Differential microRNA expression between hepatitis B and hepatitis C leading disease progression to hepatocellular carcinoma

Authors


  • Potential conflict of interest: Nothing to report.

Abstract

MicroRNA (miRNA) plays an important role in the pathology of various diseases, including infection and cancer. Using real-time polymerase chain reaction, we measured the expression of 188 miRNAs in liver tissues obtained from 12 patients with hepatitis B virus (HBV)-related hepatocellular carcinoma (HCC) and 14 patients with hepatitis C virus (HCV)-related HCC, including background liver tissues and normal liver tissues obtained from nine patients. Global gene expression in the same tissues was analyzed via complementary DNA microarray to examine whether the differentially expressed miRNAs could regulate their target genes. Detailed analysis of the differentially expressed miRNA revealed two types of miRNA, one associated with HBV and HCV infections (n = 19), the other with the stage of liver disease (n = 31). Pathway analysis of targeted genes using infection-associated miRNAs revealed that the pathways related to cell death, DNA damage, recombination, and signal transduction were activated in HBV-infected liver, and those related to immune response, antigen presentation, cell cycle, proteasome, and lipid metabolism were activated in HCV-infected liver. The differences in the expression of infection-associated miRNAs in the liver correlated significantly with those observed in Huh7.5 cells in which infectious HBV or HCV clones replicated. Out of the 31 miRNAs associated with disease state, 17 were down-regulated in HCC, which up-regulated cancer-associated pathways such as cell cycle, adhesion, proteolysis, transcription, and translation; 6 miRNAs were up-regulated in HCC, which down-regulated anti-tumor immune response. Conclusion: miRNAs are important mediators of HBV and HCV infection as well as liver disease progression, and therefore could be potential therapeutic target molecules. (HEPATOLOGY 2009.)

MicroRNA (miRNA) is an endogenous, small, single-strand, noncording RNA consisting of 20 to 25 bases and regulates gene expression of various cell types. It plays an important role in various biological processes, including organ development and differentiation as well as cellular death and proliferation, and is also involved in various diseases such as infection and cancer.1–3

miRNAs are produced as follows. A primary miRNA with a hairpin loop structure is cleaved into a precursor miRNA and transported out of the nuclei with a carrier protein (Exportin-5). The precursor miRNA is then processed by Dicer and converted into an active single-strand RNA in the cytoplasm. The miRNA binds to a target messenger RNA in a sequence-dependent manner and induces degradation of the target messenger RNA and translational inhibition. One miRNA regulates the expression of multiple target genes; bioinformatics analyses have suggested that the expression of more than 30% of human genes is regulated by miRNAs.4–7

Infection of the human liver with hepatitis B virus (HBV) and hepatitis C virus (HCV) induces the development of chronic hepatitis (CH), cirrhosis, and in some instances hepatocellular carcinoma (HCC).8 The virological features of these two distinct viruses are completely different; however, the viruses infect the liver and cause CH, which is not distinguished by histological examination or clinical manifestations. We previously reported that gene expression profiles in chronic hepatitis B (CH-B) and chronic hepatitis C (CH-C) are different. Proapoptotic and DNA repair responses were predominant in CH-B, and inflammatory and antiapoptotic phenotypes were predominant in CH-C. However, factors inducing these differences in gene expression remain to be elucidated.9, 10

We examined miRNA expression in liver tissue with HBV-related liver disease (CH-B and HCC-B) and HCV-related liver disease (CH-C and HCC-C) and in normal liver tissue via real-time detection polymerase chain reaction (RTD-PCR). We also performed global analysis of messenger RNA expression in these tissues using complementary DNA (cDNA) microarray. These analyses allowed us to find characteristic miRNAs associated with HBV or HCV infection as well as the progression of liver disease.

Abbreviations

cDNA, complementary DNA; CH, chronic hepatitis; CH-B, chronic hepatitis B; CH-C, chronic hepatitis C; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCC-B, hepatitis B-related hepatocellular carcinoma; HCC-C, hepatitis C-related hepatocellular carcinoma; HCV, hepatitis C virus; miRNA, microRNA; RTD-PCR, real-time detection polymerase chain reaction; SVM, support vector machine.

Materials and Methods

Patients.

The study subjects included 12 patients with CH-B complicated by HCC and 14 patients with CH-C complicated by HCC. Gene expression analysis was approved by the ethics committee of the Graduate School of Medicine, Kanazawa University Hospital, Japan, between 1999 and 2004. In addition, nine normal liver tissue samples obtained during surgery for metastatic liver cancer were used as control samples. Surgically removed liver tissues were stored in liquid nitrogen until analysis. Histological classification of HCC and histological evaluation of hepatitis in noncancerous regions for each patient are shown in Table 1. HCV viremia in two patients with CH-C was persistently cleared by interferon therapy before HCC development. There were no significant differences in the histological findings of HCC and noncancerous regions, as well as in sex, age, and hepatic function between the HBV and HCV infection groups.

Table 1. Characteristics of Patients Used for Analysis of miRNA and Microarray Samples
Patient No.VirusAgeSexALTHistology of ActivityBackground Liver FibrosisHistological Grade of HCCTumor Size (mm)TNM StagingHCV-RNA (KIU/mL)HBV-DNA (LEG/mL)
  • HCV RNA was assayed via Amplicor Monitor Test (KIU/mL); HBV DNA was assayed via transcription-mediated amplification (LEG/mL).

  • Abbreviations: ALT, alanine aminotransferase; F, female; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; M, male; TNM, tumor–node—metastasis.

  • *

    Vascular invasion (+).

1HBV57M1624Moderate20II3.4
2HBV51M5712Moderate48II< 2.6
3HBV61M1714Well16II< 3.7
4HBV47M1914Moderate15I< 3.7
5HBV72M1911Well25IINA
6HBV73M6213Moderate45III5.7
7HBV42M3614Moderate18I< 3.7
8HBV63M1312Moderate15I2.8
9HBV68F5412Well56II4.1
10HBV70M1302Well40II< 3.7
11HBV58M2914Moderate35IVA*3.3
12HBV72M2214Moderate18I6
13HCV66F3324Well25II423
14HCV67M8914Well30II> 850
15HCV64M3114Moderate75III< 5 (+)
16HCV68M3004Well23II> 850
17HCV46M9823Moderate20I> 850
18HCV68F3224Moderate25III< 5 (+)
19HCV66F4624Well25II> 850
20HCV47M24613Moderate20I262
21HCV75M2713Moderate19II85.1
22HCV77M2101Moderate20II< 5 (−)
23HCV66M4622Well60II50.3
24HCV65M8911Poorly25III850
25HCV53M5401Moderate28II< 5 (−)
26HCV75F21214Well19I580
2751F1800
2878F1300
2975M2000
3034M1200
3164M3000
3278M900
3353M1900
3464F1200
3560F2000

Quantitative RTD-PCR.

