Oncogene (2002) 21, 2926 ± 2937 ã 2002 Nature Publishing Group All rights reserved 0950 ± 9232/02 $25.00 www.nature.com/onc

Identi®cation, using cDNA macroarray analysis, of distinct gene expression pro®les associated with pathological and virological features of hepatocellular carcinoma

Oona Delpuech1, Jean-Baptiste Trabut1, FrancËoise Carnot2, Jean Feuillard3, Christian Brechot1 and Dina Kremsdorf*,1

1INSERM U370, CHU Necker/Institut Pasteur, 75015, Paris, France; 2Service d'Anatomo-Pathologie, HoÃpital EuropeÂen Georges Pompidou, 75015, Paris, France; 3Service d'HeÂmatologie Biologique et EA ATHSCO, HoÃpital Avicenne, 93009, Bobigny, France

It is still unclear as to whether the gene expression human HCC (around 90%) develop in a setting of pro®le in HCV- or HBV-related HCC exhibits a degree chronic in¯ammation, with chronic active hepatitis or of speci®city and whether the development of HCC in a cirrhosis, however, a subset of HCC occurs in the context of cirrhosis in¯uences this gene pro®le. To absence of cirrhosis. Numerous aetiological factors address these issues, the expression pro®les of 15 cases have been identi®ed, including chronic Hepatitis B of HCC were analysed using cDNA macroarray. A Virus (HBV) and Hepatitis C Virus (HCV) infections, global analysis and hierarchical clustering, demonstrated alcoholism, chemical carcinogens (A¯atoxin B1), or the heterogeneity of HCC patterns, with a majority of iron overload (Stuver, 1998). down-regulated genes. Statistical analysis clearly showed A number of genetic and epigenetic alterations have a distinction between the gene expression pro®les of now been reported in HCC (Buendia, 2000; Feo et al., HCV- and HBV-related HCC. HBV-associated HCC 2000; Laurent-Puig et al., 2001). High density exhibited involvement of di€erent cellular pathways, allelotyping and comparative genomic hybridizations those controlling apoptosis, p53 signalling and G1/S have enabled the de®nition of an emerging pattern of transition. In HCV-related HCC we identi®ed a more genetic alterations. HCC appears clearly as a tumour heterogenous pattern with an over-expression of the characterized by considerable loss of heterozygosity TGF-beta induced gene. In HCC developing on non- (LOH) and a low level of microsatellite instability. cirrhotic tissues, beta-catenin encoding gene and genes LOH from chromosomes 1p, 4q, 6q, 8p, 9p, 13q, 16p, implicated in the PKC pathway were speci®cally up- 16q, and 17p has been identi®ed as the most frequent. regulated. In addition, our investigation highlighted a Point mutations have been reported in genes encoding distinct pro®les of TGF-beta superfamily encoding genes p53, beta-catenin and Axin. However, these mutations in well, moderately or poorly di€erentiated HCC. are only detected in around 20 ± 30% (p53, beta- Overall, our study supports the hypothesis that despite catenin) or 5 ± 10% (Axin) of HCCs. RB-1, E- the heterogeneity of the HCC pattern, the large-scale Cadherin, IGF II Receptor and Smad 2 and 4 genes screening of gene expression may provide data signi®cant are less frequently altered. The deregulation of cell to our understanding of the mechanism of liver cycle controlling genes such as cyclin D1, cyclin A and carcinogenesis. p16 has also been reported, decreased p16 expression Oncogene (2002) 21, 2926 ± 2937. DOI: 10.1038/sj/ being associated with hyper-methylation of the regulat- onc/1205392 ing sequences. Other genes implicated in liver regen- eration, such as TGF-a, HGF, TNF-a and IL6 Keywords: hepatocellular carcinoma; Hepatitis B encoding genes may also be implicated in the Virus; Hepatitis C Virus; cDNA macroarray; distinct development of HCC (Fausto, 1991). gene pro®les It has clearly been established that HBV and HCV ONCOGENOMICS infections are major risk factors for the development of HCC (Kasai et al., 1996). A number of genetic Introduction alterations have been described in several HCC related to HCV or HBV (Buendia, 2000; Kusano et al., 1999; Hepatocellular carcinoma (HCC) is one of the most Tornillo et al., 2000). Various carcinogenetic mechan- common malignant tumours worldwide. Most cases of isms have been described in HBV-related HCC (Brechot et al., 2000). The integration of HBV DNA can directly induce chromosomal instability and insertional mutagenesis. Chromosomal DNA instability *Correspondence: D Kremsdorf, INSERM