Long noncoding RNAs (lncRNA) play a role in carcinogenesis. However, the function of lncRNAs in human gastric cancer remains largely unknown. In this study, we identified a novel lncRNA, GClnc1, which was upregulated and associated with tumorigenesis, tumor size, metastasis, and poor prognosis in gastric cancer. GClnc1 affected gastric cancer cell proliferation, invasiveness, and metastasis in multiple gastric cancer models. Mechanistically, GClnc1 bound WDR5 (a key component of histone methyltransferase complex) and KAT2A histone acetyltransferase, acted as a modular scaffold of WDR5 and KAT2A complexes, coordinated their localization, specified the histone modification pattern on the target genes, including SOD2, and consequently altered gastric cancer cell biology. Thus, GClnc1 is mechanistically, functionally, and clinically oncogenic in gastric cancer. Targeting GClnc1 and its pathway may be meaningful for treating patients with gastric cancer.

Significance: This report documents a novel lncRNA, GClnc1, which may act as a scaffold to recruit the WDR5 and KAT2A complex and modify the transcription of target genes. This study reveals that GClnc1 is an oncogenic lncRNA in human gastric cancer. Cancer Discov; 6(7); 784–801. ©2016 AACR.

This article is highlighted in the In This Issue feature, p. 681

Gastric cancer is the fourth most diagnosed type of cancer and the third most common cause of cancer-related death worldwide (1, 2). Patients with advanced gastric cancer have a poor prognosis (3, 4). Pathologic classification is used to assess prognosis and inform the treatment of gastric cancer. Massive efforts have been made to develop the noninvasive biomarkers to detect early cancer and/or reflect an individual's cancer risk, which is essential to reducing gastric cancer mortality (5). However, there has been little success in improving the disease-free survival rate of patients. Because the pathologic mechanisms of gastric cancer progression are not fully understood, more research is needed to discover and develop effective biomarkers and targets for gastric cancer diagnosis and treatment.

Long noncoding RNAs (lncRNA) are a class of noncoding transcripts greater than 200 nucleotides in length. Recent studies have revealed that lncRNAs may affect cancer progression (6, 7). For example, the lncRNAs HOTAIR, MEG3, MALAT1, H19, and GAPLINC may play a role in carcinogenesis (8–12). The role of lncRNAs and the underlying mechanisms in gastric cancer has been reported before (11–16). However, more specific mechanisms of lncRNAs in gastric cancer initiation and development need to be further teased out.

In the current study, we examined genome-wide lncRNA expression profiles in gastric tumors and paired adjacent normal tissue, from which we identified and characterized a novel lncRNA (BC041951) that was associated with prognosis in patients with gastric cancer, and therefore designated it as gastric cancer–associated lncRNA 1 (GClnc1). The biological roles of GClnc1 in gastric carcinogenesis were genetically assessed in several in vitro and in vivo models. Mechanistically, mass spectrometry combined with integrative analysis revealed that GClnc1 not only modified genomic binding of WDR5 and KAT2A, but also specified the pattern of histone modifications on target genes. Furthermore, GClnc1 upregulated the transcription of superoxide dismutase 2 mitochondrial (SOD2), one of the target genes of the WDR5/KAT2A complex, by acting as a scaffold to recruit the WDR5 and KAT2A complex to the SOD2 promoter and increasing H3K4 trimethylation and H3K9 acetylation levels in the SOD2 promoter region. Thus, our study has identified a novel lncRNA, GClnc1, with a biological, mechanistic, and clinical impact on human gastric cancer.

LncRNA Candidate BC041951 Is Clinically Relevant in Gastric Cancer

The Arraystar Human LncRNA Microarray V2.0 (8 × 60K; Arraystar) was used to profile lncRNA expression in 10 paired gastric cancer tissues and paired adjacent normal tissues. In total, 33,045 lncRNAs and 30,215 coding transcripts were collected from the most authoritative databases, such as RefSeq, the UCSC Known Genes dataset (Known Genes 4), Ensembl 37.59, H-invDB 7.0, and RNAdb 2.0, as described in the manufacturer's instructions, and the raw data can be accessed via GSE50710 (12). In order to study these microarray data and rediscover potential biomarkers not yet mined completely, more stringent filtering criteria (raw signal intensity > 1,500, fold change > 2, P < 0.01, Supplementary Fig. S1A) were used in our investigation. We found that eight candidate lncRNAs were significantly increased in gastric cancer tissues compared with adjacent normal tissues (Supplementary Table S1). To validate microarray data, we analyzed the expression of the eight candidate lncRNAs in 165 cases of fresh gastric cancer and adjacent tissues (cohort 1; Supplementary Table S2). Real-time PCR revealed that four of the eight lncRNAs were significantly increased and two of the eight lncRNAs were significantly decreased in cancer versus adjacent tissues of cohort 1 (Supplementary Fig. S1B). These data indicate that a set of lncRNAs is aberrantly expressed in gastric cancer tissues.

We next analyzed the correlation between the four lncRNA candidates and the clinical outcome in cohort 1. The Kaplan–Meier analyses showed that the lncRNA candidates AK094163, ENST00000430239, and NR_024373 have no predictive value for the clinical outcome of patients with gastric cancer, whereas high expression of the lncRNA candidate BC041951 was significantly associated with a poor prognosis in these patients (Fig. 1A; Supplementary Fig. S1C). Furthermore, real-time PCR showed that only BC041951 expression gradually increased from normal gastric tissue to intestinal metaplasia (IM), to dysplasia (DYS), and to gastric cancer (Fig. 1B; Supplementary Fig. S1D). The expression of BC041951 was significantly increased in gastric cancer tissues compared with normal tissues in patients enrolled in Renji Hospital (Supplementary Fig. S1E) and in an external gastric cancer dataset from the Gene Expression Omnibus (GEO) database repository (Supplementary Fig. S1F). In laser capture microdissection (LCM) assays, BC041951 was mainly expressed and was significantly increased in the epithelial cells of gastric cancer tissues compared with that in normal gastric epithelial cells (Supplementary Fig. S1G) and tumor stromal cells (Supplementary Fig. S1H). Thus, an increase in BC041951 may be an early event in the multistep progression of gastric carcinogenesis.

Figure 1.

LncRNA candidate BC041951 is clinically relevant in gastric cancer. A, survival was analyzed and compared between patients with low and high levels of BC041951 in 165 patients with gastric cancer (cohort 1), log-rank test. B, statistical analysis of lncRNA BC041951 expression in normal, IM, DYS, and cancerous gastric tissues, nonparametric Mann–Whitney test. All the bars represent SE. C, prediction error curves of the different predictors in cohort 1. Apparent error (AE) and 10-fold cross-validated cumulative prediction error (PE), computed for a 5-year follow-up using Kaplan–Meier estimation as reference. D, comparing different histologic stages, positive or negative vascular invasion, tumor size, and AJCC stage between GClnc1 high-expression and GClnc1 low-expression tumors of cohort 1. The heatmap illustrates the association of different clinical characters with GClnc1 high- and low-expression tumors. Statistical significance was performed by the χ2 test. PD, pathologic differentiation. E, multivariable analysis was performed in cohort 1. All the bars correspond to 95% confidence intervals. F, representative images of GClnc1 expression in gastric cancer and adjacent gastric tissues using ISH analysis in cohort 2; n = 105. G, survival was analyzed and compared between patients with high and low levels of GClnc1 expression in tumor in cohort 2; n = 105, log-rank test. H, multivariable analysis was performed in cohort 2. All the bars correspond to 95% confidence intervals.

Figure 1.

LncRNA candidate BC041951 is clinically relevant in gastric cancer. A, survival was analyzed and compared between patients with low and high levels of BC041951 in 165 patients with gastric cancer (cohort 1), log-rank test. B, statistical analysis of lncRNA BC041951 expression in normal, IM, DYS, and cancerous gastric tissues, nonparametric Mann–Whitney test. All the bars represent SE. C, prediction error curves of the different predictors in cohort 1. Apparent error (AE) and 10-fold cross-validated cumulative prediction error (PE), computed for a 5-year follow-up using Kaplan–Meier estimation as reference. D, comparing different histologic stages, positive or negative vascular invasion, tumor size, and AJCC stage between GClnc1 high-expression and GClnc1 low-expression tumors of cohort 1. The heatmap illustrates the association of different clinical characters with GClnc1 high- and low-expression tumors. Statistical significance was performed by the χ2 test. PD, pathologic differentiation. E, multivariable analysis was performed in cohort 1. All the bars correspond to 95% confidence intervals. F, representative images of GClnc1 expression in gastric cancer and adjacent gastric tissues using ISH analysis in cohort 2; n = 105. G, survival was analyzed and compared between patients with high and low levels of GClnc1 expression in tumor in cohort 2; n = 105, log-rank test. H, multivariable analysis was performed in cohort 2. All the bars correspond to 95% confidence intervals.

Close modal

To evaluate the pathologic and clinical value of the four overexpressed lncRNA candidates with the American Joint Committee on Cancer (AJCC) stage for prognosis, we computed their accuracy by prediction error curves. Predication error over time was calculated using the Brier Score. The predication error was the lowest using the combination of BC041951 and AJCC stage in cohort 1 (Fig. 1C), but not with the other three lncRNA candidates (Supplementary Fig. S1I). Further, ROC analysis showed that the AUC of the combination of BC041951-based prediction and AJCC-based model (0.732) was higher than the AJCC-based model alone (0.695; Supplementary Fig. S1J). The data indicate that the combination of BC041951 and AJCC stage is more precise in predicting clinical outcome than AJCC stage alone. Therefore, we focused our research on BC041951, and henceforth named this lncRNA candidate gastric cancer–associated lncRNA1 (GClnc1).

We next evaluated and compared GClnc1 expression with different clinicopathologic features in cohort 1. We found that GClnc1 expression positively correlated with pathologic differentiation, vascular invasion, tumor size, and AJCC stage (Fig. 1D). Univariate and multivariate regression analyses of cohort 1 demonstrated that GClnc1 expression was an independent predictor of gastric cancer aggressiveness with significant HRs for predicting clinical outcome. Its predictive value was comparable to that of the AJCC stage (Supplementary Fig. S1K and S1L; Fig. 1E).

