Skip Navigation


Carcinogenesis Advance Access originally published online on July 16, 2008
Carcinogenesis 2008 29(10):1901-1910; doi:10.1093/carcin/bgn170
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow Supplementary Data
Right arrow All Versions of this Article:
29/10/1901    most recent
bgn170v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (1)
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Gao, W.
Right arrow Articles by Sekido, Y.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Gao, W.
Right arrow Articles by Sekido, Y.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

© The Author 2008. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org

Variable DNA methylation patterns associated with progression of disease in hepatocellular carcinomas

Wentao Gao1,2, Yutaka Kondo1,*, Lanlan Shen3, Yasuhiro Shimizu4, Tsuyoshi Sano4, Kenji Yamao5, Atsushi Natsume6, Yasuhiro Goto1, Motokazu Ito1, Hideki Murakami1, Hirotaka Osada1, Jiexin Zhang7, Jean-Pierre J. Issa3 and Yoshitaka Sekido1

1 Division of Molecular Oncology, Aichi Cancer Center Research Institute, 1-1 Kanokoden, Chikusa-Ku, Nagoya 464-8681, Japan
2 Department of General Surgery, First Affiliated Hospital of Nanjing Medical University, Nanjing 210029, China
3 Department of Leukemia, The University of Texas at M.D. Anderson Cancer Center, Houston, TX 77030, USA
4 Department of Gastroenterological Surgery
5 Department of Gastroenterology, Aichi Cancer Center Hospital, 1-1 Kanokoden, Chikusa-Ku, Nagoya 464-8681, Japan
6 Department of Neurosurgery, Nagoya University School of Medicine, Showa-ku, Nagoya 466-8550, Japan
7 Department of Bioinformatics and Computational Biology, The University of Texas at M.D. Anderson Cancer Center, Houston, TX 77030, USA

* To whom correspondence should be addressed. Tel: +81 52 764 2993; Fax: +81 52 764 2993; Email: ykondo{at}aichi-cc.jp


   Abstract
Top
 Abstract
Introduction
 Materials and methods
 Results
 Discussion
 Supplementary material
 Funding
 References
 
Hepatocellular carcinoma (HCC) most commonly arises from chronic inflammation due to viral infection, as a result of genetic and epigenetic abnormalities. A global picture of epigenetic changes in HCC is lacking. We used methylated CpG island amplification microarrays (MCAMs) to study 6458 CpG islands in HCC and adjacent preneoplastic tissues [chronic hepatitis (CH) or liver cirrhosis (LC)] in comparison with normal liver tissues where neither viral infection nor hepatitis has existed. MCAM identified 719 (11%) prominent genes of hypermethylation in HCCs. HCCs arising from LC had significantly more methylation than those arising from CH (1249 genes or 19% versus 444 genes or 7%, P < 0.05). There were four patterns of aberrant methylation: Type I (4%, e.g. matrix metalloproteinase 14) shows a substantially high methylation level in adjacent tissue and does not increase further in cancer. Type II (55%, e.g. RASSF1A) shows progressively increasing methylation from adjacent tissue to HCC. Type III (4%, e.g. GNA14) shows decreased methylation in adjacent tissue but either similar or increased methylation in HCC. Type IV (37%, e.g. CDKN2A) shows low levels of methylation in normal tissue and adjacent tissue but high levels in HCC. These DNA methylation changes were confirmed by quantitative pyrosequencing methylation analysis in representative 24 genes and were analyzed for correlation with clinicopathological parameters in 38 patients. Intriguingly, methylation in the Type IV genes is characteristic of moderately/poorly differentiated cancer. Our global epigenome analysis reveals distinct patterns of methylation that are probably to represent different pathophysiologic processes in HCCs.

Abbreviations: CH, chronic hepatitis; DNMT, DNA methyltransferase; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; LC, liver cirrhosis; MCA, methylated CpG island amplification; MCAM, methylated CpG island amplification microarray; MMP14, matrix metalloproteinase 14; PCR, polymerase chain reaction; TSS, transcription start sites


   Introduction
Top
 Abstract
 Introduction
Materials and methods
 Results
 Discussion
 Supplementary material
 Funding
 References
 
Hepatocellular carcinoma (HCC) is one of the most common human malignancies worldwide. It is mostly associated with viral infection, which causes chronic inflammation and might be related to genetic or epigenetic alterations. There is accumulating evidence for the multistep nature of HCC (1). A long-term follow-up study revealed that liver cell dysplasia or adenomatous hyperplasia, both of which are known to be precancerous lesions of the liver with the histological findings of chronic hepatitis (CH) and/or liver cirrhosis (LC) transformed to cancer (2). Mutations in the critical tumor suppressor genes, TP53 or Rb gene, were observed in a late stage of HCC (3), whereas loss of heterozygosity was detected even in CH or LC (4,5). Notably, not all these genetic alterations can be detected in HCC, suggesting that there might be multiple carcinogenesis pathways contributing to HCCs.

Aberrant DNA methylation of promoter CpG islands has been described in human cancers. This epigenetic alteration is associated with gene silencing, and together with point mutations and deletions, serves as a mechanism leading to the inactivation of tumor suppressor genes and critical cancer-related genes in human cancers. Dysregulation of these hypermethylated genes has been connected with essential tumor properties such as tumor cell proliferation, antiapoptosis, neo-angiogenesis, invasive behavior and chemotherapy resistance (6,7). Previously, we demonstrated aberrant methylation even in adjacent histologically normal liver tissues, CH or LC, regardless of the viral status (5). DNA methylation in adjacent non-cancerous liver tissue of HCC has also been reported from other groups and is now accepted as early and ubiquitous events in cancer development (8,9). The profiling of CpG island methylation in human cancers implies that some genes are more frequently methylated than others and that each tumor type is associated with a unique set of methylated genes (10). Despite numerous examples of methylation-associated gene-silencing events in human cancers including HCC, little is known about the global picture of hypermethylated genes in HCCs.

This leaves the question as to which and how the DNA methylation target genes behave and contribute to tumorigenesis; in other words, to what extent these early epigenetic changes are retained in the developed cancers and which subclasses are most affected. To address this question, we conducted genome-wide screening for aberrantly methylated genes in cancerous tissue and corresponding adjacent non-cancerous tissue (CH and LC) from HCC patients using methylated CpG island amplification microarray (MCAM) technique. This technique simultaneously reduces complexity and increases specificity by targeting methylated CpG islands before amplification. Although there are intrinsic limitations of MCAM in that methylated CpG island amplification (MCA) theoretically can cover 80% of CpG islands and the microarray platform has limited gene representation, we found that MCAM provides reproducible results with a high validation rate and could successfully detect the genes methylated in normal tissues as well as cancer cells (11,12).

