Different signaling pathways in the livers of patients with chronic hepatitis B or chronic hepatitis C


  • Potential conflict of interest: Nothing to report.


The clinical manifestations of chronic hepatitis B (CH-B) and chronic hepatitis C (CH-C) are different. We previously reported differences in the gene expression profiles of liver tissue infected with CH-B or CH-C; however, the signaling pathways underlying each condition have yet to be clarified. Using a newly constructed cDNA microarray consisting of 9614 clones selected from 256,550 tags of hepatic serial analysis of gene expression (SAGE) libraries, we compared the gene expression profiles of liver tissue from 24 CH-B patients with those of 23 CH-C patients. Laser capture microdissection was used to isolate hepatocytes from liver lobules and infiltrating lymphoid cells from the portal area, from 16 patients, for gene expression analysis. Furthermore, the comprehensive gene network was analyzed using SAGE libraries of CH-B and CH-C. Supervised and nonsupervised learning methods revealed that gene expression was correlated more with the infecting virus than any other clinical parameters such as histological stage and disease activity. Pro-apoptotic and DNA repair responses were predominant in CH-B with p53 and 14-3-3 interacting genes having an important role. In contrast, inflammatory and anti-apoptotic phenotypes were predominant in CH-C. These differences would evoke different oncogenic factors in CH-B and CH-C. In conclusion, we describe the different signaling pathways induced in the livers of patients with CH-B or CH-C. The results might be useful in guiding therapeutic strategies to prevent the development of hepatocellular carcinoma in cases of CH-B and CH-C. (HEPATOLOGY 2006;44:1122–1138.)

The human liver infected with hepatitis B virus (HBV) and hepatitis C virus (HCV) develops chronic hepatitis, cirrhosis, and in some instances, hepatocellular carcinoma (HCC).1–3 The virological features of these 2 viruses are completely different. HBV is a DNA virus that integrates into the host genome.4, 5 HBV proteins, which have been reported to have transcriptional transactivator activity, may be related to the occurrence of HCC.6–9 By contrast, HCV is a positive stranded RNA virus that replicates in the cytoplasm.2 There are some reports that HCV proteins localize to the nucleus or interact with nuclear proteins.10, 11 Nevertheless, both viruses infect the liver and cause chronic hepatitis, which is not distinguishable by histological examination or clinical manifestations.12 In chronic viral hepatitis, increased numbers of immunoregulatory cells infiltrate the liver, but the functional relevance of these cells to the pathogenesis of chronic hepatitis is not known.

We previously reported that the gene expression profiles in the livers of patients with chronic hepatitis B (CH-B) or chronic hepatitis C (CH-C) are different, and revealed some characteristic features of each disease.13 However, the independent expression profiles of infiltrated lymphocytes and hepatocytes have yet to be clarified, as do the detailed signaling pathways underlying these 2 conditions.

In this study, we investigated the signaling pathways underlying CH-B and CH-C using cDNA microarray and serial analysis of gene expression (SAGE) techniques. Using laser capture microdissection (LCM), we selectively isolated hepatocytes from liver lobules and infiltrating lymphoid cells from the portal area, from biopsy specimens, and analyzed their gene expression profiles.


CH-B, chronic hepatitis B; CH-C, chronic hepatitis C; SAGE, serial analysis of gene expression; HBV, hepatitis B virus; HCV, hepatitis C virus; HCC, hepatocellular carcinoma; GO, gene ontology; LCM, laser capture microdissection; ALT, alanine aminotransferase; aRNA, antisense RNA; CTL, cytotoxic lymphocyte; Cy, cyanine; EGFR, epidermal growth factor receptor; cDNA, complementary DNA; IFN, interferon; NF-κB, nuclear factor-κB; NK cells, natural killer cells.

Patients and Methods


The subjects were 27 patients with CH-B and 26 with CH-C at the Graduate School of Medicine, Kanazawa University Hospital, Japan, between 1999 and 2003 (Table 1). Informed consent was obtained from all patients and ethics approval for the study was obtained from the ethics committee for human genome/gene analysis research at Kanazawa University Graduate School of Medicine. Liver biopsy samples were taken from 24 CH-B patients and 23 CH-C patients, and were divided into 3 portions: one was immersed in formalin for histological assessment, another was immediately frozen in liquid nitrogen for further RNA isolation, and the final portion was frozen in OCT compound for LCM analysis and stored at −80°C until use. Tissue samples from the remaining 6 patients with HCC were surgically obtained from the noncancerous parts of the liver and immediately frozen in liquid nitrogen for SAGE analysis. For normal liver, surgically obtained tissue samples of 6 patients who showed no clinical signs of hepatitis were used, as described.13

Table 1. Characteristics of Patients, as Used for Analyses of Whole Liver Biopsy, LCM, and SAGE Samples
Patient No.VirusAgeSexALTAFViral load (LEG/mL, KIU/mL)HCV serotypeHBeAgLCM HepLCM Ly
  1. Abbreviations: na., not applicable; LCM, laser capture microdissection; ALT, alanine aminotransferase; SAGE, serial analysis of gene expression; A, activity; Hep., hepatocyte obtained by LCM; Ly., lymphocyte obtained by LCM; F, fibrosis; No., if the sample was obtained from the same patient, the new sample number is shown with the old one; HCV RNA was assayed by Amplicor Monitor Test (KIU/mL); HBV DNA was assayed by transcription-mediated amplification (LEG/mL).

Whole liver biopsy samples
LCM samples
40HBV68F41225.5na.+ +
41HBV29M14022>8.7na.+ +
42HBV40M8022<3.7na. +
43HBV45M83236.1na.+ +
48HCV67M8022114II +
49HCV73M7122>850II +
50HCV67M7022>851I +
51HCV59F4323>852I +
SAGE samples

The grading and staging of chronic hepatitis were histologically assessed according to the method described by Desmet et al.14 (Table 1). There were no significant differences in the degree of histological activity or staging, nor in the sex or age of patients with CH-B or CH-C (Table 1).