To further validate the pathologic and clinical significance of GClnc1 expression in gastric cancer, we detected and compared GClnc1 expression by in situ hybridization (ISH) in an additional 105 paraffin-embedded gastric cancer and adjacent tissues (cohort 2; Supplementary Table S3). GClnc1 expression was higher in gastric cancer tissues than in adjacent tissues (Fig. 1F). Consistent with the results in cohort 1, high levels of GClnc1 expression were significantly associated with poor survival (Fig. 1G) in univariate and multivariate regression analyses (Supplementary Fig. S1M and S1N; Fig. 1H). Because H. pylori infections may be detected in 78% of all cases of patients with gastric cancer (17), we measured and found similar levels of GClnc1 expression in H. pylori–positive and H. pylori–negative tumor tissues (Supplementary Fig. S1O).

In order to determine whether GClnc1 is a novel lncRNA, we first analyzed its sequence. Northern blot revealed that the size of GClnc1 was approximately 2 kb in length in gastric cancer cell lines and gastric cancer tissues (Supplementary Fig. S1P and S1Q). The 5′ rapid amplification of cDNA ends (RACE) PCR (Supplementary Fig. S1R) and the chromatin immunoprecipitation (ChIP) assay with anti-RNA Pol II (Supplementary Fig. S1S) were performed to identify the 5′ ends and the transcription start site of GClnc1. Sequencing of PCR products revealed the boundary between the universal anchor primer and GClnc1 (Supplementary Fig. S1R). ChIP assays demonstrated that RNA Pol II bound to the predicted bind site of the 5′ ends of GClnc1 (Supplementary Fig. S1S). Thus, GClnc1 is an RNA Pol II transcript. Analysis of the GClnc1 sequences by ORF Finder from the National Center for Biotechnology Information failed to predict a protein (Supplementary Fig. S1T). GClnc1 did not contain a valid Kozak sequence. It was identified to be an lncRNA rather than a protein-coding transcript by CNCI software (18). We performed a codon substitution frequency analysis using PhyloCSF (19). GClnc1 had a very low codon substitution frequency score (–838.1; Supplementary Fig. S1U), indicating that it was unlikely to encode any protein product. We confirmed that GClnc1 was an ncRNA by in vitro translation analysis (Supplementary Fig. S1V). We also separated the nuclear and cytoplasm fractions of BGC823 and MKN45 gastric cancer cells and performed real-time PCR. We found that GClnc1 was mainly located in the nucleus of BGC823 and MKN45 cancer cells (Supplementary Fig. S1W). Collectively, GClnc1 is a novel lncRNA and highly expressed in gastric cancer tissues.

GClnc1 Is an Oncogenic LncRNA in Gastric Cancer

To elucidate whether GClnc1 plays a role in gastric cancer tumorigenesis, an RNA sequencing (RNA-seq) analysis was performed to compare the gene expression profiles of GClnc1 shRNA and control shRNA transfectants. A total of 2,050 downregulated genes and 1,894 upregulated genes (≥2-fold) were detected (raw data accessible via GEO number GSE63765) after knockdown of GClnc1 in gastric cancer cells (Supplementary Table S4). Gene set enrichment analysis (GSEA) revealed that the gene sets related to Benporath_Proliferation (cell proliferation), Provenzani_Metastasis_Up (metastasis), Kim_Gastric_Cancer_Chemosensitivity (upregulated genes in chemoresistant gastric cancer cells), and Vecchi_Gastric_Cancer_Early_Up (gastric cancer–specific signature) negatively correlated with GClnc1 downregulation in gastric cancer cells (Fig. 2A–D). The top-scoring genes recurring in the four gene sets included key cancer genes, c-MYC and CDK1. Real-time PCR confirmed that alteration of GClnc1 expression dramatically affected the key tumorigenesis gene signatures (Supplementary Fig. S2A). To gain further insight into the biological pathways involved in gastric cancer pathogenesis, based on the median of GClnc1 expression levels, we performed GSEA analysis in an independent public dataset from GEO (GSE27342; Supplementary Fig. S2B). GSEA showed that the gene signatures of cell proliferation, cell cycle, cancer-related pathways, metastasis, and chemotherapy resistance pathways were enriched in patients with high GClnc1 expression, but not in patients with low GClnc1 expression (Fig. 2E). These data suggest that GClnc1 may be an important modulator in gastric tumorigenesis.

Figure 2.

The correlation between tumorigenic gene sets and GClnc1 expression levels is assessed via GSEA. A–D, overview of GSEA used to identify the differential gene profiles between shGClnc1 gastric cancer cells and controls. E, GSEA comparison between patients with high GClnc1 expression (red) and low GClnc1 expression (blue). Median split, n = 80 (GSE 27342). Distinct pathways and biological processes were illustrated between the two patient populations. Cytoscape and Enrichment map were used for visualization of the GSEA results (1% FDR; P = 0.005). Nodes represent enriched gene sets, which were grouped and annotated by their similarity according to the related gene sets. Enrichment results were mapped as a network of gene sets (nodes). Node size was proportional to the total number of genes within each gene set. Proportion of shared genes between gene sets was presented as the thickness of the green line between nodes. This simplified network map was manually curated by removing general and uninformative subnetworks, as shown in Supplementary Fig. S2B. Enrichment plots were shown for a set of activated genes related to cell proliferation, cell cycle, cancer-related pathways, and metastasis and chemotherapy resistance pathways in 80 patients with gastric cancer. The enrichment score (ES, green line) means the degree to which the gene set was overrepresented at the top or bottom of the ranked list of genes. Black bars indicate the position of genes belonging to the gene set in the ranked list of genes included in the analysis. A positive value indicates more correlation with “high GClnc1 expression,” and a negative value indicates more correlation with “low GClnc1 expression.” NES, normalized enrichment score.

Figure 2.

The correlation between tumorigenic gene sets and GClnc1 expression levels is assessed via GSEA. A–D, overview of GSEA used to identify the differential gene profiles between shGClnc1 gastric cancer cells and controls. E, GSEA comparison between patients with high GClnc1 expression (red) and low GClnc1 expression (blue). Median split, n = 80 (GSE 27342). Distinct pathways and biological processes were illustrated between the two patient populations. Cytoscape and Enrichment map were used for visualization of the GSEA results (1% FDR; P = 0.005). Nodes represent enriched gene sets, which were grouped and annotated by their similarity according to the related gene sets. Enrichment results were mapped as a network of gene sets (nodes). Node size was proportional to the total number of genes within each gene set. Proportion of shared genes between gene sets was presented as the thickness of the green line between nodes. This simplified network map was manually curated by removing general and uninformative subnetworks, as shown in Supplementary Fig. S2B. Enrichment plots were shown for a set of activated genes related to cell proliferation, cell cycle, cancer-related pathways, and metastasis and chemotherapy resistance pathways in 80 patients with gastric cancer. The enrichment score (ES, green line) means the degree to which the gene set was overrepresented at the top or bottom of the ranked list of genes. Black bars indicate the position of genes belonging to the gene set in the ranked list of genes included in the analysis. A positive value indicates more correlation with “high GClnc1 expression,” and a negative value indicates more correlation with “low GClnc1 expression.” NES, normalized enrichment score.

Close modal

To functionally validate the pathway findings, we transfected GClnc1 siRNAs into the gastric cancer cell lines BGC823 and MKN45. The two cell lines expressed higher levels of GClnc1 (Supplementary Fig. S3A) and exhibited higher proliferation (Supplementary Fig. S3B) and invasion capacities (Supplementary Fig. S3C) compared with normal GES-1 gastric cells (Supplementary Fig. S3A–S3C). The results showed that knockdown of GClnc1, but not the lncRNA ENST00000430239 (ENST430239), significantly impaired gastric cancer cell proliferation and colony formation in BGC823 cells and MKN45 cells (Fig. 3A; Supplementary Fig. S3D–S3F). Knockdown of GClnc1 dramatically reduced BGC823 and MKN45 tumor growth (Fig. 3B and C; Supplementary Fig. S3G and S3H) and tumor weight (Fig. 3D; Supplementary Fig. S3I) in xenograft mouse tumor models. In support of the protumor role of GClnc1, Ki67 staining revealed that downregulation of GClnc1 decreased tumor cell proliferation in vivo (Supplementary Fig. S3J and S3K). Furthermore, knockdown of GClnc1 sensitized BGC823 and MKN45 cells to 5′-fluorouracil and cisplatin treatment (Fig. 3E and F; Supplementary Fig. S3L and S3M). The data suggest that GClnc1 may be an oncogenic lncRNA in gastric cancer and control gastric cancer cell proliferation and chemoresistance.

Figure 3.

GClnc1 is an oncogenic lncRNA in gastric cancer. A, BrdUrd assay was performed in BGC823 and MKN45 cells transfected with GClnc1 siRNA1/2; n = 3, nonparametric Mann–Whitney test. B, representative data of tumors in nude mice bearing gastric cancer cells transfected with nothing, control shRNA adenovirus, and GClnc1 shRNA1 adenovirus; n = 8. C, tumor volume was measured after GClnc1 shRNA1 adenovirus treatments in the xenograft mouse model; n = 8; *, P < 0.01, nonparametric Mann–Whitney test. D, tumor weight was measured in mice after different treatments; n = 8, nonparametric Mann–Whitney test. E, dose–response curve of a representative experiment showing relative 5′-fluorouracil sensitivity determined by BrdUrd incorporation. BGC823 cells were transfected with control or GClnc1 siRNA1/2 and were treated with 5′-fluorouracil; n = 3, nonparametric Mann–Whitney test, compared with control siRNA group. F, dose–response curve of a representative experiment showing relative cisplatin sensitivity determined by BrdUrd incorporation. BGC823 cells were transfected with control or GClnc1 siRNA1/2 and treated with cisplatin; n = 3, nonparametric Mann–Whitney test, compared with control siRNA group. G, Transwell Matrigel invasion assay was performed in BGC823 cells transfected with control siRNA, GClnc1 siRNA1/2, and lnc ENST430239 siRNA 1/2, respectively; n = 3. H, survival analysis was performed in mice bearing gastric cancer transfected with GClnc1 shRNA1 adenovirus, control shRNA adenovirus, or nothing (PBS injection), respectively; n = 10, log-rank test. I, representative hematoxylin and eosin staining and summarized data on tumor lung foci in nude mice at 13 weeks after injection with GClnc1 shRNA1 adenovirus; n = 10.

Figure 3.