Approximately 1 mg of each liver tissue sample stored in liquid nitrogen was ground with a homogenizer while still frozen, and total RNA containing miRNA was isolated according to the protocol of the mirVana miRNA Isolation kit (Ambion, Austin, TX) and stored at −80°C until analysis. miRNA expression levels were quantitated using the TaqMan MicroRNA Assays Human Panel Early Access kit (Applied Biosystems, Foster City, CA). cDNA was prepared via reverse transcription using 10 ng each of the isolated total RNA and 3 μL each of the reverse transcription primers with specific loop structures. Reverse transcription was performed using the TaqMan MicroRNA Reverse Transcription kit (Applied Biosystems) according to the manufacturer's protocol. Then, a mixture of 6.67 μL of nuclease-free water, 10 μL of TaqMan 2 × Universal PCR Master Mix (No AmpErase UNG; Applied Biosystems), and 2 μL of TaqMan MicroRNA Assay Mix, which was included in the kit, was prepared for each sample on a 384-well plate; 1.33 μL of the reverse transcription product was added to the mixture, and amplification reaction was performed on an ABI PRISM 7900HT (Applied Biosystems). Expression levels of 188 miRNAs in each sample were quantitated.

Analysis of RTD-PCR Data.

The measured 188 miRNAs included RNU6B, which is commonly used as a control for miRNA. β-Actin and glyceraldehyde 3-phosphate dehydrogenase were also measured simultaneously for correcting RNA amount. The mean Ct values and standard deviations of each miRNA were calculated from expression data of all patients obtained by RTD-PCR. miRNA with the lowest expression variation was used as the internal control. Ct values of each miRNA were then corrected by the Ct value of the internal control to yield −ΔCt values defined as relative miRNA expression levels and used for analyses. Statistical analyses and hierarchical cluster analyses of expression data were performed using BRB ArrayTools (http://linus.nci.nih.gov/BRB-ArrayTools.html). Relative miRNA expression levels were further normalized using the median over the all patients so that the normalized expression levels of each patient have a median log ratio of 0. A class prediction method was used for classifying two patient groups based on the supervised learning method, and a binary tree classification method was used for classifying three or more patient groups with a statistical algorithm of the support vector machine (SVM). Class prediction was performed using SVM incorporating genes differentially expressed at a univariate parametric significance level of P = 0.01. The prediction rate was estimated via cross-validation and the bootstrap method for small sample data.11 (It is worth noting that the prediction rate may be likely an overestimate of the true rate, given the weaknesses of cross-validation and bootstrapping methods in a strict sense.)

Microarray Analysis.

cDNA microarray slides (Liver chip 10k) were used as described.10 RNA isolation, amplification of antisense RNA, labeling, and hybridization were performed according to the protocols described.9, 10 Quantitative assessment of the signals on the slides was performed by scanning on the ScanArray 5000 (General Scanning, Watertown, MA) followed by image analysis using GenePix Pro 4.1 (Axon Instruments, Union City, CA) as described.10

Preliminary Survey of Independency of Paired Samples from the Same Patient.

CH and HCC expression data were derived from the same patient. Before further analysis, we examined whether the miRNA expression of paired samples was similar or independent. We compared differences in the expressions of paired and nonpaired CH and HCC samples using the Dunnett test12 (Supplementary Data). All possible tests performed for data pairs represented no dependency due to the paired data from the same patients. For data analysis, we used the standard pairwise class comparison and prediction tool in BRB ArrayTools.

Identification of Candidate miRNA Target Genes.

Candidate target genes predicted to be regulated by miRNAs based on sequence comparison were selected using MIRANDA Pro3.0 (Sanger Institute). Of the selected genes, those represented on a microarray chip were then examined for expression (Fig. 4). The number of genes showing a significant (P <0.05) expression difference among the candidate target genes represented on the chip was statistically analyzed to evaluate the significance of expression regulation by miRNAs. Analysis of significance was performed using Hotelling T2 test (BRB ArrayTools).

Figure 4.

Analysis of miRNA expression data. Target genes of miRNAs were predicted using MIRANDA Pro3.0; candidate target genes spotted on microarray were identified; number of genes that actually exhibit significant (P < 0.05) changes in expression among the genes was determined; and signal pathways involving genes regulated by the miRNAs that had exhibited differential expression between each group were analyzed using MetaCore (Table 4).

Pathway Analysis.

Of the candidate miRNA target genes, those showing a significant (P < 0.01) expression difference between N, CH-B, HCC-B, CH-C, and HCC-C samples were analyzed for pathways involving these genes using MetaCore software suite (GeneGo, St. Joseph, MI). Significance probability was calculated using the hypergeometrical distribution based on gene ontology terms. Because one gene is frequently involved in multiple pathways, all pathways corresponding to the genes with significance probability were listed.

Verification of Regulation of Candidate Target Genes by miRNAs.

Small interfering RNAs (Ambion) specific to 13 miRNAs (has-miR-17*, has-miR-20a, has-miR-23a, has-miR-26a, has-miR-27a, has-miR-29c, has-miR-30a, has-miR-92, has-miR-126, has-miR-139, has-miR-187, has-miR-200a, and has-miR-223) showing significant differences in expression were transfected into Huh7 cells using TransMessenger transfection reagent (QIAGEN, Valencia, CA), and loss of function of each miRNA was evaluated. Similarly, precursor miRNAs of five miRNAs (has-miR-23a, has-miR-26a, has-miR-27a, has-miR-92, and has-miR-200a) were also transfected into Huh7 cells, and gain of function of each miRNA was evaluated. The loss- and gain-of-function of miRNAs were evaluated via RTD-PCR. In addition, different gene expressions regulated by miRNAs were also evaluated via RTD-PCR.

HBV/HCV Infection Model Using Cultured Cells.