Treatment of Cultured Cells With Interferon-α.

Huh-7 cells were treated with recombinant interferon-α (IFN-α) (Schering-Plough Corp., Osaka, Japan) at a concentration of 1000 IU/mL for 6 hours, and were harvested for analysis of induced gene expression by cDNA microarray.

Preparation of cDNA Microarray Slides.

In addition to the in-house cDNA microarray slides consisting of 1080 cDNA clones as described,13, 15–19 we made a new cDNA microarray slide for a detailed analysis of the signaling pathways involved in metabolism and enzyme function in liver disease. Besides cDNA microarray analysis, a total of 256,550 tags were obtained from hepatic SAGE libraries (derived from normal liver, CH-C, CH-C related HCC, CH-B, and CH-B related HCC), including 52,149 unique tags. Among these, 16,916 tags with more than 2 hits were selected to avoid the effect of sequencing errors in the libraries. From these candidate genes, 9614 nonredundant clones were obtained from Incyte Genomics (Incyte Corp., Beverly, MA), Clontech (Nippon Becton Dickinson, Tokyo, Japan), and Invitrogen (Invitrogen Japan K.K., Tokyo, Japan). Each clone was sequence validated and PCR amplified by Dragon Genomics (Takara Bio, Otsu, Japan), and the cDNA microarray slides (Liver chip 10k) were constructed using SPBIO 2000 (Hitachi Software, Fukuoka, Japan) as previously described.13, 15–19

Laser Capture Microdissection.

Hepatocytes in liver lobules and infiltrated lymphoid cells in the portal area were isolated by LCM using a CRI-337 LCM system (Cell Robotics, Albuquerque, NM)18 (Fig. 1). Frozen liver biopsy specimens in OCT compound were sliced into sections 8 μm thick, immediately fixed in methanol for 5 minutes, and kept on dry ice. Tissue samples were quickly stained with toluidine blue and dissected. Around 500 lymphoid cells and a similar number of hepatocytes were excised from 3 slides and immersed in a denaturing solution. Dissection was completed within 5 minutes for each slide.

Figure 1.

Optimization of LCM and cDNA microarray analysis. (A) Toluidine blue staining of liver biopsy specimens before (left) and after (right) LCM. (B) Electrophoresis of isolated RNA using an Agilent 2001 bioanalyzer. (C) Two round-amplified aRNA from 102-104 excised hepatocytes. (D) Typical hybridization result from LCM samples. (E) Correlation of signal intensity between first and second amplified genes. Two values were significantly correlated (P < .001, r2 = .97) within 2-fold differences.

RNA Isolation and Antisense RNA Amplification.

Total RNA was isolated from liver biopsy samples using an RNA extraction kit (Micro RNA Extraction Kit, Stratagene, La Jolla, CA). Aliquots of total RNA (5 μg) were subjected to amplification with antisense RNA (aRNA) using a Message Amp aRNA kit (Ambion, Austin, TX) as recommended by the manufacturer. About 25 μg of aRNA was amplified from 5 μg of total RNA, assuming that 500-fold amplification of mRNA was obtained. Total RNA from LCM samples was isolated with a carrier nucleic acid (20 ng poly C) using RNAqueous-Micro (Ambion). The quality and degradation of the isolated RNA were estimated after electrophoresis using an Agilent 2001 bioanalyzer (Agilent Technologies, Palo Alto, CA) (Fig. 1B). RNA isolation typically yielded 20-40 ng total RNA from 500 cells. Half of the obtained RNA was amplified twice as described above to yield 20-40 μg aRNA. Antisense RNA (20 μg) was used for further labeling procedures. The optimum conditions of LCM and reproducibility of data were assessed repeatedly.

Hybridization on cDNA Microarray Slides and Image Analysis.

As a reference for each microarray analysis, aRNA samples prepared from the normal liver tissue from 1 of the patients were used. Test RNA samples fluorescently labeled with cyanine 5 (Cy5) and reference RNA labeled with Cy3 were used for microarray hybridization as described.13, 15–19 Quantitative assessment of the signals on the slides was carried out by scanning on a ScanArray 5000 (General Scanning, Watertown, MA) followed by image analysis using GenePix Pro 4.1 (Axon Instruments, Union City, CA) as described.

Processing of cDNA Microarray Data.

Hierarchical clustering of gene expression was performed by BRB-ArrayTools (http://linus.nci.nih.gov/BRB-ArrayTools.html). The filtered data were log-transferred, normalized, centered, and applied to the average linkage clustering with centered correlation. A class prediction was performed by compound covariate predictor incorporating genes that were differentially expressed at the P = .002 significance level as assessed by the random variance t test (BRB-ArrayTools). The univariate t test values for comparing the classes were used as the weights. The cross-validated misclassification rate was computed and at least 2,000 permutations were performed for a valid permutation P value. The Fisher and Kolmogorov-Smirnov tests were performed for gene ontology (GO) comparison (P < .005) (BRB-ArrayTools).

Pathway Analysis of Expression Data.

The pathway analysis of the differentially expressed genes was performed using MetaCore software suite (GeneGo, St. Joseph, MI). Possible networks were created according to the list of the differentially expressed genes using the MetaCore database, a unique, curated database of human protein-protein and protein-DNA interactions; transcription factors; and signaling, metabolic, and bioactive molecules. The P value was calculated as:

equation image

where N is total number of nodes in the MetaCore database, R the number of network objects corresponding to the genes list, n the total number of nodes in each small network generated from the genes list, and r the number of nodes with data in each small network generated from the genes list. Moreover, direct interactions among the differentially expressed genes were examined. Each connection represents a direct, experimentally confirmed, physical interaction.