GClnc1 is an oncogenic lncRNA in gastric cancer. A, BrdUrd assay was performed in BGC823 and MKN45 cells transfected with GClnc1 siRNA1/2; n = 3, nonparametric Mann–Whitney test. B, representative data of tumors in nude mice bearing gastric cancer cells transfected with nothing, control shRNA adenovirus, and GClnc1 shRNA1 adenovirus; n = 8. C, tumor volume was measured after GClnc1 shRNA1 adenovirus treatments in the xenograft mouse model; n = 8; *, P < 0.01, nonparametric Mann–Whitney test. D, tumor weight was measured in mice after different treatments; n = 8, nonparametric Mann–Whitney test. E, dose–response curve of a representative experiment showing relative 5′-fluorouracil sensitivity determined by BrdUrd incorporation. BGC823 cells were transfected with control or GClnc1 siRNA1/2 and were treated with 5′-fluorouracil; n = 3, nonparametric Mann–Whitney test, compared with control siRNA group. F, dose–response curve of a representative experiment showing relative cisplatin sensitivity determined by BrdUrd incorporation. BGC823 cells were transfected with control or GClnc1 siRNA1/2 and treated with cisplatin; n = 3, nonparametric Mann–Whitney test, compared with control siRNA group. G, Transwell Matrigel invasion assay was performed in BGC823 cells transfected with control siRNA, GClnc1 siRNA1/2, and lnc ENST430239 siRNA 1/2, respectively; n = 3. H, survival analysis was performed in mice bearing gastric cancer transfected with GClnc1 shRNA1 adenovirus, control shRNA adenovirus, or nothing (PBS injection), respectively; n = 10, log-rank test. I, representative hematoxylin and eosin staining and summarized data on tumor lung foci in nude mice at 13 weeks after injection with GClnc1 shRNA1 adenovirus; n = 10.

Close modal

Next, we examined the effects of GClnc1 on gastric cancer cell invasion and metastasis. In the invasion assay, we showed that downregulation of GClnc1, but not the lncRNA ENST430239, significantly reduced the invasion ability in gastric cancer cells (Fig. 3G; Supplementary Fig. S3N). In a gastric cancer metastatic model, mice inoculated with GClnc1 shRNA-expressing tumor cells had a longer overall survival time than mice that received control shRNA–expressing tumor cells or phosphate-buffered solutions (PBS; Fig. 3H; Supplementary Fig. S3O). There were fewer metastatic foci in the lungs of nude mice at 13 weeks after injection of GClnc1 shRNA adenovirus, when compared with control groups (Fig. 3I; Supplementary Fig. S3P). Overexpression of GClnc1 increased cell proliferation and invasion ability of GES-1 and MGC803 cells in vitro and in vivo (Supplementary Fig. S3Q–S3V). The data strongly suggest that GClnc1 may promote gastric cancer progression by regulating gastric cancer cell proliferation and metastasis, and increasing chemotherapy resistance.

GClnc1 Interacts with WDR5 and KAT2A Epigenetic Modification Complex

LncRNAs may function by physically interacting with transcriptional factors, histone regulators, and other cellular factors (20–23). To explore the mechanism of GClnc1-mediated carcinogenesis in gastric cancer, we sought to identify intracellular GClnc1-binding factors using an unbiased approach. Biotinylated GClnc1 or antisense GClnc1 RNA (negative control) was incubated with total protein extracts from BGC823 cells and pulled down with streptavidin (Fig. 4A). The associated proteins were analyzed by SDS-PAGE and silver staining. There were two specific bands in the GClnc1 pull-down samples (Fig. 4B). We excised and analyzed the two bands by mass spectrometry and identified 94 and 101 potential binding proteins in bands one and two, respectively (Supplementary Table S5). Among these binding proteins, we selected 10 proteins that may likely participate in transcription regulation (24–32) for further binding validation. The 10 proteins included eukaryotic translation initiation factor 3 subunit I (EIF3I), Zinc finger RAN-binding domain-containing protein 2 (ZRANB2), eukaryotic translation initiation factor 3 subunit G (EIF3G), WD repeat-containing protein 5 (WDR5), methyl-CpG-binding domain protein 3 (MBD3), AP-1 complex subunit gamma-1 (AP1G1), nucleolar transcription factor 1 (UBTF), histone acetyltransferase (HAT) KAT2A (KAT2A), eukaryotic translation initiation factor 3 subunit B (EIF3B), and transcription initiation factor TFIID subunit 5 (TAF5) proteins. Western blot analysis showed that only WDR5 and KAT2A bound specifically to GClnc1 (Fig. 4C; Supplementary Fig. S4A). The data suggest that the two proteins may form a complex to interact with GClnc1. WDR5 is a core subunit of the human mixed-lineage leukemia (MLL) and SET domain containing 1A (SET1, hCOMPASS) histone H3 Lys4 (H3K4) methyltransferase complexes, and an “effector” of H3K4 methylation in gene transactivation (33, 34). KAT2A is a HAT that functions primarily as a transcriptional activator, via modulating histone H3 Lys9 (H3K9) acetylation levels in the gene promoter region (35). To identify the WDR5/KAT2A-interacting region of GClnc1, we constructed and biotinylated four fragments of GClnc1 (1–500Δ1, 501–1,000Δ2, 1,001–1,500Δ3, 1,501–2,155Δ4) and used them in the pull-down assay with BGC823 cell lysates. We found that the 5′ fragment of GClnc1 mediated the interaction with WDR5 and KAT2A (Fig. 4D). To substantiate the observation, anti-WDR5 and anti-KAT2A antibodies were used to immunoprecipitate endogenous WDR5 and KAT2A from nuclear extracts of BGC823 cells. To this end, RNAs bound to WDR5 and KAT2A were extracted and analyzed. PCR data revealed that GClnc1 directly bound with WDR5 and KAT2A in gastric cancer cells (Fig. 4E). We also detected approximately 5.5-fold and 4-fold enrichments of GClnc1, but not GAPDH RNA, in the anti-WDR5 and anti-KAT2A immunoprecipitates, respectively, compared with the IgG control (Fig. 4F and G). Thus, GClnc1 may specifically bind with WDR5 and KAT2A in gastric cancer cells. Furthermore, we observed that WDR5 and KAT2A interacted with each other in gastric cancer cells (Fig. 4H). Downregulation of GClnc1 had no effect on the expression of WDR5 and KAT2A (Supplementary Fig. S4B). However, knockdown of GClnc1 or RNAase treatment abolished the interaction between WDR5 and KAT2A (Fig. 4I). Thus, GClnc1 is important for WDR5 and KAT2A complex interaction.

Figure 4.

GClnc1 interacts with WDR5 and KAT2A epigenetic modification complex. A, experimental design for pull-down assays and identification of GClnc1-associated cellular proteins. GClnc1 and anti-GClnc1 RNA were biotinylated by in vitro transcription, refolded, and incubated with BGC823 total cell lysates. B, silver staining of biotinylated GClnc1-associated proteins. Two GClnc1-specific bands (arrows) were excised and analyzed by mass spectrometry. C, Western blot of the proteins from antisense GClnc1 and GClnc1 pull-down assays; n = 3. D, Western blot of WDR5 and KAT2A in samples pulled down by full-length (FL) or truncated GClnc1 (Δ1: 1–500, Δ2: 501–1,000, Δ3: 1,001–1,500, Δ4: 1,501–2,155); n = 3. E, RNA immunoprecipitation experiments were performed using anti-WDR5 or anti-KAT2A antibody, and specific primers were used to detect GClnc1 or GAPDH; n = 3. F and G, nuclear lysates of BGC823 were immunoprecipitated with anti-WDR5 antibodies (F), anti-KAT2A antibody (G), or control IgG. Aliquots of nuclear lysates (20% of input) and the WDR5 or KAT2A immunoprecipitates were separated by SDS-PAGE, and the specific immunoprecipitation of WDR5 and KAT2A was confirmed by Western blot (WB). The complexes were analyzed for the presence of GClnc1 or GAPDH by real-time PCR. Signals were normalized to actin mRNA. Results are mean ± SE of three independent experiments. H, coimmunoprecipitation detected the interaction of WDR5 and KAT2A in the BGC823 cells. The 20% of input (cell lysate) and WDR5 or KAT2A immunoprecipitates were separated by SDS-PAGE. The specific immunoprecipitation of WDR5 and KAT2A was confirmed by Western blot; n = 3. I, immunoprecipitation assay was performed to detect the interaction between WDR5 and KAT2A after transfection of GClnc1 siRNA (top) or RNase treatment (bottom); n = 3.

Figure 4.

GClnc1 interacts with WDR5 and KAT2A epigenetic modification complex. A, experimental design for pull-down assays and identification of GClnc1-associated cellular proteins. GClnc1 and anti-GClnc1 RNA were biotinylated by in vitro transcription, refolded, and incubated with BGC823 total cell lysates. B, silver staining of biotinylated GClnc1-associated proteins. Two GClnc1-specific bands (arrows) were excised and analyzed by mass spectrometry. C, Western blot of the proteins from antisense GClnc1 and GClnc1 pull-down assays; n = 3. D, Western blot of WDR5 and KAT2A in samples pulled down by full-length (FL) or truncated GClnc1 (Δ1: 1–500, Δ2: 501–1,000, Δ3: 1,001–1,500, Δ4: 1,501–2,155); n = 3. E, RNA immunoprecipitation experiments were performed using anti-WDR5 or anti-KAT2A antibody, and specific primers were used to detect GClnc1 or GAPDH; n = 3. F and G, nuclear lysates of BGC823 were immunoprecipitated with anti-WDR5 antibodies (F), anti-KAT2A antibody (G), or control IgG. Aliquots of nuclear lysates (20% of input) and the WDR5 or KAT2A immunoprecipitates were separated by SDS-PAGE, and the specific immunoprecipitation of WDR5 and KAT2A was confirmed by Western blot (WB). The complexes were analyzed for the presence of GClnc1 or GAPDH by real-time PCR. Signals were normalized to actin mRNA. Results are mean ± SE of three independent experiments. H, coimmunoprecipitation detected the interaction of WDR5 and KAT2A in the BGC823 cells. The 20% of input (cell lysate) and WDR5 or KAT2A immunoprecipitates were separated by SDS-PAGE. The specific immunoprecipitation of WDR5 and KAT2A was confirmed by Western blot; n = 3. I, immunoprecipitation assay was performed to detect the interaction between WDR5 and KAT2A after transfection of GClnc1 siRNA (top) or RNase treatment (bottom); n = 3.

