Gene profiling, biomarkers and pathways characterizing HCV-related hepatocellular carcinoma

Background Hepatitis C virus (HCV) infection is a major cause of hepatocellular carcinoma (HCC) worldwide. The molecular mechanisms of HCV-induced hepatocarcinogenesis are not yet fully elucidated. Besides indirect effects as tissue inflammation and regeneration, a more direct oncogenic activity of HCV can be postulated leading to an altered expression of cellular genes by early HCV viral proteins. In the present study, a comparison of gene expression patterns has been performed by microarray analysis on liver biopsies from HCV-positive HCC patients and HCV-negative controls. Methods Gene expression profiling of liver tissues has been performed using a high-density microarray containing 36'000 oligos, representing 90% of the human genes. Samples were obtained from 14 patients affected by HCV-related HCC and 7 HCV-negative non-liver-cancer patients, enrolled at INT in Naples. Transcriptional profiles identified in liver biopsies from HCC nodules and paired non-adjacent non-HCC liver tissue of the same HCV-positive patients were compared to those from HCV-negative controls by the Cluster program. The pathway analysis was performed using the BRB-Array- Tools based on the "Ingenuity System Database". Significance threshold of t-test was set at 0.001. Results Significant differences were found between the expression patterns of several genes falling into different metabolic and inflammation/immunity pathways in HCV-related HCC tissues as well as the non-HCC counterpart compared to normal liver tissues. Only few genes were found differentially expressed between HCV-related HCC tissues and paired non-HCC counterpart. Conclusion In this study, informative data on the global gene expression pattern of HCV-related HCC and non-HCC counterpart, as well as on their difference with the one observed in normal liver tissues have been obtained. These results may lead to the identification of specific biomarkers relevant to develop tools for detection, diagnosis, and classification of HCV-related HCC.


Introduction
Hepatocellular carcinoma (HCC) is the most common liver malignancy as well as the third and the fifth cause of cancer death in the world in men and women, respectively [1][2][3]. As for other types of cancer, the etiology and pathogenesis of HCC is multifactorial and multistep [4]. The main risk factor for development of HCC are the hepatitis B and C virus (HBV and HCV) infection [5][6][7][8]. Non viral causes, such as toxins and drugs (i.e., alcohol, aflatoxins, microcystin, anabolic steroids), metabolic liver diseases (i.e., hereditary haemochromatosis, α1-antitrypsin deficiency), steatosis and non-alcoholic fatty liver diseases as well as diabetes, play a role in a minor number of cases [9][10][11]. The prevalence of HCC in Italy, and in Southern Italy in particular, is significantly higher compared to other Western countries. Hepatitis virus infection, long-term alcohol and tobacco consumption account for 87% of HCC cases in Italian population and, among these, 61% of HCC are attributable to HCV. In particular, a recent seroprevalence surveillance study conducted in the general population of Southern Italy Campania Region reported a 7.5% positivity for HCV infection which peaked at 23.2% positivity in the 65 years or older age group [12]. The multistep progression to HCC, in particular the one associated to hepatitis virus, is characterized by a process including chronic liver injury, tissue inflammation, cell death, cirrhosis, regeneration, DNA damage, dysplasia and finally, HCC. In this multistep process, the cirrhosis represents the preneoplastic stage showing regenerative, dysplastic as well as HCC nodules [13].
The precise molecular mechanism underlying the progression of chronic hepatitis viral infections to HCC is currently unknown. Activation of cellular oncogenes, inactivation of tumor suppressor genes, overexpression of growth factors, telomerase activation and defects in DNA mismatch repair may contribute to the development of HCC [14][15][16]. In this framework, differential gene expression patterns accompanying different stages of growth, disease initiation, cell cycle progression, and responses to environmental stimuli provide important clues to this complex process.
DNA microarray enables investigators to study expression profile and activation of thousands of genes simultaneously. In particular, the identification of cancer-related stereotyped expression patterns might allow the elucida-tion of molecular mechanisms underlying cancer progression and provides important molecular markers for diagnostic purposes. This strategy has been recently used to profile global changes in gene expression in liver samples obtained from patients with HCV-related HCC [17][18][19]. Several of these studies identified gene sets that may be useful as potential microarray-based diagnostic tools. However, the direct or indirect HCV role in HCC pathogenesis is still a controversial issue and additional efforts need to be made aimed to specifically dissect the relationship between stages of HCV chronic infection and progression to HCC.
The present study has been focused on investigating genes and pathways involved in viral carcinogenesis and progression to HCC in HCV-chronically infected patients.

