Colon cancer is one of the most common cancers in the world. In developing countries, approximately one-quarter of the patients suffering from colon cancer are in an advanced stage at presentation and have lost the opportunity for radical surgery [
Noncoding RNAs (NcRNAs) have appealed to researchers due to the modulating effect on the biological behaviors of tumor cells [
The competing endogenous RNA (ceRNA) hypothesis was first proposed by Salmena and colleagues, who suggested that protein-coding RNAs and NcRNAs act as ceRNAs by competing for miRNAs through shared miRNA response elements to mutually regulate their expression [
Here, we used an integrative computational method to identify lncRNA-mRNA-related crosstalk networks mediated by miRNAs based on the “ceRNA hypothesis” using data from The Cancer Genome Atlas (TCGA). Candidate prognostic lncRNA biomarkers in colon cancer were identified. The expression of candidate lncRNAs and target genes was also confirmed in clinical colon cancer tissues and colon cancer cell lines.
Data from patients with colon cancer were downloaded from TCGA data portal website (available at
The human miRNA and targeted gene data were collected from miRTarBase (version 6.1) [
The differentially expressed mRNAs, lncRNAs, and miRNAs in normal colon tissues and colon cancer tissues in the TCGA were analyzed using edgeR software packages. The false discovery rate (FDR) was controlled at the 0.01 threshold (Benjamini and Hochberg algorithm). The fold-change (FC) threshold was set at more than 2.0.
miRTarbase is an information resource for experimentally validated miRNA-target interactions that provides the expression profile of a miRNA and its target gene. For the differentially expressed genes in TCGA, high-quality experimentally validated miRNA-target gene interaction relationships from published low-throughput experiments were selected. A total of 1523 pairs of miRNA-gene interactions containing 345 miRNAs and 531 genes were selected.
LncBase provides experimentally supported and in silico predicted interactions of miRNA and lncRNAs. For the differentially expressed lncRNAs, target miRNA-lncRNA in LncBase with a
The Pearson correlation coefficient (PCC) for miRNA-mRNA and miRNA-lncRNA was calculated using paired miRNA, mRNA, and lncRNA expression profile data. A candidate pair of lncRNA-miRNA-mRNA was constructed based on the “ceRNA hypothesis” as follows: (i) mRNAs and lncRNAs share the same miRNAs; (ii) mRNAs and lncRNAs suggest a positive correlation when the PCC is higher than 0.3 and
All of the functional pairs were integrated to form a miRNA-mediated, lncRNA-associated ceRNA network. The degree values represent the number of genes with which the lncRNA can interact. The higher the degree is, the more centrally the lncRNA occurs within the network.
The Gene Ontology (GO) analysis was used to determine the potential roles of aberrantly expressed lncRNAs in colon cancer. GO enrichment was performed using BiNGO: a Cytoscape plugin (
The Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis (
A Kaplan-Meier survival analysis was performed for patients with different lncRNA expressions. Statistical significance was assessed using the log-rank test. Patients were assigned to high-/low-expression groups according to the median of the expression level of lncRNAs. A flowchart of the ceRNA network construction and analysis is shown in Figure
Tissue samples of colon cancer were obtained from surgical specimens at Nanfang Hospital. The specimens were snap-frozen in liquid nitrogen after excision. Thirty-five samples of colon cancer tissues and paired adjacent normal colon tissues were used for the lncRNA microarray analysis. The experimental protocols were approved by the Ethics Committee of Nanfang Hospital.