Close modal

GClnc1 Coordinates the Localization of WDR5 and KAT2A Genome-Wide

To address whether GClnc1 modulates WDR5 and KAT2A genomic binding genome-wide, we performed chromatin immunoprecipitation coupled with high-throughput sequencing (ChIP-seq) for WDR5 and KAT2A in BGC823 cells. WDR5 and KAT2A ChIP-seq data revealed 5,549 and 6,485 called peaks, respectively, in BGC823 cells transfected with control shRNA. GClnc1 shRNA virus transfection caused reduced WDR5 and KAT2A occupancies in these two histone modulator binding DNA regions (Fig. 5A and B). Further analysis showed that WDR5 and KAT2A occupied 966 and 734 gene promoters, respectively (Fig. 5C). Nearly 20% of KAT2A-occupied genes (147 genes) were also occupied by WDR5, revealing a significant overlap (P < 0.01, hypergeometric distribution). Knockdown of GClnc1 led to a concordant decrease in WDR5 and KAT2A occupancies in 80% of the WDR5 and KAT2A co-occupied gene promoter regions (Supplementary Fig. S5A and S5B; Supplementary Table S6). The promoter regions of these genes correspondingly decreased the levels of H3K4me3 and H3K9Ac, the respective histone methylation and acetylation products of the WDR5 and KAT2A complex (Supplementary Fig. S5C). To further verify these findings, we performed ChIP for WDR5, KAT2A, H3K4me3, and H3K9Ac in BGC823 and MKN45 cells transfected with control shRNA or GClnc1 shRNA virus. We randomly selected 10% of the overlapped genes (147 genes) to experimentally validate the effect of GClnc1-mediated WDR5/KAT2A complex genomic binding. We found that knockdown of GClnc1 resulted in a decrease in the binding efficiency of WDR5 and KAT2A and the levels of H3K4me3/H3K9Ac in approximately 80% of the selected binding genes in both BGC823 and MKN45 cells (Supplementary Fig. S5D and S5E). Moreover, overexpression of GClnc1 dramatically increased the binding efficiency of WDR5 and KAT2A and the levels of H3K4me3/H3K9Ac of the same measured genes in normal GES-1 cells (Supplementary Fig. S5F).

Figure 5.

GClnc1 coordinates the localization of WDR5 and KAT2A genome-wide. A, heatmap of WDR5 global genomic binding at the target sites in BGC823 cells after transduction of control and GClnc1 shRNA adenovirus. A 2-kb interval centered on the called WDR5 peak is shown. B, heatmap of KAT2A global genomic binding at the target sites in BGC823 cells after transduction of control and GClnc1 shRNA adenovirus. A 2-kb interval centered on the called KAT2A peak is shown. To demonstrate accurately the ChIP-seq data, the same control panels (IgG) were used in A and B. The antibodies were from rabbit with the same isotype. The DNA samples for WDR5 and KAT2A ChIP experiments were prepared under identical conditions from the same cells with different experimental treatments. C, Venn diagram shows the gene promoters occupied by WDR5 (966 genes), KAT2A (734 genes), or both (147 genes). D, heatmap showing the expression change of WDR5/KAT2A co-occupied genes in BGC823 cells after transfection of GClnc1 shRNA and control shRNA. Gene expression is shown as RPKM after normalization. E, GSEA data showing the enrichment of ChIP-seq promoter peaks with significant loss of WDR5/KAT2A binding for the downregulated genes in BGC823 cells after GClnc1 shRNA virus infection, compared with control shRNA virus. NES, normalized enrichment score; FWER, family wise error rate.

Figure 5.

GClnc1 coordinates the localization of WDR5 and KAT2A genome-wide. A, heatmap of WDR5 global genomic binding at the target sites in BGC823 cells after transduction of control and GClnc1 shRNA adenovirus. A 2-kb interval centered on the called WDR5 peak is shown. B, heatmap of KAT2A global genomic binding at the target sites in BGC823 cells after transduction of control and GClnc1 shRNA adenovirus. A 2-kb interval centered on the called KAT2A peak is shown. To demonstrate accurately the ChIP-seq data, the same control panels (IgG) were used in A and B. The antibodies were from rabbit with the same isotype. The DNA samples for WDR5 and KAT2A ChIP experiments were prepared under identical conditions from the same cells with different experimental treatments. C, Venn diagram shows the gene promoters occupied by WDR5 (966 genes), KAT2A (734 genes), or both (147 genes). D, heatmap showing the expression change of WDR5/KAT2A co-occupied genes in BGC823 cells after transfection of GClnc1 shRNA and control shRNA. Gene expression is shown as RPKM after normalization. E, GSEA data showing the enrichment of ChIP-seq promoter peaks with significant loss of WDR5/KAT2A binding for the downregulated genes in BGC823 cells after GClnc1 shRNA virus infection, compared with control shRNA virus. NES, normalized enrichment score; FWER, family wise error rate.

Close modal

Next, we used the RNA-seq data from BGC823 cells transfected with control or GClnc1 shRNA virus (GSE63765) to characterize the relationship between WDR5/KAT2A complex binding and GClnc1-mediated changes in gene expression. After identifying a gene signature with highly significant changes in expression (Supplementary Table S6D), we intersected this signature with the ChIP-seq data. We observed that a substantial subset of genes which exhibit decreased binding by the WDR5/KATA2 complex were dysregulated after knockdown of GClnc1 (Fig. 5D and Supplementary Table S6E). Integrative GSEA of the RNA-seq and WDR5/KAT2A ChIP-seq data demonstrated significant enrichment for genes that were repressed when GClnc1 was downregulated (Fig. 5E). Thus, knockdown of GClnc1 impairs formation of the WDR5 and KAT2A complex. This suggests that GClnc1 may act as a scaffold for this complex, thereby controlling its ability to regulate histone modification, gene expression, and biological activity.

GClnc1 Promotes Gastric Cancer Progression via SOD2

According to the ChIP-seq and RNA-seq data, SOD2 is one of the target genes of the WDR5/KAT2A complex (Supplementary Table S6E). Furthermore, we found the strongest association between SOD2 and GClnc1 expressions (Supplementary Table S7 and Fig. 6A) among 12 genes. The positive correlation between SOD2 and GClnc1 was also validated in the 165 cases of gastric cancer in our cohort 1 (Fig. 6B; Supplementary Fig. S6A).

Figure 6.

GClnc1 promotes gastric cancer progression via SOD2. A, GClnc1 subnetwork in the gastric cancer coexpression network. This subnetwork consists of GClnc1 (center) and its 12 direct neighbors. Genes colored in light blue are protein-coding RNAs with unknown function. Genes colored in light red are protein-coding RNAs involved in tumor growth and metastasis. Node size represents the association P value between the neighbor gene and GClnc1. B, correlation between SOD2 and GClnc1 RNA level in 165 human gastric cancer tissues. C, schematic representation of exon composition of the SOD2 gene and the GClnc1 loci on human chromosome 6q25.3. D, real-time PCR and Western blot assays were performed in BGC823 cells after transfection of GClnc1, SOD2, and lncRNA ENST430239 siRNAs; n = 3, nonparametric Mann–Whitney test. E, luciferase reporter vectors were generated by inserting the promoter region (–1,000 to 0 bp) of the SOD2 gene. The reporter vectors were then cotransfected into BGC823 gastric cancer cells with either GClnc1 siRNA, lncENST430239 siRNA, or control siRNA. Cells were harvested for luciferase activity assay. Results are shown as the mean ± SE of triplicate determination from three independent experiments. Nonparametric Mann–Whitney test. F, cell proliferation assay was assessed by BrdUrd incorporation after transfection of the SOD2 siRNA, the plasmid encoding GClnc1, and control in BGC823 cells. Results are shown as the mean ± SE of triplicate determination from three independent experiments. Nonparametric Mann–Whitney test. G, Transwell Matrigel invasion assay was performed in BGC823 cells after transfection of the SOD2 siRNA, the plasmid encoding GClnc1, and control in BGC823 cells; n = 3, nonparametric Mann–Whitney test.

Figure 6.

GClnc1 promotes gastric cancer progression via SOD2. A, GClnc1 subnetwork in the gastric cancer coexpression network. This subnetwork consists of GClnc1 (center) and its 12 direct neighbors. Genes colored in light blue are protein-coding RNAs with unknown function. Genes colored in light red are protein-coding RNAs involved in tumor growth and metastasis. Node size represents the association P value between the neighbor gene and GClnc1. B, correlation between SOD2 and GClnc1 RNA level in 165 human gastric cancer tissues. C, schematic representation of exon composition of the SOD2 gene and the GClnc1 loci on human chromosome 6q25.3. D, real-time PCR and Western blot assays were performed in BGC823 cells after transfection of GClnc1, SOD2, and lncRNA ENST430239 siRNAs; n = 3, nonparametric Mann–Whitney test. E, luciferase reporter vectors were generated by inserting the promoter region (–1,000 to 0 bp) of the SOD2 gene. The reporter vectors were then cotransfected into BGC823 gastric cancer cells with either GClnc1 siRNA, lncENST430239 siRNA, or control siRNA. Cells were harvested for luciferase activity assay. Results are shown as the mean ± SE of triplicate determination from three independent experiments. Nonparametric Mann–Whitney test. F, cell proliferation assay was assessed by BrdUrd incorporation after transfection of the SOD2 siRNA, the plasmid encoding GClnc1, and control in BGC823 cells. Results are shown as the mean ± SE of triplicate determination from three independent experiments. Nonparametric Mann–Whitney test. G, Transwell Matrigel invasion assay was performed in BGC823 cells after transfection of the SOD2 siRNA, the plasmid encoding GClnc1, and control in BGC823 cells; n = 3, nonparametric Mann–Whitney test.

