Genomic and epigenomic <i>EBF1</i> alterations modulate <i>TERT</i> expression in gastric cancer

Manjie Xing; Wen Fong Ooi; Jing Tan; Aditi Qamra; Po-Hsien Lee; Zhimei Li; Chang Xu; Nisha Padmanabhan; Jing Quan Lim; Yu Amanda Guo; Xiaosai Yao; Mandoli Amit; Ley Moy Ng; Taotao Sheng; Jing Wang; Kie Kyon Huang; Chukwuemeka George Anene-Nzelu; Shamaine Wei Ting Ho; Mohana Ray; Lijia Ma; Gregorio Fazzi; Kevin Junliang Lim; Giovani Claresta Wijaya; Shenli Zhang; Tannistha Nandi; Tingdong Yan; Mei Mei Chang; Kakoli Das; Zul Fazreen Adam Isa; Jeanie Wu; Polly Suk Yean Poon; Yue Ning Lam; Joyce Suling Lin; Su Ting Tay; Ming Hui Lee; Angie Lay Keng Tan; Xuewen Ong; Kevin White; Steven George Rozen; Michael Beer; Roger Sik Yin Foo; Heike Irmgard Grabsch; Anders Jacobsen Skanderup; Shang Li; Bin Tean Teh; Patrick Tan

doi:10.1172/JCI126726

Genomic and epigenomic EBF1 alterations modulate TERT expression in gastric cancer

Manjie Xing,^1,2,3 Wen Fong Ooi,² Jing Tan,^4,5 Aditi Qamra,^2,3 Po-Hsien Lee,⁶ Zhimei Li,⁵ Chang Xu,^1,6 Nisha Padmanabhan,¹ Jing Quan Lim,⁷ Yu Amanda Guo,⁸ Xiaosai Yao,⁹ Mandoli Amit,¹ Ley Moy Ng,⁶ Taotao Sheng,^1,10 Jing Wang,¹ Kie Kyon Huang,¹ Chukwuemeka George Anene-Nzelu,^11,12 Shamaine Wei Ting Ho,^1,6 Mohana Ray,¹³ Lijia Ma,¹³ Gregorio Fazzi,¹⁴ Kevin Junliang Lim,¹ Giovani Claresta Wijaya,⁵ Shenli Zhang,¹ Tannistha Nandi,² Tingdong Yan,¹ Mei Mei Chang,⁸ Kakoli Das,¹ Zul Fazreen Adam Isa,² Jeanie Wu,¹ Polly Suk Yean Poon,² Yue Ning Lam,² Joyce Suling Lin,² Su Ting Tay,¹ Ming Hui Lee,¹ Angie Lay Keng Tan,¹ Xuewen Ong,¹ Kevin White,^13,15 Steven George Rozen,^1,16 Michael Beer,^17,18 Roger Sik Yin Foo,^11,12 Heike Irmgard Grabsch,^14,19 Anders Jacobsen Skanderup,⁸ Shang Li,^1,20 Bin Tean Teh,^{1,5,6,9,16,20} and Patrick Tan^{1,2,6,16,21,22,23}

Authorship note: MX, WFO, and JT contributed equally to this work.

Published May 4, 2020 - More info

View PDF

Abstract

Transcriptional reactivation of telomerase catalytic subunit (TERT) is a frequent hallmark of cancer, occurring in 90% of human malignancies. However, specific mechanisms driving TERT reactivation remain obscure for many tumor types and in particular gastric cancer (GC), a leading cause of global cancer mortality. Here, through comprehensive genomic and epigenomic analysis of primary GCs and GC cell lines, we identified the transcription factor early B cell factor 1 (EBF1) as a TERT transcriptional repressor and inactivation of EBF1 function as a major cause of TERT upregulation. Abolishment of EBF1 function occurs through 3 distinct (epi)genomic mechanisms. First, EBF1 is epigenetically silenced via DNA methyltransferase, polycomb-repressive complex 2 (PRC2), and histone deacetylase activity in GCs. Second, recurrent, somatic, and heterozygous EBF1 DNA–binding domain mutations result in the production of dominant-negative EBF1 isoforms. Third, more rarely, genomic deletions and rearrangements proximal to the TERT promoter remobilize or abolish EBF1-binding sites, derepressing TERT and leading to high TERT expression. EBF1 is also functionally required for various malignant phenotypes in vitro and in vivo, highlighting its importance for GC development. These results indicate that multimodal genomic and epigenomic alterations underpin TERT reactivation in GC, converging on transcriptional repressors such as EBF1.

Introduction

Gastric cancer (GC) is a leading cause of global cancer mortality (1), with a particularly high prevalence in East Asia. With the exception of Japan and South Korea, where the absolute GC incidence is sufficiently high to enable cost-effective endoscopic screening, today most patients with GC are diagnosed at advanced stages of the disease, resulting in low overall survival (2). Targeted agents approved for clinical use in GC include trastuzumab for human epidermal growth factor receptor 2–positive (HER2-positive) GC (3), ramicurumab (VEGFR2) (4), and immuno-oncology agents such as pembrolizumab (5) and nivolumab (6). However, these treatments are typically only effective in small subsets of patients with GC, and clinical trials evaluating alternative GC molecular targets have proved disappointing (7–10). Investigating the basic mechanisms of GC tumorigenesis may therefore elucidate the early steps in gastric malignancy and lead to new therapeutic interventions.

Reactivation of telomerase is a pivotal step in cancer initiation. Telomere sequences at chromosome ends are lost during successive rounds of DNA replication, and critically short telomeres can induce cellular senescence and apoptosis. Telomerase is an RNA-dependent DNA polymerase consisting of an RNA component (TERC) and a protein catalytic subunit (TERT) that can lengthen a chromosome’s telomeres. TERT is normally expressed in human stem and germ cells but is silenced in differentiated somatic cells and transcriptionally reactivated in up to 90% of human malignancies (11, 12). TERT reactivation is a driver of oncogenic alteration, contributing to cellular transformation, immortalization, and oncogenic development in many cancer types. Clinically, high TERT expression levels have been associated with aggressive disease. In GC specifically, high TERT levels have been associated with poor prognosis, advanced disease stage, and lymphatic metastasis and invasion (13–15).

Despite their importance, the cellular mechanisms driving the transcriptional reactivation of TERT in cancers remain poorly described, particularly in the case of GC. In certain cancers, TERT promoter somatic mutations (C250T and C228T) have been described as a major mechanism of telomerase reactivation (16, 17), however, these promoter hotspot mutations appear to be infrequent in GC (<1%) (18). Besides promoter point mutations, large-scale genomic alterations such as enhancer hijacking (19) and copy number alterations (20) affecting TERT have been reported, and at the trans-regulatory level, oncogenic transcriptional activators (Sp1, c-MYC, AP2, and HIF-1) and tumor-suppressive transcriptional repressors (p53, Menin, and WT1) may also regulate TERT (21). To date, however, specific mechanisms underlying TERT reactivation in GC remain largely unknown.

In this study, we focused on investigating molecular processes that govern TERT expression in GC. By integrating genomic data from primary GCs and GC cell lines, we identified the early B cell factor 1 (EBF1) transcription factor (TF) as a candidate TERT repressor and validated its function through in vitro and in vivo studies. We provide evidence that in GC, EBF1 function is suppressed by various mechanisms, including epigenetic inactivation, dominant-negative mutations, and large-scale genomic alterations affecting EBF1 cis-binding regions at the TERT promoter. EBF1 inhibition may therefore contribute to TERT reexpression and activity in GC.

Results

EBF1 TF expression is negatively associated with TERT. To investigate TERT expression in GC, we queried RNA-Seq profiles of primary GCs from The Cancer Genome Atlas (TCGA) consortium (n = 415 GCs, n = 35 normal samples), in-house databases of primary tumors (n = 18 tumor/normal [T/N] pairs of GCs), and GC cell lines (n = 63). We found that TERT was significantly upregulated in GC samples compared with normal samples (P = 1.9, Figure 1A × 10^–10; Supplemental Figure 1A; supplemental material available online with this article; https://doi.org/10.1172/JCI126726DS1) and exhibited variable expression across GC lines (Supplemental Figure 1B). TERT overexpression was greater in GCs with chromosomal instability (CIN) or microsatellite instability (MSI) compared with genome-stable (GS) or EBV-positive tumors (Figure 1A). These results are consistent with previous independent studies reporting increased TERT expression in GC (see Discussion and refs. 22–24). To determine the prevalence of TERT promoter point mutations in GC, we surveyed whole-genome sequences (WGSs) of 212 GCs and matched normal samples (25) but detected no mutations at previously reported TERT mutational hotspots (C228T and C250T), consistent with previous reports that TERT promoter hotspot mutations are rare in GC (18).

Figure 1

EBF1 TF expression is negatively associated with TERT. (A) Expression of TERT in normal gastric (n = 35) and GC samples (n = 415) from TCGA consortium. Q values were determined by Wilcoxon’s rank-sum test with FDR multiple testing correction. RSEM, RNA-Seq by expectation-maximization (https://github.com/deweylab/RSEM). (B) Correlation of mRNA levels between TERT and TFs (ENCODE ChIP-Seq database) at genomic regions flanking the TERT TSS. Correlation coefficients were computed using the normalized expression matrix from TCGA. *Q < 0.05, **Q < 0.01, and ***Q < 0.001, by Pearson’s correlation test with FDR multiple testing correction. (C) A significant negative correlation was found between TERT and EBF1 mRNA levels in the South Korean and Singapore cohorts. Top: microarray data for 96 GC samples from the Korean cohort (median-centered, log-transformed). Bottom: microarray data for 185 GC samples from the Singapore (SG) cohort (log₂ expression level). P values were determined by Pearson’s correlation test. (D) Graph shows EBF1 and TERT expression levels based on RNA-Seq data for 63 GC lines. FPKM, fragments per kilobase per million mapped reads. (E) TaqMan qPCR validation of EBF1 expression levels using in-house normal stomach tissues. (F) EBF1 expression levels in normal gastric organoids (microarray). RMA, robust multiarray average. (G) Western blot analysis of EBF1 protein expression in GC lines. Ctrl, control. (H) Expression of EBF1 in normal gastric (n = 35) and GC samples (n = 415) from TCGA. P value was determined by Wilcoxon’s rank sum test. (I) EBF1 immunohistochemistry for GC tissues and normal gastric epithelium. Left: normal gastric epithelium with EBF1 expression (brown); middle: GC cells with EBF1 expression in 15% of the tumor cells (brown); right: intestinal-type GC cells with complete loss of EBF1 expression, endothelial cells and lymphocytes in the tumor stroma were EBF1⁺ (brown). Original magnification, x40. Red arrows indicate stained cells. (J) Distribution of immunohistochemical scores for EBF1 protein expression in 51 GCs and matched normal tissues, connected by black lines. P value was determined by 2-sided t test. (A and H) For the box-and-whisker plots, the lines within boxes indicate the median, the bounds of the boxes indicate the upper and lower quartiles, the whiskers indicate the minimum and maximum values, and the separated points indicate outliers.