Total RNA isolated from each of 3 patients with CH-B or CH-C was mixed to 200 μg in total, and polyadenylated RNA was extracted using a FastTrac mRNA Purification Kit (Invitrogen). The SAGE protocol was as described.20, 21 SAGE libraries were sequenced at random using an ABI Prism 377 DNA Sequencer and BigDye Terminator Cycle Sequencing Kit (PE Applied Biosystems, Foster City, CA). Sequenced files were analyzed with the SAGE version 1.00 software.

Quantitative Real-time Detection PCR.

We performed quantitative real-time detection PCR (RTD-PCR) using TaqMan Universal Master Mix (PE Applied Biosystems). Primer pairs and probes for MxA, IP10, IFI15, OAS2, GZMA, TP53, PDECGF, IFNG, DIABLO, FGFB, BGA2, CASP9, PEX5, ANGPT1, VEGF, and β-actin were obtained from TaqMan assay reagents library. Results were expressed as means ± SEM. Significance was tested by 1-way ANOVA with Bonferroni's methods and differences were considered statistically significant at P < .05.


Optimization of LCM and cDNA Microarray Analysis.

Before analysis of region-specific gene expression, the sensitivity and reliability of linear aRNA amplification was examined. The quality and degradation of the isolated RNA were estimated after electrophoresis using an Agilent 2001 bioanalyzer (Fig. 1B). We successfully amplified aRNA from 102-104 excised hepatocytes with 2 rounds of amplification (Fig. 1C). The estimated amount of isolated RNA from around 150 excised hepatocytes (Fig. 1A) was 5-10 ng, and 10-20 μg of aRNA was obtained by 2 rounds of amplification, assuming that a 25 × 104-fold amplification (500-fold by single amplification) was carried out. A typical hybridization result is shown in Fig. 1D. Cytotoxic T lymphocyte-associated protein 4 was predominantly expressed in liver-infiltrating lymphocytes, whereas albumin was predominantly expressed in hepatocytes (Fig. 1D). To determine whether multiple amplifications affected the original gene expression, the signal intensities of first- and second-round amplified genes were compared. There was a significant correlation between the 2 values (P < .001, r2 = .97), within a 2-fold difference (Fig. 1E), suggesting that the linear amplification procedure maintained the original level of gene expression.

Identification of Genes Differentially Expressed in Hepatocytes and Liver-Infiltrating Lymphocytes.

Pair-wise t test comparisons were applied and differentially expressed genes were identified in lymphocytes and hepatocytes in 4 patients with CH-B and 4 patients with CH-C (Supplementary Table 1-1). In hepatocytes, liver-specific proteins and enzymes such as fibrinogen, afamin, and cytochrome P450 were all expressed. In lymphocytes, cytokines, chemokines, and lymphocyte surface markers such as interleukin-7 receptor, chemokine (C-X-C motif) receptor 4, CD83 antigen, and CD69 antigen were all expressed (Supplementary Table 1-2). Hierarchical clustering analysis of gene expression in hepatocytes and liver-infiltrating lymphocytes demonstrated clear differences in gene expression (Fig. 2). Representative differentially expressed genes in lymphocytes and hepatocytes in CH-B and CH-C are summarized in Supplementary Tables 2-1, 2-2, 3-1, and 3-2.

Figure 2.

(A) Hierarchical clustering analysis of gene expression in hepatocytes and liver-infiltrating lymphocytes. Hep, hepatocyte; Ly, lymphocyte; B, hepatitis B; C, hepatitis C. (B) Hierarchical clustering analysis of 1,360 filtered genes (we excluded genes with an expression level within 1.5-fold of median value in more than 80% of samples) demonstrated more clear clusters of CH-B and CH-C.

Supervised and Nonsupervised Learning Methods to Classify Gene Expression Profiling According to Different Clinical Parameters.

To examine which clinical parameters contributed to the changes in gene expression, supervised and nonsupervised learning methods were applied to classify gene expression profiles. The gene expression profiles of whole liver biopsy specimens, obtained from 19 patients with CH-B and 18 with CH-C, were analyzed. Hierarchical clustering analysis; a nonsupervised learning method, using 9641 nonfiltered genes, clearly demonstrated 2 clusters in CH-B and CH-C with a few exceptions (data not shown). Hierarchical clustering analysis with 1360 filtered genes (we excluded genes with an expression level within 1.5-fold of the median value in more than 80% of samples) demonstrated clearer clusters in CH-B and CH-C (Fig. 2B). Supervised learning methods based on the compound covariate predictor revealed that, among various clinical parameters including etiology (HBV or HCV), histological stage (F1F2 or F3F4), activity (A0A1 or A2A3), age (≥50 or <50 years), and alanine aminotransferase (ALT) level at biopsy (≥80 or <80 IU/mL), only etiology significantly classified these patients (Table 2). Thus, HBV or HCV infection determines gene expression to a greater degree than any other clinical parameters, such as histological stage and disease activity.

Table 2. Supervised Learning Methods to Differentiate CHB and CHC
Classifier CategoryClinical GroupsTotal Number of CasesNumber of Cases MisclassifiedClassifier P ValuesNumber of Genes in the Classifiers (P < .002)
HBV versus HCVHBV191<0.001160
Histological stageF1F217100.40255
Histological activityA0A11360.173106
ALT at biopsy≥801470.20021

Differentially Expressed Genes in CH-B and CH-C Hepatic Lesions.