U370, Faculte de is indeed an important consequence of HBV DNA Me decine Necker Enfants Malades, 156 rue de Vaugirard, 75730 integration leading to LOH or translocations. The cis- Paris Cedex 15, France; E-mail: [email protected] Received 24 August 2001; revised 17 January 2002; accepted 8 activation of cellular genes, described to date as a rare February 2002 event in human HCC, may in fact be more prevalent, Macroarray analyses of hepatocellular carcinoma O Delpuech et al 2927 this having been revealed thanks to major advances in 2000; Clark et al., 2000; Golub et al., 1999; Perou et the human genome sequencing project and the larger al., 2000; Welsh et al., 2001). number of HCC cases analysed (Gozuacik et al., 2001). Some recent studies based on cDNA microarray Viral proteins, such as HBx or PreS2/St, are also technology have been performed in HCC. Certain probably implicated in liver cell carcinogenesis. There studies investigated HCC without associated viral is evidence that HBx modulates cell growth and infection, or, in contrast, with HBV- or HCV-related viability, many of its biological e€ects being associated HCC (Lau et al., 2000; Shirota et al., 2001; Xu et al., with its trans-activation and protein-protein interaction 2001); however only one study has so far compared abilities (Murakami, 1999). The HCV genome, a HBV with HCV-related HCC (Okabe et al., 2001). member of the Flaviviridae, does not integrate the Thus it is still unclear whether the gene expression cellular genome (Shimotohno, 2000). Some viral pro®le in HCV- or HBV-related HCCs exhibits a proteins, such as the core, NS3 and NS5A proteins, degree of speci®city, and whether the development of have been reported to a€ect cell proliferation, viability HCC in a setting of cirrhosis in¯uences this gene and response to interferon signalling (Delhem et al., pro®le. To address these issues, we have analysed a 2001; Hayashi et al., 1999; Podevin et al., 2001). In series of HCC cases associated with HCV- or HBV- particular, there is some evidence that HCV core infections and developing on cirrhotic or non-cirrhotic protein, a transcriptional regulator of several cellular tissues. In addition, so as to identify the genes related genes, may induce liver cell transformation under to the progression of carcinogenesis, we also compared certain experimental conditions (Ray and Ray, 2001). the gene expression pro®les of well, moderately or Despite all these ®ndings, no unifying mechanisms poorly di€erentiated HCCs. have been proposed which could account for liver carcinogenesis. There is evidence for an association between certain chromosomal alterations and the HBV Results or HCV status of HCC, but the molecular basis of these observations remains unclear (Buendia, 2000; cDNA macroarray hybridizations and RT ± PCR Tornillo et al., 2000). Furthermore, apart from the validations determination of alpha-fetoprotein serum levels, cur- rently used for the diagnosis of HCC, we still lack So as to avoid individual genetic variations, we reliable markers for early tumour detection (Pateron et compared the gene expression pro®les of tumour (T) al., 1994). Finally, although some prognostic markers and non-tumour (NT) tissues obtained from the same have been reported in certain cases of HCC there is patient (Table 1). To evaluate the reproducibility of clearly a need for identifying more reliable markers. cDNA macroarray data, we performed two indepen- cDNA microarray is a powerful technology which dent experiments for each patient. When the duplicate enables the large-scale screening of gene expression. It experiments were compared, a percentage of reprodu- has already been used to analyse several human cibility ranging from 67 to 90% was observed. Our cancers. cDNA microarray with or without hierarchical results indicated a global reproducibility of results, clustering allows the correlation of gene expression with some discrepancies in the genes expressed. We pro®le with clinical factors, the classi®cation or therefore decided to perform the di€erent analyses de®nition of di€erent tumour types, and the identi®ca- taking account of all the experiments. tion of genes or networks of genes involved in Semi-quantitative RT ± PCR was performed to verify carcinogenesis (Alizadeh et al., 2000; Bertucci et al., macroarray hybridization data. Four paired T and NT