Patient and Tissue Samples
Liver biopsies from fourteen HCV-positive HCC patients and seven HCV-negative non-liver cancer control patients (during laparoscopic cholecystectomy) were obtained with informed consent at the liver unit of the INT "Pascale" in Naples. In particular, from each of the HCV-positive HCC patients, a pair of liver biopsies from HCC nodule and non-adjacent non-HCC counterpart were surgically excised. All liver biopsies were stored in RNA Later at -80°C (Ambion, Austin, TX). Confirmation of the histopathological nature of the biopsies was performed by the Pathology lab at INT before the processing for RNA extraction. The non-HCC tissue from HCV-positive patient were an heterogeneous sample representing the prevalent liver condition of each subject (ranging from persistent HCV-infection to cirrhotic lesions). Furthermore, laboratory analysis confirmed that the 7 controls were seronegative for hepatitis C virus antibodies (HCV Ab).

Preparation of RNA, probe preparation, and microarray hybridization
Samples were homogenized in disposable tissue grinders (Kendall, Precision). Total RNA was extracted by TRIzol solution (Life Technologies, Rockville, MD), and purity of the RNA preparation was verified by the 260:280 nm ratio (range, 1.8-2.0) at spectrophotometric reading with Nan-oDrop (Thermo Fisher Scientific, Waltham, MA). Integrity of extracted RNA was evaluated by Agilent 2100 Bioana-lyzer (Agilent Technologies, Palo Alto, CA), analyzing the presence of 28S and 18S ribosomal RNA bands as well as the 28S/18S rRNA intensity ratio equal or close to 1.5. In addition, phenol contamination was checked and a 260:230 nm ratio (range, 2.0-2.2) was considered acceptable.
Double-stranded cDNA was prepared from 3 μg of total RNA (T-RNA) in 9 μl DEPC -treated H 2 O using the Super script II Kit (Invitrogen) with a T7-(dT15) oligonucleotide primer. cDNA synthesis was completed at 42°C for 1 h. Full-length dsDNA was synthesized incubating the produced cDNA with 2 U of RNase-H (Promega) and 3 μl of Advantage cDNA Polymerase Mix (Clontech), in Advantage PCR buffer (Clontech), in presence of 10 mM dNTP and DNase-free water. dsDNA was extracted with phenolchloroform-isoamyl, precipitated with ethanol in the presence of 1 μl linear acrylamide (0.1 μg/μl, Ambion, Austin, TX) and aRNA (amplified-RNA) was synthesized using Ambion's T7 MegaScript in Vitro Transcription Kit (Ambion, Austin, TX). aRNA recovery and removal of template dsDNA was achieved by TRIzol purification. For the second round of amplification, aliquots of 1 μg of the aRNA were reverse transcribed into cDNA using 1 μl of random hexamer under the conditions used in the first round. Second-strand cDNA synthesis was initiated by 1 μg oligo-dT-T7 primer and the resulting dsDNA was used as template for in vitro transcription of aRNA in the same experimental conditions as for the first round [20]. 6 μg of this aRNA was used for probe preparation, in particular test samples were labeled with USL-Cy5 (Kreatech) and pooled with the same amount of reference sample (control donor peripheral blood mononuclear cells, PBMC, seronegative for hepatitis C virus antibodies (HCV Ab)) labeled with USL-Cy3 (Kreatech). The two labeled aRNA probes were separated from unincorporated nucleotides by filtration, fragmented, mixed and co-hybridized to a custom-made 36 K oligoarrays at 42°C for 24 h. The oligo-chips were printed at the Immunogenetics Section Department of Transfusion Medicine, Clinical Center, National Institutes of Health (Bethesda, MD). After hybridization the slides were washed with 2 × SSC/ 0.1%SDS for 1 min, 1 × SSC for 1 min, 0.2 × SSC for 1 min, 0.05 × SSC for 10 sec., and dried by centrifugation at 800 g for 3 minutes at RT.