The 160 genes were differentially expressed in CH-B and CH-C by class prediction analysis (P < .005); representative genes (greater than 3-fold difference in t value) are listed in Tables 3 and 4. Based on the expression profiles of hepatocytes and lymphocytes isolated using LCM, genes expressed in both hepatocytes and lymphocytes are described as Hep/Ly (Tables 3 and 4). Genes expressed at a significantly greater level in hepatocytes than lymphocytes were described as Hep. Genes expressed at a significantly greater level in lymphocytes than hepatocytes were described as Ly. In CH-B, genes involved in cell cycle arrest and induction of apoptosis were preferentially expressed. Several hepatocyte-specific and apoptosis-inducing genes such as Diablo homolog (cytochrome c/apaf-1/caspase-9 pathway activator) and BCL2-associated athanogene 2 (inhibitor of heat shock protein 70) were upregulated (Table 3, Fig. 7). In CH-C, cell cycle accelerating, immune-related, and antigen-presenting genes were preferentially upregulated. Many type 1 IFN-induced genes such as IFN-α-inducible protein 27 and IFN-α-inducible protein (clone IFI-15K) were upregulated in CH-C. The induction of these genes was confirmed by examining gene expression in Huh-7 cells treated with recombinant IFN-α (Tables 3 and 4, Fig. 7).

Table 3. Differentially Upregulated Genes in Liver of Chronic Hepatitis B
GeneGenBank IDP Valuet Value HBV/HCV*Hep/LyGO: Molecular Function
  • *

    The univariate t-statistics for comparing the classes are used as the weights.