We sought to identify trans-acting TFs that regulate TERT in GC. Hypothesizing that TERT-regulatory TFs should exhibit genomic occupancy of the TERT promoter region and also systematic relationships with TERT expression, we proceeded to correlate TERT expression levels (inferred from TCGA GCs) against the expression patterns of 26 TFs predicted to bind around the TERT transcription start site (TSS) (± 10 kb). These 26 factors were identified on the basis of their predicted ability to bind to the TERT promoter region using ChIP-Seq data from the ENCODE ChIP-Seq database (https://genome.ucsc.edu/ENCODE/3). Confirming the utility of this approach, this analysis reidentified several previously known TERT regulators, such as MYC (TERT activator) and MXI1 (TERT repressor) (26, 27). Notably, we observed that TERT expression levels were negatively correlated with the EBF1 TF, ranking second in significance among the putative TERT-regulatory TFs (R = –0.20, P = 1.2 × 10^–5; Figure 1B). We extended our analysis to 2 other GC expression microarray cohorts and again observed similar significant negative correlations between TERT and EBF1, thus validating this observation in independent cohorts (Figure 1C).

Historically studied as a TF in B-lineage immune cells, EBF1 expression has been recently described in solid tumors such as breast cancer, cholangiocarcinoma, and glioblastoma (28, 29). To explore EBF1 expression in stomach tissues, we mined the Genotype–Tissue Expression (GTEx) database (https://www.gtexportal.org/home/) and observed that EBF1 is expressed in many human tissues and cell types including stomach (Supplemental Figure 1C and ref. 30). To confirm that EBF1 is expressed in GC epithelial cells, we compared EBF1 and TERT RNA-Seq expression levels across a panel of 63 GC cell lines. GC cell lines with high EBF1 expression exhibited no or low TERT transcript levels, whereas the majority of cell lines lacking EBF1 expression displayed high TERT expression (Figure 1D). EBF1 mRNA expression in stomach was further confirmed by TaqMan quantitative PCR (qPCR) analysis of a diverse collection of gastric-related samples, including primary normal gastric tissues (n = 6), GC cell lines (n = 4); gene expression analysis of TCGA normal gastric tissues (RNA-Seq, n = 39); data from 2 additional public data sets (31, 32); and analysis of normal gastric organoids (n = 6) (Figure 1, E and F, Supplemental Figure 1, D–G, and ref. 33) . To confirm EBF1 expression at the protein level, we performed Western blotting using anti-EBF1 antibodies validated by both predicted protein size (Figure 1G, left and middle) and siRNA-knockdown assays (Figure 1G, right). We confirmed endogenous EBF1 protein expression in cell lines showing high EBF1 transcript levels and low EBF1 protein expression in cell lines expressing low levels of EBF1 transcripts.

Compared with normal gastric tissues, EBF1 transcripts were significantly downregulated in GC (TCGA samples, P = 0.012; Figure 1H) — a pattern opposite that of TERT transcript levels. To confirm EBF1 protein expression in primary GCs, we performed EBF1 immunohistochemical analysis on a panel of GCs and matched normal tissues. This analysis revealed EBF1 expression in scattered nuclei of normal gastric epithelium and GC cells (Figure 1I), ranging from 0% to 25% of cells (Figure 1J). Similar to the gene expression data, primary GCs exhibited significantly lower EBF1 immunohistochemical scores than did normal gastric tissues (P = 0.0008, n = 51 T/N pairs; Figure 1J). When compared with GCs with different clinicopathological characteristics, EBF1 loss was more frequent in diffuse-type GCs and poorly differentiated GCs (P = 0.001 and P = 0.002 respectively, n = 109 patients; Supplemental Table 1). We also observed a trend toward more frequent EBF1 loss in patients with a higher number of lymph node metastases (P = 0.052, n = 109 patients; Supplemental Table 1).

EBF1 is transcriptionally silenced in GC via repressive histone modifications and DNA methylation. To explore potential mechanisms underlying EBF1 downregulation in GC, we found a promoter occupancy/expression correlation similar to that depicted in Figure 1B, but this time analyzing the EBF1 instead of the TERT promoter region. Correlation of TFs predicted to bind the EBF1 promoter region (± 2.5 kb) to EBF1 mRNA expression levels revealed a strong negative correlation of EBF1 with 2 polycomb-repressive complex 2 (PRC2) subunits: EZH2 (R = –0.44, P < 0.00001) and SUZ12 (R = –0.29, P = 2.4 × 10^–10) (Figure 2A). We queried EZH2 and SUZ12 ChIP-Seq data from the ENCODE database and found EZH2 and SUZ12 occupancy at the EBF1 promoter in various cell types (Figure 2B). As PRC2 is known to catalyze the H3K27me3-repressive histone mark, we performed H3K27me3 ChIP-Seq analysis and confirmed H3K27me3 enrichment at the EBF1 TSS locus in EBF1-low (AGS, IM95, and SNU484) but not EBF1-high GC cells (YCC11) (Figure 2B). Given that histone modifications and DNA methylation often cooperate to regulate gene expression (34), we also evaluated DNA methylation levels at the EBF1 promoter using Illumina 450K DNA Methylation Arrays. We found that EBF1-low cell lines had high to moderate levels of EBF1 promoter DNA methylation (high, AGS and IM95; moderate, SNU484), whereas the EBF1-high cell line YCC11 showed no EBF1 promoter DNA methylation.

Figure 2

EBF1 is transcriptionally silenced in GC via repressive histone modifications and DNA methylation. (A) Correlation of mRNA expression levels between EBF1 and TFs at genomic regions flanking the EBF1 TSS. Correlation coefficients were computed using the normalized expression matrix from TCGA (n = 415 tumors, n = 35 normal samples). *Q < 0.05, **Q < 0.01, and ***Q < 0.001, by Pearson’s correlation test with FDR multiple testing correction. (B) H3K27me3, EZH2, and SUZ12 signals flanking the EBF1 TSS. Top: RNA-Seq tracks depict EBF1 expression in GC lines. Bottom left: H3K27me3, EZH2, and SUZ12 enrichment. Yellow bar indicates the EBF1 promoter (TSS ± 2.5 kb). Bottom right: DNA methylation levels of GC lines. Colored graph shows the average β value of 12 CpG probes in the 5′-UTR and TSS of EBF1. EZH2 and SUZ12 ChIP-Seq data are from the ENCODE database. Layered signals from multiple cell lines are shown for EZH2 (H1 human embryonic stem cells [H1-hESCs], human mammary epithelial cells [HMECs], human umbilical vein endothelial cells [HUVECs], normal human astrocytes [NHAs], normal human epidermal keratinocytes [NHEKs], and normal human lung fibroblasts [NHLFs]). The SUZ12 profile is from H1-hESCs. (C) EZH2 and EBF1 expression levels in SNU484 cells after drug treatment. (D) DNA methylation 450K array data from TCGA. Data were plotted for the top and bottom 25th percentiles of EBF1-expressing samples (n = 191). (E) In-house DNA methylation 450K array data for 18 T/N GC pairs. Heatmap shows GCs with or without loss of EBF1. (D and E) Data are from 12 CpG probes in the 5′-UTR and TSS region of EBF1. (F) Response patterns of EBF1 and TERT mRNA levels following drug treatments. (C and F) Control cells were treated with DMSO. *Q < 0.05, by 2-sided t test with FDR multiple testing correction. qPCR results are shown as the mean ± SD of technical triplicates. All data are representative of 3 independent experiments. (G) H3K27me3 enrichment at the EBF1 locus from GC cell lines with DMSO, 5-aza, and GSK126 treatment. Red arrows indicate a decrease in H3K27me3 enrichment.

We explored whether targeting EZH2, the catalytic subunit of PRC2, might reactivate EBF1. Knockdown of EZH2 by siRNA in the GC lines AGS and SNU484 caused a 3.2- to 7.5-fold increase in EBF1 mRNA levels (Supplemental Figure 2, A and B). We then tested the effects of EZH2 pharmacologic inhibition using 2 distinct EZH2-inhibitory compounds, GSK126 and GSK343 (Figure 2C and Supplemental Figure 2C). In SNU484 cells, EZH2 inhibition was accompanied by a 40- to 87-fold increase in EBF1 expression levels. However, in AGS cells, EZH2 monoinhibition did not induce similarly strong changes in EBF1 expression. To further explore this observation, we noted recent studies have shown that multiple epigenetic pathways often act in a coordinated and combinatorial manner to repress target promoters. For example, EZH2 requires histone deacetylases (HDACs) to silence targets (35), and EZH2 can directly control DNA methylation by recruiting DNA methyltransferase (DNMT) (34). We thus explored whether EBF1 might also be regulated by DNMTs and HDACs. Examination of methylation array data from 191 GCs from the TCGA consortium revealed higher levels of EBF1 promoter DNA methylation in GCs with low EBF1 expression (i.e., higher β values), whereas GCs with high EBF1 expression exhibited lower DNA methylation levels (i.e., lower β values; Figure 2D). We observed a similar trend in a pairwise fashion in an in-house cohort of GCs with matched normal samples (n = 18 T/N pairs; Figure 2E and Supplemental Figure 2D; the same 12 CpG probes were measured in both cohorts). The degree of the β value methylation differences is fully consistent with definitions of hypermethylation by other independent studies (36, 37). For orthogonal validation of the β value DNA methylation levels, we performed quantitative DNA methylation assays (MethylLight) in 8 T/N pairs to determine EBF1 promoter DNA methylation levels (4 pairs with the highest methylation levels in GCs and 4 pairs randomly selected from GCs with low methylation levels; Supplemental Figure 2E). The quantitative DNA methylation results were concordant with patterns from the 450K DNA methylation β values. Taken collectively, these results suggest that approximately half (40%–55%) of primary GCs with loss of EBF1 expression have high levels of EBF1 promoter methylation.

To determine whether epigenetic drug combinations might affect EBF1 expression, we performed single-, double-, and triple-agent combination treatments of AGS and SNU484 cells with EZH2 inhibitors (GSK126), DNMT inhibitors (5-aza-2′-deoxycytidine [5-aza]), and the HDAC inhibitor trichostatin A (TSA). Treatment of AGS and SNU484 cells with 5-aza caused reexpression of EBF1 that was approximately 4-fold and 22-fold over levels in control cells, respectively (Figure 2F and Supplemental Figure 2F), suggesting that the magnitude of EBF1 reexpression induced by 5-aza is cell line specific, which is supported by an independent study of GC cell lines (38). For HDAC inhibition, similar EBF1 increases were achieved by treatment with either TSA or vorinostat, a more specific HDAC inhibitor (Supplemental Figure 2G and ref. 39). Notably, compared with single-agent treatments, combinatorial epigenetic therapies consistently induced substantially higher levels of EBF1 expression, demonstrating that EBF1 is inactivated in GC through diverse epigenetic pathways (PRC2, DNA methylation, and HDAC). Supporting this hypothesis, in AGS cells a triple combination of epigenetic therapies (5-aza plus TSA plus GSK126) induced the most dramatic EBF1 upregulation relative to single- and dual-agent treatments (Figure 2F).