Close modal

GClnc1 is located in chromosome 6 (chr6:159679116-159681261) with a 2155 nt length. A 195 nt–length DNA fragment of GClnc1 at the 3′ terminal overlaps with the last exon of the two transcript variants of the SOD2 gene (NM_001024465, NM_001024466). The other part of GClnc1 is completely overlapped with the last intron of the SOD2 gene (Fig. 6C). We next designed the sense-specific PCR primers (Supplementary Fig. S6B) to determine whether GClnc1 and SOD2 were different transcripts. If GClnc1 and SOD2 were different transcripts, the PCR products of primer 1 + primer 2 should be approximately 150 bp (part of the GClnc1 fragment) and the PCR products of primer 4 + primer 2 should be approximately 200 bp (part of the SOD2 fragment). In addition, there should be no PCR product to be detected by primer 3 + primer 4. If GClnc1 and SOD2 were the same transcripts, in addition to the 150 bp and 200 bp PCR products, there should be an additional PCR product (around 3,260 bp) to be detected by primer 3 and primer 4. Our results showed that the 150 bp and 200 bp PCR products were detectable; however, the 3,260 bp PCR product was not detectable (Supplementary Fig. S6C). Furthermore, Northern blot analysis confirmed the sense-specific RT-PCR data (Supplementary Fig. S6D). Thus, GClnc1 and SOD2 are different transcripts. MiRNA is not found in the GClnc1 coding region. We next explored the mechanism by which GClnc1 regulates the expression of SOD2. Real-time PCR and Western blot data showed that SOD2 expression was significantly decreased in BGC823 cells transfected with two different GClnc1 siRNAs, but not with lncRNA ENST430239 (Fig. 6D). Similar results were observed in MKN45 cells (Supplementary Fig. S6E). Real-time PCR and Western blot data (Supplementary Fig. S6F) showed that SOD2 expression was significantly increased in GClnc1-overexpressing GES-1 cells. We next examined the effect of GClnc1 on the transcriptional activity of the SOD2 gene. Luciferase assays revealed that knockdown of GClnc1, not lncRNA ENST430239, impaired the transcriptional level of the SOD2 promoter in BGC823 cells (Fig. 6E) and in MKN45 cells (Supplementary Fig. S6G). The data suggest that GClnc1 may positively regulate SOD2 transcription in gastric cancer cells. It has been reported that high levels of SOD2 expression correlated with poor clinical outcome in gastric cancer (36), and SOD2 may regulate cancer cell proliferation and invasion (37, 38). In addition, gastric cancer cell proliferation and invasion ability were compared after knockdown of GClnc1 and SOD2. We found that downregulation of SOD2 and GClnc1 similarly impaired cell proliferation and invasion (Supplementary Fig. S6H).

We next hypothesized that SOD2 mediated the biological function of GClnc1 in gastric cancer. To test this hypothesis, we transfected SOD2 siRNA into gastric cancer cells and examined its effects on cancer cell biological function. Knockdown of SOD2 expression significantly reduced gastric cancer cell proliferation induced by GClnc1 in BGC823 cells (Fig. 6F) and MKN45 cells (Supplementary Fig. S6I). We next used the CRISPR/Cas9 system to knock out the SOD2 gene in BGC823 and MKN45 cells. We obtained similar results in these cells (Supplementary Fig. S6J). Furthermore, inhibition of SOD2 expression significantly blocked tumor growth in GClnc1-overexpressing xenograft mice (Supplementary Fig. S6K–S6M). Downregulation of SOD2 also markedly blocked the GClnc1-induced cell invasion in BGC823 cells (Fig. 6G; Supplementary Fig. S6N) and MKN45 cells (Supplementary Fig. S6O and S6P). Thus, SOD2 mediates the regulatory function of GClnc1 in gastric cancer cells.

GClnc1 Is a Molecular Link between WDR5/KAT2A and SOD2

H3K4 methylation and H3K9 acetylation may regulate the transcription of the SOD2 gene promoter (39). Real-time PCR and Western blot showed that knockdown of WDR5 or KAT2A significantly blocked GClnc1-induced SOD2 upregulation in BGC823 cells (Fig. 7A; Supplementary Fig. S7A). The data suggest that WDR5, KAT2A, and GClnc1 may cooperate to regulate SOD2 gene transactivation via histone modification. The ChIP assay indicated that WDR5 and KAT2A directly bound to the promoter region of SOD2 (Fig. 7B), and knockdown of WDR5 or KAT2A significantly decreased the binding efficiency of WDR5 or KAT2A, respectively, whereas knockdown of GClnc1 decreased the binding efficiencies of both WDR5 and KAT2A (Fig. 7C). These data indicate that GClnc1 may play an important role in the binding progression of WDR5 or KAT2A to the SOD2 gene promoter region. Furthermore, we observed that H3K4 trimethylation and H3K9 acetylation existed in the same promoter region of SOD2 (Fig. 7D). Knockdown of GClnc1 significantly decreased the H3K9 acetylation and H3K4 methylation levels in the SOD2 promoter region, but not the H3K27 trimethylation level (Fig. 7E). Overexpression of GClnc1 dramatically increased the H3K9 acetylation and H3K4 methylation levels at the promoter region of the SOD2 gene (Supplementary Fig. S7B). The chromatin isolation by RNA purification (ChIRP) assay revealed that GClnc1 directly bound to the SOD2 promoter region (Fig. 7F). The combined analyses strongly suggest that GClnc1 may act as a scaffold lncRNA to recruit WDR5 and KAT2A to the promoter of the SOD2 gene and transactivate the transcription of the SOD2 gene.

Figure 7.

GClnc1 is a molecular link among WDR5/KAT2A and SOD2. A, real-time PCR assay was performed to measure SOD2 mRNA level after transfection of WDR5/KAT2A siRNA, GClnc1-expressing plasmid, and control siRNA plus control plasmid in BGC823 cells. Results are shown as the mean ± SE of triplicate determination from three independent experiments. Nonparametric Mann–Whitney test. B, SOD2 DNA was detected in the chromatin sample immunoprecipitated from BGC823 cells using an antibody against WDR5 or KAT2A; n = 3. C, real-time PCR of the ChIP samples shows the binding efficiency of WDR5 or KAT2A to the SOD2 gene promoter after transfection of WDR5, KAT2A, or GClnc1 siRNA in BGC823 cells, respectively. Results are shown as the mean ± SE of triplicate determination from three independent experiments. D, the SOD2 DNA was detected in the chromatin sample immunoprecipitated from BGC823 cells using an antibody against H3K4me3 or H3K9Ac; n = 3. E, real-time PCR of the ChIP samples shows the H3K4 trimethylation or H3K9 acetylation and H3K27 trimethylation levels of the SOD2 gene promoter after knockdown of WDR5, KAT2A, or GClnc1 in BGC823 cells, respectively. Results are shown as the mean ± SE of triplicate determination from three independent experiments. F, ChIRP analysis of GClnc1 binding to the SOD2 promoter is shown. ChIRP was performed as previously described (49). Results are expressed as: 1, the fold enrichment of SOD2 promoter fragments in GClnc1 ChIRP analysis with GClnc1 asDNA probe; 2, the fold enrichment of GAPDH promoter fragment in LacZ ChIRP analysis with LacZ asDNA probe; 3, the fold enrichment of SOD2 promoter fragments in GClnc1 ChIRP analysis with GClnc1 sDNA probe; 4, the fold enrichment of SOD2 intron 2 fragment in GClnc1 ChIRP analysis with GClnc1 asDNA probe; n = 3. G, a schematic model of lncRNA GClnc1 functions in gastric carcinogenesis. GClnc1 may function as a scaffold lncRNA to recruit WDR5 and KAT2A and transactivate the transcription of target genes, including the SOD2 gene, and finally promote gastric cancer initiation and progression.

Figure 7.

GClnc1 is a molecular link among WDR5/KAT2A and SOD2. A, real-time PCR assay was performed to measure SOD2 mRNA level after transfection of WDR5/KAT2A siRNA, GClnc1-expressing plasmid, and control siRNA plus control plasmid in BGC823 cells. Results are shown as the mean ± SE of triplicate determination from three independent experiments. Nonparametric Mann–Whitney test. B, SOD2 DNA was detected in the chromatin sample immunoprecipitated from BGC823 cells using an antibody against WDR5 or KAT2A; n = 3. C, real-time PCR of the ChIP samples shows the binding efficiency of WDR5 or KAT2A to the SOD2 gene promoter after transfection of WDR5, KAT2A, or GClnc1 siRNA in BGC823 cells, respectively. Results are shown as the mean ± SE of triplicate determination from three independent experiments. D, the SOD2 DNA was detected in the chromatin sample immunoprecipitated from BGC823 cells using an antibody against H3K4me3 or H3K9Ac; n = 3. E, real-time PCR of the ChIP samples shows the H3K4 trimethylation or H3K9 acetylation and H3K27 trimethylation levels of the SOD2 gene promoter after knockdown of WDR5, KAT2A, or GClnc1 in BGC823 cells, respectively. Results are shown as the mean ± SE of triplicate determination from three independent experiments. F, ChIRP analysis of GClnc1 binding to the SOD2 promoter is shown. ChIRP was performed as previously described (49). Results are expressed as: 1, the fold enrichment of SOD2 promoter fragments in GClnc1 ChIRP analysis with GClnc1 asDNA probe; 2, the fold enrichment of GAPDH promoter fragment in LacZ ChIRP analysis with LacZ asDNA probe; 3, the fold enrichment of SOD2 promoter fragments in GClnc1 ChIRP analysis with GClnc1 sDNA probe; 4, the fold enrichment of SOD2 intron 2 fragment in GClnc1 ChIRP analysis with GClnc1 asDNA probe; n = 3. G, a schematic model of lncRNA GClnc1 functions in gastric carcinogenesis. GClnc1 may function as a scaffold lncRNA to recruit WDR5 and KAT2A and transactivate the transcription of target genes, including the SOD2 gene, and finally promote gastric cancer initiation and progression.

Close modal

Multiple oncogenic pathways may contribute to gastric cancer carcinogenesis (40, 41); however, the potential involvement of lncRNAs is poorly defined in human gastric cancer. Through a combination of genomic, biochemical, and cell-biological analyses, we have demonstrated that GClnc1 is a novel oncogenic lncRNA in gastric cancer. GSEA analyses have demonstrated that cell proliferation, metastasis, and chemoresistance pathways in cancer are significantly enriched in response to GClnc1 alteration in patients with gastric cancer. The bioinformatic analyses have been functionally validated in several in vitro and in vivo experimental models. Downregulation of GClnc1 markedly suppresses cell growth and metastasis and increases gastric cancer cell chemosensitivity in cultured gastric cancer cells and xenograft mouse models. Thus, the data consistently point to the notion that high GClnc1 expression is a decisive factor in controlling human gastric cancer aggressiveness.