The plasmid pHBV 1.2 coding the 1.2-fold length of the HBV genome was transfected into Huh7.5 cells using Fugene6 transfection reagent (Roche Applied Science, Indianapolis, IN). HBeAg production in culture medium was measured using Immunis HBeAg/Ab EIA (Institute of Immunology Co., Ltd., Tokyo, Japan).13 The amount of HBV-DNA was measured via RTD-PCR (Supplementary Fig. 1A,B). JFH1-RNA was transfected into Huh7.5 cells using TransMessenger transfection reagent (QIAGEN) and the expression of the core protein was examined via immunofluorescence staining using anti-HCV core antibody (Affinity BioReagent, CO).14, 15 HCV-RNA amount was also measured via RTD-PCR (Supplementary Fig. 1A,B). JFH1/GND was used as a negative control. miRNA expression was quantitated by RTD-PCR 48 hours after transfection.

Figure 1.

(A) miRNA-specific RTD-PCR using sheet hairpin primers. (B) miRNA amplification curves by RTD-PCR.

Results

Expression of miRNA in Liver Tissue.

A panel of miRNA was successfully amplified from liver tissues via RTD-PCR. The representative amplification profile of miRNA as determined with RTD-PCR is shown in Fig. 1. To assess the reliability and reproducibility of this assay system, we first measured RNU6B in duplicate from all samples in different plates. The mean difference in Ct values of RNU6B expression within the same samples was 0.08 ± 0.05 (mean ± standard deviation), indicating the high reproducibility of this assay. All Ct values from each reaction were collected, and Ct variation obtained by each probe from all patients was calculated. Although RNU6B was frequently used as the internal control, the standard Ct variation was relatively high (Ct, 27 ± 1.94), suggesting that the variances in its value depend on the state of liver disease (N, CH and HCC). Therefore, we selected has-miR-328 as the internal control with the smallest standard deviation (Ct, 30 ± 0.60). The relative expression ratio of individual miRNA to has-miR-328 was calculated and applied to the following analysis using a BRB-array tool.

Hierarchical cluster analysis revealed that the expression profiles of the 188 miRNAs from each patient were roughly classified into normal liver, HBV-infected liver (CH-B+HCC-B; HBV group), and HCV-infected liver (CH-C+HCC-C; HCV group) (Fig. 2A). HCV viremia in two patients with CH-C was persistently cleared by interferon therapy before HCC development. The background liver of one of these patients was clustered in the normal group and those of others in the HCV group. Although these two patients were not clearly differentiated from others, some miRNAs such as miR-194, miR-211, and miR-340 that were down-regulated in the HCV group were significantly up-regulated in two patients (Fig. 3, cluster 2).

Figure 2.

(A) Hierarchical cluster analysis using total miRNA. Chronic hepatitis is indicated by histological stage and grade (F, fibrosis; A, activity) and type of infecting virus (B or C). HCC is indicated by histological grade (w, well differentiated; m, moderately differentiated; p, poorly differentiated) and type of infecting virus (B or C), with the patient number added at the end. (B) Relationship between five classes divided by binary tree classification. Expression profiles were first classified into normal liver and non-normal liver groups (node 1), then into HBV and HCV groups (node 2). The HBV group was further divided into HCC-B and CH-B (node 3), and the HCV group into HCC-C and CH-C (node 4).

Figure 3.

Cluster 1: Eight miRNAs specifically differentiated node 1 classification. Cluster 2: Nineteen miRNAs specifically differentiated node 2 classification. Cluster 3: Twenty-three miRNAs differentiated CH-B and HCC-B as well as CH-C and HCC-C.

The present CH and HCC expression data were obtained from the same patient; however, each sample clustered irrespective of pairs in all but two patients. miRNA expression profiling was therefore more dependent on the disease condition than on the paired condition, as also confirmed by the Dunnett test.12 We then attempted to classify the expression profiles into HBV and HCV groups using supervised learning methods (Table 2-1). HBV and HCV groups were significantly differentiated at an 87% accuracy (P < 0.001). The normal liver and CH (CH-B + CH-C) and CH and HCC (HCC-B + HCC-C) were also significantly differentiated at a 90% rate of accuracy. These results suggest that different stages of liver disease (normal, CH, and HCC) can be differentiated from each other based on the miRNA expression profile, as well as HBV and HCV infection.

Table 2-1. Class Prediction
No.ClassPrediction (%)No. of PredictorsP Value
  1. Class prediction algorithm was used for the classification of two groups of patients. Feature selection was based on the univariate significance level (alpha = 0.01). The support vector machine classifier was used for class prediction.

  2. Abbreviations: CH, nontumor lesion of HCC; HCC, hepatocellular carcinoma; N, normal.

1HBV versus HCV8732<0.001
2N versus CH (B+C)91260.007
3CH (B+C) versus HCC (B+C)92340.003

To examine the relationship among five categories of groups, namely, N, CH-B, CH-C, HCC-B and HCC-C, we attempted to differentiate the five groups using a supervised learning algorithm (binary tree classification) used for classifying three or more groups. SVM was used as a prediction method. Expression profiles were first classified into groups N (normal) and non-N (non-normal) (CH-C, CH-B, HCC-C, and HCC-B) (node 1) (P < 0.01). The non-N group was then classified into HBV and HCV (node 2) (P < 0.01). The HBV group was further classified into CH-B and HCC-B (node 3) (P < 0.01), and the HCV group was further classified into CH-C and HCC-C (node 4) (P < 0.01) (Fig. 2B, Table 2-2). Thus, the findings support the notion that differences in miRNA expression between HBV and HCV are as distinct as those between CH and HCC.

Table 2-2. Binary Tree Classification
NodeGroup 1 ClassGroup 2 ClassNo. of PredictorsMisclassification Rate (%)
  1. Binary tree classification algorithm was used for the classification of each category of patients. Feature selection was based on the univariate significance level (alpha = 0.01). The support vector machine classifier was used for class prediction. There were four nodes in the classfication tree.