Table 1 Characteristics of the 15 analysed patients Histology of the Histological Case Gender Anti-HCV HBs-Ag non-tumorous tissue grading of tumoral tissue

1M + 7 Chronic active hepatitis Moderately differentiated 2F 7 + Cirrhosis Poorly differentiated 3M 7 + Chronic active hepatitis Well differentiated 4M 7 + Cirrhosis Moderately differentiated 5M + 7 Cirrhosis Moderately differentiated 6M + 7 Cirrhosis Well differentiated 7M 7 + Cirrhosis Moderately differentiated 8M + 7 Cirrhosis Well differentiated 9F + 7 Cirrhosis Well differentiated 10 F + + Cirrhosis Poorly differentiated 11 M 7 + Chronic active hepatitis Moderately differentiated 12 M + 7 Chronic active hepatitis Moderately differentiated 13 M 7 + Cirrhosis Moderately differentiated 14 F + 7 Cirrhosis Well differentiated 15 F + 7 Cirrhosis Moderately differentiated

Oncogene Macroarray analyses of hepatocellular carcinoma O Delpuech et al 2928

Figure 1 Semiquantitative RT ± PCR. The gene expression pro®le of six genes (p-21, c-myc, SHP1, ISPK1, Egr1, beta-actin) was analysed in a setting of four paired tumour (T) and non-tumour (NT) tissues from HCC patients (5, 9, 11 and 12). For each gene, the same dilution of ampli®ed cDNA is shown. Macroarray data are indicated: down-regulated gene (down), up-regulated gene (up), equal gene expression (equal). Di€erent data obtained from duplicate macroarray experiments are indicated (e.g. equal/down). The data were compared after being normalized by the intensity of the 40S ribosomal encoded gene, used as internal control. Discrepancies between results are indicated in italics (p21, patient 5, SHP1, patient 11)

tissues from patients with HCC were used, and the results suggested an association between HCC gene level of RNA expression in seven selected genes were expression pro®les and HBV/HCV status. In addition, evaluated in paired tissue samples (Figure 1). Ampli®ed branch III only included patients who had developed fragments from T and NT tissues were compared after HCC on cirrhotic tissue and not on non-cirrhotic being normalized by the intensity of the 40S ribosomal tissue, independent of their viral status. Hierarchical encoding gene, used as an internal control. In most clustering showed multidimensional variations in gene cases (22/24, 92%), the signal intensity ratio obtained expression pro®les, illustrating the heterogeneity of the from macroarray data correlated well with semi- di€erent HCC patterns (Figure 2a). Despite this quantitative RT ± PCR data. heterogeneity, most genes were found to be down- regulated in HCC, and two distinct clusters of down- regulated genes were observed in the gene dendogram. Hierarchical clustering In the ®rst cluster (Figure 2c), the majority of genes Hierarchical clustering was performed, organising the belonged to the immune system family. This probably 30 experiments and selected genes (105) on the basis of re¯ected the in¯ammatory cellular in®ltrate present in similarities in their gene patterns (Figure 2). Thirteen of non-tumourous tissue. In addition to these genes, two the 15 patients analysed in duplicate clustered together others were found to be down-regulated. Alpha-2- on the terminal branches of the dendogram, indicating macroglobulin, a plasma glycoprotein known to act as that their gene expression pro®les were comparable. an irreversible inhibitor of a variety of proteinases, The resulting dendogram showed four cluster branches binds numerous growth factors, cytokines and hor- (I ± IV) (Figure 2b). The distances between branches I mones, such as TGF-beta (Feige et al., 1996). The and II and between branches III and IV re¯ected the interferon-inducible protein 9-27 (Leu-13) is associated relatedness of the samples. A majority of HBV-related with CD19/CD21/CD81/TAPA-1 on the surface of B HCCs were linked to branches I and II, and a majority cells, generating signal transduction activations and of HCV-related HCCs to branches III and IV. These controlling cell growth (Bradbury et al., 1992). The

