An integrative meta-analysis of microRNAs in hepatocellular carcinoma

Abstract

We aimed to shed new light on the roles of microRNAs (miRNAs) in liver cancer using an integrative in silico bioinformatics analysis. A new protocol for target prediction and functional analysis is presented and applied to the 26 highly differentially deregulated miRNAs in hepatocellular carcinoma. This framework comprises: (1) the overlap of prediction results by four out of five target prediction tools, including TargetScan, PicTar, miRanda, DIANA-microT and miRDB (combining machine-learning, alignment, interaction energy and statistical tests in order to minimize false positives), (2) evidence from previous microarray analysis on the expression of these targets, (3) gene ontology (GO) and pathway enrichment analysis of the miRNA targets and their pathways and (4) linking these results to oncogenesis and cancer hallmarks. This yielded new insights into the roles of miRNAs in cancer hallmarks. Here we presented several key targets and hundreds of new targets that are significantly enriched in many new cancer-related hallmarks. In addition, we also revealed some known and new oncogenic pathways for liver cancer. These included the famous MAPK, TGFβ and cell cycle pathways. New insights were also provided into Wnt signaling, prostate cancer, axon guidance and oocyte meiosis pathways. These signaling and developmental pathways crosstalk to regulate stem cell transformation and implicate a role of miRNAs in hepatic stem cell deregulation and cancer development. By analyzing their complete interactome, we proposed new categorization for some of these miRNAs as either tumor-suppressors or oncomiRs with dual roles. Therefore some of these miRNAs may be addressed as therapeutic targets or used as therapeutic agents. Such dual roles thus expand the view of miRNAs as active maintainers of cellular homeostasis.

Keywords: Cancer hallmarks; Hepatocellular carcinoma; Integrative bioinformatics; Target prediction; miRNA.

Figure 1

A portrayal of the important miRNA targets linked to the hallmarks of cancer The deregulated miRNA targets in HCC were assigned based on GO annotations using DAVID tool, highlighting the impact of miRNA deregulation on carcinogenesis and metastasis for HCC and other common cancers.

Figure 2

A flow chart illustrating our new improved protocol for the miRNA target prediction steps and functional analysis.

Figure 3

Secondary structure hybridization and MFE of different miRNA-target pairs Shown are the examples for hybridization between miRNAs and their respective target genes for miR-195 and FGF-7 (A), miR-195 and GHR (B), miR-122a and GIT1 (C), miR-199a-3p and FOXQ1 (D), miR-182 and RASA1 (E), miR-182 and MTSS1 (F), miR-199a-5p and CELSR1 (G), miR-224 and NUP153 (H), indicating different modes of target recognition exhibited by miRNAs (canonical for some key targets that would undergo degradation similar to siRNA mode of action, while 5′ dominant and 3′UTR compensatory for targets that would undergo translational suppression). All these examples show the fertility of our approach of unified target prediction, as all targets have MFE < −20 kcal/mol (I). Green represents miRNA and red represents the target sequence. MFE stands for minimum free energy.

Figure 4

GO functional categories of the targets for miR-122a, miR-199 a-3p, miR-182, miR-195, miR-221, miR-224 and miR-96 that are differentially expressed in HCC The functional categories that are enriched in response to miRNA deregulation in HCC were analyzed using DAVID with P < 0.05. The result shows deregulation in transcriptional-related processes such as activity of transcription factors, gene expression and cellular biosynthetic process.

Figure 5

Highly enriched pathways and GO terms for the miRNA target gene set The functional categories that are enriched in response to miRNA deregulation in HCC were analyzed using GeneTrail with the enrichment analysis option (P < 0.05) and Bonferroni and FDR corrections.

Figure 6

Important pathways in HCC Shown is the mTOR/AKP/PIP3 pathway that contributes to transformation of nodules into metastatic counterparts . The target genes are indicated in green and key pathway phenotypes are shown at the bottom. Deactivation of the MAPK pathway through inhibition/repression of the up-regulating miRNAs or activation of suppressing miRNAs might be useful as an alternative therapeutic intervention strategy. Genes are indicated in pink ovals; FGF signaling pathway, actin cytoskeleton pathway and mTOR pathway are represented with lines in black, red and green, respectively.