We also asked whether epigenetic treatment affected EBF1-associated histone modifications and/or DNA methylation. Using H3K27me3 ChIP-Seq, we confirmed decreased H3K27me3 signals at the EBF1 locus after treatment with either GSK126 or 5-aza in 2 cell lines (Figure 2G). Quantitative DNA methylation analysis of both cell lines before and after 5-aza treatment also confirmed significant decreases in DNA methylation levels after treatment with 5-aza but not GSK126 (Supplemental Figure 2H). However, we did not observe convincing alterations in H3K27ac signals after epigenetic treatment (Supplemental Figure 2I). It is possible that our experimental conditions and selection time points may have been too short to observe changes in H3K27ac or, alternatively, that inhibition of EZH2 (by GSK126) or DMNT (by 5-aza) may not be sufficient to alter H3K27ac patterns. Taken collectively, these results suggest that EBF1 is transcriptionally silenced in GC through a combination of repressive histone modifications and DNA methylation.

Besides the EBF1 promoter, we also investigated distal cis-regulatory element regions upstream and downstream of EBF1 (TSS +100 kb/–1000 kb). We identified 43 potential enhancers associated with the EBF1 locus. Of these 43, we found that 5 predicted enhancers (Supplemental Figure 2J, blue highlights) had differential H3K27ac signals between EBF1-high cell lines (LMSU and Hs746T) and EBF1-low cell lines (AGS and SNU484) (P < 0.05, 1-sided Student’s t test). Supporting their designation as enhancer elements, these 5 enhancers were also associated with increased H3K4me1 and ATAC-Seq (assay for transposase-accessible chromatin using sequencing) signals in EBF1-high cell lines, but not in EBF1-low cell lines, and 3 of these enhancers exhibited significant interactions with the EBF1 promoter in public databases of Promoter Capture Hi-C (pcHi-C) data (Supplemental Figure 2J; arc lines; P < 0.01; promoter/capture point: green bar; ref. 40).

The EBF1 TF negatively regulates TERT and oncogenic development in vitro and in vivo. Consistent with EBF1 being a TERT repressor, enhancement of EBF1 reexpression by combinatorial epigenetic therapies also caused TERT silencing in both cell lines (Figure 2F and Supplemental Figure 2F). To specifically test whether EBF1 regulates TERT, siRNA-mediated depletion of endogenous EBF1 using either pooled or independent EBF1 siRNAs in YCC11 cells, which express high EBF1 levels, resulted in 2.3- to 3-fold increases in TERT RNA levels, as determined by TaqMan qPCR assays (Figure 3A and Supplemental Figure 3A). Notably, this degree of TERT induction is comparable to expression levels associated with TERT promoter hotspot mutations (41). Using telomeric repeat amplification protocol (TRAP) assays, we also confirmed that EBF1 knockdown caused 1.8- to 7.9-fold increases in telomerase activity (Figure 3B). These results were replicated and confirmed in 2 additional GC cell lines (Hs746T and HGC27), each using 2 independent EBF1 shRNAs (Supplemental Figure 3B). In the reciprocal experiment, we induced overexpression of EBF1 in AGS, SNU484, and IM95 cells, which express low or no EBF1. Western blotting confirmed increased levels of EBF1 protein in the EBF1-transduced cell lines (Figure 3C), and EBF1 overexpression resulted in a 26% to 46% decrease in TERT mRNA levels (Figure 3D). Similarly, using TRAP assays, we confirmed that EBF1 overexpression caused a 30% to 40% decrease in telomerase activity (Figure 3E and Supplemental Figure 3C). These findings suggest that EBF1 is a negative regulator of TERT in GC.

Figure 3

EBF1 TF negatively regulates TERT and oncogenic development in vitro and in vivo. (A) Individual si-EBF1 silencing caused TERT overexpression in YCC11. Two individual EBF1 siRNAs were used (left) and resulted in TERT upregulation (right) in YCC11. (B) TRAP assay showed an increase in telomerase activity caused by overexpression of TERT via EBF1 knockdown. (C) Western blotting showed increased EBF1 protein levels after overexpression in GC lines. (D and E) EBF1 overexpression caused TERT reductions and a decrease in telomerase activity. (F and G) Cell proliferation capacity after EBF1 overexpression in GC cell lines. Proliferation assays and monolayer colony formation assays showed a dramatic decrease in cell proliferation capacity after EBF1 overexpression. (H and I) In vivo effects of EBF1 overexpression on tumorigenesis. (H) NOD/SCID mouse tumor volumes after injection of either endogenous EBF1 (vector, blue) or EBF1-overexpressing (EBF1-OE) AGS GC cells. Tumor volumes were measured every 3 days. n = 6 mice per group. (I) Image of harvested tumors after an experimental period of 10 weeks. Tumor sizes are shown in centimeters. ^#Q < 0.05, by 2-sided t test with FDR multiple testing correction (A, B, and D). *P < 0.05, **P < 0.01, and ***P < 0.001, by 2-sided t test (E, F, and H). Error bars indicate the SD. RT-PCR results are shown as the mean ± SD of technical triplicates. All data are representative of 3 independent experiments.

To assess whether EBF1 plays a functional role in GC, we performed proliferation and monolayer colony formation assays and found that overexpression of EBF1 in multiple cell lines dramatically reduced cellular proliferation and monolayer colony formation (Figure 3, F and G). In vivo, we tested NOD/SCID xenograft mice injected with AGS cells with or without EBF1 overexpression. After 41 days, the tumors formed from EBF1-overexpressing GC cells were significantly smaller than those formed from control cells (Figure 3, H and I). The tumor growth–inhibiting effect of EBF1 was sustained throughout the entire experimental period of approximately 10 weeks. To identify biological pathways affected by EBF1 overexpression, we sequenced the transcriptomes of EBF1-overexpressing cells. Using the Genomic Regions Enrichment of Annotations Tool (GREAT) algorithm (42), we found that genes downregulated by EBF1 overexpression were enriched in pathways related to chromosomal organization, reassuringly including TERT (Supplemental Figure 3D, left). Reciprocally, we found that genes upregulated by EBF1 overexpression, including FAS, HIF1A, and AIFM2, were enriched in pathways such as cell death regulation (Supplemental Figure 3D, right).

EBF1 exhibits dominant-negative mutations in GC. To obtain further evidence of EBF1 disruptions in GC, we investigated whether EBF1 exhibits somatic DNA mutations that might alter EBF1 function. We queried exome- and whole-genome sequencing data for 459 GCs from public and in-house data sets and identified 14 GCs with EBF1 coding missense mutations (Figure 4A). Of these, 50% of the mutations were localized to the EBF1 DNA–binding domain (DBD), targeting highly conserved sequence residues (Supplemental Figure 4A). To distinguish passenger from driver mutations, we individually cloned the EBF1 DBD mutations and expressed the EBF1 mutants in 2 GC cell lines that normally do not express EBF1 (AGS and SNU484). In both cell lines, Western blotting confirmed comparable expression levels of both WT and mutant EBF1 proteins (Figure 4B and Supplemental Figure 4B). Proliferation and monolayer colony formation assays showed that WT EBF1 dramatically reduced cellular proliferation and monolayer colony formation (Figure 4, C and D, and Supplemental Figure 4, C and D). In contrast, the EBF1 mutants Q196P and R209Q had a proliferative capacity similar to that of empty vector controls, demonstrating that Q196P and R209Q mutations are sufficient to abolish EBF1 activity. I139T, R111W, and R242W/Q mutants exhibited a proliferative capacity similar to that of WT EBF1 and were designated passengers.

Figure 4

EBF1 exhibits dominant-negative somatic mutations in GC. (A) Distribution of EBF1 coding mutations in 14 GC patients with annotated functional domains. Each dot represents 1 patient. (B) Western blotting revealed equal levels of EBF1 protein abundance after overexpression of WT and mutant EBF1 in AGS cells. (C and D) Proliferation assays (C) and monolayer colony formation assays (D) showed cell proliferation capacity after overexpression of WT and mutant EBF1 in GC cells. Error bars indicate SD. (E) 3D structural analysis of EBF1 and missense mutations. The x-ray crystal structure reveals that 2 EBF1 protomers (green and cyan ribbons) are predicted to dimerize and bind to DNA (orange ribbon with blue ladders). Purple sticks and labels indicate the positions of GC-associated missense mutations on 3 functional domains. HLH, helix-loop-helix. (F and G) Alterations of DNA-binding loop flexibility by EBF1 mutations. (F) EBF1 DBDs are represented by green ribbons. Purple sticks indicate the position of 4 missense mutations. The 3 orange beads mark Cα atoms of R63, N172, and P205. Double helix shows the relative orientation of DNA in the complex. (G) Distance distributions of specified Cα atoms obtained from MD simulations of unbound EBF1 DBD. Vertical dashed lines indicate the distances measured from x-ray crystal structures. (H) EMSAs showed that Q196P or R209Q mutated recombinant EBF1 protein did not have DNA binding to probes containing EBF1-binding motifs. (I) Western blots showed similar protein levels of WT EBF1, Q196P-mutant EBF1, and combined WT plus Q196P-mutant EBF1 after overexpression. Proliferation assay (J) and monolayer colony formation assay (K) showed cell proliferation capacity after overexpression. Error bars indicate SD. (L) Gene expression levels (RNA-Seq) of AGS cells after retroviral infection with either empty vector (black), WT EBF1 (blue), Q196P-mutant EBF1 (red), or combined WT plus Q196P-mutant EBF1 (purple). Expression levels of EBF1 and TERT are shown.

To investigate the structural impact of Q196P and R209Q mutations on EBF1 activity, we performed 3D structural analysis and molecular dynamic (MD) simulations. Mapping of the Q196P and R209Q mutations to EBF1-DNA complex structures revealed that these residues are probably not directly involved in DNA contact or DNA recognition (Figure 4, E and F). Using MD simulations to investigate the effect of these mutations at the atomic level, we determined that both WT and mutant EBF1 DBDs are predicted to exhibit similar integrity and stability under both unbound and DNA-bound states (Supplemental Figure 4, E and F). However, during the unbound EBF1 simulations, the Q196P and R209Q mutants showed loss of intramolecular hydrogen bonds in the DBD GH loop (aa 193–210), which would be predicted to cause greater loop flexibility and a longer time to attain a proper DNA-binding conformation (Figure 4G). Also, compared with the WT EBF1 DBD, we found that the Q196P and R209Q mutants had energetically less favorable associations with DNA (Supplemental Table 2). These results suggest that Q196P and R209Q EBF1 DBDs are likely to bind less stably to DNA compared with WT EBF1. To confirm this prediction, we performed EMSAs using recombinant EBF1 proteins. Compared with WT EBF1, Q196P- or R209Q-mutated EBF1 did not display DNA binding to probes containing EBF1-binding motifs (Figure 4H).