Oncogene Macroarray analyses of hepatocellular carcinoma O Delpuech et al 2929

Figure 2 Hierarchical clustering of genes and patients. Hierarchical clustering was performed on data from duplicate experiments concerning the 15 patients tested (a). Red and green colours illustrate over-expressed and under-expressed genes in tumoral tissues, respectively; black and grey represent unchanged or absent gene expression in tumoral and non-tumoral tissues. The dendrogram and coloured images were produced as described by Eisen et al. (1998); the colour scale ranged from saturated green for log ratios 73.417 to saturated red for log ratios 3.417. (b) The resulting dendogram showed four cluster branches (I ± IV), showing the similarities in the expression patterns of experimental samples. Black bars beneath the gene dendogram correspond to an enlarged view of the selected cluster, with down-regulated (c and e) or up-regulated genes (d). B and C: HBV and HCV infections, respectively. PmD and WD: poorly/moderately and well di€erentiated, respectively. Ci and NC: cirrhotic and non-cirrhotic tissue, respectively

Oncogene Macroarray analyses of hepatocellular carcinoma O Delpuech et al 2930

Figure 3 Hierarchical clustering of genes. Patients were arranged in terms of their viral aetiology and histological features. Red and green colour illustrate over-expressed and under-expressed genes in tumoral tissues, respectively; black and grey represent unchanged or absent of gene expression in tumoral and non-tumoral tissues. The dendrogram and coloured images were produced as described by Eisen et al. (1998); the colour scale ranged from saturated green for log ratios 73.417 to saturated red for log ratios 3.417. (a) Group of nine clustered genes di€erentially expressed in HCV- or HBV-related HCC (HBV/HCV). (b) Group of three genes preferentially down-regulated in moderately/poorly di€erentiated (PmD) HCC and up-regulated in well di€erentiated (WD) HCC

second cluster included nine genes that were down- a cluster of three genes which were preferentially down- regulated, mostly in patients clustering on branches I regulated in moderately/poorly di€erentiated HCC and and II of the dendogram (Figure 2e). Among these up-regulated in well di€erentiated HCC (Figure 3b). genes (aminoacylase 1, STAT-induced STAT inhibitor One of these genes was aldehyde oxidase, which is 3, haemoglobin, metallothionein-III), some have al- involved in NAD oxidation (Mira et al., 1995). The ready been described as being modulated in HCC or down regulation of plasminogen had been found cancer-derived cell lines (Garrett et al., 1999; Miller et previously in a global HCC analysis, but no data was al., 1989; Okabe et al., 2001). In addition to these available concerning a possible association with the clusters of down-regulated genes, a cluster with a histological di€erentiation of HCC status (Okabe et al., majority of up-regulated genes was observed (Figure 2001; Shirota et al., 2001). No clear di€erence in gene 2d). Two of the four up-regulated genes corresponded expression in the cluster could be seen between HCC to those implicated in DNA repair and in the developing in the context of cirrhosis or not (data not ubiquitin/proteasome pathway (Ubiquitin-conjugating shown). However, interpretation of this hierarchical enzyme E2, UV excision repair protein-Rad23), dereg- clustering analysis was hampered by the small number ulation of which may contribute to carcinogenesis of HCC cases developing on non-cirrhotic tissues. (Schauber et al., 1998; Weissman, 2001). We then performed a second hierarchical clustering, Identification of genes differentially expressed in HCCs based on gene analysis alone and not on both genes and patients, to test for the possible association of gene Because hierarchical clustering highlighted multidimen- expression pro®les and virological or pathological sional variations in HCC gene patterns, we examined features of HCC (Figure 3). The hierarchical clustering genes that were up- or down-regulated in more than of genes with patients arranged according to their 30% of the 30 comparisons so as to better de®ne genes HBV/HCV status, highlighted a group of nine which are di€erentially expressed in HCC. Of the 1098 clustered genes di€erentially expressed in HCV- or genes modulated in HCC, 14 or the 553 up-regulated HBV-related HCC (Figure 3a). This gene cluster was genes and 30 of the 545 down-regulated genes were identical to one of those described above in the global found to be modulated in at least 30% of the cluster analysis (Figure 2e). This was consistent with an comparisons (Table 2). Among those genes, 10 were association between the down-regulation of these genes found to be down-regulated in at least 50% of the and HBV infection. Hierarchical clustering of genes comparisons. As previously illustrated by the hierarch- based on the di€erentiation status of HCC highlighted ical clustering, a large number of genes were down-