Data Analysis
Hybridized arrays were scanned at 10-μm resolution with a GenePix 4000 scanner (Axon Instruments) at variable photomultiplier tube (PMT) voltage to obtain maximal signal intensities with less than 1% probe saturation. Image and data files were deposited at microarray data base (mAdb) at http://nciarray.nci.nih.gov and retrieved after median centered, filtering of intensity (>200) and spot elimination (bad and no signal). Data were further analyzed using Cluster and TreeView software (Stanford University, Stanford, CA).

Statistical Analysis Unsupervised Analysis
For this analysis, a low-stringency filtering was applied, selecting the genes differentially expressed in 80% of all experiments with a >3 fold change ratio in at least one experiment. 7'760 genes were selected for the analysis including the three groups of analyzed samples (the HCVrelated HCC, their non-HCC counterpart, as well as samples from the controls); 5'473 genes were selected for the analysis including the HCV-related HCC and normal control samples; 6'069 genes were selected for the analysis including the HCV-related non-HCC paired tissue and normal control samples. Hierarchical cluster analysis was conducted on these genes according to Eisen et al. [21]; differential expressed genes were visualized by Treeview and displayed according to the central method [22].

Supervised Analysis
Supervised class comparison was performed using the BRB ArrayTool developed at NCI, Biometric Research Branch, Division of Cancer Treatment and Diagnosis. Three subsets of genes were explored. The first subset included genes upregulated in HCV-related HCC compared to normal control samples, the second subset included genes upregulated in the HCV-related non-HCC counterpart compared with normal control samples, the third subset included genes upregulated in HCV-related HCC compared to the non-HCC paired liver tissue samples. Paired samples were analyzed using a two-tailed paired Student's t-test. Unpaired samples were tested with a two-tailed unpaired Student's t-test assuming unequal variance or with an F-test as appropriate. All analyses were tested for an univariate significance threshold set at a pvalue < 0.01 for the first subset of genes and at a p-value < 0.001 for the second subset. Gene clusters identified by the univariate t-test were challenged with two alternative additional tests, an univariate permutation test (PT) and a global multivariate PT. The multivariate PT was calibrated to restrict the false discovery rate to 10%. Genes identified by univariate t-test as differentially expressed (p-value < 0.001 and p-value < 0.01) and a PT significance <0.05 were considered truly differentially expressed. Gene function was assigned based on Database for Annotation, Visualization and Integrated Discovery (DAVID) and Gene Ontology http://www.geneontology.org/.

Ingenuity pathway analysis
The pathway analysis was performed using the gene set expression comparison kit implemented in BRB-Array-Tools. The human pathway lists determined by "Ingenuity Purity and integrity quality control of total extracted RNA System Database" was selected. Significance threshold of t-test was set at 0.001. The Ingenuity Pathways Analysis (IPA) is a system that transforms large data sets into a group of relevant networks containing direct and indirect relationships between genes based on known interactions in the literature.

Quality Control
The quality of extracted total RNA was verified by Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA), showing discrete 28S and 18S rRNA bands ( Figure 1A) as well as a 28S/18S rRNA intensity ratio equal or close to 1.5 which is considered appropriate for total RNA extracted from liver tissue biopsies ("Assessing RNA Quality", http:/ /www.ambion.com/techlib/tn/111/8.html). Based on this parameter, all extracted total RNA samples met the quality control criteria ( Figure 1B).