In addition, we performed quantitative pyrosequencing DNA methylation analysis in representative 24 identified genes in 38 HCC patients and assessed the relationship between DNA methylation status and clinicopathological background in those patients. Our extensive analyses revealed that aberrant DNA methylation is a frequent event in both early and late stages during the liver malignant transformation and that CpG island promoters appear to be methylated in different patterns during progression of the disease, which may represent different pathophysiologic processes. Moreover, these data suggest that different mechanisms for acquisition of this epigenetic change act during cancer formation.


   Materials and methods
Top
 Abstract
 Introduction
 Materials and methods
Results
 Discussion
 Supplementary material
 Funding
 References
 
Tissue samples
Paired samples of adjacent non-cancerous liver tissue and cancerous tissue were obtained from 38 patients with HCC who underwent surgical resection at the Aichi Cancer Center Hospital, Nagoya, Japan in accordance with the institutional policy (supplementary Table 1 is available at Carcinogenesis Online). All patients provided written informed consent. Histologic examination of the non-cancerous liver tissues revealed no remarkable findings (N) or findings compatible with CH or LC in 2, 19 and 17 cases, respectively. Hepatitis B virus (HBV) surface (HBs) antigen and anti-hepatitis C virus (HCV) antibody were measured serologically. In addition, samples of normal liver tissue were also obtained from eight patients without HBV or HCV infection who underwent partial hepatectomy for liver metastasis of primary colon cancer. Tissue samples were flash frozen and stored at –80°C. Genomic DNA was extracted using a standard phenol–chloroform method.

Cell lines and culture conditions
Two hepatoma cell lines, HepG2 and Huh7, were grown in Dulbecco’s modified Eagle’s medium (Invitrogen, Carlsbad, CA), plus 10% fetal bovine serum in plastic tissue culture plates in a humidified atmosphere containing 5% CO2 at 37°C. Huh7 was the kind gift of Dr. Tetsuro Suzuki (National Institute of Infectious Diseases, Tokyo, Japan). HepG2 was obtained from the American Type Culture Collection (Manassas, VA).

Trichostatin A and 5-aza-2'-deoxycytidine treatment of cells
Cells were seeded 12-24 hours before treatment. Cells were then treated with either: (i) 1 µM 5-aza-2'-deoxycytidine (5Aza-dC; Sigma-Aldrich, St Louis, MO, USA) or phosphate-buffered saline for 72 h. Media containing 5Aza-dC or phosphate-buffered saline was changed every 24 h; or (ii) 300 nM trichostatin A (TSA; MP Biomedicals, Solon, OH) or an identical volume of ethanol for 24 h.

MCAM
MCA was carried out using DNA from Huh7, paired samples of adjacent non-cancerous liver tissue and cancerous tissue from 10 HCC patients (P1, P3, P9, P16, P17, P18, P22, P25, P26 and P33, supplementary Table 1 is available at Carcinogenesis Online) and two normal liver samples (Figure 1A). Six of 10 patients developed cancer from LC and 4 patients developed cancer from CH. Nine patients have HCV infection and one patient has neither HBV nor HCV infection.


Figure 1
View larger version (50K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 1. MCAM analysis in HCC samples. (A) Outline of MCAM. DNA from cancerous tissue, adjacent tissue or normal liver was sequentially digested with SmaI and XmaI. After amplification of methylated DNA fragments, MCA product was labeled with Cy5 (red) for the adjacent tissue or cancerous tissue and Cy3 (green) for the normal liver and then applied for a promoter microarray. (B) Number of methylated genes in each patient. Patient number (P) in white box and gray box represents CH patient and LC patient, respectively. Black and white bars indicate number of methylated probes in cancerous tissue and adjacent tissue, respectively. (C) Number of methylated probes on each chromosome. Methylated probe indicates the probe on which DNA methylation was detected in >3 patients. Black and white bars represent the same as in (B). Total probe numbers on each chromosome are shown in gray bars at the bottom. (D) Dendrogram and heat map overview of hierarchical cluster analysis (unsupervised learning) of DNA methylation data in cancerous and adjacent tissues from HCCs and Huh7 cell line using significant genes (y axis; average signal ratio of these genes in cancerous tissues >2.0). Each cell in the matrix represents the DNA methylation status of a gene in an individual sample. Red and green in cells reflect high and low methylation level, respectively, as indicated in the scale bar below the matrix (left matrix, log2-transformed scale). Sample numbers in gray box and in white box indicate sample from LC patient and that from CH patient, respectively. C and Aj indicate cancerous tissue and adjacent tissue, respectively. Length and subdivision of branches indicate the relatedness of each sample. Cancerous tissue from LC patient makes a distinct cluster by DNA methylation status.

 
A detailed protocol of MCA was described previously (13). Briefly, 2 µg of genomic DNA was digested with 100U of methylation-sensitive restriction endonuclease SmaI (New England Biolabs, Ipswich, MA) for 8 h at 20°C two times, which cuts unmethylated DNA and leaves blunt ends (CCC/GGG). Subsequently, the DNA was digested with 20U of methylation-insensitive restriction endonuclease XmaI for 9 h at 37°C, which creates sticky ends (C/CCGGG). Five hundred nanogram of digested DNA was ligated. After filling in the overhanging ends of the ligated DNA fragments at 72°C, DNA was amplified at 95°C for 3 min followed by 25 cycles of 1 min at 95°C and 3 min at 77°C using 100 pmol of RMCA24 primer. MCA products were labeled with Cy5 (red) for adjacent non-cancerous liver tissue, cancerous tissue or Huh7 DNA and Cy3 (green) for reference DNA from gender-matched normal liver tissues using a random primed Klenow polymerase reaction (Invitrogen) at 37°C for 3 h. Labeled samples were then hybridized to the 88K human promoter array (G4475A, AMADID 013902 and 013903, Agilent Technologies, Santa Clara, CA) in the presence of human Cot-1 DNA for 40 h at 65°C. This human promoter array contained 60mer oligonucleotide probes covering the region close to the transcription start sites (TSSs) (–1.0 to +0.3 kb) for 18 300 annotated human genes (12). Probe information was downloaded from Agilent website (http:\\www.agilent.com). Each probe is blast against all sequences in SmaI/XmaI database using standalone blastall v2.2.8 program downloaded from National Center for Biotechnology Information (/data/bioinfo/SequenceTools/blast/). Probes with multiple BLAT hits were excluded from further study. Probes residing in SmaI/XmaI fragments are identified with the annotation for fragment length, CpG island and repetitive sequences. The criteria for CpG islands used in this study were defined previously (14). Finally, we selected probes, of which at least one of the encompassing SmaI/XmaI sites is on CpG island. We calculated each SmaI/XmaI site relative to the TSS and found that >90% of these SmaI/XmaI sites on CpG island were within 1 kb from TSS. After washing the array according to the manufacturer’s protocol, arrays were scanned on an Agilent scanner and analyzed using Feature Extraction software and normalization was done using a linear per-array algorithm according to the manufacturer’s protocol (Agilent Technologies) (12).