Oncogene Macroarray analyses of hepatocellular carcinoma O Delpuech et al 2931 Table 2 Genes up or down-regulated in HCCs tumour tissuesa Category Gene Name GenBank accession

Up-regulated genes in HCCs tumor tissues Extracellular matrix and cytoskeleton Fibronectinc X02761; K00799; K02273 Tubulin alpha 1 subunit K00558 Matrix metalloproteinase 14 D26512; X83535 Osteonectin; SPARC J03040 DNA damage repair protein UV excision repair protein; RAD23A D21235 Ubiquitin-conjugating enzyme E2 M74524 Trafficking protein Neutrophil gelatinase-associated lipocalin precursor lipocalin 2 X99133 TRAM protein X63679 ADP/ATP carrier protein J02683 Transcription factor High mobility group protein M23619 Growth factor Insulin-stimulated protein kinase 1 U08316 GTP binding protein Transforming protein rhoA H12 L25080 IFN response Interferon gamma antagonist A25270 Cytokines Macrophage inhibitory cytokine 1 AF019770 Down-regulated genes in HCCs tumor tissues Immune system IgG, IgG Kb M63438+U72063 IgG3; IgG1 L; IgG1 K; IgG1 Fcb D78345+Y14737 IgA1; IGHAb J00220+S71043 IgC mu heavy chain constant regionb X57086; X57331 IgG receptor FC large subunit P51 U12255 Lymphocyte antigen M81141 Metabolic pathway Betaine-homocysteine S-methyltransferaseb U50929 Methylenetetrahydrofolate dehydrogenase J04031 Aldehyde oxidase L11005 Uridine diphosphoglucose pyrophosphorylase U27460 Metalloproteinase inhibitor 1 X03124 Metallothionein-IIIb D13365; M93311 Growth factor alpha-2-macroglobulinb M11313 alpha-2-macroglobulin receptor-associated protein M63959 TGF-betac S81439 hepatocyte growth factor-like protein D49742; S83182 NGF-inducible anti-proliferative protein PC3 U72649 Early growth response protein 1b X52541; M62829 Insulin-like growth factor-binding protein 3c M31159; M35878 Insulin-like growth factor binding protein 4 M62403 IFN response Interferon-induced protein MXA M33882 Interferon-inducible protein 9-27 J04164 Hormones, receptors Nuclear hormone receptor L76571 Extracellular matrix Plasminogenb,c X05199 Haemoglobin alpha subunitb,c V00491 Decorin M14219 Tumour associated c-myc oncogene V00568 Cell cycle cyclin-dependent kinase inhibitor 1 U09579; L25610 DNA-binding and chromatin protein DNA-binding protein CPBP U44975 GTP binding protein Transforming protein rhoB X06820 aReport of genes up- or down-regulated in more than 30% of the comparison ± tumour versus non tumour. bGenes down-regulated in more than 50% of the HCCs. cGenes previously reported in the literature regulated when compared with over-expressed genes. together, these results support the hypothesis that the Several of these genes may be integrated in the deregulation of genes encoding proteins associated with networks of genes acting at di€erent levels of cell the cytoskeleton plays a role in liver carcinogenesis (Le signalling. Bail et al., 1997). It should be noted that, osteonectin, A ®rst group of genes involved in the immune an extra-cellular matrix-associated glycoprotein, reg- response (immunoglobulins, lymphocyte antigen) was ulates the activation of matrix metalloproteinase 2, found to be down-regulated, as illustrated by the collagen type I and TGF-beta 1, and has been reported hierarchical clustering (Figure 2c). A second group of to be involved in oesophageal and breast cancers genes, regulating the composition of the extra-cellular (Francki et al., 1999; Gilles et al., 1998; Porte et al., matrix and the cytoskeleton (Fibronectin, Tubulin 1998). Like beta-catenin, osteonectin is involved in alpha1, matrix metalloproteinase 14, osteonectin regulating many of the extra-cellular matrix proteins SPARC, RhoA) was found to be up-regulated. These implicated in the development of cancer. A third group genes have been described as being up-regulated in of genes, related to cell growth and the regulation of variety of carcinomas and in HCC (Kamai et al., 2001; di€erentiation, was found to be down-regulated in Polette and Birembaut, 1998; Porte et al., 1998; HCC. The expression of NGF-inducible anti-prolifera- Takano et al., 2001; Takayasu et al., 2001). Taken tive protein PC3 (BTG2) and insulin-like growth factor