LncRNAs may participate in chromatin remodeling complexes (42–46); however, the underlying molecular mechanisms remain unknown. The mass spectrometry and RNA pull-down data have demonstrated that GClnc1 directly interacts with WDR5, the key component of histone methyltransferase complex, and the histone acetyltransferase KAT2A. Furthermore, GClnc1 can link WDR5 and KAT2A by acting as a potential modular scaffold in gastric cancer cells. This notion is supported by three lines of experimental evidence: (i) GClnc1 directly binds with WDR5 and KAT2A via its 5′ domain; (ii) genetic deficiency of GClnc1 or RNase treatment of the immunoprecipitation products abrogates the interaction between WDR5 and KAT2A; and (iii) knockdown of GClnc1 significantly disrupts the binding of the WDR5/KAT2A complex to the promoter regions of WDR5/KAT2A co-occupied genes. In support of our observation, the lncRNAs HOTAIR and HOTTIP assist epigenetic complexes, including PRC2 and MLL, facilitate their genomic binding, and enhance their functions (20, 21, 46). Similarly, GClnc1 modulates the interaction of the WDR5 and KAT2A complex to further alter the pattern of histone modification and transactivate target genes in gastric cancer cells (Fig. 7G).

When we explored the mechanisms by which GClnc1 contributes to gastric carcinogenesis, we found the involvement of SOD2. It has been reported that SOD2 expression is increased in gastric cancers (36). The expression of SOD2 can be epigenetically silenced in human breast cancer cells, and H3K4 methylation and H3K9 acetylation may be the major histone regulatory pattern in SOD2 gene transcription (39). Consistent with these studies, we show that GClnc1 activates SOD2 transcription and that SOD2 is functionally responsible for GClnc1-mediated gastric carcinogenesis. Notably, as GClnc1 may be directly and indirectly linked to gene regulatory networks, in addition to SOD2, we do not rule out the involvement of other genes in GClnc1-associated biologic function.

Given the clinical, genetic, biochemical, and functional significance of GClnc1 in gastric cancer, we conclude that GClnc1 and its associated pathway is crucial for gastric carcinogenesis, and targeting this pathway may be pivotal in the prevention or treatment of gastric cancer.

Patient Specimens

Tumors and the adjacent gastric tissues were obtained from patients with gastric cancer who underwent surgery at Shanghai Renji Hospital from January 2000 to January 2005. The study protocol was approved by the ethics committee of Shanghai Jiao Tong University School of Medicine, Renji Hospital. Written informed consent was obtained from all participants in this study. All the research was carried out in accordance with the provisions of the Declaration of Helsinki of 1975. None of these patients had received radiotherapy or chemotherapy prior to surgery. The percentage of tumor cellularity in the tissue section was at least 70% as determined by a pathologist.

Human gastric mucosal tissues (normal tissues, tissues diagnosed with IM or DYS) were collected from patients by gastroscope inspection in Renji Hospital with written informed consent. None of the patients had taken nonsteroidal anti-inflammatory drugs, H2 receptor antagonists, proton pump inhibitors, antimicrobials, or bismuth compounds 4 weeks prior to the study. We used 20 samples per group to ensure adequate statistical power to distinguish phenotypes between the groups. The extent of inflammation in these tissues was examined according to the updated Sydney System (International Workshop on the Histopathology of Gastritis, Houston, 1994).

Bioinformatics Analysis

LncRNAs were identified from the gene expression data which can be accessed by GEO Series accession number GSE50710. The RNA-seq and ChIP-seq data can be accessed by GEO Series accession number GSE63765. The detailed Bioinformatics analyses are described in the Supplementary Materials and Methods.

Cell Culture and Treatment

The human gastric epithelial cell line GES-1 and the gastric cancer cell lines MKN45 and AGS were obtained from Dr. Jun Yu at the Chinese University of Hong Kong. The human gastric cancer cell lines BGC823, SGC7901, and MGC803 were obtained from Dr. Shuiping Tu of Rutgers, the State University of New Jersey. All cell lines were genotyped for identity by Beijing Microread Genetics Co., Ltd and tested routinely for Mycoplasma contamination. Cell lines were cultured in RPMI-1640 medium (Gibco) supplemented with 10% FBS at 37°C in an atmosphere of 5% CO2. Cell lines were transfected with different siRNA and GClnc1-expressing plasmids for genetic and functional assays. The detailed cell line treatments are described in the Supplementary Materials and Methods.

Sense-Specific RT-PCR

The sense-specific RT-PCR (Supplementary Fig. S6B) assay was developed after pilot and optimizing experiments. The detailed sense-specific RT-PCR is described in the Supplementary Materials and Methods.

5′ Rapid Amplification of cDNA Ends

5′-RACE was performed to determine the transcriptional initiation site of GClnc1 using a SMARTer RACE cDNA Amplification Kit (Clontech). The gene-specific primer used for the PCR of the RACE analysis was 5′-TATGTACACAGTGTTGCTCACC-3′ (5′RACE).

Cell Proliferation and Transwell Assay

Cell proliferation was assessed by the BrdUrd incorporation assay (Roche Molecular Biochemicals). Briefly, control and treated gastric cancer cells were seeded onto 96-well plates at an initial density of 5 × 103 cells per well. BrdUrd labeling solution (10 μL/well) was added to the cells at specified time points. After incubating for 2 hours, the reaction product was quantified according to the manufacturer's instructions.

Transwell assays were performed using 24-well transwells (8-μm pore size; Millipore) precoated with Matrigel (BD Biosciences). Cells at logarithmic phase were transfected with GClnc1 siRNA or control siRNA. Transfected cells were then harvested, and 1 × 105 cells were seeded in serum-free medium into the upper chamber, whereas medium supplemented with 20% FBS was applied to the lower chamber as a chemoattractant. After 48 hours of incubation, the migrated cells at the bottom surface of the filter were fixed, stained, and counted.

Hybridization In Situ and Immunohistochemical Staining

The in situ detection of GClnc1 was performed on 6-μm formalin-fixed, paraffin-embedded sections using DIG-labeled miRCURYTM Detection probe (Exiqon). Positive controls (RNU6B; Exiqon) and scrambled control RNAs were included for each hybridization procedure and analyzed using a Nikon 80i microscope with Nikon NIS-Elements F 2.3 software (Nikon).

The expression of Ki67 was examined with an anti-Ki67 antibody (Ki67; Abcam; 1:100) using the LSAB+ Kit (DakoCytomation). Slides were examined independently by two investigators. Protein expression was quantified using a visual grading system based on the extent of staining (the percentage of positive tumor cells on a scale of 0–4: 0, none; 1, 1%–25%; 2, 26%–50%; 3, 51%–75%; 4, >75%) and the intensity of staining (graded on a scale of 0–3: 0, no staining; 1, weak staining; 2, moderate staining; 3, strong staining). The product of the extent and intensity grades was used to define the Ki67 protein expression.

Luciferase Assay

Cell lines were transfected with the plasmids expressing the designated combinations of pGL3-SOD2PWT and other relevant siRNAs at 1.0 μg and 100 ng of phRL (Renilla luciferase) with Lipofectamine 2000 (Invitrogen) or DharmaFECT 1 siRNA transfection reagents (Dharmacon). Twenty-four hours after transfection, the cells were collected to detect luciferase activity using the Dual-Luciferase reporter assay system (Promega). Luciferase activity was measured by using a BD Monolight 3010 luminometer (BD Biosciences). Transfection efficiency was normalized by dividing the luciferase activity of the construct by the corresponding Renilla luciferase activity.

RNA Analysis, Extraction, and Quantitative Real-Time PCR

The RNA expression levels were measured using a real-time quantitative PCR system. Total RNA was extracted by TRIzol reagent (Invitrogen), and 1 μg of total RNA was reverse-transcribed using the PrimeScriptP RT Reagent Kit (Perfect Real Time; Takara). The amplified transcript level of each specific gene was normalized to that of 18S of GAPDH. The primers (Supplementary Table S8) were provided by Shenggong Company.

Western Blot

Western blot analysis was performed using standard techniques (47). An anti-GAPDH antibody was used as a control for whole-cell lysates. Antibodies were purchased mainly from three manufacturers: Cell Signaling Technology, Inc. (KAT2A, WDR5, GAPDH), Abcam (SOD2, ZRANB2, EIF3B, TAF5), and Sigma (EIF3I, EIF3G, MBD3, AP1G1, UBTF). The dilution of primary antibodies was 1:1,000.

Northern Blot

Northern blot analysis was performed with 300 to 600 ng of purified poly (A) mRNA. RNAs were resolved by denaturing agarose gel electrophoresis (Ambion) and transferred to Hybond-XL membranes (GE Healthcare). LncRNAs were detected using 32P-labeled DNA probes.

In Vitro Transcription and Translation

LncRNA GClnc1 and LacZ were cloned into pBluescript KSII downstream of the T7 promoter. The plasmid was linearized by single digestion with KpnI. Linear pCDNA3.1-GClnc1 (250 ng) was transcribed in vitro (Epicentre), and 700 ng of purified GClnc1 RNAs were translated in vitro using biotinylated leucine tRNA (Promega). Biotinylated proteins were detected using a BrightStar BioDetect Kit (Ambion). LacZ mRNA was used as a positive translation control, and mock-translated samples (no RNA template) were used as negative controls.

RNA Pull-Down Assay

Cell nuclear lysates were incubated with biotinylated RNA and streptavidin beads for RNA pull-down incubation. Then beads were collected by centrifugation. RNA-associated proteins were eluted and resolved by SDS-PAGE followed by silver staining (Bio-Rad). The detailed RNA pull-down assay is described in the Supplementary Materials and Methods.

LC-MS Analysis, Database Search, and Protein Identification

To identify specific GClnc1 interactors, GClnc1 and Antisense GClnc1 pull-down eluates were compared. The bands that were predominantly represented in the GClnc1 pull-down sample were chosen. The bands were excised to perform LC-MS analysis. The proteins sequenced with at least two peptides were considered a reliable identification. The detailed LC-MS, database, and protein identification are described in the Supplementary Materials and Methods.

RNA Immunoprecipitation

RNA immunoprecipitation (RIP) experiments were performed using the Megna RIP RNA-binding Protein Immunoprecipitation Kit (Millipore). The anti-WDR5 antibody and anti-KAT2A antibody used for RIP were purchased from CST (Cell Signaling Technology). The co-precipitated RNAs were detected by reverse transcription PCR. To demonstrate that the detected RNA signals specifically bind to WDR5 or KAT2A, total RNA (input controls) and normal rabbit IgG controls were simultaneously assayed. The gene-specific primers used for detecting GClnc1 are displayed in Supplementary Table S8.