Criteria for DNA methylation pattern
To assess the DNA methylation pattern from the microarray signal, we determined the criteria for each methylation pattern as follows. Type I: early pattern genes, either the average signal ratio for each gene (Cy5/Cy3) in both cancerous tissue (Ca-R) and adjacent tissue (Adj-R), were >2.0 or Ca-R > 2.0, and the difference between Ca-R and Adj-R was <15%; Type II: progressing pattern genes, Ca-R > 2.0, Adj-R > 1.1 (at least methylation level of Adj-R methylation level is higher than that of normal liver) and Ca-R > Adj-R, exclude the early pattern genes; Type III: contrary pattern genes, Ca-R > 2.0 and signal ratio of Cy3/Cy5 (normal/adjacent) in adjacent tissue >1.5 and Type IV: cancer-specific pattern, Ca-R > 2.0 and Adj-R ≤ 1.1. These criteria are consistently matched with the types, which were classified by the average methylation level of 24 genes by pyrosequencing quantitative analysis in 10 patients, confirming the validity of these criteria.

Method of unsupervised two-dimensional clustering
Cluster analysis was performed using an agglomerative hierarchical clustering algorithm (15). For specimen clustering, pairwise similarity measures among specimens were calculated using Cluster 3.0 software (http://rana.lbl.gov/EisenSoftware.htm) based on DNA methylation intensity measurements across all significant 719 genes (average signal ratio of the cancer samples >2.0). A dendrogram and heat map were constructed using TreeView software (http://rana.lbl.gov/EisenSoftware.htm).

Pathway analysis using MetaCore
We analyzed the hypermethylated genes whose signal ratio (Cy5/Cy3) is >2.0 using a functional mapping tool, MetaCore (GeneGO, St Joseph, MI), to identify novel gene networks composed of biological pathways of precancerous and cancerous stages (16). MetaCore is a web-based computational platform designed primarily for the analysis of high-throughput experimental data in the context of human regulatory networks and pathways. It includes a curated database of human protein interactions, metabolism and bioactive compounds. For a network of a given size, MetaCore can be used to calculate statistical significance based on the probability of the network’s assembly from a random set of nodes (genes) and the same size as the input list (P value).

Bisulfite pyrosequencing methylation analysis
We performed bisulfite treatment as reported previously (17,18). Briefly, 2 µg of genomic DNA was incubated with 3 M sodium bisulfite (pH 5.0) for 16 h at 50°C. DNA was purified using a Wizard Miniprep Column (Promega, Madison, WI), precipitated with ethanol and resuspended in 30 µl of diluted water. DNA methylation level was measured by a highly quantitative method using PyrosequencingTM technology (Pyrosequencing AB, Uppsala, Sweden) (19). For pyrosequencing, the biotinylated polymerase chain reaction (PCR) product was captured on streptavidin-coated beads (Amersham Biosciences, Uppsala, Sweden) and run on the PSQHS pyrosequencing system (Biotage, Uppsala, Sweden) to obtain the degree of methylation. Primer sequences are shown in supplementary Table 2 (available at Carcinogenesis Online). All the primers were designed to assay the methylation status of CpGs within 0.4 kb from the TSS. Global DNA methylation was measured by the LINE1 methylation status (18). hMLH1 is used as a negative control for methylation analysis (5). The methylation levels at different C sites measured by pyrosequencing were averaged to represent the degree of methylation in each sample for each gene. For each assay, setup includes positive controls (samples after SssI treatment; New England Biolabs) and negative controls (samples after whole-genomic amplification using GenomiPhi V2; GE Healthcare, Piscataway, NJ), mixing experiments to rule out bias and repeat experiments to assess reproducibility. Optimizing annealing temperature for PCR is used to overcome PCR bias (20). Methylation status determined by pyrosequencing assay was analyzed as both continuous variable (methylation level) and categorical variable (methylation negative: methylation level <12%, methylation positive: level ≥12%). The 12% cutoff was selected because lower values could not be easily distinguished from background.

Reverse transcription and quantitative PCR analyses
Total RNA was isolated using TRIzol (Invitrogen). Two micrograms was reverse transcribed with moloney murine leukemia virus reverse transcriptase (Promega) to make complementary DNA. TaqMan quantitative PCRs and SYBR GREEN quantitative PCRs were carried out in duplicate for the target genes using the following primers: RASSF1A (forward CCTCTGTGGCGACTTCATCTG, reverse TAGTGGCAGGTGAACTTGCAA, probe CCTGCAGTGCGCGC); P16 (forward CGCTGCCCATCATCATGA, reverse CCAACGCACCGAATAGTTACG, probe CTGGATCGGCCTCC); matrix metalloproteinase 14 (MMP14) (forward GCAGAAGTTTTACGGCTTGCA, reverse TCTGGAACACCACATCGGG); TBX4 (forward GCCGCGGAGCAGACC, reverse GGTGCCCGCCTCGTG); GNA14 (forward GGAGATCGAGCGACAGCTT, reverse TTCGTGAACCCCTTTCTGTC) and CCNA1 (forward GGGCTCCCAGATTTCGTCTT, reverse GACCTCGGGCCACTGTAGC) (Applied Biosystems, Foster City, CA).

Mutation analysis
P53 mutations were examined in complementary DNA from HCC samples by direct sequencing using primers, P53-cDNAf, TGCATTCTGGGACAGCCA, and P53-cDNAr, CCACGGATCTGAAGGGTGAA, which amplify from exon 4 to exon 9. Genomic DNA sequence was also examined by direct sequencing using the following primers: P53-Ex5f, CACTTGTGCCCTGACTTTCA; P53-Ex5r, AACCAGCCCTGTCGTCTCT; P53-Ex6f, AGGCCTCTGATTCCTCACTG; P53-Ex6r, GCCACTGACAACCACCCTTA; P53-Ex7f, AAAAGGCCTCCCCTGCTT; P53-Ex7r, GTGTGCAGGGTGGCAAGT; P53-Ex8f, TGATTTCCTTACTGCCTCTTGC and P53-Ex8r, AAAAGTGAATCTGAGGCATAACTG.

Statistical analysis
All statistical analyses were done using GraphPad Prism software for Windows (version 3.03; GraphPad Software, San Diego, CA). Associations between methylation status and clinicopathologic variables were analyzed by Fisher’s exact test or Kruskal–Wallis test. All reported P values were two sided, and P < 0.05 was considered statistically significant.


   Results
Top
 Abstract
 Introduction
 Materials and methods
 Results
Discussion
 Supplementary material
 Funding
 References
 
Validation of MCAM analysis in HCC samples
In an initial validation of the MCAM method in HCC samples, first, we examined signal intensity on each probe located inside of SmaI/XmaI fragments (length from 20 bp to 10 kb) to evaluate efficacy of the MCA technique (Figure 1A and supplementary Figure 1 is available at Carcinogenesis Online). Significantly increased signal intensity was observed on probes corresponding to fragment sizes of ≤2 kb in MCA product, implying that probes hybridized with MCA product of these sizes can be analyzed accurately on our system (supplementary Figure 1A is available at Carcinogenesis Online). We further annotated all the restriction enzyme (SmaI/XmaI) sites for CpG islands and repetitive elements over the whole genome and excluded probes corresponding to repetitive elements. Finally, we predicted that 16434 probes corresponding to 6458 unique genes on the 88K promoter array fitted the above criteria and would be informative for our MCAM analysis (see Materials and Methods).