  • 3.9-fold induction,

  • 7.7-fold induction, and

  • §

    1.8-fold induction by IFN-α in Huh-7 cells

Viral genome     
HBV-coreX015870.0006.69HepViral genome
Cell cycle and growth related     
V-ets erythroblastosis virus E26 oncogene homolog 2NM_0052390.0013.97Hep/Lyskeletal development
RAP2A, member of RAS oncogene familyAI6983760.0003.91Hep/Lysignal transduction
Melanoma antigen, family C, 1NM_0054620.0013.76Hep/Lyregulation of transcription
Cell division cycle 27NM_0012560.0013.54Hep/Lycell proliferation
Cyclin HNM_0012390.0003.10Hep/LyDNA repair
Immune response     
Interferon regulatory factor 6NM_0061470.0003.80Hepregulation of transcription, DNA-dependent
Proteoglycan 2, bone marrowR283360.0013.65Hep/Lydefense response to bacteria
Chemokine (C-C motif) ligand 16AW8271470.0013.49Hep/Lychemokine activity
Janus kinase 2 (a protein tyrosine kinase)NM_0049720.0013.48LyJAK-STAT cascade
Chemokine (C-X-C motif) receptor 3NM_0015040.0003.03Hep/LyG-protein coupled receptor protein signaling pathway
Cell death     
BCL2-associated athanogene 2NM_0042820.0003.95Hepapoptosis
Fas (TNFRSF6) associated factor 1AA8318370.0013.74Hep/Lyapoptosis
Proline dehydrogenase (oxidase) 1R885910.0003.73Hep/Lyinduction of apoptosis by oxidative stress
Caspase 9, apoptosis-related cysteine proteaseNM_0329960.0013.58Hep/Lyapoptotic program
Purinergic receptor P2X, ligand-gated ion channel, 1NM_0025580.0033.52Hep/Lyapoptosis
Tumor suppressing subtransferable candidate 1NM_0033100.0023.35Hep/Lyapoptosis
Tumor necrosis factor (ligand) superfamily, member 11NM_0330120.0023.25Hepcell differentiation
Diablo homolog (Drosophila)NM_0198870.0043.04Hepapoptosis
Cell communication     
Nexilin (F actin binding protein)NM_1445730.0004.15Hep/Lyunknown
Neurogranin (protein kinase C substrate, RC3)NM_0061760.0004.09Hepsignal transduction
Collagen, type XV, alpha 1NM_0018550.0004.08Hep/Lyextracellular matrix
Chromogranin B (secretogranin 1)NM_0018190.0013.47Hep/Lyhormone activity
Prostaglandin I2 (prostacyclin) receptor (IP)NM_0009600.0013.42LyG-protein signaling
Integral membrane protein 2CNM_0309260.0023.36Lyintegral to membrane
Sperm autoantigenic protein 17NM_0174250.0023.26Hep/LycAMP-dependent protein kinase regulator activity
Talin 2AF007154 3.18Lycell adhesion
Cadherin 16, KSP-cadherinAI2413190.0033.11Hepcell adhesion
Syntaxin binding protein 6 (amisyn)AA2814490.0043.03Lycell adhesion
Stress response     
RAD51-like 1 (S. cerevisiae)NM_0028770.0003.78Hep/LyDNA repair
Metallothionein 1XBC0538820.0013.44Hepelectron transport
Siah-interacting proteinAA0693220.0023.08Hep/Lyubiquitin cycle
Metallothionein 2ANM_0059530.0043.03Hepcopper ion homeostasis
F-box and leucine-rich repeat protein 2NM_0121570.0003.01Hep/Lyubiquitin cycle
Wolf-Hirschhorn syndrome candidate 1NM_1333350.0014.51Hep/Lymorphogenesis
Homeo box B2AI2920430.0013.87Hep/Lydevelopment
Neurogenic differentiation 1NM_0025000.0003.38Hep/Lycell differentiation
Opiate receptor-like 1NM_0009130.0043.29Hep/LyG-protein coupled receptor protein signaling pathway
Wingless-type MMTV integration site family, member 2BNM_0244940.0023.14Hep/Lyfrizzled-2 signaling pathway
Cell motility     
Oligophrenin 1R819420.0013.80Hep/Lyrho GTPase activator activity
ATP-binding cassette, subfamily C, member 9H161930.0043.06Heptransporters
Sodium channel, voltage gated, type VIII, alphaNM_0141910.0043.78Hep/Lycation transport
HMT1 hnRNP methyltransferase-like 6 (S. cerevisiae)NM_0181370.0014.44Hep/Lys-adenosylmethionine-dependent methyltransferase
Chymotrypsin-likeNM_0019070.0013.74Hep/Lynegative regulation of blood coagula
Aspartoacylase (aminocyclase) 3§NM_0806580.0053.26Hep/Lymetabolism
Transcription and signal transduction     
Hepatocyte nuclear factor 4, gammaAW2730650.0004.38Hep/Lyregulation of transcription
Nuclear receptor coactivator 6NM_0140710.0003.98Hep/LyDNA recombination
Protein kinase C, gammaNM_0027390.0013.88Hep/Lyintracellular signaling cascade
T-box 2NM_0059940.0003.82Hep/Lydevelopment
Zinc finger protein 167NM_0186510.0033.49Hep/Lyregulation of transcription, DNA-dependent
Small nuclear ribonucleoprotein polypeptide AAI4918620.0023.37Hep/Lyintracellular signaling cascade
Zinc finger protein 266NM_1980580.0023.03Lyregulation of transcription, DNA-dependent
Table 4. Differentially Upregulated Genes in Liver of Chronic Hepatitis C
GeneGenBank IDP Valuet Value HCV/HBVHep/LyIFN inducedGO: biological process
Cell cycle and growth related      
Hect domain and RLD 5NM_0163230.0004.50Hep/Ly7.7regulation of cyclin dependent protein kinase activity
Inhibitor of growth family, member 4NM_1982870.0013.50Hep/Ly grow arest
Phosphoinositide-3-kinase, class 3AI4461840.0013.42Hep/Ly inositol or phosphatidylinositol kinase activity
Non-metastatic cells 1, protein (NM23A) expressed inNM_0002690.0023.28Hep CTP biosynthesis
Mitogen-activated protein kinase kinase kinase 10AI9916210.0033.23Hep/Ly JNK cascade
Immune responces      
Interferon, alpha-inducible protein 27NM_0055320.0006.29Hep2.4response to pest, pathogen or paras
Interferon, alpha-inducible protein (clone IFI-15K)NM_0051010.0004.65Hep/Ly27.9cell-cell signaling
Myxovirus (influenza virus) resistance 1NM_0024620.0004.28Hep/Ly49.9 
Cold autoinflammatory syndrome 1NM_1833950.0004.14Ly inflammatory response
Interferon-stimulated transcription factor 3, gamma 48kDNM_0060840.0003.89Hep/Ly1.8immune response
Beta-2-microglobulinNM_0040480.0013.63Hep/Ly2.7antigen presentation, endogenous antigen
2′-5′-oligoadenylate synthetase 2 (69-71 kD)AA7311480.0013.49Hep/Ly3.3immune response
Interferon-induced protein 44-likeNM_0068200.0013.42Ly4.5immune response
Apolipoprotein L, 3AW0027660.0033.23Ly inflammatory response
Immunoglobulin kappa constantBC0627320.0043.04Ly immune response
Cell death      
Defender against cell death 1NM_0013440.0004.11Hep/Ly apoptosis
HIV-1 Tat interactive protein 2, 30kDaNM_0064100.0043.03Hep/Ly induction of apoptosis
Cell communication      
Major histocompatibility complex, class I, CNM_0021170.0013.74Hep/Ly antigen presentation
CD97 antigenNM_0784810.0013.72Ly cell adhesion
Major histocompatibility complex, class I, BNM_0055140.0023.38Hep/Ly1.9antigen presentation
Carcinoembryonic antigen-related cell adhesion molecule 5NM_0043630.0023.30Hep/Ly integral to plasma membrane
Major histocompatibility complex, class II, DQ beta 1NM_0021230.0023.25Ly antigen presentation
Major histocompatibility complex, class II, DR beta 4NM_0225550.0023.25Hep/Ly antigen presentation
Dystroglycan 1 (dystrophin-associated glycoprotein 1)AI6840760.0033.14Hep/Ly extracellular matrix
Dipeptidylpeptidase 6NM_1307970.0043.11Hep/Ly integral to membrane
Ubiquitin and proteasome system      
Proteasome (prosome, macropain) subunit, beta type, 8U174960.0004.55Hep/Ly immune response
Ubiquitin DNM_0063980.0033.13Ly2.1antimicrobial humoral response
Proteasome (prosome, macropain) 26S subunit, non-ATPase, 2NM_0028080.0043.05Hep/Ly regulation of cell cycle
Eukaryotic translation elongation factor 1 beta 2AI2625060.0004.46Ly protein biosynthesis
Eukaryotic translation initiation factor 1A, Y-linkedNM_0046810.0033.19Hep/Ly5.3protein biosynthesis
Lipid metabolism      
Diacylglycerol O-acyltransferase homolog 1 (mouse)NM_0120790.0023.31Hep/Ly O-acyltransferase activity
24-dehydrocholesterol reductaseNM_0147620.0033.19Hep cholesterol biosynthesis
Carnitine palmitoyltransferase IINM_0000980.0053.01Hep/Ly fatty acid beta-oxidation
Nucleotide metabolism      
Adenosine deaminase, RNA-specificNM_0158410.0013.46Hep/Ly RNA editing
Topoisomerase (DNA) IJ032500.0033.22Hep/Ly DNA unwinding
THO complex 1L365290.0033.15Hep/Ly nuclear mRNA splicing, via spliceosome
Karyopherin alpha 3 (importin alpha 4)NM_0022670.0033.14Hep/Ly NLS-bearing substrate-nucleus import
Nicotinamide nucleotide adenylyltransferase 1NM_0227870.0043.06Hep/Ly NAD biosynthesis
Nuclear autoantigenic sperm protein (histone-binding)M978560.0053.00Hep/Ly DNA packaging
Ribonucleotide reductase M2 polypeptideNM_0010340.0053.00Ly DNA replication
G protein binding protein      
Regulator of G-protein signalling 10NM_0029250.0023.38Hep/Ly signal transduction
Transcription and signal transduction      
Staphylococcal nuclease domain containing 1NM_0143900.0004.60Hep/Ly development
Ring-box 1NM_0142480.0013.61Ly protein ubiquitination
TrophininNM_1775580.0013.44Ly embryo implantation
Forkhead box F1AI4533330.0013.18Hep/Ly2.5regulation of transcription, DNA-dependent
Nuclear antigen Sp100M606180.0033.02Hep/Ly5.8regulation of transcription, DNA-dependent
Zinc finger protein 211NM_1988550.0043.02Ly regulation of transcription, DNA-dependent
GA binding protein transcription factor, beta subunit 2, 47kDaNM_1814270.0043.02Hep/Ly regulation of transcription, DNA-dependent
LIM protein (similar to rat protein kinase C-binding enigma)AI4455920.0043.01Hep/Ly heart development
Hematopoietic cell-specific Lyn substrate 1NM_0053350.0053.00Ly intracellular signaling cascade
ADP-ribosylation factor 5M575670.0053.00Hep/Ly intracellular protein transport