Human colon cancer cell lines HT-29 and HCT116 were obtained from Shanghai Advanced Research Institute, Chinese Academy of Sciences. The cells were cultured using conventional methods. For transient transfection of siRNA, the cells were seeded in 6-well plates and transfected with siRNA using Lipofectamine 2000 (Invitrogen, USA) according to the manufacturer’s instructions. The target sequences for siRNAs are shown in Table
Total RNA was extracted from the frozen tissues of patients or treated cells using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) following the manufacturer’s protocol. Purified total RNA was reverse transcribed using the PrimeScript RT reagent kit (Takara, Japan). Quantitative real-time PCR (qPCR) of the top 5 lncRNAs was performed using the Power SYBR Green PCR Master Mix (lnRcute lncRNA qPCR detection kit) according to the manufacturer’s guidelines. The specific lncRNA primers are listed in Table
Total RNA from cultured cell lines was isolated using the TRIzol reagent (Invitrogen). Reverse transcriptase reactions were performed using Taqman Reverse Transcription Reagents (P/N: N808-0234, Applied Biosystems). Each reaction sample contains total RNA, 50 nM stem-loop RT primer, 1x RT buffer, 0.25 mM dNTPs, 3.33 U/
A total of 439 tumorous tissues and 42 nontumorous adjacent tissues were included. A total of 16266 genes, 2758 lncRNAs, and 298 miRNAs were included in our study. In total, 115 miRNAs, 4188 mRNAs, and 663 lncRNAs were identified as differentially expressed between tumor tissues and normal tissues. Of the significantly differentially expressed lncRNAs, 441 were upregulated, and 222 were downregulated. Of the aberrantly expressed mRNAs, 2174 were upregulated, and 2014 were downregulated. In the significantly differentially expressed miRNAs, 71 were upregulated, and 44 were downregulated.
An experimental interaction network among miRNAs, mRNAs, and lncRNAs was constructed. Positively and negatively correlated mRNA/miRNA pairs are both interesting from a purely statistical framework [
A lncRNA-miRNA-mRNA ceRNA network containing 181 ceRNA crosstalk candidates was identified comprising 97 mRNAs, 31 lncRNAs, and 34 miRNAs (Doc
Colon cancer-specific lncRNA-associated ceRNA network. Graphical view of the mRNA-lncRNA network in colon cancer. The size of the nodes represents the power of the interrelation among the nodes. In the network, genes are colored in green, and lncRNAs are colored in red. The more edges a lncRNA has, the more genes that connect to it and the more central a role the lncRNA plays within the network.
Characteristics of the ceRNA network. (a) Degree distribution of the nodes in the colon cancer-associated ceRNA network. The degree of a node is the number of edges connecting to other nodes. (b) Degree of the lncRNAs in the colon cancer-associated ceRNA network.
The competing genes of lncRNAs in the ceRNA network.
lncRNA | Degree | Gene 1 | Gene 2 | Gene 3 | Gene 4 | Gene 5 | Gene 6 | Gene 7 | Gene 8 | Gene 9 | Gene 10 | Gene 11 | Gene 12 | Gene 13 | Gene 14 | Gene 15 | Gene 16 | Gene 17 | Gene 18 |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
RP11-284N8.3 | 18 | PIK3CG | SEMA6A | IGF1 | THRB | DPYD | IL6R | LPAR1 | CXCL12 | FAS | MT2A | PPARGC1A | CROT | CPEB3 | FGF9 | ESR1 | BCL2L11 | KAT2B | MXD1 |
RP11-399O19.9 | 17 | FAS | CYP19A1 | IL10 | CASP7 | BMP2 | CDKN1A | BCL2L11 | TNFSF12 | NPAS3 | HSPB2 | TP53INP1 | BCL2 | EPAS1 | KAT2B | PHLPP2 | PRKG1 | KIT | |
LINC00641 | 13 | LPAR1 | CXCL12 | ATAT1 | FAS | PPARGC1A | IL6R | SOX5 | FGF7 | BCL2 | RECK | AKT3 | BRCA1 | FGFR1 | |||||
MAGI2-AS3 | 12 | CXCL12 | LPAR1 | IL6R | FAS | MT2A | ESR1 | CSF1 | MEOX2 | ZEB1 | PDGFRA | TP53INP1 | BCL2 | ||||||
RP11-2C24.4 | 10 | CCNE1 | AXIN2 | VEGFA | CCND1 | CHEK1 | BIRC5 | CDK4 | CDC25A | BRCA1 | HMGA1 | ||||||||
