Zinc Finger Protein CTCF Regulates Extracellular Matrix (ECM)-Related Gene Expression Associated With the Wnt Signaling Pathway in Gastric Cancer

Abstract

Gastric cancer (GC), a leading cause of cancer-related death, is a heterogeneous disease. We aim to describe clinically relevant molecular classifications of GC that incorporate heterogeneity and provide useful clinical information. We combined different gene expression datasets and filtered a 7-gene signature related to the extracellular matrix (ECM), which also exhibited significant prognostic value in GC patients. Interestingly, putative CCCTC-binding factor (CTCF) regulatory elements were identified within the promoters of these ECM-related genes and were confirmed by chromatin immunoprecipitation sequencing (ChIP-Seq). CTCF binding sites also overlapped with histone activation markers, indicating direct regulation. In addition, CTCF was also correlated with the Wnt signaling pathway. A comparison of human GC cell lines with high or low expression of ECM-related genes revealed different levels of tumor aggressiveness, suggesting the cancer development-promoting functions of ECM-related genes. Furthermore, CTCF regulated COL1A1 and COLA31 expression in vitro. Silencing CTCF or COL1A1/COL1A3 markedly inhibited cell growth and migration in the metastatic GC cell line BGC823. Collectively, this ECM-related 7-gene signature provides a novel insight for survival prediction among GC patients. The zinc finger protein CTCF regulates ECM-related genes, thereby promoting GC cell growth and migration.

Keywords: Wnt signaling; extracellular matrix (ECM); gastric cancer; histone modification; zinc finger protein CTCF.

Figure 1

Flow chart of the systematic comprehensive analysis of differentially expressed genes (DEGs) from The Cancer Genome Atlas (TCGA) cohort and the Gene Expression Omnibus (GEO) series of gastric cancer (GC) patients and correlation of DEG expression with overall survival (OS). Further investigation revealed that the extracellular matrix (ECM) signature regulates gastric cancer (GC) progression through the Wnt pathway, and this result was validated by mRNA expression, cell viability and migration assays using a siRNA strategy in aggressive GC cell lines.

Figure 2

Identification of the extracellular matrix (ECM) subtype. (A) Venn diagrams of the differentially expressed genes (DEGs) between the GSE79973 dataset and the The Cancer Genome Atlas (TCGA) gastric cancer (GC) dataset. Ninety-five DEGs were upregulated and 116 DEGs were downregulated in the two datasets. (B) Venn diagrams of the overlapping pathways. A total of two enriched pathways overlapped the two datasets: ECM-receptor interaction and protein digestion and absorption. (C) Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of the DEGs in the GSE79973 dataset. (D) KEGG pathway enrichment analysis of the DEGs in TCGA GC dataset. (E) KEGG pathway enrichment analysis of overlapping DEGs between the GSE79973 dataset and the TCGA dataset. (F) KEGG pathway enrichment analysis of the genes in the protein-protein interaction (PPI) module. The overlapping pathways are marked in red.

Figure 3

Clinical pathological indicators and prognostic characteristics of elevated extracellular matrix (ECM) subtype genes: Analysis of the correlations between ECM subtype gene expression and survival and clinical characteristics. (A) Survival analysis for the ECM subtype genes COMP, SPP1, THBS2, COL1A1, COL3A1, COL10A1, and COL11A1. (B) Correlations between the seven hub genes and clinical characteristics. The numbers in each small rectangle indicate the P-value for the correlation. (C) The relationship between the hub genes and clinical stage and T stage. *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 4

Exploration of the biological relevance of expression differences in the extracellular matrix (ECM) subtype genes: (A–D) Representative immunohistochemistry (IHC) images of ECM subtype genes in gastric cancer (GC) tissues and normal tissues [Human Protein Atlas (HPA) database]. The quantification results suggested that the protein levels of COL1A1, COL3A1, SPP1, and THBS2 were significantly increased in GC tumor tissues compared to normal gastric tissues. (E) Expression levels of the ECM subtype genes were determined in seven GC cell lines (aggressive cell lines: BGC823, HGC27, LMSU, and NCCStCK140; and less aggressive cell lines: FU97, MKN45, and AGS).

Figure 5

CCCTC-binding factor (CTCF) transcription factor binding motifs are present in a subset of extracellular matrix (ECM) candidate genes. (A) Computational predictions of binding sites of CTCF and other transcription factors are shown upstream of the ECM-related gene transcription factor binding sites (TFBSs); COL1A1, COL3A1, and COL11A1 genes have a common transcription factor: CTCF. (B) ChIP-Seq data of two common active markers, H3K27Ac and H3K4me3, in the promoter region of ECM subtype genes that overlapped with the CTCF binding sites.

Figure 6

CCCTC-binding factor (CTCF) expression correlated with the Wnt signaling pathway: (A) The CTCF transcript was highly expressed in GC tissues compared with normal tissues. (B) The top six enriched pathways in GC samples with high CTCF expression were identified by gene set enrichment analysis (GSEA). (C) The prognostic value of the Wnt signaling pathway in gastric cancer (GC). Kaplan-Meier curves demonstrate the correlation between Wnt signaling protein expression and overall survival. (D) Gene Set Enrichment Analysis (GSEA)-generated heat map for highly enriched genes in samples with high CTCF expression vs. those with low CTCF expression. (E) Pearson’s correlation analysis of CTCF and Wnt pathway-related genes. We used the service provided by the website Gene Expression Profiling Interactive Analysis (GEPIA) (http://gepia.cancer-pku.cn/index.html) to examine the correlations among GSK3B, LRP6, NLK, APC, SMAD4, DVL3, TCF7L2, AXIN1, and CTCF.

Figure 7

CTCF, COL1A1, and COL3A1 promote gastric cancer (GC) cell viability and migration in vitro. (A) Comparison of CTCF, COL1A1, and COL3A1 mRNA expression between BGC823 and HCG27 by RT-qPCR. (B) Real-time qRT-PCR was performed to detect the expression of the indicted genes to verify the effectiveness of knockdown. (C) BGC823 cells were treated with siRNA for 24, 48, and 72 h, and the MTT index was measured based on the MTT assay. The result is presented as fold change of metabolic viability/cell and normalized with respect to control. (D) Wound−healing assays were used to determine the effects of single siRNAs targeting CTCF, COL1A1, or COL3A1 on BGC823 cell migration. The bar chart represents the wound width (%) at 24 or 48 h divided by the width at 0 h. Data are presented as the mean ± SD of three independent experiments. **P < 0.001.