Comprehensive analysis of aberrantly expressed lncRNAs and construction of ceRNA network in gastric cancer

Abstract

Gastric cancer remains fifth most common cancer often diagnosed at an advanced stage and is the second leading cause of cancer-related death worldwide. Long non-coding RNAs (lncRNAs) involved in various cellular pathways are essential for tumor occurrence and progression and they have high potential to promote or suppress the expression of many genes. In this study, we profiled 19 selected cancer-associated lncRNAs in thirty gastric adenocarcinomas and matching normal tissues by qRT-PCR. Our results showed that most of the lncRNAs were significantly upregulated (12/19). Further, we performed bioinformatic screening of miRNAs that share common miRNA response elements (MREs) with lncRNAs and their downstream mRNA targets. The prediction identified three microRNAs (miR-21, miR-145 and miR-148a) and five gastric cancer-specific target genes (EGFR, KLF4, DNMT1 and AGO4) which also showed strong correlation with lncRNAs in regression analysis. Finally, we constructed an integrated lncRNA-miRNA-mRNA interaction network of the candidate genes to understand the post-transcriptional gene regulation. The ceRNA network analysis revealed that the differentially regulated miR-21 and miR-148a were playing as central candidates coordinating sponging activity of the lncRNAs analyzed (H19, TUG1 and MALAT1) in this study and the overexpression of H19 and miR-21 could be a signature event of gastric tumorigenesis that could serve as prognostic indicators and therapeutic targets.

Keywords: ceRNA; competing endogenous RNA; gastric cancer; lncRNA; miRNA.

Figure 1. Oncoprint model of lncRNA expression in gastric tumor samples

The top two panel shows the clinicopathological features of the 30 gastric cancer samples used in the study. The bottom heat map shows the expression levels of lncRNAs in each sample. The overall expression levels of the lncRNAs are presented in the right side of the expression heat map as log2 values.

Figure 2. Expression level of lncRNAs in gastric tumors compared to adjacent normal tissues

(A) Expression of lncRNAs that are classified as very highly expressed lncRNAs (>3 folds). (B) Expression of lncRNAs classified as highly expressed and (C) lncRNAs those are downregulated in tumors than normal tissues. (D) Comparative expression of lncRNAs from present study samples and TCGA gastric cancer datasets presented as log10 values.

Figure 3. Expression level of gastric cancer-associated genes

(A) Expression level of gastric cancer associated genes DNMT1, AGO4, KLF4, EGFR and FGFR2 between tumor and normal tissues. (B) Comparative expression of gastric cancer associated genes from present study samples and TCGA gastric cancer datasets.

Figure 4. Correlation matrix of the genes screened for the expression

Correlation matrix of differentially expressed lncRNAs (A), mRNA (B) and miRNA (C) in gastric cancer tissues based on spearman's linear correlation. (D) Correlation matrix showing the extent of correlation between the expression of lncRNAs and mRNAs in gastric tumors based on spearman's linear correlation.

Figure 5. Linear regression analysis of lncRNA and mRNA

Curve estimation linear regression analysis, coefficient analysis (slope) showing that BC032469 was positively associated with KLF4, EGFR and FGFR2 ((A–C) respectively) for the 30 gastric cancers included in the study. Similarly, linear regression analysis of KLF4 with MALAT1 (D) and EGFR with MALAT1 (E), PANDA (F) and PTENP1-AS (G), respectively showed a positive association.

Figure 6. LncRNAs-miRNAs-mRNAs co-expression network for identification of novel ceRNA network

A prediction of the ceRNA hypothesis in gastric cancer by integrated analysis of the mRNAs and lncRNAs containing the same MREs which compete for a common pool of miRNAs using online prediction tools and validated reports. In the ceRNA network, the lncRNAs (blue node) can sponge the miRNAs (pink node) that target the gastric cancer driver or associated genes (orange node) and enables the genes to skip miRNA-mediated post-translation gene regulation. The miRNAs those not sponged by lncRNAs that were targeting individual genes were mentioned in the green node and miRNAs targeting two or more genes were mentioned in the yellow node.