The EBF1 mutations observed in GC were heterozygous, indicating that EBF1-mutated GCs may still express a WT copy of EBF1. However, EBF1 has been shown to function as a dimer (43), raising the possibility that the Q196P and R209Q mutations may function as dominant-negative isoforms. To test this possibility, we induced co-overexpression of WT and Q196P EBF1 in AGS cells. We found that cells coexpressing WT and Q196P EBF1 had compromised proliferation rates, decreased monolayer colony formation, and reduced TERT mRNA levels, intermediate between EBF1 WT and Q196P-mutant cells (Figure 4, I–L, and Supplemental Figure 4G). These results suggest that certain EBF1 DBD mutations have dominant-negative activity.

EBF1 binds to a TERT 5′ proximal cis-regulatory element. To determine whether EBF1 occupies cis-regulatory elements at the TERT promoter, we performed EBF1 ChIP-Seq in EBF1-expressing GC cell lines (HGC27, Hs746T, YCC11, and AGS with endogenous EBF1). We also analyzed AGS cells overexpressing either WT EBF1 or the EBF1 DNA–binding mutants. Motif analysis of the EBF1 ChIP-Seq peaks revealed that the top-ranked de novo consensus binding motif matched previously known EBF1 factors (Figure 5A and refs. 44, 45). We identified 2 EBF1-binding peaks in the TERT 5′ proximal region, at –2.3 kb and –4 kb, both with EBF1-binding motifs (hereafter referred to as peak 1 and peak 2, respectively) (Figure 5B). We observed no EBF1 binding at peaks 1 or 2 in cell lines lacking endogenous EBF1 or in cells overexpressing mutant EBF1 (Q196P or R209Q), further demonstrating that certain EBF1 DBD mutations have decreased DNA-binding capacity (Figure 5B). Besides peak 1 and peak 2 at the TERT 5′ proximal region, expanded analysis of the TERT genomic locus did not reveal significant or consistent EBF1 binding at other distal regions, suggesting that peaks 1 and 2 are the predominant EBF1-binding regions at TERT in GC (Supplemental Figure 5A).

Figure 5

EBF1 binds to a TERT 5′ proximal cis-regulatory element. (A) De novo binding motif analysis of 2 GC cell lines revealed significant enrichment of EBF1-binding motifs. An EBF1 consensus motif from the HOMER database is shown. P values were calculated using a hypergeometric test. (B) EBF1 ChIP-Seq data showed EBF1 binding to the TERT 5′ proximal region (highlighted by a yellow bar) in 4 GC cell lines (HGC27, Hs746T, YCC11, and AGS with endogenous EBF1) and also in AGS cells overexpressing WT EBF1 but not Q196P- or R209Q-mutant EBF1. DNA-binding motifs (dashed lines) of EBF1 were identified within these regions. (C) Gene expression (RNA-Seq) and H3K27ac and H3K4me1 enrichment signals at the TERT locus for primary tumor and matched normal samples and GC cell lines. The green and yellow highlights indicate EBF1-binding peaks 1 and 2 in the TERT 5′ proximal region, respectively. Gray highlight indicates a previously described distal TERT regulatory element. (D) Line plots show detected cis interactions from the capture point (blue) at the TERT TSS to EBF1-binding peaks 1 and 2 (green and yellow bars, respectively) and the distal TERT-regulatory element. Interaction strength in units of reads per million (RPM) is represented in the y axis. (E) CapSTARR-Seq showed reporter activity at EBF1-binding peak 2. Black bar indicates the cloned TERT 5′ proximal region. Peaks within the yellow highlighted area indicate reporter activity in 2 GC cell lines. Q values were calculated using a Poisson distribution statistical test.

To determine whether the TERT 5′ proximal region containing peaks 1 and 2 has properties of an enhancer element, we then mapped potential cis-regulatory elements and chromatin features associated with the TERT 5′ proximal region using Nano-ChIP-Seq histone profiles (46). Specifically, we compared GC cell lines expressing high and low TERT transcription levels and also primary GCs relative to matched normal tissues (Figure 5C and sequencing statistics in Supplemental Table 3). Concentrating on H3K27ac and H3K4me1 modifications corresponding to activity and enhancer marks, respectively (47), we reidentified a previously described TERT-regulatory element 21 kb upstream of the TERT TSS, supporting the reliability of our data (Figure 5C, gray bar, and ref. 48). Notably, we observed increased H3K27ac and H3K4me1 signals at peak 2 in primary GCs compared with matched normal tissue (Figure 5C, yellow bar) and also in cell lines with high TERT expression (KATO III and SNU16) compared with cell lines with low TERT expression (SNU719, YCC11, and YCC10). H3K27ac signals were also observed at peak 1 in the TERT-expressing lines KATO III and SNU16. To determine whether chromatin accessibility at the TERT locus correlates with TERT transcription, we performed ATAC-Seq on 6 GC cell lines with high or low TERT expression. We observed higher levels of chromatin accessibility across the TERT locus (including the 5′ proximal region containing peaks 1 and 2) in cell lines with high TERT expression (Supplemental Figure 5B, blue bar) and lower levels of chromatin accessibility in cell lines with low TERT expression. To assess whether DNA modifications, such as 5-methylcytosine (5mC) and 5-hydroxymethyl (5hmC), also correlate with TERT expression, we next performed 5mC and 5hmC profiling on cell lines with high or low TERT expression. However, we did not observe 5mC or 5hmC signals in this proximal element with EBF1-binding peaks, in cell lines with either high or low TERT expression (Supplemental Figure 5D and see Discussion). It is possible that TERT expression in GC is uncorrelated with DNA methylation levels or, alternatively, that because TERT is a low-expression gene (49), changes in TERT DNA methylation were below the limits of detection of the 5mc and 5hmc profiling methods used in this study.

We asked whether the TERT TSS might interact with the peak 1 and 2 regions via chromatin interactions (50). Applying Capture-C technology on 4 GC cell lines (AGS, SNU16, SNU484, and YCC11) targeting the TERT TSS, we confirmed that both peaks 1 and 2 had chromatin interactions with the TERT TSS (Figure 5D and see Methods). However, despite occurring more distal to the TERT TSS, peak 2 showed stronger TSS interactions compared with peak 1. This suggests that although both peaks 1 and 2 may contribute to TERT regulation, peak 2 may play a stronger role, consistent with peak 2 having stronger epigenomic signals of enhancers (Figure 5C). To determine whether peak 2 had properties of a functional enhancer, we also applied CapSTARR-Seq technology, a high-throughput technique for functional testing of candidate enhancer elements (see Methods). Using this approach, peak 2 was confirmed to have enhancer activity (SNU16, Q = 1.58 × 10^–8; OCUM1, Q = 7.9 × 10^–12; Figure 5E). Interestingly, although YCC11 is a cell line expressing low levels of TERT, we observed similar levels of cis interactions compared with cell lines expressing high levels of TERT (SNU16, SNU484, and AGS), supporting recent studies showing that many cis interactions are largely stable and preformed (51, 52). To explore global chromatin interaction landscapes in the TERT locus, we also performed Hi-C chromatin conformation profiling in cell lines with high (SNU16) and low (SNU719) TERT expression. Hi-C interactions at the TERT locus were higher in SNU16 cells than in SNU719 cells (Supplemental Figure 5C).

Genomic alterations of the TERT 5′ proximal region in primacy GC. In other cancers, large-scale genomic alterations at the TERT locus (e.g., translocations, copy number variants) have been shown to deregulate TERT expression (19, 20). To investigate whether the TERT 5′ proximal region containing peaks 1 and 2 might be associated with genomic alterations in vivo, we examined structural variants (SVs) at the TERT locus in 212 primary GCs and 62 cell lines (25). We identified 2 GCs (P990489 and HK-PFG144T) and 1 GC cell line (LMSU) that had somatic deletions or rearrangements affecting the TERT 5′ proximal region including peaks 1 and 2 (Figure 6). In GC P990489, the deletion was a highly focal 1-kb deletion, occurring at –3.5 to –4.4 kb, coinciding with peak 2 (Figure 6A), whereas LMSU cells harbored a genomic rearrangement deleting both peak 1 and peak 2 (Figure 6B). For both GC P990489 and LMSU cells, the 5′ TERT alterations were orthogonally confirmed by Sanger sequencing (Figure 6, C and D). At the expression level, GC P990489, harboring a deletion of the TERT 5′ proximal region, had the second-highest level of TERT overexpression in our in-house cohort (>8 fold higher compared with matched normal samples) (Figure 6E), and LMSU GC cells had high TERT mRNA expression, ranking 11th in our in-house panel of 63 gastric cell lines (Supplemental Figure 1B). We also performed telomere length measurements and observed an increase in GC P990489 telomere lengths compared with those of matched normal tissue (Figure 6F). The third GC (HK-PFG144T) also harbored a deletion affecting both peaks (Supplemental Figure 6A). These results suggest that in certain GCs, the TERT 5′ proximal region containing peaks 1 and 2 is associated with genomic deletions and/or rearrangements, which are predicted to delete and rearrange EBF1-binding sites.

Figure 6

Genomic alterations of the TERT 5′ proximal region in primacy GC. (A) Somatic large-scale genomic deletion affecting EBF1-binding peak 2 in GC T990489. Tracks 1 and 3: aligned read coverage from WGS profiles. Tracks 2 and 4: distribution of aligned reads. Red reads in track 4 have larger-than-expected inferred sizes and therefore indicate possible deletions. Lower coverage in track 3 with respect to track 1 at the same loci supports the existence of a somatic genomic deletion. Purple box indicates the genomic deleted region. (B) Translocation disrupting the TERT 5′ proximal region (including both EBF1-binding peaks 1 and 2) in LMSU cells. Blue bars indicate breakpoints of the translocation. Dark pink tracks depict H3K27ac signals. (C and D) Validation of somatic genomic deletion/translocation of the TERT 5′ proximal region in GC T990489 and LMSU lines via Sanger sequencing. (E) Log-transformed fold change of TERT expression in 18 GCs with respect to matched normal samples. GC P990489 is highlighted in yellow. (F) Telomere lengthening was observed in GC T990489 by Southern blotting and densitometric analysis. (G) The CRISPR-deleted region is indicated by scissors (top). Graph shows a comparison of TERT expression between WT YCC11 clones (black bars) and clones (C1, C2, C3) with CRISPR-mediated homozygous deletions at the TERT 5′ proximal region (brown bars). (H) Comparison of TERT expression between WT clones (black) and clones with mutations in at least 1 allele in 1 or both EBF1-binding motifs (blue). (I) Deletions/translocations that overlapped with the TERT 5′ proximal region are shown for 2 GC cases, 1 GC cell line, and 10 cases of other cancer types. Vertical red and yellow lines highlight EBF1-binding peaks 1 and 2. *Q < 0.05, by 2-sided t test with FDR multiple testing correction (G and H). RT-PCR results are shown as the mean ± SD of technical triplicates. Figures are representative of 3 independent experiments.