Oncogene Macroarray analyses of hepatocellular carcinoma O Delpuech et al 2932 binding protein 3 (IGFBP3) has been reported to be This was consistent with the more heterogeneous regulated by p53 (Rouault et al., 1996; Shen et al., pattern of gene expression in HCC developing in a 1999). This observation may re¯ect a down-regulation context of cirrhosis, compared with the pattern seen in of the p53 pathway; interestingly, we also observed the non-cirrhotic tissue. This may re¯ect the fact that down-regulation of early growth response protein 1, numerous factors are involved in the development of implicated in p53 trans-activation (Liu et al., 1998). A cirrhosis and thereafter of HCC. In non-cirrhotic fourth group of genes involved in protein tracking tissues, we also noted an over-expression of alpha1- was found to be over-expressed, including ADP/ATP and beta-catenin and type II cytoskeletal 8 keratin carrier protein and TRAM encoding genes (Hegde et encoding genes, which has been reported in di€erent al., 1998). It should be noted that an alteration to types of cancers, including HCC (Endo et al., 2000; Xu ADP/ATP carrier protein expression, inducing carrier et al., 2001). In addition, the down-regulation of a dysfunction, has been described during some viral PKC pathway inhibitor (protein kinase C inhibitor 1) infections (Schulze and Schultheiss, 1995). Finally, a and the up-regulation of a PKC-regulated gene (L- ®fth group of genes controlling di€erent metabolic lactate dehydrogenase M subunit) suggested activation pathways was found to be under-expressed, including of the PKC pathway (Juengel et al., 1998; Jungmann et methylene-tetrahydrofolate dehydrogenase and the al., 1998). metalloenzymes (metallothionein III, betaine-homocys- In well di€erentiated HCC, four of the 33 up- tein S-methyltransferase, metalloproteinase inhibitor 1) regulated and two of the 34 down-regulated genes were (Hidalgo and Carrasco, 1998; Hol et al., 1998; Iredale speci®cally modulated (Table 5). One of the 16 up- et al., 1992; Millian and Garrow, 1998). regulated genes and seven of the 35 down-regulated genes were found to be speci®cally modulated in moderately/poorly di€erentiated HCC (Table 5). Association of gene expression patterns with virological Among these genes, macrophage inhibitory cytokine and pathological features of HCC 1, a member of the TGF-beta superfamily, was found We attempted to establish links between gene expres- to be speci®cally down-regulated in 60% and up- sion and viral status (HBV/HCV) (Table 3), the regulated in 40% of cases of well and moderately/ development of HCC in a context of cirrhosis (Table poorly di€erentiated HCC, respectively (Fairlie et al., 4), and the di€erentiation status of HCC (Table 5). 