  2. Abbreviations: CH-B, non-tumor lesion of HCC-B; CH-C, nontumor lesion of HCC-C; HCC-B, hepatitis B virus-related hepatocellular carcinoma; HCC-C, hepatitis C virus-related hepatocellular carcinoma; N, normal

1HCC-B, HCC-C, CH-B, CH-CN204.9
2HCC-B, CH-BHCC-C, CH-C1913.5
3HCC-BCH-B1529.2
4HCC-CCH-C1417.9

Out of 20 miRNAs that differentiated node 1 classification (Table 2-2), 12 also differentiated node 3 or node 4 classification. The remaining eight miRNAs specifically differentiated node 1 classification. They were down-regulated in the HBV and HCV groups compared with the normal group (Fig. 3, cluster 1). Nineteen miRNAs differentiated node 2 classification (Table 2-2) and the hierarchical clustering using these miRNAs clearly differentiated the HBV and HCV groups (Fig. 3, cluster 2). There were 15 and 14 miRNAs that differentiated node 3 and 4 classifications, respectively (Table 2-2). Hierarchical clustering using these miRNAs revealed that these miRNAs differentiated CH-B and HCC-B as well as CH-C and HCC-C, respectively; 17 miRNAs were down-regulated in HCC, and six were up-regulated in HCC (Fig. 3, cluster 3).

These results indicate that there were two types of miRNAs—one associated with HBV and HCV infection (cluster 2), the other associated with the stages of liver disease (clusters 1 and 2) that were irrelevant to the differences in HBV and HCV infection.

Differential miRNAs and Their Candidate Target Genes and Signaling Pathways.

Differentially expressed miRNAs are shown in Table 3. In addition to the expression ratios of miRNAs in each group, the number of genes analyzed on the microarray predicted to be the target genes of miRNAs and that which actually showed significant (P < 0.05) differences in expression are also shown. Based on the frequencies and levels of expression of differential genes, the significance of regulation of these gene groups by miRNAs was evaluated using Hotelling T2 test (BRB ArrayTools) (Table 3). The representative candidate target genes and their signaling pathways by each miRNA were shown one by one (Table 3). The signaling pathways regulated by all differential miRNAs in each category of groups are shown in Table 4.

Eight miRNAs were down-regulated in the HBV and HCV groups compared with the normal group (Table 3-1; Fig. 3, cluster 1). These miRNAs were associated with an increased expression of genes related to cell adhesion, cell cycle, protein folding, and apoptosis (Tables 3-1, 4-1), and possibly with the common feature of CH irrespective of the differences in HBV and HCV infection.

Nineteen miRNAs clearly differentiated the HBV and HCV groups (Fig. 3, cluster 2, Table 3-2). Thirteen miRNAs exhibited a decreased expression in the HCV group, and six showed a decreased expression in the HBV group. miRNAs exhibiting a decreased expression in the HCV group regulate genes related to immune response, antigen presentation, cell cycle, proteasome, and lipid metabolism. On the other hand, those exhibiting a decreased expression in the HBV group regulate genes related to cell death, DNA damage and recombination, and transcription signals. These findings reflected the differences in the gene expression profile between CH-B and CH-C described (Tables 3-2, 4-2).10 Interestingly, although these miRNAs were HBV and HCV infection–specific, some of them were reported to be tumor-associated miRNAs, suggesting the possible involvement of infection-associated miRNAs in HCC development.

Table 3-1. Representative miRNAs That Were Commonly Repressed in CH-B, CH-C, HCC-B, and HCC-C Compared with Normal Liver (Cluster)
miRNAParametric P ValueRatio*No. of Significant Genes/Predicted Target GenesHotelling Test P ValueDifferentially Expressed Target Genes§Pathway of Regulated Genes
  • *

    Ratio of HCC-B, HCC-C, CH-B, and CH-C to normal.

  • The number of significant genes (P < 0.05) out of predicted target genes in which expression was evaluated in microarray.

  • Statistical assessment of presence of differentially expressed genes out of predicted target genes of miRNAs.

  • §

    Representative differentially expressed genes out of predicted target genes of miRNAs.

  • Representative pathway of differentially expressed genes out of predicted target genes of miRNAs.

hsa-miR-2197.3E-050.2825/1092.59E-04Glypican-3, ERP5, PLK2, HIRA, HMG2Regulatory T cell differentiation
     ACOX1Fatty acid beta-oxidation
     NF-X1MHC class II biosynthetic process
hsa-miR-3209.8E-050.5026/883.50E-06Vimentin, ALP (N-acetyltransferase-like), SEC61 beta, G-protein alpha-i2, Filamin AProtein kinase cascade
     Rac1, RhoGOrganelle organization and biogenesis
     Vinexin beta, Profilin I, Ca-ATPase3Actin cytoskeleton organization and biogenesis
hsa-miR-1542.7E-040.1522/705.40E-06OTR, NET1(TSPAN1), NAP1, Vimentin, PDIA3, cytochrome P-450 reductaseRegulation of apoptosis
     DLX2Morphogenesis
     GUAC, ACAT1Branched chain family amino acid catabolic process
hsa-miR-29c1.8E-030.5553/1331.00E-06FBX07, ASPP1, HSPA4, Cathepsin O, PDF, COL4A1, HSPA4, TIP30, CXADRCell-substrate adhesion
     NS1-BP, ALP (N-acetyltransferase-like), ACTR10, Beclin 1Transcription, DNA-dependent
     SMAD6, LTBR(TNFRSF3), ENPP7Apoptosis
hsa-miR-3385.2E-030.4630/1013.60E-06ID3, GATA-4, NFIA, FR-beta, CREST, HYOU1Developmental process
     G3ST1, CAD,Immune effector process
     FKBP12, LZIP, PDIA3, Schwannomin (NF2), CRESTImmune system process
hsa-miR-26a6.3E-030.7037/1192.64E-05LIG4, c-FLIP, GADD45 beta, DAPK1, PRDX4, LRP130Response to stimulus
     Cyclin E, ZDHHC6, Tx1, ATG8 (GATE-16), WASP, C1sDNA replication initiation
     COPG1Ion transport
hsa-miR-1268.1E-030.6527/1014.04E-03ANP32B (april), HSPA4, RLI, LIV-1 (SLC39A6), PTP-MEG2, CD97, DHPRRegulation of cellular protein metabolic process
     NFKBIA, NMI, MDH1, PDCD2Response to stress
     SMAD6, ATP6AP2, ANP32B (april), NMI, HSPA4Apoptosis
hsa-miR-3258.7E-030.2018/632.03E-04TRADD, CREST, NEDD8, annexin IV, GPX2, PDF, TNFAIP1Developmental process
     Glypican-3, ID1, PC-TP,Multicellular organismal development
     SNRPB (Sm-B)RNA splicing
Table 4-1. Pathway Analysis of Targeted Genes by miRNAs that Were Commonly Repressed in CH-B, CH-C, HCC-B, and HCC-C Compared with Normal Liver (Cluster 1)
No.Pathway NameP Value
Down-regulated miRNA in CH-B,HCC-B,CH-C and HCC-C (possibly up-regulating target genes) 
1Cell adhesion_Platelet-endothelium-leukocyte interactions1.11E-02
2Cell cycle_S phase2.18E-02
3Protein folding_Protein folding nucleus2.43E-02
4Cell cycle_G1-S3.07E-02
5Development_Cartilage development3.89E-02
6Protein folding_Folding in normal condition3.89E-02
7Proteolysis_Connective tissue degradation3.99E-02
8Proteolysis_Proteolysis in cell cycle and apoptosis4.31E-02
9Signal Transduction_BMP and GDF signaling5.81E-02
10Immune_Antigen presentation6.05E-02
Table 3-2. Differentially Expessed miRNA Between HCC-B, CH-B, and HCC-C, CH-C, and Their Representative Target Genes (Cluster 2)
miRNAParametric P ValueRatio*No. of Significant Genes/Predicted Target GenesHotelling Test P ValueDifferentially Expressed Target Genes§Pathway of Regulated Genes
  • *