Unsupervised analysis is concordant with Pathological Classification
The gene expression profiles of tissue samples from the three groups of analyzed samples (the HCV-related HCC, their non-HCC counterpart, as well as samples from con- trol patients) were compared by an unsupervised analysis. No clear separation of the 3 different groups was observed, although control samples clustered mainly with samples from HCV-related non-HCC paired tissue, which includes dysplastic lesion in cirrhotic liver, representing a pre-neoplastic step (Figure 2A).

Unsupervised hierarchical clustering
In order to identify genes differentially modulated in HCV-related lesions compared to normal liver tissue samples, an unsupervised analysis was then performed including only paired samples from HCV-related HCC and normal control samples or from the HCV-related non-HCC counterpart and control samples (Figures 2B and  2C). According to filtering described in Material and Methods, HCV-related HCC and normal control samples showed 5'473 genes differentially expressed, with a per-fect clustering according to histological characteristics ( Figure 2B). Similarly, HCV-related non-HCC tissue and normal control samples showed 6'069 genes differentially expressed with a perfect clustering according to histological characteristics also in this case ( Figure 2C). The only exception to this pattern is represented by the normal control sample (CTR#80) which did not fall in the control cluster (CTR).

Supervised analysis
The supervised analysis was performed comparing pairs of gene sets using an unpaired Student's t-test with a cut-off set at p < 0.01.
The analysis comparing gene sets in liver tissues from HCV-related HCC and normal controls identified 825  Figure 3A). The first 40 genes showing the highest fold of up-regulation are listed in Table 1.
The analysis comparing gene sets in liver tissues from HCV-related non-HCC tissue and controls identified 151 genes differentially expressed. Among them, 127 were shown to be up-regulated and 24 down-regulated in HCVrelated non-HCC liver tissues ( Figure 3B). The first 40 genes showing the highest fold of up-regulation are listed in Table 2.
The analysis comparing gene sets in liver tissues from HCV-related HCC and HCV-related non-HCC counterpartidentified 383 genes differentially expressed. Among them, 83 were shown to be up-regulated and 300 downregulated in HCV-related HCC liver tissues ( Figure 3C). The first 40 genes showing the highest fold of up-regulation are listed in Table 3.

Ingenuity pathway analysis
The pathway analysis was performed including the genes found up-regulated in the supervised comparisons, using the gene set expression comparison kit implemented in BRB-Array-Tools. The human pathway lists determined Heat map of the gene signature, identified by Class Comparison Analysis