1999). In addition, transforming growth factor beta- Among the 90 and 43 genes whose expression was induced encoding gene was found to be up-regulated in modulated in HBV- and HCV-related HCC, respec- more than 50% of well-di€erentiated HCCs. This may tively, 37 exhibited speci®cally modulated expression re¯ect a di€erential regulation of the TGF-beta pattern in HBV-related HCC, as opposed to only one superfamily encoding genes at di€erent stages of liver in the HCV-related HCC (Table 3). This was consistent carcinogenesis. The expression of hepatoma-derived with the more heterogeneous gene expression pro®le growth factor (HDGF), a regulator of liver cell growth, seen in HCV, rather than HBV-related HCC, as was found to be up-regulated in more than 50% of illustrated by the gene hierarchical clustering (Figure well di€erentiated HCC and may also contribute to 3a). However, transforming growth factor beta-induced liver cancer progression (Nakamura et al., 1989). encoding gene was found to be up-regulated in 44% of Another network of genes, implicated in several HCV-related HCCs. An over-expression of TGF-beta metabolic pathways, was found to be altered in has been reported in HCC, including HCV-related moderately or poorly di€erentiated HCCs. In parti- HCC (Abou-Shady et al., 1999; Kim et al., 2000; cular, cytosolic superoxide dismutase 1 (SOD1) expres- Thorgeirsson et al., 1998). Our result suggested the sion was found to be down-regulated. It should be modulation of the TGF-beta pathway in HCV-related noted that, SOD2 was reported to be modulated HCC. In HBV-related HCC, six of the 25 up-regulated during another study of HCC patterns (Yamashita et and 31 of the 65 down-regulated genes were found to al., 2001). Finally, plasminogen and haemoglobin be speci®cally modulated. In particular, they included encoding genes have also exhibited a di€erential genes implicated in the control of G1/S transition expression in moderately/poorly di€erentiated HCC, (cyclin-dependent kinase regulatory subunit, cyclin- as previously reported in two studies of HCC using dependent kinase inhibitor 17(p21)), DNA damage cDNA microarray technology (Okabe et al., 2001; and apoptosis (PIG7, growth arrest and DNA-damage- Shirota et al., 2001). inducible protein (GADD45)) and growth factor synthesis (hepatocyte growth factor activator, growth factor receptor-bound protein 2 (GRB2)) (Demetrick et Discussion al., 1996; Dotto, 2000; Lowenstein et al., 1992; Okajima et al., 1997; Polyak et al., 1997; Zhan et al., 1994). The most important ®nding of this study is that despite In HCC developing on non-cirrhotic tissue, 12 of the considerable heterogeneity in gene expression pro®les 32 up-regulated and ®ve of the 31 down-regulated in HCC, a link could be seen between certain gene genes were speci®cally modulated (Table 4). Only four expression patterns and pathological or virological of the 30 down-regulated genes, and none of the 12 up- features of HCC. The heterogeneity of our macroarray regulated genes were found to be speci®cally modu- pattern did not result from individual genetic varia- lated in HCC developing on cirrhotic tissue (Table 4). tions, since we compared tumour and non-tumour