STAG3L5P-PVRIG2P-PILRB | 10 | BRCA1 | HMGA1 | VEGFA | AXIN2 | CCNE1 | CCND1 | CHEK1 | BIRC5 | CDC25A | SALL4 | ||||||||
PWAR6 | 9 | CPEB3 | ESR1 | MEIS1 | ZEB2 | ZEB1 | MITF | CPEB1 | NPAS3 | TP53INP1 | |||||||||
RP11-305E6.4 | 9 | EZH2 | RCBTB1 | CHEK1 | CKS2 | DNMT3B | HMGA1 | MYC | CCNE1 | CDC6 | |||||||||
RP11-458F8.4 | 8 | CDC25A | CCND1 | HMGA1 | TNFRSF10B | E2F1 | EZH2 | MYC | AURKB | ||||||||||
RP13-942N8.1 | 8 | UBE2C | RBL1 | MYC | LIMK1 | PTPRO | CCND1 | VEGFA | E2F1 | ||||||||||
GAS5 | 7 | EZH2 | CKS2 | HMGA1 | RCBTB1 | CHEK1 | IARS | TRIB3 | |||||||||||
TRG-AS1 | 7 | HMOX1 | NRP2 | DYRK2 | CCR7 | RAVER2 | VDR | CDKN1A | |||||||||||
ATP2B1-AS1 | 6 | ZEB2 | PTGER4 | MEIS1 | ZEB1 | CPEB1 | MITF | ||||||||||||
LINC-PINT | 6 | E2F5 | PROX1 | RCBTB1 | CHEK1 | EZH2 | ESR1 | ||||||||||||
AC010226.4 | 5 | PIK3CG | THRB | SEMA6A | IGF1 | DPYD | |||||||||||||
RP11-545I5.3 | 5 | CXCL12 | LPAR1 | PPARGC1A | FAS | IL6R | |||||||||||||
RP11-834C11.4 | 5 | COL1A2 | DYRK2 | ITGB3 | CCR7 | VDR | |||||||||||||
LINC01197 | 4 | NOTCH3 | SLC8A1 | MYOCD | HAND2 | ||||||||||||||
RP11-120E11.2 | 3 | CPEB3 | FGF9 | ESR1 | |||||||||||||||
KCNQ1OT1 | 2 | ESR1 | CYP19A1 | ||||||||||||||||
MALAT1 | 2 | ESR1 | NPAS3 | ||||||||||||||||
RP11-588 K22.2 | 2 | SNCG | CAV1 | ||||||||||||||||
RP11-6O2.3 | 2 | JPH2 | MXI1 | ||||||||||||||||
RP11-798M19.6 | 2 | SNCG | CAV1 | ||||||||||||||||
RP1-239B22.5 | 2 | CDC25A | CDC25C | ||||||||||||||||
UCA1 | 2 | HSPD1 | TKT | ||||||||||||||||
ALMS1-IT1 | 1 | E2F5 | |||||||||||||||||
LINC00294 | 1 | CDKN1A | |||||||||||||||||
MIR17HG | 1 | CCNA2 | |||||||||||||||||
PVT1 | 1 | GPX2 | |||||||||||||||||
RP11-29G8.3 | 1 | KIF20B |
Notably, lncRNA TRG-AS1 and RP11-399O19.9 in the network were significantly downregulated in the tumor tissues compared with normal colon tissues, and expression of the two lncRNAs was markedly lower in late-stage patients (stages III and IV) than in early-stage patients (stages I and II). Some other lncRNAs (UCA1, RP11-458F8.4, RP1-239B22.5, and ALMS1-IT1) that were upregulated in cancers showed a higher expression in the late stage than in the early stage (Table
Differential expression analysis of lncRNAs in tumor vs. normal tissues and in early-stage vs. late-stage tumors in the network.
lncRNA | Tumor vs |
Stages III and IV vs | |||||
---|---|---|---|---|---|---|---|
lncRNA_ID | lncRNA | logFC | FDR | logFC | FDR | ||
ENSG00000281103 | TRG-AS1 | -1.423908012 | -0.493426442 | 0.001872111 | |||
ENSG00000261438 | RP11-399O19.9 | -1.076885782 | -0.422363333 | 0.002745267 | |||
ENSG00000214049 | UCA1 | 3.895863041 | 0.698484894 | 0.000322105 | 0.009535683 | ||
ENSG00000273142 | RP11-458F8.4 | 1.199175473 | 0.218373189 | 0.010661844 | 0.090116662 | ||
ENSG00000260196 | RP1-239B22.5 | 1.162274768 | 0.187325891 | 0.02032783 | 0.132090353 | ||
ENSG00000230002 | ALMS1-IT1 | 1.425424943 | 0.257786466 | 0.020982907 | 0.135386216 |
To verify the reliability of the aberrant lncRNAs found in the TCGA database, a qPCR assay was used to detect the expression levels of the top five lncRNAs with the highest degree in 35 paired tumor tissues and adjacent normal tissues from patients with colon cancer. The results indicate that the RP11-2C24.4 levels increased while the RP11-284N8.3, RP11-399O19.9, LINC00641, and MAGI2-AS3 levels decreased in the tumor tissues (Figure
Validation by qPCR of the top five candidate lncRNAs and the correlation between target genes. (a) The relative expression levels of the top 5 lncRNAs with the highest degree in the ceRNA network were detected by qPCR in the tumor tissues and paired adjacent normal colon tissues from 35 patients with colon cancer. The data are presented as the relative expression level in tumor tissues compared with normal control tissues.