To functionally test the role of peak 1 and peak 2 in TERT regulation, we performed CRISPR-mediated genomic deletion of both peaks 1 and 2 in a GC cell line (YCC11) expressing low levels of TERT. Data from multiple single-cell clones, with homozygous deletion of both TERT regions, consistently showed upregulated TERT expression compared with intact clones (P < 0.05; Figure 6G). We further introduced CRISPR-mediated point mutations to disrupt the EBF1-binding motifs of both regions in YCC11 cells (see Methods). After sequence verification (Supplemental Figure 6B), we analyzed single-cell clones with mutations in at least 1 allele in either or both of the peak 1 and peak 2 EBF1-binding motifs (i.e., peak 1, heterozygous; peak 2, heterozygous; peaks 1 and 2, heterozygous; peak 2, homozygous). Using qPCR to access TERT expression, we detected higher TERT expression levels across all the mutated clones compared with levels in WT clones (P < 0.05; Figure 6H). These results suggest that both peaks probably contribute to TERT regulation.

To determine whether the TERT 5′ proximal region is also associated with genomic alterations in other cancers, we surveyed published studies across 32 cancer types (19, 53) and identified 10 more cases with genomic rearrangements affecting the TERT 5′ proximal region. These included 2 chromophobe renal cell carcinomas (KICH, TCGA), 1 urothelial bladder cancer (BLCA, TCGA), 2 sarcomas (SARC, TCGA), 2 hepatocellular carcinomas (LIHC, TCGA), and 3 neuroblastomas (NBL) (Figure 6I), suggesting that the TERT 5′ proximal region is associated with genomic alterations in other cancer types besides GC. Notably, in a total of 13 cases, genomic alterations affected both peaks 1 and 2 in 11 cases and peak 2 in 2 cases, indicating the importance of both peaks in contributing to TERT regulation.

Discussion

TERT overexpression is a signature hallmark of cancer. In GC, Gigek et al. reported that TERT expression was detected in 80% of gastric tumors but not in normal gastric mucosa (24), and other studies have also reported significant TERT overexpression in GC (22, 23). In this study, we performed a comprehensive analysis of TERT expression in GC. We queried RNA-Seq data from TCGA cohort and confirmed that TERT was significantly overexpressed in GC and that all 4 GC molecular subtypes exhibited significantly higher TERT expression levels than those in normal samples. We further replicated these findings in an independent RNA-Seq data set of 18 T/N pairs. Studies in different tumor types have highlighted distinct mechanisms for TERT transcriptional reactivation. At the cis-regulatory level, TERT promoter mutations at C250T and C228T have been primarily observed in glioblastoma (89%), melanoma (72%), and bladder cancer (70%) (16, 17), creating de novo binding motifs for ETS TFs and activating TERT (41). In other tumor types such as neuroblastoma, lung cancer, colon cancer, and cervical cancer, studies have reported large-scale SVs and copy number alterations involving TERT that can also cause high expression of TERT by affecting noncoding regulatory elements at the TERT locus (19, 20). However, specific mechanisms underlying TERT reactivation in GC remain largely unknown. In this study, we identified the TF EBF1 as a direct repressor of TERT expression. Our results also suggest that in GC, TERT is reactivated by abolishing EBF1 function via at least 3 distinct mechanisms: (a) epigenetic silencing of EBF1 expression via PRC2, DNA methyltransferase, and HDAC activity; (b) somatic EBF1 mutations resulting in dominant-negative isoforms; and (c) deletion or rearrangement of EBF1-binding sites proximal to the TERT promoter.

The discovery of EBF1 as a TERT repressor is notable. To date, 4 EBF family TFs have been described (EBF1, EBF2, EBF3, and EBF4) and shown to play important developmental roles in the regulation of cell fate decisions, differentiation, and migration (54). Besides B cell development (55), EBF1 has been reported to regulate adipocyte differentiation and morphology (56, 57) and also the differentiation of sensory neurons (58), highlighting a role for EBF1 beyond the hematopoietic system. Supporting these findings, in vivo studies of EBF1 in Ebf1^+/− mice on a high-fat diet revealed white adipose tissue (WAT) hypertrophy and insulin resistance (57). In addition, late Ebf1^−/− embryos were shown to have specific increases in striatal cell death and reductions in size after birth (58). In cancer, EBF1 has been proposed as a potential tumor suppressor (28) in malignancies such as breast cancer, in which EBF1 genomic deletions have been reported in 18.9% of breast cancer lines (29). EBF1 has also been proposed as an important epigenetic modifier in breast cancer (59), and in cholangiocarcinoma (CCA), EBF1 downregulation through prolonged oxidative stress was reported to induce tumorigenic properties (60). Interestingly, besides GC, analysis of RNA-Seq data across other TCGA cancer types revealed similar negative correlations between EBF1 and TERT (Supplemental Figure 6C). EBF1 may thus regulate TERT in other cancer types, although this hypothesis will require further investigation.

Our data suggest that in GC, EBF1 is downregulated in primary tumors particularly through epigenetic silencing. Specifically, at the level of DNA promoter hypermethylation, approximately half (40%–55%) of primary GCs with loss of EBF1 expression exhibited high levels of EBF1 promoter methylation.

Besides being silenced by epigenetic regulation, EBF1 was also found to exhibit somatic mutations resulting in dominant-negative isoforms. We observed EBF1 somatic point mutations in 3.1% of GCs, half of which occurred within the highly conserved EBF1 DBD. Functional experiments support the notion that some of these mutations (e.g., Q196P) strongly abrogate the ability of EBF1 to inhibit cell proliferation. Notably, although these EBF1 mutations were heterozygous, we found that they had dominant-negative activity, which is likely explained by the fact that EBF1 protein acts as a dimer (43). In the literature, other examples of “single-hit” tumor suppressors include FBXW7 (61) and p53 (62), which also result in the creation of dominant-negative isoforms. We also note that the presence of functional somatic EBF1 mutations in GC provides further support for EBF1 activity in the stomach.

A third and distinct mechanism of abrogation of EBF1 function occurred through deletions and rearrangements of EBF1-binding sites proximal to the TERT promoter. Specifically, we identified a regulatory element close to the TERT promoter that interacts with the TERT TSS, containing EBF1-binding sites and demonstrating EBF1 occupancy. CRISPR-mediated deletion of this region, or CRISPR-induced mutations of EBF1-binding motifs in either or both peaks, resulted in increased TERT expression in GCs expressing low levels of TERT. Epigenomic analysis revealed that this region was silenced in GCs with low TERT expression and active in GCs with high TERT expression, raising the possibility that this TERT 5′ proximal region may function as a regulatory “rheostat,” recruiting transcriptional repressors or activators in GCs expressing low and high levels of TERT, respectively. Importantly, this region exhibited somatic deletions and rearrangements in certain GCs and other tumor types, confirming its disruption in vivo. Interestingly, for LMSU cells, analysis of in-house H3K27ac ChIP-Seq data on GC cell lines suggested that, besides deleting this TERT 5′ proximal region, genomic rearrangements in these cells may also cause genomic fusion of the 5′ TERT region with 2 distal superenhancers (Figure 6B). Genomic disruptions of this region may be required for high TERT expression, as the TERT upstream region is well known to show strong epigenetic repression across multiple cell types (63).

In conclusion, telomerase has attracted interest as a target for cancer therapy, as most somatic cells have no or only low-level telomerase activity, whereas up to 90% of human cancers have transcriptionally reactivated TERT. Numerous telomerase-based therapeutic strategies are under investigation, such as the telomerase inhibitor imetelstat (GRN163L) and telomerase-derived anticancer peptide vaccines (GV1001 and GRNVAC1) (64). However, recent clinical trial results for GRN163L have shown significant dose-dependent hematological side effects, liver function abnormalities, and other adverse effects (65), which may be due to simultaneous inhibition of telomerase activity in both cancer cells and normal tissue–specific stem/progenitor cells. Thus, there remains an urgent need to elucidate the basic mechanisms of telomerase reactivation in cancer, which may highlight new strategies to inhibit telomerase. In particular, combinatorial inhibition of epigenetic pathways regulating EBF1 expression may prove to be an interesting area for further study.

Methods

Cell lines. AGS, KATO III, SNU16, and Hs746T cells were obtained from the American Type Culture Collection (ATCC). SNU719 and SNU484 cells were obtained from the Korean Cell Line Bank (KCLB). LMSU, GSS, and HGC27 cells were obtained from the RIKEN cell bank. IM95 cells were obtained from the Japanese Collection of Research Bioresources (JCRB) Cell Bank. YCC10 and YCC11 cells were gifts from Yonsei Cancer Centre (Seoul, South Korea). Cell line identities were confirmed by short tandem repeat (STR) DNA profiling performed at the Centre for Translational Research and Diagnostics (Cancer Science Institute of Singapore, Singapore), and STR profiles were assessed according to the standard American National Standards Institute/ATCC (ANSI/ATCC) ASN-0002-2011 nomenclature. The profiles of cell lines showed greater than 80% similarity to reference databases. All cell lines were negative for mycoplasma contamination.

Statistics. For Capture-C data analysis, r3Cseq was used to identify significant interactions of the viewpoint against a scaled background (Q < 0.05). For CapSTARR-Seq data analysis, an enhancer peak was called when there was significant enrichment of fragments from 1 region in the output library compared with the representation of that region in the input library based on Poisson distribution using MACS2 (https://github.com/taoliu/MACS/). Histone ChIP-enriched peaks were detected using the control-based ChIP-Seq analysis tool CCAT (https://github.com/vanbug/cluster/tree/master/cluster/CCAT/CCAT3.0) (FDR <5%). TF ChIP-enriched peaks were detected using MACS2 with the following thresholds: Q < 0.05 for paired-end sequencing and P < 0.00005 for single-end sequencing. TF-binding motifs were predicted using detected TF-binding sites from ChIP-Seq and HOMER (Hypergeometric Optimization of Motif EnRichment; http://homer.ucsd.edu/homer/motif/) with default parameters. For the immunohistochemistry studies in Supplemental Table 1, a Mann-Whitney U test was used for comparisons between 2 groups, whereas a Kruskal-Wallis test was used for comparisons of more than 2 groups. For cell phenotype assays, tumor xenograft studies, qPCR assays, real-time TRAP assays, and immunohistochemical analyses, P values were calculated using a 2-sided t test. A P value of less than 0.05 was considered statistically significant. FDR-based multiple testing correction was applied for multiple comparisons when applicable. A Q value of less than 0.05 was considered statistically significant. For comparison of gene expression levels between tumor samples and normal samples in Figure 1A, a Wilcoxon’s rank-sum test with multiple testing correction using the FDR procedure was performed against normal expression levels. For comparison of gene expression levels between all tumor samples and normal samples in Figure 1H and Supplemental Figure 1A, a Wilcoxon’s rank-sum test was performed against normal expression levels. For the correlation analysis in Figure 1C and Supplemental Figure 6C, P values for correlation were calculated using Pearson’s correlation test. For correlation analysis in Figure 1B and Figure 2A, P values for correlation were calculated using Pearson’s correlation test with multiple testing correction with the FDR procedure.