Location in relation to a TSS is now recognized to be extremely important (21). Depletion of nucleosomes just upstream of TSSs (nucleosome-free region) has recently been observed in genome-wide screens (22). It seems probably that the absence of at least one nucleosome (corresponding to 150–200 bp) is necessary for gene transcription (23). To evaluate the location of SmaI/XmaI sites in relation to TSS, we analyzed the distance from the TSS of the reference gene to the closest SmaI/XmaI sites of each probe and found that ~44% of at least one of the encompassing SmaI/XmaI sites were within 200 bp away from the TSS (supplementary Figure 1B is available at Carcinogenesis Online).

Next, to determine the criteria for most accurate identification of hypermethylated loci, we performed bisulfite pyrosequencing on 56 genes, which were selected from the array with a different range of signal intensity and ratio of Cy5/Cy3 (supplementary Figure 1C is available at Carcinogenesis Online). We found a good correlation between the microarray and pyrosequencing results. Specificity and sensitivity were calculated by each signal ratio from the microarray analyses and revealed a signal ratio of >2.0 as an optimal criterion for a methylation-positive spot on our microarray analysis (supplementary Figure 1D is available at Carcinogenesis Online).

Analyses of DNA methylation targets by MCAM in paired adjacent non-cancerous tissue and cancerous tissue from 10 HCC patients
To assess the global DNA methylation targets in HCCs, we performed MCAM in paired adjacent non-cancerous tissue and cancerous tissue from 10 patients (Figure 1). Histologic examination of the non-cancerous liver tissues revealed findings compatible with hepatitis or cirrhosis. We selected the patients whose non-cancerous tissues had typical pathological findings of either CH or LC. In particular, we carefully collected the sample of cirrhotic nodule from cirrhotic liver. Each HCC averaged 927 hypermethylated genes (range 371–2314), whereas non-cancerous samples averaged 444 hypermethylated genes (range 78–860). The number of positive genes was obviously different between patients who developed cancer from CH and those who developed cancer from LC (Figure 1B) in both cancerous tissues (average 444 genes and 1249 genes in HCC-CH and HCC-LC, respectively; P < 0.05) and in adjacent tissues (average 275 genes and 557 genes in CH and LC, respectively; P < 0.05). These hypermethylated genes were almost equally distributed across all the chromosomes in both adjacent tissue (range 2–7% of total probes) and cancerous tissue (range 7–16% of total probes; Figure 1C).

An unsupervised, hierarchical clustering algorithm revealed that the cancerous tissue samples can be divided into two distinct groups on the basis of DNA methylation status (Figure 1D). One group with a large number of methylated loci consisted of only patients who developed cancers from LC background (four HCC-LC patients), whereas another group with less methylated loci consisted of four HCC-CH patients and two HCC-LC patients. The latter group is closer to the clusters of adjacent tissues. The Huh7 cell line was clustered closer to the former group than the latter. However, this could be due not only to its original pathological status but also to accumulation of methylated genes during long passage.

The number of positive genes and cluster analysis lead to the hypothesis of two distinctive pathways in HCC, which might reflect the histological findings of the non-cancerous background liver where the cancer cell arose. HCC-LC tends to be associated with a high frequency of DNA methylation, whereas the other (HCC-CH and a part of HCC-LC cases) is less dependent on DNA methylation.

Hypermethylated genes were classified into four distinct patterns in paired adjacent non-cancerous tissue and cancerous tissue from HCC patients
MCAM preferentially and validly detect hypermethylated loci, as methylated loci of DNA are selectively amplified with MCA (11–13). In the current study, we first focused on hypermethylated genes in cancers, the genes that might be associated with cancer formation through inactivation of the genes by DNA methylation. We then evaluated the methylation status of those genes in adjacent tissues to see how DNA methylation behaves during progression of disease. At least four distinct signal patterns were recognized on the microarrays: (i) hypermethylation (i.e. red) and hypomethylation (i.e. green); (ii) hypermethylation and the same level of methylation as normal controls (i.e. black); (iii) hypermethylation and slight hypermethylation (i.e. light red) and (iv) hypermethylation and hypermethylation in the cancerous tissue and the adjacent tissues, respectively (Figure 2A). To gain insight into these DNA methylation behaviors, 24 genes were selected from the 56 genes, which were used for validation of the microarray. These 24 genes were methylated in cancer samples with a high frequency and showed different patterns of signal ratio in the cancerous tissues and the adjacent tissues (supplementary Figure 1 is available at Carcinogenesis Online, the diagrams of the promoters of 24 genes were shown in supplementary Figure 2 is available at Carcinogenesis Online). The methylation level of the 24 genes in 38 HCC patients and in eight normal liver controls where neither viral infection nor hepatitis had existed was examined by pyrosequencing assay (Figure 2B and supplementary Table 3 is available at Carcinogenesis Online). Notably, a lower level of DNA methylation in LINE1 (24), which represents the overall methylation status (18), was observed in cancerous tissues compared with the adjacent tissues or normal liver (supplementary Table 3 is available at Carcinogenesis Online). Consistent with the microarray analysis, we observed the above four distinct types of DNA methylation in HCC samples by pyrosequencing analysis.


Figure 2
View larger version (23K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 2. Four distinct methylation patterns in HCCs. (A) Twenty-four genes were selected, which showed four distinct patterns on the microarray [i.e. red and green, red and black, red and light red and red and red in cancerous tissue (C) and adjacent tissue (Aj), respectively]. Each cell in the matrix represents the same as in Figure 1. (B) Quantitation of methylation level by bisulfite pyrosequencing in 38 HCC samples classified the hypermethylated genes into four distinct types. Dots represent methylation level of the indicated gene in each sample. Red lines in the graph represent averages in each group. N, samples from normal livers. Average methylation levels in normal livers and adjacent non-cancerous tissues reflect the fraction of cells with methylation among similar cells in polyclonal tissues, whereas in cancers, which are theoretically monoclonal, average methylation level reflects more cases with high level of methylation in those groups. (C) Reactivation of hypermethylated genes by inhibition of DNA methylation or histone deacetylation. Gene expression for representative eight genes was measured by quantitative reverse transcription–polymerase chain reaction in Huh7 and HepG2 cell line treatment with DNA methylation inhibitor 5-aza-2'-deoxycytidine (D; black) or HDAC inhibitor trichostatin A (T; gray). Relative values for messenger RNA expression of each gene to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) are shown in y axis. All silenced genes with high DNA methylation level were effectively reactivated by DNA methylation inhibition. Parentheses below control (C) indicate methylation level before treatment.