The expression of the involved miRNAs, which were predicted to be “sponged” by the top five lncRNAs in the cell lines, was verified. We showed that all of the miRNAs are expressed in HT-29 and HCT116 cells (Figure
To characterize the function of our ceRNA network, a GO enrichment analysis and KEGG pathway analysis were performed (Doc
Functional characterization of colon cancer-specific ceRNA network. (a) GO enrichment analysis of the ceRNA network. The top twenty significantly enriched GO terms of biological processes are shown. (b) KEGG pathway analysis of the ceRNA network. The top twenty pathways are shown. The adjusted
To preliminarily explore the potential functions of the dysregulated lncRNAs in colon cancer, a functional enrichment analysis for the target genes of the top 5 lncRNAs with the highest degree was performed based on GO terms. We showed that the mRNA targets for lncRNA RP11-399O19.9 included a variety of tumor-associated genes, such as the apoptosis-related genes FAS, BCL2, and BCL2L11 (Figure
Analysis for lncRNAs in the ceRNA network. (a) Network map for lncRNA RP11-399O19.9-related genes and miRNAs in the ceRNA network. Genes are colored in blue, and miRNAs are colored in purple. (b) KEGG analysis for the top five lncRNAs in the ceRNA network. (c) Significantly enriched KEGG pathway for lncRNA RP11-399O19.9. The top 10 enriched pathways are listed.
GO function for lncRNA RP11-284N8.3.
GO ID | GO function | FDR | |
---|---|---|---|
60526 | Prostate glandular acinus morphogenesis | ||
60527 | Prostate epithelial cord arborization involved in prostate glandular acinus morphogenesis | ||
23033 | Signaling pathway | ||
48608 | Reproductive structure development | ||
8584 | Male gonad development | ||
60736 | Prostate gland growth | ||
35468 | Positive regulation of signaling pathway | ||
60442 | Branching involved in prostate gland morphogenesis | ||
60525 | Prostate glandular acinus development | ||
10647 | Positive regulation of cell communication | 7 | |
46546 | Development of primary male sexual characteristics | ||
10740 | Positive regulation of intracellular protein kinase cascade | ||
46661 | Male sex differentiation | ||
30879 | Mammary gland development | ||
48522 | Positive regulation of cellular process | ||
23052 | Signaling | ||
42592 | Homeostatic process | ||
10907 | Positive regulation of glucose metabolism | ||
9967 | Positive regulation of signal transduction | ||
10676 | Positive regulation of cellular carbohydrate metabolism | ||
45913 | Positive regulation of carbohydrate metabolism | ||
23056 | Positive regulation of signaling process | ||
48878 | Chemical homeostasis | ||
8633 | Activation of proapoptotic gene products | ||
42981 | Regulation of apoptosis | ||
43067 | Regulation of programmed cell death | ||
2690 | Positive regulation of leukocyte chemotaxis | ||
42531 | Positive regulation of tyrosine phosphorylation of STAT protein | ||
10941 | Regulation of cell death | ||
3006 | Reproductive developmental process | ||
35295 | Tube development | ||
60740 | Prostate gland epithelium morphogenesis | ||
48518 | Positive regulation of biological processes | ||
32787 | Monocarboxylic acid metabolism | ||
60512 | Prostate gland morphogenesis | ||
8406 | Gonad development | ||
30335 | Positive regulation of cell migration |
Mechanisms of the dysregulated lncRNAs in colon cancer were further explored using KEGG pathway analysis. The pathway analysis revealed that these lncRNAs participated in several cancer-related signaling pathways including apoptosis, the PI3K-AKT signaling pathway, and the EGFR signaling pathway (Figure
We further evaluated the prognostic value of the identified lncRNAs in colon cancer. We found that, compared with lower expression, higher expression of ALMS1-IT1 and RP13-942N8.1 was significantly correlated with poor prognosis of colon cancer (Figures
Prognostic value of lncRNAs for assessing the clinical outcome of colon cancer. Kaplan-Meier survival curves for colon cancer patients using the lncRNA signature. Patients were assigned to high-/low-expression groups according to the median of the expression level of lncRNAs.