Coimmunoprecipitation

Coimmunoprecipitation was performed as described previously (48). Briefly, both input and IP samples were analyzed by Western blot using various antibodies at the indicated dilutions: anti-WDR5 antibody (1:1,000; Cell Signal Technology), anti-KAT2A antibody (1:1,000; Cell Signal Technology), and normal rabbit IgG.

ChIRP

GClnc1 asDNA (antisense DNA), LacZ asDNA (antisense DNA), and GClnc1 sDNA (sense DNA) probes were designed using the online probe designer at singlemoleculefish.com. Oligonucleotides were biotinylated at the 3′ end with an 18-carbon spacer arm. BGC823 cells were collected and subjected to ChIRP as previously described (49).

In Vivo Experiments

In order to clarify the effect of GClnc1 on tumor growth and invasion in vivo, 4-week-old male BALB/c nude mice (Experimental Animal Centre of Shanghai Institutes for Biological Sciences) were used. Different numbers of cells were inoculated into the mice. Ten days after subcutaneous inoculation, mice were treated with a different virus. Tumor growth and invasion were monitored. All experimental procedures were approved by the Institutional Animal Care and Use Committee of Ren Ji Hospital, School of Medicine, Shanghai Jiao Tong University. The detailed in vivo experiments are described in the Supplementary Materials and Methods.

Adenovirus and Plasmids Construction

The control shRNA, GClnc1 shRNA1/2, SOD2 shRNA, GClnc1-overexpressing adenovirus, and all plasmids were constructed by Shanghai Obio Technology Company.

ChIP and High-Throughput Sequencing

Chromatin immunoprecipitation and high-throughput sequencing (ChIP-seq) was performed as follows. Chromatin was sonicated and immunoprecipitated with different antibodies. Real-time PCR was performed in triplicate. Each PCR assay was carried out in a 20 μL reaction volume using the eluted immunoprecipitated DNA. The amount of genomic DNA coprecipitated with the specific antibody was calculated as follows: CBTB = CBTB (genomic input) − CBTB (specific antibody), where CBTB (genomic input) and CBTB (specific antibody) were the mean threshold cycles of PCR performed in triplicate on DNA samples from the genomic input samples and the specific antibody samples, respectively. The forward and reverse primers are listed in Supplementary Table S9. High-throughput RNA and ChIP-seq was performed after knockdown of GClnc1 in BGC823 cells. The detailed ChIP and High-Throughput Sequencing are described in the Supplementary Materials and Methods.

Drug Sensitivity Assays to 5′-Fluorouracil and Cisplatin

Cell proliferation was assessed by the BrdUrd incorporation assay (Roche Molecular Biochemicals). Briefly, control and siRNA-transfected cells, which were seeded onto 96-well plates at an initial density of 5 × 103 cells per well, were treated with different doses of 5′-fluorouracil and cisplatin, and BrdUrd labeling solution (10 μL/well) was added to the cells at specified time points. After 2-hour incubation, culture medium was removed and the cells were fixed. Then DNA was denatured by adding FixDenat (200 μL/well), and anti–BrdUrd-POD working solution (100 μL/well) was added to the cells and incubated for 90 minutes. The immune complexes were detected by the subsequent substrate reaction. The reaction product was quantified by measuring the absorbance at 370 nm (reference wavelength: approximately 492 nm).

Tissues and LCM

Gastric cancer tissues and adjacent normal tissues were obtained from 20 patients enrolled in Renji Hospital. Tissues were embedded in Tissue-Tek OCT medium (Sakura) and snap-frozen on dry ice. The frozen sections were fixed in 70% ethanol for 30 seconds and stained with hematoxylin and eosin, followed by three dehydration steps of 5 seconds each in 70%, 95%, and 99.5% ethanol and a final 5-minute dehydration in xylene. Once air-dried, the stained tissues were laser-capture microdissected by a PixCell II LCM system (Arcturus Engineering). Normal gastric epithelial cells, gastric cancer cells, and stroma cells were selectively captured by microscopic visualization.

Generation of SOD2 Knockout BGC823 (BGC823KO) and (MKN45KO) Cell Lines

LentiCRISPR (Addgene plasmid #49535) is a lentiviral vector expressing Cas9 and single-guide RNA (sgRNA; ref. 50). Genome engineering experiments using CRISPR/Cas9 systems were performed as described previously (51). Briefly, three sgRNAs targeting SOD2 were cloned into the lentiCRISPR vector. Lentivirus was produced in HEK 293T cells. BGC823 and MKN45 cells were transduced with lentivirus and selected with 2 μg/mL puromycin. The SOD2 knockout cell clones were isolated by limited serial dilution. SOD2 deletion was confirmed by sequencing and immunoblotting analysis. The sgRNA sequences targeting SOD2 were as follows: 5′-TTCCAGGGCGCCGTAGTCGT-3′ (site 1); 5′-AAGCTGACGGCTGCATCTGT-3′ (site 2); and 5′-CTTGGTTTCAATAAGGAACG-3′ (site 3; ref. 50).

Statistical Analysis

The detailed statistical analysis is described in the Supplementary Materials and Methods.

No potential conflicts of interest were disclosed.

Conception and design: T.-T. Sun, J. Hong, W. Zou, J.-Y. Fang

Development of methodology: J. Hong

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): T.-T. Sun, J. He, Q. Liang, L.-L. Ren, T.-T. Yan, T.-C. Yu, J.-Y. Tang, Y.-J. Bao, Y. Hu, Y. Lin, D. Sun, Y.-X. Chen

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. He, Q. Liang, H. Chen

Writing, review, and/or revision of the manuscript: T.-T. Sun, J. Hong, W. Zou, J.-Y. Fang

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y.-X. Chen, W. Zou

Study supervision: H. Chen

This project was supported (in part) by grants from the National Natural Science Foundation of China (81421001, 81320108024, 31371420, 91129724, 31271366, 31371273, 81522008, 81402347, and 81530072); the Shanghai Natural Science Foundation (grant no. 13ZR14244000); the Program for Professor of Special Appointment (Eastern Scholar No. 201268 and 2015 Youth Eastern Scholar NO.QD2015003) at Shanghai Institutions of Higher Learning; Shanghai Municipal Education Commission—Gaofeng Clinical Medicine Grant Support (no. 20152512); and Chenxing Project of Shanghai Jiao-Tong University to H. Chen and J. Hong.