 
The first type represents methylation that develops substantially at the transition between normal tissue and adjacent tissue and does not increase further in cancer (Type I, early pattern). The putative promoter region of microRNA hsa-mir-219-2 (mir-219), MMP14 and cell death-inducing DNA fragmentation factor alpha-like effector A (CIDEA), reflected this pattern. The second type shows progressively increasing methylation from adjacent tissue to HCC (Type II, progressing pattern). A well-known tumor suppressor gene, RASSF1A, TBX4, carbonic anhydrase X (CA10), proenkephalin (PENK), developing brain homeobox 1 (DBX1), protein tyrosine phosphatase receptor type T (PTPRT) and PRDI-BF1 and RIZ homologous (PR) domain containing protein 14 (PRDM14) showed this pattern. In the third type, an unexpected pattern, a moderate or high level of methylation, was observed in normal livers, whereas the methylation level was low in adjacent tissue but increased in cancerous tissue (Type III, contrary pattern). Guanine nucleotide-binding protein alpha-14 subunit (GNA14), solute carrier family 16 member 5 (SLC16A5) and hypothetical protein, C9orf106, typified this pattern. In the fourth type, a high level of DNA methylation was observed only in the cancerous tissue, in contrast to little or no DNA methylation in adjacent tissue and normal liver tissue (Type IV, cancer-specific pattern). This pattern was found from important cell cycle regulators, CDKN2A (p16INK4A) and cyclin A1 (CCNA1), tumor necrosis factor receptor superfamily member 10C (TNFRSF10C; also known as decoy receptor 1, DcR1), bone morphogenetic protein 6 (BMP6) and several tumor-related genes (supplementary Table 3 is available at Carcinogenesis Online). Among 719 prominent hypermethylated genes with an average signal ratio of >2.0 in the cancer samples, 32 genes (4%), 396 genes (55%), 27 genes (4%) and 264 genes (37%) were classified into Type I, II, III and IV, respectively, implying that more than half the aberrant DNA methylation on CpG island had emerged in the precancerous stage, whereas ~40% of DNA methylation was specific to cancer [gene names are available in supplementary Table 4 (available at Carcinogenesis Online)].

Reactivation of silenced genes by epigenetic drug
To examine whether the gene-silencing mechanism was related to DNA methylation, we treated the two hepatoma cell lines by DNA methylation inhibitor 5-aza-2'-deoxycytidine or the histone deacetylase inhibitor, trichostatin A, and examined gene expression in a candidate gene of DNA methylation (Figure 2C). Genes with little or no DNA methylation were all expressed in the two cell lines (data not shown). In contrast, genes with >80% of the methylation level were all silenced in the two cell lines, which were effectively reactivated by DNA methylation inhibition but responded little to trichostatin A treatment, consistent with the previous findings that only inhibitor of DNA methylation can reactivate the typical transcriptional repressive states associated with DNA methylation (25).

DNA methylation status in association with clinicopathological parameters in 38 HCC patients
Next, by focusing on the selected 24 genes, we examined DNA methylation status in association with clinicopathological parameters in 38 HCC patients using pyrosequencing analyses (Table 1 and supplementary Table1 is available at Carcinogenesis Online). In cancerous tissue, DNA methylation was more frequently observed in HCC-LC than in HCC-CH (14.9 genes versus 10.2 genes, P = 0.01) as we found in the microarray analyses. As for virus status, there were significantly more methylated genes in cancers with HCV infection (9.8 genes, 14.6 genes and 11.3 genes in HBV, HCV and HBs antigen and anti-HCV antibody negative, respectively; P = 0.019).


View this table:
[in this window]
[in a new window]

  		Table I. Relationship between DNA methylation and clinicopathological features

 
To clarify these relations more clearly, the histological findings of adjacent liver tissues, viral status and methylation frequency of individual patients were evaluated (Figure 3A). The highest frequency of DNA methylation (17.2 genes) was detected in HCC-LC with HCV infection compared with the other types of histological or viral status (P = 0.007). These data support the idea that DNA methylation is accumulated in LC with a long period of chronic HCV infection and finally lead to tumorigenesis.


Figure 3
View larger version (22K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 3. Association of DNA methylation with clinical background of HCC patients. (A) Methylation frequency in association with histological findings of adjacent tissue and virus status. Twenty-four genes (see Figure 2) were examined in 38 HCCs. Y axis indicates methylation frequency. HBV, HCV and HBs antigen and anti-HCV antibody negative represent HBV-positive patient, HCV-positive patient and patients negative for both HBV and HCV, respectively. Paired gray bar and black bar indicate sample from adjacent tissue and cancerous tissue in each patient, respectively. (B) Methylation frequencies of cancerous tissue from HCC patient in different methylation pattern groups. Box and whisker plot represents the same as in Figure 2. Genes in the cancer-specific pattern (Type IV) were more frequently methylated in moderately (M) or poorly (P) differentiated cancers than in well (W)-differentiated cancers (left). By contrast, genes in the other types (I, II and II) or all groups (total) were methylated almost equally regardless of cancer differentiation (middle and right). (C) Genes in early pattern (Type I) were more highly methylated in LC than in N or CH. *P < 0.05.

 
Having identified the four distinct types of methylation patterns during progression of the disease, we then assessed the association of these groups with clinicopathological parameters (Figure 3B and C). A clear difference in the number of methylated genes was observed between well-differentiated HCCs and moderately or poorly differentiated HCCs among 11 genes of the cancer-specific pattern type (3.2 genes versus 5.1 genes, P < 0.05). Interestingly, the difference was eliminated in comparison among the other types (I, II and III) or among the total 24 genes. In the early pattern type, mir-219, MMP14 or CIDEA, a significantly higher level of methylation was observed in LC compared with N or CH (mir-219, 87.0 versus 78.7%, P = 0.012; MMP14, 28.4 versus 16.4%, P = 0.017; CIDEA, 34.9 versus 24.9%, P = 0.03; Figure 3C and supplementary Table 3 is available at Carcinogenesis Online). This is consistent with our global analysis showing more methylated loci in LC than in CH (Figure 1B). The difference was not observed in the other group of genes, suggesting that DNA methylation detected on the genes in this group might reflect the long period of chronic inflammation.

