This article is mentioned in:

Introduction

Intrahepatic cholangiocarcinoma (ICC) is the second most frequently occurring primary liver cancer with a high mortality rate (1), which affects 1–2 per 100,000 patients (2). Various factors have previously been suggested to contribute to the progression of ICC, including sclerosing cholangitis and hepatobiliary flukes (3). Ultrasound scans may provide diagnostic information (4); however, specific diagnostic criteria for patients with ICC remain to be elucidated. Therefore, numerous patients present with ICC for diagnosis at an advanced stage, which may contribute to its poor prognosis (5). It is therefore important to identify potential biomarkers of ICC to aid prevention and identification of therapeutic strategies.

The progression of ICC is associated with numerous genetic factors, including gene mutations and dysregulation of gene expression (6). Various technological advances have identified molecular targets. Weber et al (7) demonstrated that low frequency alterations of phosphatase and tensin homolog, cyclin-dependent kinase inhibitor 2A, and breast cancer 1/2 may explain the mutation spectrum in ICC, based on clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated protein 9-based targeted somatic multiplex-mutagenesis. Lee et al (8) profiled the expression of 315 genes and screened 400 alterations in 84 genes, via hybridization capture, which may act as therapeutic targets of ICC (8). Various drugs or therapeutic methods have been developed based on the aforementioned identified biomarkers. Lovastatin, which is a 3-hydroxy-3-methylglutaryl-coenzyme-CoA reductase inhibitor, may inhibit the proliferation, migration and adhesive activities of ICC cells, by inhibiting the expression of transforming growth factor-β1, cyclooxygenase-2, and intracellular adhesion molecule-2 (9). Furthermore, various genes have previously been demonstrated to affect drug sensitivity and resistance in ICC. Tepsiri et al (10) demonstrated that the expression of multidrug resistance-associated protein 3 was significantly associated with the half maximal inhibitory concentration of etoposide in ICC. Various biomarkers have been identified; however, the underlying mechanism remains to be fully elucidated, to improve the prognosis and therapy of ICC.

Gene microarrays profile the expression of thousands of genes simultaneously and have been incorporated in numerous cancer and ICC associated investigations (11–13). The present study compared gene and microRNA (miRNA/miR) expression data in ICC samples with healthy liver tissues, and differentially expressed genes (DEGs) and miRNAs (DEMs) were identified. Functional and pathway analyses were conducted, and subsequently, a miRNA-gene regulation network was constructed to explore potential biomarkers of ICC to improve its prognosis.

Materials and methods

Microarray data

The miRNA expression dataset GSE32957 and mRNA expression dataset GSE32879 (14) were downloaded from the Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo/). A total of 16 ICC, 7 combined hepatocellular-cholangiocarcinoma (CHC) and 2 hepatic adenoma samples were included in the mRNA and miRNA expression datasets. In addition, the miRNA expression dataset contained 3 focal nodular hyperplasia (FNH) and 5 healthy liver tissue, and 2 cholangiocarcinoma cell lines, whereas the mRNA expression dataset contained 5 FNH and 7 healthy liver tissue samples. Hybridizations of miRNA and mRNA were performed on GPL14732 Nanostring nCounter Human microRNA Expression Platform (NanoString Technologies, Inc., Seattle, WA, USA) and GPL6244 [HuGene-1_0-st] Affymetrix Human Gene 1.0 ST Array [transcript (gene) version; Affymetrix, Inc., Santa Clara, CA, USA], respectively.

Differential expression analysis

The free R package NanoStringNorm (15) was used for the data preprocessing of GSE32957, whereas the affy package (https://bioconductor.org/packages/release/bioc/html/affy.html) was selected for analysis of GSE32879, due to the background correction and quartile normalization qualities it exhibits. DEGs and DEMs were identified in ICC samples compared with healthy liver tissues, via the limma package (http://bioconductor.org/packages/release/bioc/html/limma.html) in R. DEGs and DEMs were screened with the thresholds of Benjamini-Hochberg adjusted P<0.05 and |log2(fold change)|>1.