The frequent pathway processes observed in CH-B and CH-C using MetaCore are shown in Table 5. Induction of genes related to apoptosis (caspase activation via cytochrome C), transcription, and fibrosis (intermediate filament-based process and TGF-β receptor signaling pathway) were upregulated in CH-B, whereas genes related to immune reaction (defense response, antigen presentation, Golgi vesicle transport, and ubiquitin cycle), lipid metabolism (regulation of cholesterol absorption), and epidermal growth factor receptor (EGFR) signaling were upregulated in CH-C. This suggests that there are different signaling pathways in CH-B and CH-C.

Table 5. Pathway Analysis
Frequent Pathway ProcessP Value
Whole liver tissue in CHB (n = 19)
Caspase activation via cytochrome c7.04E-11
Regulation of transcription, DNA-dependent1.66E-12
Intermediate filament-based process1.24E-07
Calcium ion transport9.08E-08
Regulation of blood pressure2.94E-07
Protein amino acid phosphorylation4.04E-07
Regulation of angiogenesis5.35E-09
TGF-beta receptor signaling pathway8.08E-11
Whole liver tissue in CHC (n = 20)
Defense response3.27E-06
Antigen presentation, endogenous antigen6.79E-06
Golgi vesicle transport5.22E-07
Lipid catabolism6.61E-06
Regulation of cell cycle2.43E-08
Regulation of cholesterol absorption1.02E-05
EGF receptor signaling pathway1.59E-09
Ubiquitin cycle4.71E-05

Go Comparison of Expressed Genes in CH-B and CH-C Hepatic Lesions.

The analysis of differentially expressed genes could underestimate the presence of mean full signaling pathways that were coordinately upregulated or downregulated, with subtle differences at an individual gene level. The biological significance of these coordinately regulated signaling pathways has recently been demonstrated.22 Therefore, we applied the GO comparison tool to expressed genes in CH-B and CH-C hepatic lesions. The comparison tool provided a list of GO categories that were coordinately regulated between CH-B and CH-C.

In accordance with pathway analysis, antigen-presenting major histocompatibility complex molecules and IFN-α-induced genes were preferentially upregulated in CH-C (Table 6, Fig. 3). Genes related to apoptosis, DNA repair and cell death were upregulated in CH-B. DNA repair and apoptosis-related transcription factors were upregulated in CH-B, whereas anti-apoptosis and cell proliferation-related transcription factors were upregulated in CH-C. Platelet activating factor was upregulated in CH-C. As for metabolism-related gene regulation, peroxisome-associated genes were upregulated in CH-B, whereas cholesterol biosynthesis was upregulated in CH-C.

Table 6. Gene Ontology Comparison
GO DescriptionNumber of GenesLS Permutation (P Value)KS Permutation (P Value)HBVHCVReference
Whole liver tissue      
Antigen presenting150.001050.0341.011.490.81
IFN-alpha induced71<1 × 10−50.0001.492.091.16
Cell death340.0050.0191.351.150.99
DNA repair620.0050.0411.511.101.11
G1/S transition of mitotic cell cycle180.0010.0091.251.411.23
Transcription factor binding740.0170.0011.331.331.30
Cholesterol biosynthesis120.0290.0021.111.441.30
Peptidyl-prolyl cis-trans isomerase activity90.0020.0011.311.481.15
Single-stranded DNA binding160.0190.0031.851.341.27
IFN-alpha induced770.0040.1461.625.771.35
Immunological synapse120.0020.0036.383.783.31
Induction of apoptosis via deathdomain receptors70.0040.0181.531.021.07
Figure 3.

One-way hierarchical clustering of whole liver samples with representative genes (P < .05) included in each GO category which was significantly different in CH-B and CH-C (P < .005). Green text denotes genes expressed predominantly in hepatocytes, and blue text denotes genes expressed predominantly in lymphocytes.

To investigate these findings in more detail, lymphocytes and hepatocytes were separately isolated by LCM and their gene expression was examined (Table 6, Fig. 4A, Fig. 7). Cyclophilin A and cyclophilin C, encoding peptidyl-prolyl cis-trans isomerases, were upregulated in CH-C. A recent report describes inhibition of HCV replication in Huh-7 cells by cyclophilin.23, 24 The upregulation of ssDNA-binding genes, such as p53 and RAD, and the relative downregulation of mitochondrial genes in hepatocytes, in CH-B, reflect a strong DNA damage response inducing apoptosis. Many IFN-α-induced genes were upregulated in hepatocytes rather than lymphocytes in CH-C.

Figure 4.

(A) One-way hierarchical clustering of LCM samples with representative genes (P < .05). (B) One-way hierarchical clustering of liver-infiltrating lymphocytes, featuring specific gene sets of immune function. (c) One-way hierarchical clustering of whole liver sample gene sets of chemokines.