    Ratio of HCC-B, CH-B, to HCC-C,CH-C.

  • The number of significant genes (p <0.05) out of predicted target genes in which expression was evaluated in microarray.

  • Statistical assesment of presence of differentially expressed genes out of predicted target genes of miRNAs.

  • §

    Representative differentially expressed genes out of predicted target genes of miRNAs.

  • Representative pathway of differentially expressed genes out of predicted target genes of miRNAs.

hsa-miR-1901.2E-052.0621/684.47E-02Chk1, C2orf25, VRK2, USP16, STAF65(gamma)Regulation of cell cycle
     AP1S2, RNASE4Mitotic cell cycle
     PPP2R1B, ARHGAP15, UBPYNegative regulation of apoptosis
hsa-miR-1342.3E-045.7411/583.40E-06VKDGC, SH2B, MALS-1, DDB2Multicellular organismal process
     BCRP1Regulation of viral reproduction
     DDB2Lipid biosynthetic process
hsa-miR-1512.8E-041.8212/626.41E-01RGS2, UFO, AK2, USP7G-protein signaling
     elF4G2, USP7Regulation of translation
     SLC22A7Organic anion transport
hsa-miR-1935.0E-041.6723/959.30E-01G-protein alpha-11, p130CAS, VAV-1, PDCD11Cell motility
     Colipase, ACSAEnergy coupled proton transport
     DCORIntracellular signaling cascade
hsa-miR-133b1.7E-032.4220/973.69E-02DDB2, Bcl-3, Cystatin BProteasomal protein catabolic process
     Rab-3, RAG1AP1, KCNH2, DCORRegulation of biological quality
     AL1B1Carbohydrate metabolic process
hsa-miR-324-5p2.9E-031.5127/1211.90E-06SKAP55, VAV-1, DDB2, E2A, NIP1Cellular developmental process
     MEMO (CGI-27), Rab-3Cellular structure morphogenesis
     COPG1, GPX3, OAZ2Glutathione metabolic process
hsa-miR-182*3.1E-032.2328/123< 1e-07Alpha-endosulfine, HCCR-2, Thioredoxin-like 2, TPT1, USP7Translation initiation in response to stress
     DDB2, TPT1Cellular developmental process
     JIP-1JNK cascade
hsa-miR-1054.6E-034.3818/684.74E-05Beta-2-microglobulin, HLA-B27Antigen processing and presentation
     PIMT, IL-17RCImmune response
     MHC class I, CDK9, ERG1, Desmocollin 3 
hsa-miR-2115.3E-0325.6110/562.00E-04PSMD5, SLC26A6Proteasomal protein catabolic process
hsa-miR-205.7E-031.5227/1135.28E-03Noelin, SC4MOL, Thioredoxin-like 2, CCL5, NALP3Regulation of apoptosis
     Hic-5/ARA55, USP16, MAP4, Ferroportin 1Positive regulation of cellular process
     TOP3A, PLRP1Oxygen transport
hsa-miR-1916.7E-031.3925/797.55E-04CDK9, GPS2, CLTA, LXR-alphaNucleic acid metabolic process
     ACSAAcetyl-CoA biosynthetic process
     UGCGL1, SGPP1Metal ion transport
hsa-miR-3408.5E-031.4817/813.73E-03FKBP12, DCOR,Calcium ion transport
     Gelsolin, VAV-1, ARF6Actin cytoskeleton organization and biogenesis
     HXK3Glucose catabolic process
hsa-miR-1948.7E-031.6713/745.90E-01Cyclin B1, SerglycinM phase of mitotic cell cycle
     PTE2Acyl-CoA metabolic process
     SLC7A6Carbohydrate utilization
hsa-miR-23a1.9E-040.4614/97< 1e-07RGL2, MANR, MEK1 (MAP2K1), Caspase-3, AZGP1Protein kinase cascade
     FRK, Pyk2(FAK2), CSE1LCellular developmental process
     AZGP1Defense response
hsa-miR-142-5p4.9E-040.4025/899.10E-06Sirtuin4, PAI2, PSAT, RIL, CDC34, SPRY1Metabotropic glutamate receptor
     E4BP4, DNAJC12, WWP1, PAIP1, PASK, rBATRegulation of gene expression
     VCAM1, CaMK I, WWP1, FHL3Cell-matrix adhesion
hsa-miR-34c5.1E-040.2031/1297.30E-06Diacylglycerol kinase, zeta, PLC-delta 1, ATP2C1, PAI2Manganese ion transport
     MLK3(MAP3K11), MEK1(MAP2K1), CDC25C, MRF-1, XPCProtein kinase cascade
     GNT-IVInflammatory cell apoptosis
hsa-miR-124b8.6E-040.3225/1207.10E-05E2F5, Rad51, Jagged1Muscle development
     MLK3(MAP3K11), RGS1Intracellular signaling cascade
     COL16A1MAPKKK cascade
hsa-let-7a1.0E-030.4528/1369.35E-04RAD51C, CoAA, hASH1, Cockayne syndrome B, Caspase-1, PP5Response to DNA damage stimulus
     PLC-delta 1, MANR, ACADVLFibroblast proliferation
     HGF, NGFCellular developmental process
hsa-miR-27a3.9E-030.5918/1081.19E-02COL16A1, RIL, RhoGDI gamma, ANP32B (april)Cytoskeleton organization and biogenesis
     VE-cadherin, NTH1, GATA-2, E4BP4Response to external stimulus
     RAD51CDNA recombination
Table 4-2. Pathway Analysis of Targeted Genes by Differentially Expressed miRNAs Between HBV-Related Liver Disease (CH-B,HCC-B) and HCV Related Liver Disease (CH-C,HCC-C Cluster 2)
No.Pathway NameP Value
Down-regulated miRNA in CH-C,HCC-C (possibly up-regulating target genes) 
1Immune_Phagosome in antigen presentation5.80E-04
2Muscle contraction1.05E-03
3Immune_Antigen presentation5.75E-03
4Cell cycle_Meiosis1.49E-02
5Reproduction_Male sex differentiation2.06E-02
6Cell adhesion_Platelet aggregation2.77E-02
7Transport_Synaptic vesicle exocytosis3.56E-02
8Inflammation_Kallikrein-kinin system3.73E-02
9Inflammation_IgE signaling4.10E-02
10Development_Skeletal muscle development5.02E-02
Down-regulated miRNA in CH-B,HCC-B (possibly up-regulating target genes) 
1Signal Transduction_Cholecystokinin signaling1.15E-04
2Inflammation_NK cell cytotoxicity5.29E-03
3Signal transduction_CREM pathway5.31E-03
4Reproduction_GnRH signaling pathway7.80E-03
5DNA damage_DBS repair1.02E-02
6Cell cycle_G2-M1.63E-02
7Development_Neuromuscular junction2.07E-02
8Apoptosis_Apoptosis mediated by external signals2.42E-02
9Reproduction_FSH-beta signaling pathway2.92E-02
10Cell adhesion_Amyloid proteins3.81E-02