Alternative splicing (AS) analysis

AS, which is prevalent in the mammalian genome, may generate a complex transcriptome from a finite genome and thus, result in a diverse range of proteins with various functions (16,17). The present study used AltAnalyze (18) software to infer AS via the interpretation of alternative exon inclusion, in ICC samples. Briefly, raw CEL microarray data were normalized using the FIRMA method implemented in AltAnalyze, followed by the identification of alternative splicing genes (ASGs) with all of the default parameters of AltAnalyze. In addition, hierarchical clustering of ASGs in ICC and healthy liver tissue samples were conducted.

Functional and pathway enrichment analysis

The Database for Annotation, Visualization and Integrated Analysis (DAVID; https://david.ncifcrf.gov/) is a widely used web-based tool for functional and pathway enrichment analysis (19). DAVID was used to conduct functional and pathway enrichment analysis of DEGs. P<0.05 was used as the criterion to identify significantly enriched Gene Ontology (GO; http://geneontology.org/) terms and Kyoto Encyclopedia of Genes and Genomes (KEGG; http://www.genome.jp/kegg/) pathways.

Screening of target genes of DEMs

TargetScan (20) (http://www.targetscan.org/) is a web-based software that was used to predict the target genes of DEMs. Overlaps among these target genes, DEGs and ASGs were obtained, and regulatory associations between the overlaps and DEMs were retrieved to construct the miRNA-gene regulation network, which was visualized using Cytoscape software (21).

Results

Identification of DEGs, DEMs and ASGs

A total of 2,327 DEGs; 1,214 of which were downregulated and 1,113 of which were upregulated, and 70 DEMs, including 5 down- and 65 upregulated, were identified in ICC samples compared with healthy liver tissues. A total of 623 genes exhibited alternative splicing in ICC samples compared with healthy liver tissue samples. Hierarchical clustering of ASGs in ICC and healthy liver tissue samples was conducted, as presented in Fig. 1.

Figure 1. Hierarchical clustering of alternative splicing genes in ICC and healthy liver tissue samples, using AltAnalyze software. The red and blue colors indicate a high and low expression value of alternative splicing genes, respectively. ICC, intrahepatic cholangiocarcinoma.

Functional and pathway enrichment analysis of DEGs

Functional and pathway enrichment analysis was performed for the 2,327 DEGs. The majority of enriched GO terms were involved in the activity of the cell, and the most significantly enriched were presented in cell-matrix adhesion. Ten of the GO terms that were randomly selected at P<0.05, are presented in Fig. 2. A total of 51 KEGG pathways associated with metabolism and degradation were revealed to be enriched in DEGs; 10 randomly selected pathways are presented in Table I.

Figure 2. DEG functional and pathway enrichment analysis. Ten GO terms randomly selected from an enrichment analysis of 2327 DEGs using the Database for Annotation, Visualization and Integrated Analysis. P<0.05. DEGs, differently expressed genes.

Table I. A total of 10 Kyoto Encyclopedia of Genes and Genomes pathways enriched in differentially expressed genes.

Table I.

A total of 10 Kyoto Encyclopedia of Genes and Genomes pathways enriched in differentially expressed genes.

Pathway name	P-value	Gene number
hsa04610: Complement and coagulation cascades	5.92E-24	49
hsa00071: Fatty acid metabolism	1.89E-13	28
hsa00260: Glycine, serine and threonine metabolism	5.08E-13	24
hsa00280: Valine, leucine and isoleucine degradation	5.72E-12	28
hsa00380: Tryptophan metabolism	1.99E-11	26
hsa03320: PPAR signaling pathway	2.63E-10	34
hsa00982: Drug metabolism	1.22E-09	31
hsa00330: Arginine and proline metabolism	1.10E-08	27
hsa00980: Metabolism of xenobiotics by cytochrome P450	1.22E-08	29
hsa00250: Alanine, aspartate and glutamate metabolism	7.26E-08	19

[i] PPAR, peroxisome proliferator-activated receptor.