Discussion
The pathogenetic mechanisms leading to HCC development in HCV chronic infection are not yet fully elucidated. In particular, besides inducing liver tissue inflammation and regeneration, which ultimately may result in cellular transformation and HCC development, HCV may play a more direct oncogenic activity inducing an altered expression of cellular genes. To this aim, global gene expression profile can identify specific genes differentially expressed and provide powerful insights into mechanisms regulating the transition from pre-neoplastic to fully blown neoplastic proliferation [23,24]. In the present study, the differential gene expression was evaluated by microarray analysis on liver tissues obtained from fourteen HCV-positive HCC patients and seven HCV-negative control patients. In particular, from each of the HCV-positive HCC patients, a pair of liver biopsies from HCC nodule and non-HCC non adjacent counterpart were surgically excised.
The unsupervised analysis didn't show a clear separation of samples from the 3 different groups (HCV-related HCC, their non-HCC counterpart, as well as control patients), suggesting the lack of a clear-cut distinct gene signature pattern. Nevertheless, normal control samples, with the exception of CTR#76 sample, grouped in a single cluster close to samples from HCV-related paired non-HCC samples. The latter, in fact, comprise several non-HCC pathological stages including dysplastic, not fully transformed lesions, representing pre-neoplastic step in the progression to HCC and should still retain a gene signature pattern closer to normal than to transformed cell physiology. On the contrary, the unsupervised analysis including only one of the HCV-related liver tissues (HCC or non-HCC counterpart) and normal controls showed a clear-cut segregation of the pathological from the control cluster, indicating the identification of specific gene signature patterns peculiar to the HCV-related pre-neoplastic (non-HCC) and neoplastic (HCC) tissues compared to normal controls. A supervised analysis was performed by pairwise comparison between samples of the three groups analyzed in the present study. The results indicated that the HCV-related HCC liver tissues showed 825 genes differentially expressed compared to controls, of which 465 were upregulated and 360 down-regulated. The HCV-related non-HCC liver tissues showed 151 genes differentially expressed compared to controls, of which 127 were upregulated and 24 down-regulated. The HCV-related HCC liver tissues showed 383 genes differentially expressed compared to HCV-related non-HCC counterpart, of which 83 were up-regulated and 300 down-regulated. In each of these independent class comparison analysis, the differentially expressed genes were selected based on a 3fold difference at a significance p-value < 0.01.
The up-regulated genes identified within the individual class comparison analysis were further evaluated and classified by a pathway analysis, according to the "Ingenuity System Database".
The genes up-regulated in samples from HCV-related HCC are classified in metabolic pathways, and the most represented are the Aryl Hydrocarbon receptor signaling (AHR) and, protein Ubiquitination pathways, which have been previously reported to be involved in cancer, and in particular in HCC, progression.
The Aryl Hydrocarbon receptor signal transduction Pathway (AHR) is involved in the activation of the cytosolic aryl hydrocarbon receptor by structurally diverse xenobiotic ligands (including dioxin, and polycyclic or halogenated aromatic hydrocarbons) and mediating their toxic and carcinogenic effects [25,26]. More recently AHR pathway has been shown to be involved in apoptosis, cell cycle regulation, mitogen-activated protein kinase cascades [27]. In particular, studies on liver tumor promotion have shown that dioxin-induced AHR activation mediates clonal expansion of initiated cells by inhibiting apoptosis and bypassing AHR-dependent cell cycle arrest [28]. Furthermore, it has been shown that changes in mRNA expression of specific genes in the AHR pathway are linked to progression of HCV-associated hepatocellular carcinoma [29]. Moreover, the HCV-induced AHR signal transduction pathway, could be directly involved in the increased severity of hepatic lesions in patients with chronic hepatitis C induced by smoking [30,31].
The ubiquitin and ubiquitin-related proteins of the ubiquitination pathway play instrumental roles in cell-cycle regulation [32] as well as cell death/apoptosis [33] through modification of target proteins. In particular, ubiquitin-like proteins, i.e. FAT10, has been reported to bind non-covalently to the human spindle assembly checkpoint protein, MAD2 [34], which is responsible for maintaining spindle integrity during mitosis [35] and whose inhibited function has been associated with chromosomal instability [36,37]. Moreover, FAT10 overexpression has been previously shown in hepatocellular carcinoma [38].
The genes up-regulated in samples from HCV-related non-HCC tissue are classified in several pathways prevalently associated to inflammation and native/adaptive immunity and most of the overexpressed genes belong to the Antigen Presentation pathway. Considering the chronic HCV infection, these result could be unexpected and contradictory, since a reduced native and/or adaptive specific immune response would represent a very much favorable environment for the virus. Nevertheless, these findings, which confirm also a recent report by others [39], could explain the generic massive inflammation and immunopathological tissue damage characteristic of HCV-related cirrhosis [40].
In this study, informative data on the global gene expression pattern in HCV-related HCC as well as HCV-related non-HCC counterpart liver tissues have been obtained compared to normal controls. These data, which need further confirmation studies on a larger set of samples and also at protein level, may be extremely helpful for the identification of exclusive activation markers to characterize gene expression programs associated with progression of HCV-related lesions to HCC.