Histone ChIP-Seq, TF ChIP-Seq, Capture-C, CapSTARR-Seq, and RNA-Seq data sets are available in the Gene Expression Omnibus (GEO) database (GEO GSE121140). The public data sets used are described in the Supplemental Methods.

Study approval. Patients’ primary samples were obtained from the SingHealth tissue repository with approval from the SingHealth Centralised IRB and signed patient informed consent. Animal studies were conducted in compliance with animal protocols approved by the SingHealth IACUC of Singapore. Additional details, including the methods associated with WGS and structural variant detection, ChIP-Seq, RNA-Seq, telomere length measurements, Capture-C, CRISPR/Cas9, silencing and overexpression of EBF1, tumor xenografts, DNA methylation analysis, real-time TRAP assays, drug treatments, CapSTARR-Seq, MD simulations, MethylLight assays, ATAC-Seq, EMSAs, detection of EBF1 coding mutations, EBF1 immunohistochemistry, Western blotting, Sanger sequencing, cell phenotype assays, qPCR, and data availability are described in the Supplemental Methods.

Acknowledgments

We thank Yoshiaki Ito and Shing Leng Chan for helpful discussions. We thank Angie Tan, Beatrice Tan, and Hui Hoon Chua for their administrative support. We also thank members of the Duke-NUS Genome Biology Facility and the Sequencing and Scientific Computing teams at the Genome Institute of Singapore for providing sequencing services and data management capabilities. This work was supported by National Medical Research Council grants TCR/009-NUHS/2013, OFLCG18May-0003, NMRC/STaR/0026/2015, and OFIRG15nov072; National University Cancer Institute, Singapore (NCIS) grant NR13NMR111OM; A*STAR A*ccelerate GAP grant ETPL/15-R15 GAP-0021; Ministry of Education Academic Research Tier 2 grant MOE2017-T2-1-105; Guangdong Innovative and Entrepreneurial Research Team Program grants 2016ZT06S638 and 2016ZT06S252; and National Natural Science Foundation of China grant 61661146004. Funding was also provided by the Cancer Science Institute of Singapore, NUS, under the National Research Foundation Singapore and the Singapore Ministry of Education under its Research Centres of Excellence initiative, and block funding was received from Duke-NUS Medical School.

Address correspondence to: Patrick Tan, Duke-NUS Medical School, 8 College Road #07-19, Singapore 169857. Phone: 65.6516.1783; Email: gmstanp@duke-nus.edu.sg.

References

Ferlay J, et al. Cancer incidence and mortality worldwide: sources, methods and major patterns in GLOBOCAN 2012. Int J Cancer. 2015;136(5):E359–E386.
View this article via: PubMed CrossRef Google Scholar

Seyedin S, Wang PC, Zhang Q, Lee P. Benefit of adjuvant chemoradiotherapy for gastric adenocarcinoma: a SEER population analysis. Gastrointest Cancer Res. 2014;7(3-4):82–90.
View this article via: PubMed Google Scholar

Bang YJ, et al. Trastuzumab in combination with chemotherapy versus chemotherapy alone for treatment of HER2-positive advanced gastric or gastro-oesophageal junction cancer (ToGA): a phase 3, open-label, randomised controlled trial. Lancet. 2010;376(9742):687–697.
View this article via: PubMed CrossRef Google Scholar

Fuchs CS, et al. Ramucirumab monotherapy for previously treated advanced gastric or gastro-oesophageal junction adenocarcinoma (REGARD): an international, randomised, multicentre, placebo-controlled, phase 3 trial. Lancet. 2014;383(9911):31–39.
View this article via: PubMed CrossRef Google Scholar

Schnipper LE, et al. American Society of Clinical Oncology statement: a conceptual framework to assess the value of cancer treatment options. J Clin Oncol. 2015;33(23):2563–2577.
View this article via: PubMed CrossRef Google Scholar

Kang YK, et al. Nivolumab in patients with advanced gastric or gastro-oesophageal junction cancer refractory to, or intolerant of, at least two previous chemotherapy regimens (ONO-4538-12, ATTRACTION-2): a randomised, double-blind, placebo-controlled, phase 3 trial. Lancet. 2017;390(10111):2461–2471.
View this article via: PubMed CrossRef Google Scholar

Waddell T, et al. Epirubicin, oxaliplatin, and capecitabine with or without panitumumab for patients with previously untreated advanced oesophagogastric cancer (REAL3): a randomised, open-label phase 3 trial. Lancet Oncol. 2013;14(6):481–489.
View this article via: PubMed CrossRef Google Scholar

Lordick F, et al. Capecitabine and cisplatin with or without cetuximab for patients with previously untreated advanced gastric cancer (EXPAND): a randomised, open-label phase 3 trial. Lancet Oncol. 2013;14(6):490–499.
View this article via: PubMed CrossRef Google Scholar

Satoh T, et al. Lapatinib plus paclitaxel versus paclitaxel alone in the second-line treatment of HER2-amplified advanced gastric cancer in Asian populations: TyTAN--a randomized, phase III study. J Clin Oncol. 2014;32(19):2039–2049.
View this article via: PubMed CrossRef Google Scholar

Lordick F, Lorenzen S, Yamada Y, Ilson D. Optimal chemotherapy for advanced gastric cancer: is there a global consensus? Gastric Cancer. 2014;17(2):213–225.
View this article via: PubMed CrossRef Google Scholar

Shay JW, Bacchetti S. A survey of telomerase activity in human cancer. Eur J Cancer. 1997;33(5):787–791.
View this article via: PubMed CrossRef Google Scholar

Shay JW, Wright WE. Telomeres and telomerase in normal and cancer stem cells. FEBS Lett. 2010;584(17):3819–3825.
View this article via: PubMed CrossRef Google Scholar

La SH, Kim SJ, Kang HG, Lee HW, Chun KH. Ablation of human telomerase reverse transcriptase (hTERT) induces cellular senescence in gastric cancer through a galectin-3 dependent mechanism. Oncotarget. 2016;7(35):57117–57130.
View this article via: PubMed Google Scholar

Hu C, et al. hTERT promotes the invasion of gastric cancer cells by enhancing FOXO3a ubiquitination and subsequent ITGB1 upregulation. Gut. 2017;66(1):31–42.
View this article via: PubMed CrossRef Google Scholar

Tang B, et al. Human telomerase reverse transcriptase (hTERT) promotes gastric cancer invasion through cooperating with c-Myc to upregulate heparanase expression. Oncotarget. 2016;7(10):11364–11379.
View this article via: PubMed Google Scholar

Killela PJ, et al. TERT promoter mutations occur frequently in gliomas and a subset of tumors derived from cells with low rates of self-renewal. Proc Natl Acad Sci U S A. 2013;110(15):6021–6026.
View this article via: PubMed CrossRef Google Scholar

Vinagre J, et al. Frequency of TERT promoter mutations in human cancers. Nat Commun. 2013;4:2185.
View this article via: PubMed Google Scholar

Qu Y, et al. Low frequency of TERT promoter mutations in a large cohort of gallbladder and gastric cancers. Int J Cancer. 2014;134(12):2993–2994.
View this article via: PubMed CrossRef Google Scholar

Peifer M, et al. Telomerase activation by genomic rearrangements in high-risk neuroblastoma. Nature. 2015;526(7575):700–704.
View this article via: PubMed CrossRef Google Scholar

Cao Y, Bryan TM, Reddel RR. Increased copy number of the TERT and TERC telomerase subunit genes in cancer cells. Cancer Sci. 2008;99(6):1092–1099.
View this article via: PubMed CrossRef Google Scholar

Kyo S, Takakura M, Fujiwara T, Inoue M. Understanding and exploiting hTERT promoter regulation for diagnosis and treatment of human cancers. Cancer Sci. 2008;99(8):1528–1538.
View this article via: PubMed CrossRef Google Scholar

Hu LH, Chen FH, Li YR, Wang L. Real-time determination of human telomerase reverse transcriptase mRNA in gastric cancer. World J Gastroenterol. 2004;10(23):3514–3517.
View this article via: PubMed CrossRef Google Scholar

Chen F, Hu L, Li Y, Wang L. Expression of telomerase subunits in gastric cancer. J Huazhong Univ Sci Technol Med Sci. 2005;25(6):741–743.
View this article via: PubMed CrossRef Google Scholar

Gigek CO, et al. hTERT methylation and expression in gastric cancer. Biomarkers. 2009;14(8):630–636.
View this article via: PubMed CrossRef Google Scholar

Guo YA, et al. Mutation hotspots at CTCF binding sites coupled to chromosomal instability in gastrointestinal cancers. Nat Commun. 2018;9(1):1520.
View this article via: PubMed CrossRef Google Scholar

Lebel R, McDuff FO, Lavigne P, Grandbois M. Direct visualization of the binding of c-Myc/Max heterodimeric b-HLH-LZ to E-box sequences on the hTERT promoter. Biochemistry. 2007;46(36):10279–10286.
View this article via: PubMed CrossRef Google Scholar

O’Hagan RC, et al. Gene-target recognition among members of the myc superfamily and implications for oncogenesis. Nat Genet. 2000;24(2):113–119.
View this article via: PubMed CrossRef Google Scholar

Liao D. Emerging roles of the EBF family of transcription factors in tumor suppression. Mol Cancer Res. 2009;7(12):1893–1901.
View this article via: PubMed CrossRef Google Scholar

Neve RM, et al. A collection of breast cancer cell lines for the study of functionally distinct cancer subtypes. Cancer Cell. 2006;10(6):515–527.
View this article via: PubMed CrossRef Google Scholar

GTEx Consortium. Human genomics. The Genotype-Tissue Expression (GTEx) pilot analysis: multitissue gene regulation in humans. Science. 2015;348(6235):648–660.
View this article via: PubMed CrossRef Google Scholar

Duff MO, et al. Genome-wide identification of zero nucleotide recursive splicing in Drosophila. Nature. 2015;521(7552):376–379.
View this article via: PubMed CrossRef Google Scholar

Fagerberg L, et al. Analysis of the human tissue-specific expression by genome-wide integration of transcriptomics and antibody-based proteomics. Mol Cell Proteomics. 2014;13(2):397–406.
View this article via: PubMed CrossRef Google Scholar

Nanki K, et al. Divergent routes toward Wnt and R-spondin niche independency during human gastric carcinogenesis. Cell. 2018;174(4):856–869.e17.
View this article via: PubMed CrossRef Google Scholar

Viré E, et al. The Polycomb group protein EZH2 directly controls DNA methylation. Nature. 2006;439(7078):871–874.
View this article via: PubMed CrossRef Google Scholar

Varambally S, et al. The polycomb group protein EZH2 is involved in progression of prostate cancer. Nature. 2002;419(6907):624–629.
View this article via: PubMed CrossRef Google Scholar

Guerra JVDS, et al. Genes Controlled by DNA Methylation Are Involved in Wilms Tumor Progression. Cells. 2019;8(8):E921.
View this article via: PubMed Google Scholar

Martino D, et al. Epigenome-wide association study reveals longitudinally stable DNA methylation differences in CD4⁺ T cells from children with IgE-mediated food allergy. Epigenetics. 2014;9(7):998–1006.
View this article via: PubMed CrossRef Google Scholar