Twenty-three miRNAs clearly differentiated CH and HCC that were irrelevant to the differences in HBV and HCV infection. Seventeen miRNAs were down-regulated in HCC that up-regulated cancer-associated pathways such as cell cycle, adhesion, proteolysis, transcription, translation, and the Wnt signaling pathway (Tables 3-3, 4-3). Six miRNAs were up-regulated in HCC that down-regulated all inflammation-mediated signaling pathways, potentially reflecting impaired antitumor immune response.

Table 3-3. Differentially Expessed miRNA Between CH and HCC and Their Representative Target Genes (Cluster 3)
miRNAParametric p-valueRatio*No. of Significant Genes/Predicted Target GenesHotelling Test P ValueDifferentially Expressed Target Genes§Pathway of Regulated Genes
  • *

    Ratio of HCC to CH.

  • The number of significant genes (P<0.05) out of predicted target genes in which expression was evaluated in microarray.

  • Statistical assesment of presence of differentially expressed genes out of predicted target genes of miRNAs.

  • §

    Representative differentially expressed genes out of predicted target genes of miRNAs.

  • Representative pathway of differentially expressed genes out of predicted target genes of miRNAs.

hsa-miR-1394.50E-060.4219/1062.70E-03Cyclin B1, DHX15, MCM5, Histone H2AMitotic cell cycle
     RBCK1, SYHHProtein catabolic process
     ILK, IGFBP7, SAFB, CTR9Response to external stimulus
hsa-miR-30a-3p2.50E-050.4926/1441.73E-02GGH, Pirin, ZNF207, Annexin VIIRegulation of oxidoreductase activity
     ILK, LTA4H, ABC50, GNPATCell-matrix adhesion
     DLC1Morphogenesis
hsa-miR-130a7.00E-050.5022/1081.07E-02SPHM, PPP2R5D, RHEB2, SPHMMitotic cell cycle
     MLK3(MAP3K11), Otubain1, TIMP4Protein modification process
     NRBPCell differentiation
hsa-miR-2233.40E-040.3914/906.52E-03Ephrin-A1, Midkine, FDPSCell morphogenesis
     K(+) channel, subfamily JNotch signaling pathway
hsa-miR-1873.55E-040.1216/666.76E-04HFE2, Otubain1Negative regulation of programmed cell death
     PRSS11, SUPT5H, RAG1AP1Developmental process
     PLOD3Mitochondrial ornithine transport
hsa-miR-200a6.86E-040.1820/1412.15E-02CDC25B, KAP3, CDK2AP2, CHKACell communication
     POLDDNA replication
     CPSF4RNA splicing
hsa-miR-17-3p8.42E-040.5828/1088.98E-04MLK3(MAP3K11), Tip60, ACBD6, DOC-1R, DAX1, RBCK1Protein kinase cascade
     WNT5A, 14-3-3 gamma, DHX15BMP signaling pathway
     HFE2, MCM5DNA recombination
hsa-miR-99a1.17E-030.5333/1639.52E-03Calpain small subunit, Thoredoxin-like 2, SurvivinCytokinesis
     IBP2, DNA-PK, KAP3,Intracellular signaling cascade
     NFE2L1, PARP-1, HDAC11Regulatory T cell differentiation
hsa-miR-200b1.57E-030.1824/1472.72E-02HSP47, HMG2, NRBPRegulation of cell cycle
     SNX17Cell motility
     Ephrin-A1Receptor protein signaling pathway
hsa-miR-125b1.82E-030.5526/1141.03E-01COL4A2, TIP30, HSP47, MSP58Cell adhesion
     MLK3(MAP3K11), ERK2 (MAPK1), ERK1 (MAPK3), PLOD3Nuclear translocation of MAPK
     Otubain1, SCN4A(SkM1)Ubiquitin-dependent protein catabolic process
hsa-miR-30e2.10E-030.6524/1514.30E-02Cyclin B1, XTP3B, GAK, Annexin VII, MIC2, NRBPMitotic cell cycle
     MSS4Protein localization
     S100A10Calcium ion transport
hsa-miR-199a*4.26E-030.3511/717.16E-02BUB3, Cyclin B1, LMNBRMitotic cell cycle
     PRAMECardiac muscle cell differentiation
hsa-miR-122a6.31E-030.5111/801.01E-03JAB1, APEX, Clathrin heavy chainBase-excision repair
     PARNTranslational initiation
     DDAH2Regulation of cellular respiration
hsa-miR-199a8.77E-030.3518/943.56E-02IL-13, MLK3(MAP3K11), CLK2, ACP33Protein amino acid phosphorylation
     PAFAH beta, SPA1, CLCN4Small GTPase mediated signal transduction
hsa-miR-3269.00E-030.5729/1472.25E-01Midkine, ENT1, IP3KA, PSMC5, ANCO-1Regulation of programmed cell death
     Thy-1, MCM6, Tip60, VILIP3Cell-matrix adhesion
     COMP, Cathepsin ABlood vessel development
hsa-miR-929.60E-030.8128/1402.47E-02TUBGCP2, Fibrillin 1, PIPKI gamma, KAP3Rho protein signal transduction
     SNX15, BCAT2LDL receptor and BCAA metabolism
     IGFBP7, FZD6, COPS6Adenosine receptor signaling pathway
hsa-miR-2213.40E-063.3416/673.59E-01Lck, Kallistatin, Neuromodulin, LFA-3, PA24A, AZGP1, MSH2Immune response-activating signal transduction