CD4, CD8, linker for activation of T cells, and pro-apoptotic genes were upregulated in lymphocytes in CH-B. Despite the activated T cell responses in CH-B, chemokine expression was induced more in the lymphocytes in CH-C than lymphocytes in CH-B (Fig. 4A). To examine the functional role of liver-infiltrating lymphocytes further, LCM samples were also obtained from 4 more patients with CH-B and 4 with CH-C. Gene expression was compared for lymphocyte subsets (84 CD markers, including 26 T cell makers, 21 B cell markers, 16 myeloid cell markers, 11 NK cell markers, and 12 AD markers). Among these, many T cell markers and Th1 cytokines were significantly more upregulated in CH-B than in CH-C lymphocytes. Conversely, B cell marker, Th2 cytokines, and chemokines were preferentially induced in CH-C (Fig. 4B-C). The differences in immune reaction in CH-B and CH-C may be a reflection of their different pathogenesis.

Detailed Gene Network Analysis of Differentially Expressed Genes in CH-B and CH-C.

To obtain a detailed and comprehensive gene network underlying CH-B and CH-C, SAGE data were integrated with those from cDNA microarray analysis. We applied 361 upregulated genes in CH-B (P < .05) and 344 in CH-C (P < .05), obtained from cDNA microarray analysis, and 1924 upregulated genes in CH-B (more than 5-fold tag count differences) and 1780 in CH-C, obtained from SAGE analysis, to the construction of the knowledge-based gene network. To find the gene network among these induced genes, published results of interaction of individual genes were integrated with these results using MetaCore software. Direct interactions between individual genes were searched for. The gene network of these differentially expressed genes formed a complex interaction of individual genes; however, representative signaling pathways underlying CH-B or CH-C were identified (Fig. 5).

Figure 5.

(A) Gene network of differentially expressed genes in CH-B. (B) Gene network of differentially expressed genes in CH-C. Core transcription factors are represented by black ovals. Green ovals show genes expressed predominantly in hepatocytes and blue ovals show genes expressed predominantly in lymphocytes.

In CH-B, p53 and 14-3-3 interacting genes might play an important role in the induced signaling pathways. Transcriptional factors such as CCAAT/enhancer binding protein (C/EBP), c-JUN, and cAMP-responsive element binding protein 1 (CREB1) are possibly also important molecules regulating these signaling pathways. These molecules induced apoptosis and activated transcription and oncogenes. Such activation might activate peroxisomes in CH-B (Fig. 5). In CH-C, type 1-IFN signaling (ISGF3/STAT1) might play a major role in the induced signaling pathways. The activation of the NF-κB and epidermal growth factor receptor (EGFR) signaling pathways may reflect liver inflammation and regeneration. These activations could lead to activation of liver X receptor/retinoid X receptor (LXR/RXR), a regulator of lipid metabolism.

Based on the database of MetaCore, which covers the entire regulation of the transcriptional factors, transcriptional regulation of differentially expressed genes was analyzed (Table 7). Transcription of mothers against decapentaplegic homolog 3 (SMAD 3), activator protein-1 (AP-1), p53, CREB1, and sterol regulatory element binding transcription factor 1 (SREB-1) was induced in CH-B, whereas NF-κB, IRF-1, STAT1, and retinoid acid receptor-alpha (RARα) signaling pathways were induced in CH-C. These differences fundamentally explain the different signaling pathways in CH-B and CH-C.

Table 7. Transcription Regulation
 Frequent pathway processP value
 Chronic hepatitis B
1Mothers against decapentaplegic homolog 3 (SMAD3)5.25E-36
2Activator protein-1 (AP-1)4.24E-33
4cAMP-responsive element binding protein 1 (CREB1)2.39E-32
5v-ets erythroblastosis virus E26 oncogene homolog 1 (ETS1)3.38E-32
6Sterol regulatory element binding transcription factor 1 (SREBP1)6.73E-32
7Transcription factor binding to IGHM enhancer 3 (TFE3)9.48E-32
8Signal transducer and activator of transcription 3 (STAT3)1.33E-31
9v-ets erythroblastosis virus E26 oncogene homolog 2 (ETS2)1.88E-31
10Transcription factor 7/Lymphoid enhancer binding factor 1 [Tcf(ref)]1.88E-31
 Chronic hepatitis C 
1Nuclear factor of κ light polypeptide gene enhancer in B-cells 1 (NF-κB)1.32E-35
2Interferon regulatory factor 1 (IRF1)4.34E-33
3Splicing factor 1(SF1)9.17E-33
4Signal transducer and activator of transcription 1 (STAT1)1.28E-32
5Retinoid acid receptor- (RAR)1.81E-32
6Nuclear factor of κ light polypeptide gene enhancer in B-cells 2 (RelA)3.56E-32
7Vitamin D receptor (VDR)5.00E-32
8Wilms tumor 1(WT1)7.02E-32
9Sterol regulatory element binding transcription factor 2 (SREBP2)9.84E-32
10Epidermal growth factor receptor (EGFR)1.92E-31

To examine whether these differences in gene expression contribute the different mechanism of hepatocarcinogenesis, we compared the angiogenic factors in CH-B and CH-C. The hierarchical clustering of patients using 34 angiogenesis-related genes obtained from cDNA microarray analysis, significantly clustered patients into 2 groups of CH-B or CH-C (P = .0001) (Fig. 6A). In CH-B, VEGF-family genes, FGF, and the angiopoietin family were induced by several transcriptional factors including AP-1, c-fos, and STAT3, which were all strongly upregulated. In CH-C, inflammation-related angiogenic factors such as IL-8, IL-18, and PDGF1, induced by NF-κB, were also upregulated (Fig. 6B, Fig. 7). Thus, CH-B and CH-C showed different angiogenic properties, which implied that the tumorigenic process in CH-B and CH-C may differ.

Figure 6.

(A) Hierarchical clustering of whole liver samples using angiogenic genes. (B) Gene network of angiogenic genes in CH-B and CH-C.

Quantitative RTD-PCR.