Gene network analysis for hypermethylated genes in HCC
To identify the predominant biological networks that are disrupted by DNA methylation and might be linked to tumorigenesis in HCCs, a pathway analysis was conducted on the genes with an average signal ratio of >2.0 in cancerous tissues or adjacent tissues. A series of putative networks with significant P values are shown in Figure 4A. Five biological process subgroups were involved as a target of DNA methylation with a highly significant P value. In particular, genes related to cell adhesion were commonly the target of DNA methylation in cancerous tissue and adjacent tissue. The genes involved in this pathway might play roles in early tumorigenesis in HCC. Figure 4B shows the methylation status of each gene in significant biological process subgroups. Multiple genes in the same family, wingless-type mouse mammary tumour virus integration site (WNT) family, Frizzled (FZD) family, integrin (ITG) family, calcium channel (CACN) family, N-acetylgalactosaminyltransferase (GALNT) family and claudin (CLDN) family, were simultaneously methylated in multiple cancers (range from 10 to 90%). Intriguingly, genes in the ITG family and CLDN family were methylated specifically in cancerous tissue, whereas genes in the other family were methylated in adjacent tissues as well as cancerous tissues, supporting the idea that DNA methylation contributed to multiple stages of cancer formation.


Figure 4
View larger version (22K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 4. Gene network analysis for hypermethylated genes in HCC. (A) Hypermethylated genes were analyzed to identify gene networks composed of biological pathways in adjacent tissue (gray bar) and cancerous tissue (black bar). P values (log10 ratios) for each pathway, which were calculated based on the probability of the network’s assembly, are described above. (B) Methylation status of each gene in the significant biological process. Each biological process is described above the column. Genes corresponding to each process are shown below the column. Gray bar and black bar indicate methylation frequency of each gene in adjacent tissue and cancerous tissue calculated from microarray data, respectively.

 

   Discussion
Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
Supplementary material
 Funding
 References
 
In the present study, MCAM was used to investigate comprehensive profiling of DNA methylation in cancerous tissue and precancerous tissue from HCC patients. This technique enabled evaluation of the global methylation status on CpG island promoters with high specificity and sensitivity (11,12) and demonstrated that aberrant DNA methylation was almost equally distributed across the chromosomes. Validation by highly quantitative pyrosequencing analysis revealed that signal intensity in MCAM was well associated with the methylation level of the promoter regions. The accuracy of our MCAM data is consistent with our previous report showing high specificity and sensitivity by this technique (12). We preliminarily performed an immunocapturing approach followed by DNA microarray analysis using cell line DNA (26). However, we obtained fewer genes compared with MCAM, probably because of less efficiency of the blunt end ligation-mediated PCR during the process.

Hypermethylation was more closely associated with tumorigenesis in LC than in CH, related to HCV infection. Cancers can be classified according to their degree of methylation, and those cancers with high degrees of methylation (CpG island methylator phenotype (CIMP)) represent a clinically and etiologically distinct group that is characterized by epigenetic instability (27). Our MCAM showed that some patients (P1 and P18 in HCC-LC) are extensively methylated, suggesting CIMP exists in HCCs. However, MCAM analysis was limited to 10 patients. To establish the CpG island methylator phenotype in HCC, a more extensive analysis might be desired.

Several studies have shown that HBV-related protein could induce a methylation event (28,29). HBx expression increased total DNA methyltransferase (DNMT) activities by upregulation of DNMTs and selectively promoted regional hypermethylation of specific tumor suppressor genes (29). However, RNA levels of DNMT1 or DNMT3 have largely not proven useful in explaining aberrant patterns of DNA methylation in cancers (30). Therefore, the relationship between viral status and induction of DNA methylation remains controversial. Nevertheless, our analyses clearly revealed that CpG island promoters are methylated in different patterns during progression of the disease, suggesting that multiple mechanisms were involved in the acquisition of epigenetic changes in HCCs.

Clinicopathologically, progression from CH/LC to dysplasias and finally to invasive HCCs has been observed in a long-term follow-up study (2). This model implies progressive accumulation of genetic and epigenetic changes during continuous inflammation. Indeed, it has been suggested that stem and progenitor cells are especially at risk for cancer initiation during states of chronic inflammation (31). In such a situation, repressive histone marks and aberrant DNA methylation could prevent proper differentiation and could predispose to further malignant development (32,33). We have reported CpG island methylation in adjacent histologically normal liver tissues, CH or LC as well as cancerous tissues from HCC patients (5). Here, we expanded the previous findings by global DNA methylation analysis and revealed that DNA methylation in the gene groups (59%, Type I and II) was observed in precancerous lesions, whereas some genes (37%, Type IV) were specifically affected in progressed cancerous lesions. Among multiple molecular pathways in HCC (1,34), there might be a distinguishable pathway depending on DNA methylation during cancer development and progression.

We identified hundreds of hypermethylated genes in HCCs by MCAM, which preferentially detect hypermethylated genes in the samples. Our evaluation of the methylation status of those genes in adjacent precancerous lesions revealed that DNA methylation was not accumulating uniformly in each gene and was classified into four characteristic types (Figure 5). These variations in the DNA methylation pattern may depend on the intrinsic susceptibility to de novo methylation in each gene (35) caused by multiple mechanisms for acquisition of epigenetic changes and could reflect events that are selected during tumor progression (36).


Figure 5
View larger version (20K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 5. Schema of DNA methylation changes observed in progression of HCC. Most CpG islands in normal liver tissue from young individuals are unmethylated except a subset of genes, which was methylated according to the age or tissue specifically. In HCC, the dynamics of DNA methylation-mediated loss of gene function vary. DNA methylation in some genes starts in the earliest phases of tumor progression and is established even in non-cancerous liver tissue (Type I, early pattern), whereas that in the other genes might begin gradually (Type II, progressing pattern) or emerge only in cancer (Type IV, cancer-specific pattern). Intriguingly, in some genes, the methylation level was decreased in CH or LC and was increased again in cancer (Type III, contrary pattern). These distinct patterns of methylation probably represent different pathophysiologic processes. Open box, gray box and black box indicate methylation level in normal liver (NL), CH or LC and cancer (HCC), respectively.

 
In the early pattern (Type I), genes appeared to be methylated in the very early stage of the multistep process of HCCs. mir-219, a gene of this type, might be involved in early tumorigenesis since dysregulation of some microRNA associated with aberrant DNA methylation has been reported in human cancer cells (37). A significantly higher level of methylation was observed in LC than in CH, suggesting that DNA methylation in this type may also reflect the long clinical course of persistent inflammation. In the progressing pattern (Type II), DNA methylation appeared to accumulate according to the progression of the disease. RASSF1A, a well-known tumor suppressor, or PRDM14, a member of PR domain family, and a homolog of PRDM2, a putative tumor suppressor (38), were representative of this type. A subset of cells that acquired DNA methylation in these promoters could be prone to cancer formation. In the contrary pattern (Type III), the DNA methylation level was unexpectedly lower in the adjacent tissues than in normal liver. The reason why DNA methylation was reduced in adjacent tissue in this group is uncertain. It is possible that activation of the genes may be required during the persistent inflammation and regeneration of hepatocytes. Further study to clarify the biological significance of this type of genes may be worthwhile. Nevertheless, this pattern indicated that the DNA methylation level did not always increase during chronic inflammation with persistent viral infection in HCC patients. In the cancer-specific pattern (Type IV), genes were specifically methylated in cancerous tissues. This type includes the genes, which have been known for their key role in tumorigenesis, such as CDKN2A and CCNA1, suggesting that methylation of these genes is cancer relevant and appears to be required for the establishment and progression of cancers. Indeed, methylation in the Type IV genes is prominently associated with moderately/poorly differentiated HCC, whereas this correlation was eliminated in comparison among the other types of genes. Detection of CDKN2A promoter methylation in the adjacent tissues has been reported in two studies, including our own (5,8). However, methylation analysis in those studies was based on methylation-specific PCR, which is very sensitive (27). The discrepancy in the methylation status of adjacent tissues might be due to the different technology used for methylation analysis.