     KYNU, PMCA3DNA repair
hsa-miR-2226.50E-062.2318/851.59E-02Thrombospondin 1, Lck, MSH2, ATF-2, CITED2, KallistatinCell motility
     PGARTriacylglycerol metabolic process
     KYNUDNA replication
hsa-miR-3015.22E-051.9614/711.16E-01Beta-2-microglobulin, PPCKM, PRC, Fra-1, PPCKM, ACAT2Antigen processing and presentation
     BMPR1B, ARMER, EHM2, RBBP8Meiotic recombination
     Neuromodulin, LDLRCell motility
hsa-miR-217.67E-031.5719/811.86E-04Btk, Fra-1, MSH2, Collectrin, AdipophilinRegulation of T cell proliferation
     RNASE4, AGXT2L1Peptidyl-tyrosine phosphorylation
     SARDHNatural killer cell activation during immune response
hsa-miR-1832.46E-023.5113/863.36E-01Hdj-2, PEMT, Lck, MKP-5, Chondromodulin-I, ABCA8Cell differentiation
     IL-16, MTRR, SerRSMethionine biosynthetic process
hsa-miR-985.22E-021.3224/1302.95E-04ACAA2, LTB4DH, ACADVL, DECR, S14 protein,Fatty acid metabolic process
     Rapsyn, Kallistatin, ENPEP, Beta crystallin B1Multicellular organismal process
     CYP4F8Prostaglandin metabolic process
Table 4-3. The Pathway Analysis of Targeted Genes by Differentially Expressed miRNAs Between CH and HCC (Cluster 3)
No.Pathway NameP Value
 Down-regulated miRNA in HCC (possibly up-regulating target genes)
1Cytoskeleton_Spindle microtubules2.15E-03
2Transcription_Chromatin modification5.27E-03
3Proteolysis_Ubiquitin-proteasomal proteolysis6.43E-03
4Cell adhesion_Cell-matrix interactions7.30E-03
5Cell cycle_Meiosis7.83E-03
6DNA damage_Checkpoint1.69E-02
7Reproduction_Progesterone signaling1.94E-02
8Apoptosis_Apoptotic mitochondria3.14E-02
9Translation_Regulation of initiation4.22E-02
10Signal transduction_WNT signaling4.26E-02
 Up-regulated miRNA in HCC (possibly down-regulating target genes)
1Inflammation_IgE signaling1.05E-02
2Inflammation_Kallikrein-kinin system2.46E-02
3Inflammation_Innate inflammatory response2.51E-02
4Inflammation_Histamine signaling4.25E-02
5Inflammation_Neutrophil activation4.55E-02
6Chemotaxis4.68E-02
7Inflammation_IL-12,15,18 signaling5.16E-02
8Inflammation_NK cell cytotoxicity7.25E-02
9Cell cycle_G0-G17.53E-02
10Inflammation_Complement system7.72E-02

Relationship Between Expressions of Infection-Associated miRNA in Liver and Cultured Cells Infected with HBV and HCV.

To clarify whether the expression of infection-associated miRNA is regulated by HBV and HCV infection, we investigated the relationship between changes in miRNA in liver tissues and those in miRNA in Huh7.5 cells in which infectious HBV or HCV clones replicated. To evaluate the replication of each clones in Huh7.5 cells, we measured time-course changes in the amounts of HBV-DNA and HCV-RNA in Huh7.5 cells transfected with pHBV1.2 and JFH1-RNA, respectively, by RTD-PCR (Supplementary Fig. 1A). The expression of HBV proteins was examined by measuring the amount of HBeAg released in culture medium (Supplementary Fig. 1B). HCV protein expression was examined by evaluating the core protein expression after 48 hours by fluorescence immunostaining (Supplementary Fig. 1C). RNA was extracted from the Huh7.5 cells 48 hours after gene transfection, and miRNA expression pattern in the cells was compared with those in liver tissues. We found a strong correlation between differences in miRNA expression between liver tissues of the HBV and HCV groups, and those in miRNA expression between Huh7.5 cells transfected with HBV and HCV clones (r = 0.73, P = 0.0006) (Fig. 5). These results revealed that differences in the expression of infection-associated miRNA in the liver between the HBV and HCV groups are explained by changes in miRNA expression caused by HBV and HCV infections.

Figure 5.

Correlation between differences in miRNA expression between liver tissues infected with HBV and HCV and those in miRNA expression between cultured cell models of HBV and HCV infections. A total of 140 of 188 miRNAs were confirmed to be expressed in Huh7.5 cells. There was a significant correlation of infection-associated miRNA (closed lozenge) in vitro and in vivo (r = 0.73, P = 0.0006), but none for the other 121 miRNAs (open lozenge) (r = 0.02, P = 0.82).

Verification of Regulation of Candidate Target Genes by miRNA.