Shao L, et al. Methylation of the HOXA10 promoter directs miR-196b-5p-dependent cell proliferation and invasion of gastric cancer cells. Mol Cancer Res. 2018;16(4):696–706.
View this article via: PubMed CrossRef Google Scholar

Lauffer BE, et al. Histone deacetylase (HDAC) inhibitor kinetic rate constants correlate with cellular histone acetylation but not transcription and cell viability. J Biol Chem. 2013;288(37):26926–26943.
View this article via: PubMed CrossRef Google Scholar

Jung I, et al. A compendium of promoter-centered long-range chromatin interactions in the human genome. Nat Genet. 2019;51(10):1442–1449.
View this article via: PubMed CrossRef Google Scholar

Huang FW, Hodis E, Xu MJ, Kryukov GV, Chin L, Garraway LA. Highly recurrent TERT promoter mutations in human melanoma. Science. 2013;339(6122):957–959.
View this article via: PubMed CrossRef Google Scholar

McLean CY, et al. GREAT improves functional interpretation of cis-regulatory regions. Nat Biotechnol. 2010;28(5):495–501.
View this article via: PubMed CrossRef Google Scholar

Treiber N, Treiber T, Zocher G, Grosschedl R. Structure of an Ebf1:DNA complex reveals unusual DNA recognition and structural homology with Rel proteins. Genes Dev. 2010;24(20):2270–2275.
View this article via: PubMed CrossRef Google Scholar

Landt SG, et al. ChIP-seq guidelines and practices of the ENCODE and modENCODE consortia. Genome Res. 2012;22(9):1813–1831.
View this article via: PubMed CrossRef Google Scholar

Heinz S, et al. Simple combinations of lineage-determining transcription factors prime cis-regulatory elements required for macrophage and B cell identities. Mol Cell. 2010;38(4):576–589.
View this article via: PubMed CrossRef Google Scholar

Muratani M, et al. Nanoscale chromatin profiling of gastric adenocarcinoma reveals cancer-associated cryptic promoters and somatically acquired regulatory elements. Nat Commun. 2014;5:4361.
View this article via: PubMed Google Scholar

Farh KK, et al. Genetic and epigenetic fine mapping of causal autoimmune disease variants. Nature. 2015;518(7539):337–343.
View this article via: PubMed CrossRef Google Scholar

Yan T, et al. HoxC5 and miR-615-3p target newly evolved genomic regions to repress hTERT and inhibit tumorigenesis. Nat Commun. 2018;9(1):100.
View this article via: PubMed CrossRef Google Scholar

Kim NW, et al. Specific association of human telomerase activity with immortal cells and cancer. Science. 1994;266(5193):2011–2015.
View this article via: PubMed CrossRef Google Scholar

Jin F, et al. A high-resolution map of the three-dimensional chromatin interactome in human cells. Nature. 2013;503(7475):290–294.
View this article via: PubMed CrossRef Google Scholar

Yao X, et al. VHL Deficiency drives enhancer activation of oncogenes in clear cell renal cell carcinoma. Cancer Discov. 2017;7(11):1284–1305.
View this article via: PubMed CrossRef Google Scholar

Platt JL, et al. Capture-C reveals preformed chromatin interactions between HIF-binding sites and distant promoters. EMBO Rep. 2016;17(10):1410–1421.
View this article via: PubMed CrossRef Google Scholar

Barthel FP, et al. Systematic analysis of telomere length and somatic alterations in 31 cancer types. Nat Genet. 2017;49(3):349–357.
View this article via: PubMed CrossRef Google Scholar

Liberg D, Sigvardsson M, Akerblad P. The EBF/Olf/Collier family of transcription factors: regulators of differentiation in cells originating from all three embryonal germ layers. Mol Cell Biol. 2002;22(24):8389–8397.
View this article via: PubMed CrossRef Google Scholar

Boller S, Grosschedl R. The regulatory network of B-cell differentiation: a focused view of early B-cell factor 1 function. Immunol Rev. 2014;261(1):102–115.
View this article via: PubMed CrossRef Google Scholar

Jimenez MA, Akerblad P, Sigvardsson M, Rosen ED. Critical role for Ebf1 and Ebf2 in the adipogenic transcriptional cascade. Mol Cell Biol. 2007;27(2):743–757.
View this article via: PubMed CrossRef Google Scholar

Gao H, et al. Early B cell factor 1 regulates adipocyte morphology and lipolysis in white adipose tissue. Cell Metab. 2014;19(6):981–992.
View this article via: PubMed CrossRef Google Scholar

Garel S, Marín F, Grosschedl R, Charnay P. Ebf1 controls early cell differentiation in the embryonic striatum. Development. 1999;126(23):5285–5294.
View this article via: PubMed Google Scholar

Fernandez-Jimenez N, et al. Lowly methylated region analysis identifies EBF1 as a potential epigenetic modifier in breast cancer. Epigenetics. 2017;12(11):964–972.
View this article via: PubMed CrossRef Google Scholar

Armartmuntree N, et al. Prolonged oxidative stress down-regulates Early B cell factor 1 with inhibition of its tumor suppressive function against cholangiocarcinoma genesis. Redox Biol. 2018;14:637–644.
View this article via: PubMed CrossRef Google Scholar

Huang KK, et al. Genomic and epigenomic profiling of high-risk intestinal metaplasia reveals molecular determinants of progression to gastric cancer. Cancer Cell. 2018;33(1):137–150.e5.
View this article via: PubMed CrossRef Google Scholar

Goh AM, Coffill CR, Lane DP. The role of mutant p53 in human cancer. J Pathol. 2011;223(2):116–126.
View this article via: PubMed CrossRef Google Scholar

Roadmap Epigenomics Consortium, et al. Integrative analysis of 111 reference human epigenomes. Nature. 2015;518(7539):317–330.
View this article via: PubMed CrossRef Google Scholar

Xu Y, Goldkorn A. Telomere and telomerase therapeutics in cancer. Genes (Basel). 2016;7(6):E22.
View this article via: PubMed Google Scholar

Chiappori AA, et al. A randomized phase II study of the telomerase inhibitor imetelstat as maintenance therapy for advanced non-small-cell lung cancer. Ann Oncol. 2015;26(2):354–362.
View this article via: PubMed CrossRef Google Scholar

[1] Ferlay J, et al. Cancer incidence and mortality worldwide: sources, methods and major patterns in GLOBOCAN 2012. Int J Cancer. 2015;136(5):E359–E386.
View this article via: PubMed CrossRef Google Scholar

[2] Seyedin S, Wang PC, Zhang Q, Lee P. Benefit of adjuvant chemoradiotherapy for gastric adenocarcinoma: a SEER population analysis. Gastrointest Cancer Res. 2014;7(3-4):82–90.
View this article via: PubMed Google Scholar

[3] Bang YJ, et al. Trastuzumab in combination with chemotherapy versus chemotherapy alone for treatment of HER2-positive advanced gastric or gastro-oesophageal junction cancer (ToGA): a phase 3, open-label, randomised controlled trial. Lancet. 2010;376(9742):687–697.
View this article via: PubMed CrossRef Google Scholar

[4] Fuchs CS, et al. Ramucirumab monotherapy for previously treated advanced gastric or gastro-oesophageal junction adenocarcinoma (REGARD): an international, randomised, multicentre, placebo-controlled, phase 3 trial. Lancet. 2014;383(9911):31–39.
View this article via: PubMed CrossRef Google Scholar

[5] Schnipper LE, et al. American Society of Clinical Oncology statement: a conceptual framework to assess the value of cancer treatment options. J Clin Oncol. 2015;33(23):2563–2577.
View this article via: PubMed CrossRef Google Scholar

[6] Kang YK, et al. Nivolumab in patients with advanced gastric or gastro-oesophageal junction cancer refractory to, or intolerant of, at least two previous chemotherapy regimens (ONO-4538-12, ATTRACTION-2): a randomised, double-blind, placebo-controlled, phase 3 trial. Lancet. 2017;390(10111):2461–2471.
View this article via: PubMed CrossRef Google Scholar

[7] Waddell T, et al. Epirubicin, oxaliplatin, and capecitabine with or without panitumumab for patients with previously untreated advanced oesophagogastric cancer (REAL3): a randomised, open-label phase 3 trial. Lancet Oncol. 2013;14(6):481–489.
View this article via: PubMed CrossRef Google Scholar

[8] Lordick F, et al. Capecitabine and cisplatin with or without cetuximab for patients with previously untreated advanced gastric cancer (EXPAND): a randomised, open-label phase 3 trial. Lancet Oncol. 2013;14(6):490–499.
View this article via: PubMed CrossRef Google Scholar

[9] Satoh T, et al. Lapatinib plus paclitaxel versus paclitaxel alone in the second-line treatment of HER2-amplified advanced gastric cancer in Asian populations: TyTAN--a randomized, phase III study. J Clin Oncol. 2014;32(19):2039–2049.
View this article via: PubMed CrossRef Google Scholar

[10] Lordick F, Lorenzen S, Yamada Y, Ilson D. Optimal chemotherapy for advanced gastric cancer: is there a global consensus? Gastric Cancer. 2014;17(2):213–225.
View this article via: PubMed CrossRef Google Scholar

[11] Shay JW, Bacchetti S. A survey of telomerase activity in human cancer. Eur J Cancer. 1997;33(5):787–791.
View this article via: PubMed CrossRef Google Scholar

[12] Shay JW, Wright WE. Telomeres and telomerase in normal and cancer stem cells. FEBS Lett. 2010;584(17):3819–3825.
View this article via: PubMed CrossRef Google Scholar

[13] La SH, Kim SJ, Kang HG, Lee HW, Chun KH. Ablation of human telomerase reverse transcriptase (hTERT) induces cellular senescence in gastric cancer through a galectin-3 dependent mechanism. Oncotarget. 2016;7(35):57117–57130.
View this article via: PubMed Google Scholar

[14] Hu C, et al. hTERT promotes the invasion of gastric cancer cells by enhancing FOXO3a ubiquitination and subsequent ITGB1 upregulation. Gut. 2017;66(1):31–42.
View this article via: PubMed CrossRef Google Scholar

[15] Tang B, et al. Human telomerase reverse transcriptase (hTERT) promotes gastric cancer invasion through cooperating with c-Myc to upregulate heparanase expression. Oncotarget. 2016;7(10):11364–11379.
View this article via: PubMed Google Scholar

[16] Killela PJ, et al. TERT promoter mutations occur frequently in gliomas and a subset of tumors derived from cells with low rates of self-renewal. Proc Natl Acad Sci U S A. 2013;110(15):6021–6026.
View this article via: PubMed CrossRef Google Scholar

[17] Vinagre J, et al. Frequency of TERT promoter mutations in human cancers. Nat Commun. 2013;4:2185.
View this article via: PubMed Google Scholar

[18] Qu Y, et al. Low frequency of TERT promoter mutations in a large cohort of gallbladder and gastric cancers. Int J Cancer. 2014;134(12):2993–2994.
View this article via: PubMed CrossRef Google Scholar