We performed quantitative real-time detection PCR (RTD-PCR) using 15 TaqMan probes. The results of RTD-PCR on whole liver biopsy and LCM samples are shown in Fig. 7. In CH-B, apoptosis-inducing genes such as CASP9, IFNG, GZMA, TP53, BGA2, and DIABLO were upregulated. In CH-C, IFN-α-induced genes and chemokines such as MxA, IFI15, OAS2, and IP10 were upregulated. Angiogenic factors such as FGFB, ANGPT1, and VEGF were upregulated in CH-B, and another angiogenic factor, PDECGF, was upregulated in CH-C. The results are consistent with those from the cDNA microarray.

Figure 7.

Quantitative real-time detection PCR (RTD-PCR) using 15 TaqMan probes. The results of whole liver biopsy (HBV; 19 samples of CH-B, HCV; 18 samples of CH-C and N; 6 samples of normal liver) and LCM samples (HBV-H; 4 samples of hepatocyte in CH-B, HCV-H; 4 samples of hepatocyte in CH-C, HBV-Ly; 8 samples of lymphocyte in CH-B, HCV-Ly; 8 samples of lymphocyte in CH-C) were shown. *P < .05, **P < .01.


The biological activity of viral coding polyproteins of HBV and HCV has been extensively investigated in cell lines and in transgenic mouse models. For example, accumulated evidence shows HBV-X protein to be a transcriptional transactivator that interacts with p53 tumor suppressor protein, modulating its signaling pathway.9, 25 The transgenic mouse model with overexpression of HCV polyproteins in the liver develops steatosis and HCC.26, 27 However, these findings have not been well evaluated in clinical samples.

Using in-house cDNA microarray analysis of 1080 genes, we previously reported differing gene expression profiles of liver tissue from patients with CH-B and CH-C.13 However, the detailed signaling pathways underlying these diseases needed further clarification. In this study, we constructed a new microarray slide, liver chip 10 K, consisting of 9614 clones which were selected from unique tag sequences in our hepatic SAGE libraries, including 667,067 tag sequences (manuscript in preparation), for the purpose of analyzing gene expression profiling in liver disease. We analyzed the gene expression profiles of whole liver biopsy specimens obtained from 37 patients with CH-B and CH-C. In addition, we selectively isolated liver-infiltrating lymphocytes (16 samples) and hepatocytes (8 samples) from liver biopsy specimens using LCM (Fig. 1D) and analyzed their gene expression. Furthermore, SAGE data were obtained from pooled samples of 3 CH-B or 3 CH-C patients, and their gene expression data were integrated to reveal the comprehensive, detailed gene network involved in CH-B and CH-C, respectively.

Hierarchical clustering analysis of 37 patients grouped these patients into 2 groups with CH-B or CH-C, with a few exceptions. Moreover, gene prediction analysis significantly discriminated between CH-B and CH-C patients (P < .001). HBV or HCV was the only factor significantly involved in patient classification, and other factors such as histological stage, disease activity, age, and ALT levels were not significantly associated with the classification of these patients. This indicates that virus type, whether HBV or HCV, influences liver gene expression to a greater degree than any other clinical parameter, such as degree of fibrosis or inflammation (Table 2).

The pathway analysis and GO comparison in CH-B and CH-C using whole liver biopsy revealed that antigen-presenting genes, IFN-α-induced genes, G1/S transition genes, and cholesterol biosynthesis and platelet-derived factors were upregulated in CH-C, whereas genes related to cell death, DNA repair, and peroxisomes were upregulated in CH-B (Tables 5-6, Fig. 3). The association of HCV infection with steatosis in the liver in CH-C has been reported.28, 29 There might also be an association between HBV replication and peroxisomal activation, as reported using hepatoma-derived cell lines.30, 31 We combined SAGE data with those from cDNA microarray analysis and constructed the detailed and comprehensive gene network underlying CH-B and CH-C. In CH-B, p53-mediated and 14-3-3-mediated pro-apoptotic signaling; transcription factors such as AP-1, C/EBP, c-JUN, and CREB1; and oncogenes and peroxisomes were activated (Fig. 5). In CH-C, type 1-IFN (ISGF3/STAT1), NF-κB, EGFR, and LXR/RXR signaling were activated.

Lesion-specific gene expression analysis by LCM revealed more precise differences in gene expression between CH-B and CH-C (Fig. 4, Fig. 7), although a larger number of samples will be needed to reach concrete conclusions. Interestingly, many IFN-α-induced genes were upregulated in hepatocytes, but not in lymphocytes, in CH-C. On the other hand, DNA repair genes such as p53 and RAD were induced in hepatocytes in CH-B. Detailed analysis of lymphocyte markers revealed Th1-dominant responses in the liver in CH-B and Th2-dominant responses in the liver in CH-C.

Despite greater lymphocyte infiltration and homing in the liver, a weak T cell response and no T cell accumulation were observed in CH-C.32, 33 These contributed to the induction of various chemokines, cytokines, and growth factors, which may lead to cell proliferation and angiogenesis in CH-C. Surprisingly, gene expression profiling of angiogenic factors revealed clear differences in CH-B and CH-C. Many of the chemokines involved in angiogenesis are independent of VEGF-mediated or angiopoietin-mediated signaling pathways.34 These findings possibly reflect a different means of carcinogenesis of HCC in CH-B and CH-C (Fig. 6).

In summary, we investigated the detailed signaling pathways in CH-B and CH-C. Although our data reveal the different signaling pathways induced in CH-B and CH-C, the precise mechanisms underlining these differences must be proven experimentally in the future. Nevertheless, from the therapeutic point of view, these results might be indicative that antiviral agents will be most effective for CH-B whereas anti-inflammatory drugs, other than IFN, would be effective for CH-C, for the prevention of HCC (Fig. 8). Further studies are needed to elucidate these findings clinically and biologically.

Figure 8.

Schematic representation of different pathogenesis of hepatitis and development of HCC in CH-B and CH-C.


We thank Masami Ueda and Mikiko Nakamura for excellent technical assistance.