Target genes of DEMs

A total of 956 target genes of DEMs were obtained via TargetScan software (data not shown). In addition, 63 overlaps were identified among these target genes, DEGs and ASGs, and 243 miRNA-gene regulation pairs were retrieved between these overlaps and DEMs. The miRNA-gene regulation network is presented in Fig. 3. Furthermore, 52 miRNA-gene pairs (Table II) exhibited an opposite trend in the alteration of miRNA and gene expression values in ICC samples, compared with healthy liver tissues.

Figure 3. miRNA-gene regulation network. The network was constructed and visualized using Cytoscape software. The green and red circles represent miRNA and genes, respectively. miRNA/miR, microRNA.

Table II. A total of 52 miRNA-gene pairs presenting opposite trends in the alterations of miRNA and gene expression values in intrahepatic cholangiocarcinoma samples compared with those in healthy liver tissues.

Table II.

A total of 52 miRNA-gene pairs presenting opposite trends in the alterations of miRNA and gene expression values in intrahepatic cholangiocarcinoma samples compared with those in healthy liver tissues.

miRNA	Gene
hsa-miR-186	ADRA1A
hsa-miR-122	MGAT5
hsa-miR-132	ADRA1A
hsa-miR-96	TPRG1
hsa-miR-96	SCN9A
hsa-miR-122	LRP10
hsa-miR-96	LONP2
hsa-miR-122	ADSS
hsa-miR-96	TIAM1
hsa-miR-96	CLCN5
hsa-miR-96	GK
hsa-miR-200c	TPRG1
hsa-miR-21	TIAM1
hsa-miR-141	SCN9A
hsa-miR-21	CNTFR
hsa-miR-144	MGAT5
hsa-miR-218	TPRG1
hsa-miR-144	PTPN12
hsa-miR-144	TMEM165
hsa-miR-200c	CNTFR
hsa-miR-141	TIAM1
hsa-miR-224	TPRG1
hsa-miR-183	TIAM1
hsa-miR-141	CLCN5
hsa-miR-200c	MCC
hsa-miR-335	MTMR4
hsa-miR-144	ZNF292
hsa-miR-144	SPRED1
hsa-miR-144	C20orf194
hsa-miR-218	CLCN5
hsa-miR-1	GPR125
hsa-miR-223	LONP2
hsa-miR-144	TMEM55A
hsa-miR-186	TPRG1
hsa-miR-155	CLCN5
hsa-miR-186	SCN9A
hsa-miR-191	ATP2B2
hsa-miR-144	VLDLR
hsa-miR-144	SMARCA4
hsa-miR-144	FAM222B
hsa-miR-423-5p	CNTFR
hsa-miR-218	MCC
hsa-miR-144	RCN2
hsa-miR-144	ATP2C1
hsa-miR-24	FANCC
hsa-miR-24	CNTFR
hsa-miR-24	CLCN5
hsa-miR-224	NOTUM
hsa-miR-25	XBP1
hsa-miR-24	MCC
hsa-miR-532-5p	MTMR4
hsa-miR-186	MCC

[i] miRNA/miR, microRNA.

Discussion

In recent years, the emergence of novel diagnostics and treatments, including B-mode ultrasound and chemotherapy, has improved the prognosis of ICC. However, limitations are present in terms of its early diagnosis and cure. The present study conducted a statistical comparison of mRNA and miRNA expression profiles between ICC and healthy liver tissues samples, in order to identify DEGs, DEMs and ASGs. Functional and pathway analyses of DEGs were conducted to explore associated functions and pathways. The miRNA-gene regulation network was constructed to identify the potential biomarkers of ICC.