Small interfering RNAs (Ambion) specific to 13 miRNAs (has-miR-17*, has-miR-20a, has-miR-23a, has-miR-26a, has-miR-27a, has-miR-29c, has-miR-30a, has-miR-92, has-miR-126, has-miR-139, has-miR-187, has-miR-200a, and has-miR-223) showing significant differences in expression were transfected into Huh7 cells to examine loss of function of the miRNAs. Five miRNAs (has-miR-23a, has-miR-26a, has-miR-27a, has-miR-92, and has-miR-200a) showed a decreased expression by more than 50%. Precursor miRNAs of these miRNAs were also transfected into the cells to examine the gain of function of the miRNAs (Supplementary Fig. 2). It was confirmed that the expressions of target genes of the five miRNAs (LIG4 [by has-miR-26a]; RGL2 [by has-miR-23a]; Rad51C [by has-miR-27a]; KAP3, CDC25B, KAP3, CDK2AP2, POLD, and CPSF4 [by has-miR-200a]; and TUBGCP2, SNX15 and BCAT2 [by has-miR-92]) were increased by the suppression of the miRNAs induced by small interfering RNAs and were decreased by the overexpression of precursor miRNAs (Supplementary Fig. 3).

Discussion

miRNA plays an important role in various diseases such as infection and cancer.1–3 In this study, we examined miRNA expression profiles in normal liver and HCC, including nontumor lesions infected with HBV or HCV. Although the expression profiles of miRNAs in HCC have been reported,16–18 most of the studies were performed using a microarray system. Because we thought that miRNAs could not produce enough detection signals owing to their short length, we applied a highly sensitive and quantitative RTD-PCR method for miRNAs. Moreover, global gene expression in the same tissues was analyzed via cDNA microarray to examine whether the differentially expressed miRNAs could regulate their target genes. Because the absolute standard of miRNA is not available at present, and miRNA expression was compared within the samples and genes analyzed in this study, there might be possible errors when a larger number of samples and genes were analyzed.

Using these systems, we found that the expression profile in miRNAs was clearly different according to HBV and HCV infection for the first time. The differences were confirmed by the nonsupervised learning method, hierarchical clustering (Fig. 2A), and supervised learning methods based on SVM at an 87% accuracy (P < 0.001) (Table 2-1). As similarly described, the expression profile in miRNAs was significantly different according to the progression of liver disease (normal, CH, and HCC) in this study. The present CH and HCC expression data were derived from the same patient, and some microarray analyses suggested that the noncancerous liver tissue can predict the prognosis of HCC.19, 20 We examined whether the miRNA expression of paired samples was similar or independent using the Dunnett test12 (Supplementary Data). Our data indicated that miRNA expression profiling was more dependent on the disease condition than on the paired condition, although the issue of paired samples should be taken into account carefully.

Binary tree prediction analysis and detailed assessment of hierarchical clustering revealed two types of differential miRNAs, one associated with HBV and HCV infection, the other associated with the stages of liver disease that were irrelevant to the differences in HBV and HCV infection. We found that differences in miRNA expression between liver tissues with HBV and HCV (HBV/HCV) were strongly correlated with those in miRNA between cultured cell models of HBV and HCV infection (HBV/HCV) (r = 0.73 P = 0.0006) (Fig. 5). Thus, there exist HBV- and HCV-infection–specific miRNAs that potentially regulate viral replication and host gene signaling pathways in hepatocytes.

The pathway analysis of targeted genes by miRNAs revealed that 13 miRNAs exhibiting a decreased expression in the HCV group regulate genes related to immune response, antigen presentation, cell cycle, proteasome, and lipid metabolism. Six miRNAs showing a decreased expression in the HBV group regulate genes related to cell death, DNA damage and recombination, and transcription signals. These findings reflected differences in the gene expression profile between CH-B and CH-C as described.10 Many of the miRNAs were down-regulated in the HCV group rather than in the HBV group. It has been reported that human endogenous miRNAs may be involved in defense mechanisms, mainly against RNA viruses.21 On the other hand, it is suggested that endogenous miRNAs may be consumed and reduced by defense mechanisms, especially those against RNA viruses.

Although the expressions of these HBV- and HCV-infection–specific miRNAs were irrelevant to the differences in CH and HCC (Fig. 3, cluster 2), some of them have been reported to play pivotal roles in the occurrence of cancer. For example, has-let-7a regulates ras and c-myc genes,22 and has-miR-34 is involved in the p53 tumor suppressor pathway.23 These miRNAs were down-regulated in the HBV group, possibly participating in a more aggressive and malignant phenotype in HCC-B rather than in HCC-C. High expression of has-miR-191 was shown to be significantly associated with the worse survival in acute myeloid leukemia,24 and has-miR-191 was overexpressed in the HBV group compared with the HCV group. On the other hand, has-miR-133b, which was reported to be down-regulated in squamous cell carcinoma,25 was repressed in the HCV group compared with the HBV group. Some hematopoietic-specific miRNAs such as has-miR-142-5p were up-regulated in the HCV group. Therefore, these miRNAs were not only HBV and HCV infection–associated but also tumor-associated. These findings indicate different mechanisms of development of HCC infected with HBV and HCV (Fig. 6).

Figure 6.

Infection-associated and HCC-specific miRNAs and liver disease progression.

Following HCC development, common changes in miRNA expression between HCC-B and HCC-C appeared (Fig. 3, cluster 3). The 23 miRNAs mentioned above clearly differentiated CH and HCC that were irrelevant to the differences in HBV and HCV infections. Seventeen miRNAs were down-regulated in HCC, which up-regulated cancer-associated pathways. Six miRNAs were up-regulated in HCC that down-regulated all inflammation-mediated signaling pathways, potentially reflecting impaired antitumor immune response in HCC. These results suggest that common signaling pathways are involved in HCC development from CH, and that HBV- and HCV-specific miRNAs participate in generating HCC-specific miRNA expressions (Fig. 6). Therefore, these miRNAs might be good candidates for molecular targeting to prevent HCC occurrence, because they regulate a common signaling pathway underlying HCC-B and HCC-C development.

In conclusion, we showed that miRNAs are important mediators of HBV and HCV infections as well as liver disease progression. Further studies are needed to enable more detailed mechanistic analysis of the miRNAs identified here and to evaluate the usefulness of miRNAs as diagnostic/prognostic markers and potential therapeutic target molecules.

Acknowledgements

The authors thank Mikiko Nakamura and Nami Nishiyama for excellent technical assistance.

Ancillary

Advertisement