[19] Peifer M, et al. Telomerase activation by genomic rearrangements in high-risk neuroblastoma. Nature. 2015;526(7575):700–704.
View this article via: PubMed CrossRef Google Scholar

[20] Cao Y, Bryan TM, Reddel RR. Increased copy number of the TERT and TERC telomerase subunit genes in cancer cells. Cancer Sci. 2008;99(6):1092–1099.
View this article via: PubMed CrossRef Google Scholar

[21] Kyo S, Takakura M, Fujiwara T, Inoue M. Understanding and exploiting hTERT promoter regulation for diagnosis and treatment of human cancers. Cancer Sci. 2008;99(8):1528–1538.
View this article via: PubMed CrossRef Google Scholar

[22] Hu LH, Chen FH, Li YR, Wang L. Real-time determination of human telomerase reverse transcriptase mRNA in gastric cancer. World J Gastroenterol. 2004;10(23):3514–3517.
View this article via: PubMed CrossRef Google Scholar

[23] Chen F, Hu L, Li Y, Wang L. Expression of telomerase subunits in gastric cancer. J Huazhong Univ Sci Technol Med Sci. 2005;25(6):741–743.
View this article via: PubMed CrossRef Google Scholar

[24] Gigek CO, et al. hTERT methylation and expression in gastric cancer. Biomarkers. 2009;14(8):630–636.
View this article via: PubMed CrossRef Google Scholar

[25] Guo YA, et al. Mutation hotspots at CTCF binding sites coupled to chromosomal instability in gastrointestinal cancers. Nat Commun. 2018;9(1):1520.
View this article via: PubMed CrossRef Google Scholar

[26] Lebel R, McDuff FO, Lavigne P, Grandbois M. Direct visualization of the binding of c-Myc/Max heterodimeric b-HLH-LZ to E-box sequences on the hTERT promoter. Biochemistry. 2007;46(36):10279–10286.
View this article via: PubMed CrossRef Google Scholar

[27] O’Hagan RC, et al. Gene-target recognition among members of the myc superfamily and implications for oncogenesis. Nat Genet. 2000;24(2):113–119.
View this article via: PubMed CrossRef Google Scholar

[28] Liao D. Emerging roles of the EBF family of transcription factors in tumor suppression. Mol Cancer Res. 2009;7(12):1893–1901.
View this article via: PubMed CrossRef Google Scholar

[29] Neve RM, et al. A collection of breast cancer cell lines for the study of functionally distinct cancer subtypes. Cancer Cell. 2006;10(6):515–527.
View this article via: PubMed CrossRef Google Scholar

[30] GTEx Consortium. Human genomics. The Genotype-Tissue Expression (GTEx) pilot analysis: multitissue gene regulation in humans. Science. 2015;348(6235):648–660.
View this article via: PubMed CrossRef Google Scholar

[31] Duff MO, et al. Genome-wide identification of zero nucleotide recursive splicing in Drosophila. Nature. 2015;521(7552):376–379.
View this article via: PubMed CrossRef Google Scholar

[32] Fagerberg L, et al. Analysis of the human tissue-specific expression by genome-wide integration of transcriptomics and antibody-based proteomics. Mol Cell Proteomics. 2014;13(2):397–406.
View this article via: PubMed CrossRef Google Scholar

[33] Nanki K, et al. Divergent routes toward Wnt and R-spondin niche independency during human gastric carcinogenesis. Cell. 2018;174(4):856–869.e17.
View this article via: PubMed CrossRef Google Scholar

[34] Viré E, et al. The Polycomb group protein EZH2 directly controls DNA methylation. Nature. 2006;439(7078):871–874.
View this article via: PubMed CrossRef Google Scholar

[35] Varambally S, et al. The polycomb group protein EZH2 is involved in progression of prostate cancer. Nature. 2002;419(6907):624–629.
View this article via: PubMed CrossRef Google Scholar

[36] Guerra JVDS, et al. Genes Controlled by DNA Methylation Are Involved in Wilms Tumor Progression. Cells. 2019;8(8):E921.
View this article via: PubMed Google Scholar

[37] Martino D, et al. Epigenome-wide association study reveals longitudinally stable DNA methylation differences in CD4⁺ T cells from children with IgE-mediated food allergy. Epigenetics. 2014;9(7):998–1006.
View this article via: PubMed CrossRef Google Scholar

[38] Shao L, et al. Methylation of the HOXA10 promoter directs miR-196b-5p-dependent cell proliferation and invasion of gastric cancer cells. Mol Cancer Res. 2018;16(4):696–706.
View this article via: PubMed CrossRef Google Scholar

[39] Lauffer BE, et al. Histone deacetylase (HDAC) inhibitor kinetic rate constants correlate with cellular histone acetylation but not transcription and cell viability. J Biol Chem. 2013;288(37):26926–26943.
View this article via: PubMed CrossRef Google Scholar

[40] Jung I, et al. A compendium of promoter-centered long-range chromatin interactions in the human genome. Nat Genet. 2019;51(10):1442–1449.
View this article via: PubMed CrossRef Google Scholar

[41] Huang FW, Hodis E, Xu MJ, Kryukov GV, Chin L, Garraway LA. Highly recurrent TERT promoter mutations in human melanoma. Science. 2013;339(6122):957–959.
View this article via: PubMed CrossRef Google Scholar

[42] McLean CY, et al. GREAT improves functional interpretation of cis-regulatory regions. Nat Biotechnol. 2010;28(5):495–501.
View this article via: PubMed CrossRef Google Scholar

[43] Treiber N, Treiber T, Zocher G, Grosschedl R. Structure of an Ebf1:DNA complex reveals unusual DNA recognition and structural homology with Rel proteins. Genes Dev. 2010;24(20):2270–2275.
View this article via: PubMed CrossRef Google Scholar

[44] Landt SG, et al. ChIP-seq guidelines and practices of the ENCODE and modENCODE consortia. Genome Res. 2012;22(9):1813–1831.
View this article via: PubMed CrossRef Google Scholar

[45] Heinz S, et al. Simple combinations of lineage-determining transcription factors prime cis-regulatory elements required for macrophage and B cell identities. Mol Cell. 2010;38(4):576–589.
View this article via: PubMed CrossRef Google Scholar

[46] Muratani M, et al. Nanoscale chromatin profiling of gastric adenocarcinoma reveals cancer-associated cryptic promoters and somatically acquired regulatory elements. Nat Commun. 2014;5:4361.
View this article via: PubMed Google Scholar

[47] Farh KK, et al. Genetic and epigenetic fine mapping of causal autoimmune disease variants. Nature. 2015;518(7539):337–343.
View this article via: PubMed CrossRef Google Scholar

[48] Yan T, et al. HoxC5 and miR-615-3p target newly evolved genomic regions to repress hTERT and inhibit tumorigenesis. Nat Commun. 2018;9(1):100.
View this article via: PubMed CrossRef Google Scholar

[49] Kim NW, et al. Specific association of human telomerase activity with immortal cells and cancer. Science. 1994;266(5193):2011–2015.
View this article via: PubMed CrossRef Google Scholar

[50] Jin F, et al. A high-resolution map of the three-dimensional chromatin interactome in human cells. Nature. 2013;503(7475):290–294.
View this article via: PubMed CrossRef Google Scholar

[51] Yao X, et al. VHL Deficiency drives enhancer activation of oncogenes in clear cell renal cell carcinoma. Cancer Discov. 2017;7(11):1284–1305.
View this article via: PubMed CrossRef Google Scholar

[52] Platt JL, et al. Capture-C reveals preformed chromatin interactions between HIF-binding sites and distant promoters. EMBO Rep. 2016;17(10):1410–1421.
View this article via: PubMed CrossRef Google Scholar

[53] Barthel FP, et al. Systematic analysis of telomere length and somatic alterations in 31 cancer types. Nat Genet. 2017;49(3):349–357.
View this article via: PubMed CrossRef Google Scholar

[54] Liberg D, Sigvardsson M, Akerblad P. The EBF/Olf/Collier family of transcription factors: regulators of differentiation in cells originating from all three embryonal germ layers. Mol Cell Biol. 2002;22(24):8389–8397.
View this article via: PubMed CrossRef Google Scholar

[55] Boller S, Grosschedl R. The regulatory network of B-cell differentiation: a focused view of early B-cell factor 1 function. Immunol Rev. 2014;261(1):102–115.
View this article via: PubMed CrossRef Google Scholar

[56] Jimenez MA, Akerblad P, Sigvardsson M, Rosen ED. Critical role for Ebf1 and Ebf2 in the adipogenic transcriptional cascade. Mol Cell Biol. 2007;27(2):743–757.
View this article via: PubMed CrossRef Google Scholar

[57] Gao H, et al. Early B cell factor 1 regulates adipocyte morphology and lipolysis in white adipose tissue. Cell Metab. 2014;19(6):981–992.
View this article via: PubMed CrossRef Google Scholar

[58] Garel S, Marín F, Grosschedl R, Charnay P. Ebf1 controls early cell differentiation in the embryonic striatum. Development. 1999;126(23):5285–5294.
View this article via: PubMed Google Scholar

[59] Fernandez-Jimenez N, et al. Lowly methylated region analysis identifies EBF1 as a potential epigenetic modifier in breast cancer. Epigenetics. 2017;12(11):964–972.
View this article via: PubMed CrossRef Google Scholar

[60] Armartmuntree N, et al. Prolonged oxidative stress down-regulates Early B cell factor 1 with inhibition of its tumor suppressive function against cholangiocarcinoma genesis. Redox Biol. 2018;14:637–644.
View this article via: PubMed CrossRef Google Scholar

[61] Huang KK, et al. Genomic and epigenomic profiling of high-risk intestinal metaplasia reveals molecular determinants of progression to gastric cancer. Cancer Cell. 2018;33(1):137–150.e5.
View this article via: PubMed CrossRef Google Scholar

[62] Goh AM, Coffill CR, Lane DP. The role of mutant p53 in human cancer. J Pathol. 2011;223(2):116–126.
View this article via: PubMed CrossRef Google Scholar

[63] Roadmap Epigenomics Consortium, et al. Integrative analysis of 111 reference human epigenomes. Nature. 2015;518(7539):317–330.
View this article via: PubMed CrossRef Google Scholar

[64] Xu Y, Goldkorn A. Telomere and telomerase therapeutics in cancer. Genes (Basel). 2016;7(6):E22.
View this article via: PubMed Google Scholar

[65] Chiappori AA, et al. A randomized phase II study of the telomerase inhibitor imetelstat as maintenance therapy for advanced non-small-cell lung cancer. Ann Oncol. 2015;26(2):354–362.
View this article via: PubMed CrossRef Google Scholar

Genomic and epigenomic EBF1 alterations modulate TERT expression in gastric cancer

Article tools

Metrics

Go to

Genomic and epigenomic EBF1 alterations modulate TERT expression in gastric cancer

Article tools

Metrics

Go to

Sign up for email alerts