CpG island promoters appear to be methylated in different patterns during progression of the disease, which may represent different pathophysiologic processes. Moreover, methylation of each gene was observed in a different stage of the disease, suggesting that different mechanisms regulate acquisition of these epigenetic changes during cancer formation.

Gene ontology analysis has revealed unequal distribution of hypermethylated genes in HCC patients. Enrichment of genes classified in a specific pathway related to tumorigenesis led us to hypothesize that those genes might be a driving force in HCC. The frizzled-mediated WNT pathway is an evolutionary highly conserved pathway involved in the regulation of proliferation, motility, cell–cell interaction, organogenesis and axis formation (39). Although the oncogenic role of the FZD/β-catenin pathway has been reported in HCC (40), WNT-signaling pathways can act in an antagonistic manner (41,42) and remain unclear in hepatocarcinogenesis. Further studies on the WNT or FZD genes silenced by DNA methylation will be of interest to define their role in HCCs. Integrin-mediated adhesion to the extracellular matrix reportedly plays important roles in regulating cell survival, proliferation and motility, although DNA methylation-mediated silencing of these genes is unknown (43). Calcium signaling is involved in cell proliferation and cell death (44). Inactivation of a calcium channel gene by DNA methylation has been implicated in cancer development (45). In addition, we also found aberrant DNA methylation of several important genes, such as DUSP2 (supplementary Table 4 is available at Carcinogenesis Online) (46) and BMP6 (47), which are known to be closely associated with tumorigenesis, although not detected in a specific pathway. Early detection of aberrant DNA methylation in these critical genes in precancerous tissues might be highly beneficial for risk assessment of HCC.

In conclusion, this is the first global microarray-based DNA methylation analysis in human primary HCCs. We showed here using MCAM technique that aberrant DNA methylation is a frequent event during early to further progressed stages of malignant transformation of HCC. Although further studies are required to determine the functional significance of the hypermethylated genes, our semiquantitative MCAM analysis reveals a global picture of DNA methylation changes in HCCs and provides hundreds of promising markers that could be of diagnostic utility in HCCs and preneoplastic lesions.


   Supplementary material
Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Supplementary material
Funding
 References
 
Supplementary Tables 1–4 and Figures 1 and 2 can be found at http://carcin.oxfordjournals.org/


   Funding
Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Supplementary material
 Funding
References
 
Grant-in-Aid for Cancer Research from the Ministry of Health, Labor and Welfare of Japan; Grant-in-Aid for Scientific Research from Japan Society for Promotion of Science; Japan-China Sasakawa Medical Fellowship to W.G.


   Acknowledgments
 
We thank Ms Ikuko Tomimatsu for technical assistance and Dr Keiko Shinjo for the advice and critical reading of the manuscript.

Conflict of Interest Statement: None declared.


   References
Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Supplementary material
 Funding
 References
 

