
;
26
:
19
27
.