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.
McLean
MH
,
El-Omar
EM
. 
Genetics of gastric cancer
.
Nat Rev Gastroenterol Hepatol
2014
;
11
:
664
74
.
2.
Torre
LA
,
Bray
F
,
Siegel
RL
,
Ferlay
J
,
Lortet-Tieulent
J
,
Jemal
A
. 
Global cancer statistics, 2012
.
CA Cancer J Clin
2015
;
65
:
87
108
.
3.
Cunningham
D
,
Allum
WH
,
Stenning
SP
,
Thompson
JN
,
Van de Velde
CJ
,
Nicolson
M
, et al
Perioperative chemotherapy versus surgery alone for resectable gastroesophageal cancer
.
N Engl J Med
2006
;
355
:
11
20
.
4.
Ueda
T
,
Volinia
S
,
Okumura
H
,
Shimizu
M
,
Taccioli
C
,
Rossi
S
, et al
Relation between microRNA expression and progression and prognosis of gastric cancer: A microRNA expression analysis
.
Lancet Oncol
2010
;
11
:
136
46
.
5.
Cancer Genome Atlas Research N
. 
Comprehensive molecular characterization of gastric adenocarcinoma
.
Nature
2014
;
513
:
202
9
.
6.
Mercer
TR
,
Dinger
ME
,
Mattick
JS
. 
Long non-coding RNAs: Insights into functions
.
Nat Rev Genet
2009
;
10
:
155
9
.
7.
Ponting
CP
,
Oliver
PL
,
Reik
W
. 
Evolution and functions of long noncoding RNAs
.
Cell
2009
;
136
:
629
41
.
8.
Kogo
R
,
Shimamura
T
,
Mimori
K
,
Kawahara
K
,
Imoto
S
,
Sudo
T
, et al
Long noncoding RNA HOTAIR regulates polycomb-dependent chromatin modification and is associated with poor prognosis in colorectal cancers
.
Cancer Res
2011
;
71
:
6320
6
.
9.
Zhang
X
,
Gejman
R
,
Mahta
A
,
Zhong
Y
,
Rice
KA
,
Zhou
Y
, et al
Maternally expressed gene 3, an imprinted noncoding RNA gene, is associated with meningioma pathogenesis and progression
.
Cancer Res
2010
;
70
:
2350
8
.
10.
Lai
MC
,
Yang
Z
,
Zhou
L
,
Zhu
QQ
,
Xie
HY
,
Zhang
F
, et al
Long non-coding RNA MALAT-1 overexpression predicts tumor recurrence of hepatocellular carcinoma after liver transplantation
.
Med Oncol
2012
;
29
:
1810
6
.
11.
Yang
F
,
Bi
J
,
Xue
X
,
Zheng
L
,
Zhi
K
,
Hua
J
, et al
Up-regulated long non-coding RNA H19 contributes to proliferation of gastric cancer cells
.
FEBS J
2012
;
279
:
3159
65
.
12.
Hu
Y
,
Wang
J
,
Qian
J
,
Kong
X
,
Tang
J
,
Wang
Y
, et al
Long noncoding RNA GAPLINC regulates CD44-dependent cell invasiveness and associates with poor prognosis of gastric cancer
.
Cancer Res
2014
;
74
:
6890
902
.
13.
Endo
H
,
Shiroki
T
,
Nakagawa
T
,
Yokoyama
M
,
Tamai
K
,
Yamanami
H
, et al
Enhanced expression of long non-coding RNA HOTAIR is associated with the development of gastric cancer
.
PLoS ONE
2013
;
8
:
e77070
.
14.
Xu
TP
,
Liu
XX
,
Xia
R
,
Yin
L
,
Kong
R
,
Chen
WM
, et al
SP1-induced upregulation of the long noncoding RNA TINCR regulates cell proliferation and apoptosis by affecting KLF2 mRNA stability in gastric cancer
.
Oncogene
2015
;
34
:
5648
61
.
15.
Yang
F
,
Xue
X
,
Bi
J
,
Zheng
L
,
Zhi
K
,
Gu
Y
, et al
Long noncoding RNA CCAT1, which could be activated by c-Myc, promotes the progression of gastric carcinoma
.
J Cancer Res Clin Oncol
2013
;
139
:
437
45
.
16.
Xu
TP
,
Liu
XX
,
Xia
R
,
Yin
L
,
Kong
R
,
Chen
WM
, et al
SP1-induced upregulation of the long noncoding RNA TINCR regulates cell proliferation and apoptosis by affecting KLF2 mRNA stability in gastric cancer
.
Oncogene
2015
;
34
:
5648
61
.
17.
Sugano
K
,
Tack
J
,
Kuipers
EJ
,
Graham
DY
,
El-Omar
EM
,
Miura
S
, et al
Kyoto global consensus report on Helicobacter pylori gastritis
.
Gut
2015
;
64
:
1353
67
.
18.
Sun
L
,
Luo
H
,
Bu
D
,
Zhao
G
,
Yu
K
,
Zhang
C
, et al
Utilizing sequence intrinsic composition to classify protein-coding and long non-coding transcripts
.
Nucleic Acids Res
2013
;
41
:
e166
.
19.
Lin
MF
,
Jungreis
I
,
Kellis
M
. 
PhyloCSF: A comparative genomics method to distinguish protein coding and non-coding regions
.
Bioinformatics
2011
;
27
:
i275
82
.
20.
Gupta
RA
,
Shah
N
,
Wang
KC
,
Kim
J
,
Horlings
HM
,
Wong
DJ
, et al
Long non-coding RNA HOTAIR reprograms chromatin state to promote cancer metastasis
.
Nature
2010
;
464
:
1071
6
.
21.
Tsai
MC
,
Manor
O
,
Wan
Y
,
Mosammaparast
N
,
Wang
JK
,
Lan
F
, et al
Long noncoding RNA as modular scaffold of histone modification complexes
.
Science
2010
;
329
:
689
93
.
22.
Gong
C
,
Maquat
LE
. 
lncRNAs transactivate STAU1-mediated mRNA decay by duplexing with 3′ UTRs via Alu elements
.
Nature
2011
;
470
:
284
8
.
23.
Sun
S
,
Del Rosario
BC
,
Szanto
A
,
Ogawa
Y
,
Jeon
Y
,
Lee
JT
. 
Jpx RNA activates Xist by evicting CTCF
.
Cell
2013
;
153
:
1537
51
.
24.
Spriggs
KA
,
Bushell
M
,
Willis
AE
. 
Translational regulation of gene expression during conditions of cell stress
.
Mol Cell
2010
;
40
:
228
37
.
25.
Loughlin
FE
,
Mansfield
RE
,
Vaz
PM
,
McGrath
AP
,
Setiyaputra
S
,
Gamsjaeger
R
, et al
The zinc fingers of the SR-like protein ZRANB2 are single-stranded RNA-binding domains that recognize 5′ splice site-like sequences
.
Proc Natl Acad Sci U S A
2009
;
106
:
5581
6
.
26.
Wysocka
J
,
Swigut
T
,
Milne
TA
,
Dou
Y
,
Zhang
X
,
Burlingame
AL
, et al
WDR5 associates with histone H3 methylated at K4 and is essential for H3 K4 methylation and vertebrate development
.
Cell
2005
;
121
:
859
72
.
27.
Narlikar
GJ
,
Fan
HY
,
Kingston
RE
. 
Cooperation between complexes that regulate chromatin structure and transcription
.
Cell
2002
;
108
:
475
87
.
28.
Brown
SE
,
Szyf
M
. 
Epigenetic programming of the rRNA promoter by MBD3
.
Mol Cell Biol
2007
;
27
:
4938
52
.
29.
Gao
H
,
Dahlman-Wright
K
. 
The gene regulatory networks controlled by estrogens
.
Mol Cell Endocrinol
2011
;
334
:
83
90
.
30.
Matera
AG
,
Wu
W
,
Imai
H
,
O'Keefe
CL
,
Chan
EK
. 
Molecular cloning of the RNA polymerase I transcription factor hUBF/NOR-90 (UBTF) gene and localization to 17q21.3 by fluorescence in situ hybridization and radiation hybrid mapping
.
Genomics
1997
;
41
:
135
8
.
31.
van Riggelen
J
,
Yetil
A
,
Felsher
DW
. 
MYC as a regulator of ribosome biogenesis and protein synthesis
.
Nat Rev Cancer
2010
;
10
:
301
9
.
32.
Huisinga
KL
,
Pugh
BF
. 
A genome-wide housekeeping role for TFIID and a highly regulated stress-related role for SAGA in Saccharomyces cerevisiae
.
Mol Cell
2004
;
13
:
573
85
.
33.
Wysocka
J
,
Swigut
T
,
Xiao
H
,
Milne
TA
,
Kwon
SY
,
Landry
J
, et al
A PHD finger of NURF couples histone H3 lysine 4 trimethylation with chromatin remodelling
.
Nature
2006
;
442
:
86
90
.
34.
Ruthenburg
AJ
,
Wang
W
,
Graybosch
DM
,
Li
H
,
Allis
CD
,
Patel
DJ
, et al
Histone H3 recognition and presentation by the WDR5 module of the MLL1 complex
.
Nat Struct Mol Biol
2006
;
13
:
704
12
.
35.
Tanner
KG
,
Trievel
RC
,
Kuo
MH
,
Howard
RM
,
Berger
SL
,
Allis
CD
, et al
Catalytic mechanism and function of invariant glutamic acid 173 from the histone acetyltransferase GCN5 transcriptional coactivator
.
J Biol Chem
1999
;
274
:
18157
60
.
36.
Janssen
AM
,
Bosman
CB
,
van Duijn
W
,
Oostendorp-van de Ruit
MM
,
Kubben
FJ
,
Griffioen
G
, et al
Superoxide dismutases in gastric and esophageal cancer and the prognostic impact in gastric cancer
.
Clin Cancer Res
2000
;
6
:
3183
92
.
37.
Nelson
KK
,
Ranganathan
AC
,
Mansouri
J
,
Rodriguez
AM
,
Providence
KM
,
Rutter
JL
, et al
Elevated sod2 activity augments matrix metalloproteinase expression: evidence for the involvement of endogenous hydrogen peroxide in regulating metastasis
.
Clin Cancer Res
2003
;
9
:
424
32
.
38.
Chen
PM
,
Wu
TC
,
Wang
YC
,
Cheng
YW
,
Sheu
GT
,
Chen
CY
, et al
Activation of NF-kappaB by SOD2 promotes the aggressiveness of lung adenocarcinoma by modulating NKX2–1-mediated IKKbeta expression
.
Carcinogenesis
2013
;
34
:
2655
63
.
39.
Hitchler
MJ
,
Oberley
LW
,
Domann
FE
. 
Epigenetic silencing of SOD2 by histone modifications in human breast cancer cells
.
Free Radic Biol Med
2008
;
45
:
1573
80
.
40.
Ooi
CH
,
Ivanova
T
,
Wu
J
,
Lee
M
,
Tan
IB
,
Tao
J
, et al
Oncogenic pathway combinations predict clinical prognosis in gastric cancer
.
PLoS Genet
2009
;
5
:
e1000676
.
41.
Deng
N
,
Goh
LK
,
Wang
H
,
Das
K
,
Tao
J
,
Tan
IB
, et al
A comprehensive survey of genomic alterations in gastric cancer reveals systematic patterns of molecular exclusivity and co-occurrence among distinct therapeutic targets
.
Gut
2012
;
61
:
673
84
.
42.
Khalil
AM
,
Guttman
M
,
Huarte
M
,
Garber
M
,
Raj
A
,
Rivea Morales
D
, et al
Many human large intergenic noncoding RNAs associate with chromatin-modifying complexes and affect gene expression
.
Proc Natl Acad Sci U S A
2009
;
106
:
11667
72
.
43.
Pandey
RR
,
Mondal
T
,
Mohammad
F
,
Enroth
S
,
Redrup
L
,
Komorowski
J
, et al
Kcnq1ot1 antisense noncoding RNA mediates lineage-specific transcriptional silencing through chromatin-level regulation
.
Mol Cell
2008
;
32
:
232
46
.
44.
Nagano
T
,
Mitchell
JA
,
Sanz
LA
,
Pauler
FM
,
Ferguson-Smith
AC
,
Feil
R
, et al
The Air noncoding RNA epigenetically silences transcription by targeting G9a to chromatin
.
Science
2008
;
322
:
1717
20
.
45.
Zhao
J
,
Sun
BK
,
Erwin
JA
,
Song
JJ
,
Lee
JT
. 
Polycomb proteins targeted by a short repeat RNA to the mouse X chromosome
.
Science
2008
;
322
:
750
6
.
46.
Rinn
JL
,
Kertesz
M
,
Wang
JK
,
Squazzo
SL
,
Xu
X
,
Brugmann
SA
, et al
Functional demarcation of active and silent chromatin domains in human HOX loci by noncoding RNAs
.
Cell
2007
;
129
:
1311
23
.
47.
Xiong
H
,
Hong
J
,
Du
W
,
Lin
YW
,
Ren
LL
,
Wang
YC
, et al
Roles of STAT3 and ZEB1 proteins in E-cadherin down-regulation and human colorectal cancer epithelial-mesenchymal transition
.
J Biol Chem
2012
;
287
:
5819
32
.
48.
Uyama
N
,
Zhao
L
,
Van Rossen
E
,
Hirako
Y
,
Reynaert
H
,
Adams
DH
, et al
Hepatic stellate cells express synemin, a protein bridging intermediate filaments to focal adhesions
.
Gut
2006
;
55
:
1276
89
.
49.
Chu
C
,
Quinn
J
,
Chang
HY
. 
Chromatin isolation by RNA purification (ChIRP)
.
J Vis Exp
2012
;
61
:
3912
.
50.
Shalem
O
,
Sanjana
NE
,
Hartenian
E
,
Shi
X
,
Scott
DA
,
Mikkelsen
TS
, et al
Genome-scale CRISPR-Cas9 knockout screening in human cells
.
Science
2014
;
343
:
84
7
.
51.
Ran
FA
,
Hsu
PD
,
Wright
J
,
Agarwala
V
,
Scott
DA
,
Zhang
F
. 
Genome engineering using the CRISPR-Cas9 system
.
Nat Protoc
2013
;
8
:
2281
308
.

Supplementary data