  1. Thorgeirsson SS, et al. Molecular pathogenesis of human hepatocellular carcinoma. Nat. Genet. (2002) 31:339–346.[CrossRef][Web of Science][Medline]
  2. Lencioni R, et al. Long-term follow-up study of adenomatous hyperplasia in liver cirrhosis. Ital. J. Gastroenterol. (1994) 26:163–168.[Web of Science][Medline]
  3. Murakami Y, et al. Aberrations of the tumor suppressor p53 and retinoblastoma genes in human hepatocellular carcinomas. Cancer Res. (1991) 51:5520–5525.[Abstract/Free Full Text]
  4. Kanai Y, et al. Aberrant DNA methylation precedes loss of heterozygosity on chromosome 16 in chronic hepatitis and liver cirrhosis. Cancer Lett. (2000) 148:73–80.[CrossRef][Web of Science][Medline]
  5. Kondo Y, et al. Genetic instability and aberrant DNA methylation in chronic hepatitis and cirrhosis—A comprehensive study of loss of heterozygosity and microsatellite instability at 39 loci and DNA hypermethylation on 8 CpG islands in microdissected specimens from patients with hepatocellular carcinoma. Hepatology (2000) 32:970–979.[CrossRef][Web of Science][Medline]
  6. Baylin SB, et al. Alterations in DNA methylation: a fundamental aspect of neoplasia. Adv. Cancer Res. (1998) 72:141–196.[Web of Science][Medline]
  7. Jones PA, et al. The fundamental role of epigenetic events in cancer. Nat. Rev. Genet. (2002) 3:415–428.[Web of Science][Medline]
  8. Kaneto H, et al. Detection of hypermethylation of the p16(INK4A) gene promoter in chronic hepatitis and cirrhosis associated with hepatitis B or C virus. Gut (2001) 48:372–377.[Abstract/Free Full Text]
  9. Lee S, et al. Aberrant CpG island hypermethylation along multistep hepatocarcinogenesis. Am. J. Pathol. (2003) 163:1371–1378.[Abstract/Free Full Text]
  10. Costello JF, et al. Aberrant CpG-island methylation has non-random and tumour-type-specific patterns. Nat. Genet. (2000) 24:132–138.[CrossRef][Web of Science][Medline]
  11. Estecio MR, et al. High-throughput methylation profiling by MCA coupled to CpG island microarray. Genome Res. (2007) 17:1529–1536.[Abstract/Free Full Text]
  12. Shen L, et al. Genome-wide profiling of DNA methylation reveals a class of normally methylated CpG island promoters. PLoS Genet. (2007) 3:2023–2036.[Web of Science][Medline]
  13. Toyota M, et al. Identification of differentially methylated sequences in colorectal cancer by methylated CpG island amplification. Cancer Res. (1999) 59:2307–2312.[Abstract/Free Full Text]
  14. Takai D, et al. Comprehensive analysis of CpG islands in human chromosomes 21 and 22. Proc. Natl Acad. Sci. USA. (2002) 99:3740–3745.[Abstract/Free Full Text]
  15. Eisen MB, et al. Cluster analysis and display of genome-wide expression patterns. Proc. Natl Acad. Sci. USA. (1998) 95:14863–14868.[Abstract/Free Full Text]
  16. Ekins S, et al. Pathway mapping tools for analysis of high content data. Methods Mol. Biol. (2007) 356:319–350.[Medline]
  17. Kondo Y, et al. Alterations of DNA methylation and histone modifications contribute to gene silencing in hepatocellular carcinomas. Hepatol. Res. (2007) 37:974–983.[CrossRef][Web of Science][Medline]
  18. Yang AS, et al. A simple method for estimating global DNA methylation using bisulfite PCR of repetitive DNA elements. Nucleic Acids Res. (2004) 32:e38.[Abstract/Free Full Text]
  19. Colella S, et al. Sensitive and quantitative universal pyrosequencing methylation analysis of CpG sites. Biotechniques (2003) 35:146–150.[Web of Science][Medline]
  20. Shen L, et al. Optimizing annealing temperature overcomes bias in bisulfite PCR methylation analysis. Biotechniques (2007) 42, 48, 50, 52 passim.
  21. Li B, et al. The role of chromatin during transcription. Cell (2007) 128:707–719.[CrossRef][Web of Science][Medline]
  22. Heintzman ND, et al. Distinct and predictive chromatin signatures of transcriptional promoters and enhancers in the human genome. Nat. Genet. (2007) 39:311–318.[CrossRef][Web of Science][Medline]
  23. Lin JC, et al. Role of nucleosomal occupancy in the epigenetic silencing of the MLH1 CpG island. Cancer Cell (2007) 12:432–444.[CrossRef][Web of Science][Medline]
  24. Lin CH, et al. Genome-wide hypomethylation in hepatocellular carcinogenesis. Cancer Res. (2001) 61:4238–4243.[Abstract/Free Full Text]
  25. Cameron EE, et al. Synergy of demethylation and histone deacetylase inhibition in the re-expression of genes silenced in cancer. Nat. Genet. (1999) 21:103–107.[CrossRef][Web of Science][Medline]
  26. Weber M, et al. Chromosome-wide and promoter-specific analyses identify sites of differential DNA methylation in normal and transformed human cells. Nat. Genet. (2005) 37:853–862.[CrossRef][Web of Science][Medline]
  27. Issa JP. CpG island methylator phenotype in cancer. Nat. Rev. Cancer. (2004) 4:988–993.[CrossRef][Web of Science][Medline]
  28. Zhong S, et al. Silencing of GSTP1 gene by CpG island DNA hypermethylation in HBV-associated hepatocellular carcinomas. Clin. Cancer Res. (2002) 8:1087–1092.[Abstract/Free Full Text]
  29. Park IY, et al. Aberrant epigenetic modifications in hepatocarcinogenesis induced by hepatitis B virus X protein. Gastroenterology (2007) 132:1476–1494.[CrossRef][Medline]
  30. Eads CA, et al. CpG island hypermethylation in human colorectal tumors is not associated with DNA methyltransferase overexpression. Cancer Res. (1999) 59:2302–2306.[Abstract/Free Full Text]
  31. Baylin SB, et al. Epigenetic gene silencing in cancer—a mechanism for early oncogenic pathway addiction? Nat. Rev. Cancer. (2006) 6:107–116.[CrossRef][Web of Science][Medline]
  32. Ohm JE, et al. A stem cell-like chromatin pattern may predispose tumor suppressor genes to DNA hypermethylation and heritable silencing. Nat. Genet. (2007) 39:237–242.[CrossRef][Web of Science][Medline]
  33. Widschwendter M, et al. Epigenetic stem cell signature in cancer. Nat. Genet. (2007) 39:157–158.[CrossRef][Web of Science][Medline]
  34. Lee JS, et al. A novel prognostic subtype of human hepatocellular carcinoma derived from hepatic progenitor cells. Nat. Med. (2006) 12:410–416.[CrossRef][Web of Science][Medline]
  35. Feltus FA, et al. Predicting aberrant CpG island methylation. Proc. Natl Acad. Sci. USA. (2003) 100:12253–12258.[Abstract/Free Full Text]
  36. Myöhänen SK, et al. Hypermethylation can selectively silence individual p16ink4A alleles in neoplasia. Cancer Res. (1998) 58:591–593.[Abstract/Free Full Text]
  37. Saito Y, et al. Specific activation of microRNA-127 with downregulation of the proto-oncogene BCL6 by chromatin-modifying drugs in human cancer cells. Cancer Cell (2006) 9:435–443.[CrossRef][Web of Science][Medline]
  38. He L, et al. RIZ1, but not the alternative RIZ2 product of the same gene, is underexpressed in breast cancer, and forced RIZ1 expression causes G2-M cell cycle arrest and/or apoptosis. Cancer Res. (1998) 58:4238–4244.[Abstract/Free Full Text]
  39. Hoppler S, et al. Wnt signalling: variety at the core. J. Cell Sci. (2007) 120:385–393.[Abstract/Free Full Text]
  40. Merle P, et al. Functional consequences of frizzled-7 receptor overexpression in human hepatocellular carcinoma. Gastroenterology (2004) 127:1110–1122.[CrossRef]
  41. Westfall TA, et al. Wnt-5/pipetail functions in vertebrate axis formation as a negative regulator of Wnt/beta-catenin activity. J. Cell Biol. (2003) 162:889–898.[Abstract/Free Full Text]
  42. Topol L, et al. Wnt-5a inhibits the canonical Wnt pathway by promoting GSK-3-independent beta-catenin degradation. J. Cell Biol. (2003) 162:899–908.[Abstract/Free Full Text]
  43. Tamura M, et al. PTEN gene and integrin signaling in cancer. J. Natl Cancer Inst. (1999) 91:1820–1828.[Abstract/Free Full Text]
  44. Berridge MJ, et al. Calcium—a life and death signal. Nature (1998) 395:645–648.[CrossRef][Web of Science][Medline]
  45. Toyota M, et al. Inactivation of CACNA1G, a T-type calcium channel gene, by aberrant methylation of its 5’ CpG island in human tumors. Cancer Res. (1999) 59:4535–4541.[Abstract/Free Full Text]
  46. Yin Y, et al. PAC1 phosphatase is a transcription target of p53 in signalling apoptosis and growth suppression. Nature (2003) 422:527–531.[CrossRef][Web of Science][Medline]
  47. Du J, et al. Bone morphogenetic protein 6 inhibit stress-induced breast cancer cells apoptosis via both Smad and p38 pathways. J. Cell. Biochem. (2008) 103:1584–1597.[CrossRef][Web of Science][Medline]
Received February 11, 2008; revised July 9, 2008; accepted July 10, 2008.


Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?


This article has been cited by other articles:


Home page
 GutHome page
T Asselah, I Bieche, A Sabbagh, P Bedossa, R Moreau, D Valla, M Vidaud, and P Marcellin
Gene expression and hepatitis C virus infection
Gut, June 1, 2009; 58(6): 846 - 858.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow Supplementary Data
Right arrow All Versions of this Article:
29/10/1901    most recent
bgn170v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (1)
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Gao, W.
Right arrow Articles by Sekido, Y.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Gao, W.
Right arrow Articles by Sekido, Y.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?