The ceRNA hypothesis represents a novel posttranscriptional regulatory dimension of gene regulation [
Functional studies show that ceRNA appears to participate in several cancer-related processes such as cell proliferation, cell cycle, and cell invasion. The pathway analysis suggested that the ceRNA network is potentially involved in multiple signaling pathways, such as the PI3K-AKT, Rap1, Ras, cytokine-cytokine receptor interaction, and EGFR tyrosine kinase inhibitor resistance signaling pathways.
Studies have revealed a critical role of lncRNA dysregulation in modulating gene expression as well as tumor development and progression. A recent study reported that lncRNA H19 is a miRNA 200a sponge that inhibits miRNA 200a functions, thereby promoting cell proliferation in colorectal cancer [
In the constructed ceRNA network, lncRNAs TRG-AS1 and RP11-399O19.9 were significantly downregulated in tumor tissues, and the expression was much lower in late-stage cancer than in early-stage cancer, suggesting that they may be protective factors in colon cancer. Meanwhile, lncRNAs UCA1, RP11-458F8.4, RP1-239B22.5, and ALMS1-IT1, which were markedly upregulated in tumor tissues, have higher expression levels in late-stage cancer. The results suggest that these lncRNAs may play crucial roles in colon cancer occurrence and progression.
From our literature review, we found that several lncRNAs among the ceRNA network, such as lncRNAs UCA1, GAS5, MALAT1, and PVT1, are associated with oncogenesis and the development of colon cancer [
However, the functions of most lncRNAs enrolled in the network have not been determined. Computational methods for predicting the function of lncRNA have shown many advantages to functional annotation [
lncRNAs can be powerful predictors for the survival of cancer patients [
In conclusion, the present study represents a view of colon cancer from a concurrent analysis of lncRNAs, miRNAs, and mRNAs. The constructed colon cancer ceRNA network brings to light an unknown lncRNA regulatory network in colon cancer. Furthermore, analysis of the ceRNA network identified several lncRNAs that are possibly involved in the regulation mechanisms, progression, and prognosis of colon cancer. Future functional investigation of these lncRNAs is essential to confirm the association with colon cancer and to explore novel potential targets for therapy.
The data used to support the findings of this study are available from the corresponding author upon request.
All experimental procedures were approved by the Ethics Committee of Nanfang Hospital.
The authors declare that they have no competing interests to report.
Yang Cheng, Lanlan Geng, and Kunyuan Wang contributed equally to the work.
This work was supported by the National Natural Science Foundation of China (no. 81602646), Natural Science Foundation of Guangdong Province (no. 2016A030310254), and China Postdoctoral Science Foundation (no. 2016M600648).
Table S1: target sequences for siRNAs for the top 5 lncRNAs in the colon cancer-specific ceRNA network. Table S2: primer sequence for the QRT-PCR analysis of the top five lncRNAs in the colon cancer-specific ceRNA network. Table S3: specific gene primers of the target genes of lncRNAs. Table S4: specific primers of the target miRNAs of lncRNAs. Table S5: network characteristics of a colon cancer-specific ceRNA network. Figure S1: a flowchart of ceRNA network construction and analysis in our study. Figure S2: verification of the expression of the target miRNAs in the ceRNA network. Figure S3: functional enrichment of lncRNA RP11-399O19.9 in the ceRNA network.
Doc S1: characteristics of the patients in the TCGA COAD database.
Doc S2: ceRNA crosstalk identified in the COAD database.
Doc S3: GO enrichment analysis of the ceRNA network.
Doc S4: KEGG pathway analysis of the ceRNA network.