The GeneChip Human Gene 1.0 ST array (HuGene-1_0-st-v1) is a transcript-based array, used for the detection of gene expression values, and estimation of AS based on the well-annotated exon probes (22). The present study identified the DEGs and ASGs in ICC samples compared with healthy liver tissues, based on the data detected by HuGene-1_0-st-v1. Functional enrichment analysis indicated that the DEGs were primarily involved in the biological processes associated with cell activity, including cell adhesion and regulation of cell proliferation, which have previously been demonstrated to be associated with cancer progression (23). Pathway enrichment analysis revealed that DEGs were significantly enriched in the pathways associated with substance metabolism and degradation, including fatty acid, glycine, serine and threonine metabolism, and valine, leucine and isoleucine degradation. The extracellular matrix-receptor pathway was also demonstrated to be enriched in DEGs, consistent with the findings of Lee et al (24). The genes that were simultaneously presented in the extracellular matrix-receptor pathway and were revealed to be ASGs, included chondroadherin (CHAD), integrin subunit (ITG)-A3, ITG-B1 and laminin subunit alpha-5 (LAMA5). Three of the aforementioned genes, (ITGA3, ITGB1 and LAMA5) have previously been demonstrated to be associated with the progression of ICC (25,26). CHAD encodes a cartilage matrix protein, which may mediate adhesion of isolated chondrocytes. CHAD may be included in the phosphoinositide 3-kinase-Akt and focal adhesion signaling pathways, and may be regulated by p53, which is associated with the pathogenesis and development of numerous cancers, including ICC (http://www.genecards.org). Therefore, CHAD may be a novel biomarker of ICC that may contribute to its progression; however, further studies are required to confirm this.

miRNA are non-coding RNA molecules, which contain ~22 nucleotides, and are present in animals, plants and viral genomes. miRNAs may induce RNA-silencing and regulate gene expression at the post-transcriptional stage (27,28). The dysregulation of miRNA may result in numerous diseases, including ICC (29–31). The present study identified DEMs in ICC samples compared with healthy liver tissues samples, and the screening of DEM target genes was conducted using the TargetScan database. A miRNA-gene regulation network was constructed based on the DEMs and the overlapping genes among the DEGs, ASGs and target genes of the DEMs. In the regulation network, hsa-miR-96 regulated 22 target genes and 6 out of these regulation pairs [hsa-miR-96-tumor protein P63 regulated 1, hsa-miR-96-sodium voltage-gated channel alpha subunit 9, hsa-miR-96-lon peptidase 2, peroxisomal, hsa-miR-96-T-cell lymphoma invasion and metastasis 1 (TIAM1), hsa-miR-96-chloride voltage-gated channel 5, hsa-miR-96-glycerol kinase] revealed an opposite trend in the alterations of miRNA and mRNA expression values (upregulated miRNA and downregulated gene expression levels). hsa-miR-96 is closely associated with cell proliferation and growth, and Collins et al (32) demonstrated that hsa-miR-96 may be used to identify cholangiocarcinoma and pancreatic adenocarcinoma. Furthermore, the dysregulation of hsa-miR-96 was revealed to contribute to the progression of ICC (33). TIAM1 is a target gene of hsa-miR-96, which is important in cellular migration and remodeling of the actin cytoskeleton, and may be involved in the proliferation and migration of ICC via effects on RasGEF domain family member 1A expression (34). Further target genes of hsa-miR-96 have previously been identified, and may act as novel biomarkers of ICC, however further experimental verification is necessary (35).

In conclusion, the present study identified DEGs, ASGs and DEMs in ICC samples compared with healthy liver tissues, via bioinformatics analysis of miRNA and mRNA expression data. Functional and pathway analyses indicated that DEGs were primarily involved in biological processes or pathways associated with cancer progression or substance metabolism. A miRNA-gene regulation network was constructed based on DEMs and overlaps among DEGs, ASGs and target genes of DEMs. Furthermore, biomarkers that have previously been identified, including hsa-miR-96 and TIAM1, and numerous novel biomarkers in ICC, including anthranilate synthase component II and sodium voltage-gated channel a subunit 9, were identified. However, the functions of these biomarkers in ICC remain to be elucidated.

References