Oncogene Macroarray analyses of hepatocellular carcinoma O Delpuech et al 2933 Table 3 Genes speci®cally regulated in HBV- or HCV-related HCCsa Gene name GenBank accession

Gene specifically modulated in HBV-related HCCs Up regulated Cell cycle CDC28 protein kinase 2 X54942 CDC27HS protein U00001 Extracellular matrix Integrin beta 4 X53587; X52186 Desmoplakin I & II M77830; J05211 Metabolic pathway Procallagen C proteinase M22488+U50330 Growth factor Platelet-derived growth factor A subunit X06374 Down regulated Metabolic pathway CMP-N-acetylneuraminate-beta-galactosamide-alpha-2,6-sialytransferase D13365; M93311 Betaine-homocysteine S-methyl-transferase U50929 Dihydro-orotate dehydrogenaseb M94065 Cytidine deaminase L27943 Aldehyde oxidaseb L11005 Aminoacylase 1 L07548 Methylenetetrahydrofolate dehydrogenaseb J04031 Metallothionein-IIIb X62822 Immune system IgG receptor FC large subunit P51b U12255 Major histocompatibiity complex enhancer-binding protein MAD3 M69043 Leukocyte IgG receptor J04162 FC-epsilon-receptor gamma subunit M33195 HLA-DR antigen-associated invariant subunit X00497 Growth factor Hepatocyte growth factor activator D14012 Growth factor receptor-bound protein 2 isoform L29511; M96995 Epidermal growth factor receptor, oncogene ERBBc X00588; K03193 EGF response factor 1 X79067 IFN response STAT-induced STAT inhibitor 2b AB004903 STAT-induced STAT inhibitor 3b,c Hormone Insulin-induced protein 1b U96876 Cytokine receptor Interleukin-1 receptor antagonist protein precursor M63099 Apoptosis PIG7; gene induced by p53 AF010312 Growth arrest and DNA-damage-inducible protein M60974 Cell cycle Cyclin-dependent kinase inhibitor 1; P21 U09579; L25610 G protein Guanine nucleotide-binding protein G U31383 Transcription factor Signal transducer and activator of transcription 3b L29277 Polyhomeotic 2 homolog U89278 Liver specific Haemoglobin alpha subunitb,c V00491 Regulatory protein Gravin M96322 Binding protein DNAX activation protein 12 AF019562 Unknown function KIAA0022 Gene D14664 Gene specifically modulated in HCV-related HCCs Up regulated Growth factor Transforming growth factor, beta-inducedb M77349 aGenes speci®cally modi®ed (w2 test) in more than 30% of the comparisons. bGenes modulated in more than 50% of the comparisons. cGenes previously reported in the literature

tissues from the same patients. The heterogeneity of our understanding of liver carcinogenesis. In particular, gene expression may result from the complexity of we noted deregulation of expression of certain gene human HCC, developing in a context of di€erent encoding carrier proteins, as well as proteins implicated pathological liver conditions (i.e. cirrhotic or non- in insulin (SOCS2, IGFBP2, IGFBP3, GRB2, insulin- cirrhotic tissues), associated or not with HBV/HCV induced protein) and thyroid-dependent signal path- infections or related to other aetiological factors. ways (HMG I, ADP/ATP carrier, nuclear hormone Hierarchical clustering and cDNA macroarray-based receptor). Insulin and thyroid hormones have clearly analyses allowed us to identify a number of potential been shown to act as important regulators of liver cell gene networks which were deregulated in the context of proliferation and di€erentiation (Veneziani et al., liver carcinogenesis. Some of these genes, such as beta- 1990). Similarly, the expression of several genes catenin, p53, TGF-beta and genes involved in G1/S controlling di€erent metabolic pathways, such as transition, have previously been incriminated in HCC aldehyde oxidase, metallothionein-III, superoxide dis- or the development of other cancers (Abou-Shady et mutase 1, was modi®ed. al., 1999; Buendia, 2000; Ito et al., 1999; Sato et al., HBV and HCV both induce in¯ammation and 1999; Vousden, 2000; Wong et al., 2001; Yuspa et al., cirrhosis, but the mechanisms causing carcinogenesis 1994). In addition, our analysis has highlighted other remain unclear. Several reports have suggested an regulatory pathways which may be of importance to association between some chromosomal alterations and

Oncogene Macroarray analyses of hepatocellular carcinoma O Delpuech et al 2934 Table 4 Genes speci®cally regulated in HCC developed on cirrhotic or non cirrhotic tissuea Category Gene name GenBank accession

Genes specifically modulated in HCC developed on non cirrhotic tissue Up regulated Metabolic pathway Lactate dehydrogenase A X02152 Aldehyde oxidaseb L11005 Nucleoside-diphosphate kinase Y07604 Cytoskeleton Alpha 1 cateninb,c D13866; D14705 Beta cateninc X87838; Z19054 Collagen 6 alpha 1 subunit X15879 Type II cytoskeletal 8 keratin M34225 Growth factor, receptor Growth factor receptor-bound protein 2 and 3 isoforms L29511; M96995 Hormone Progesteron receptor-associated protein (HSC70) U28918 Tumour associated c-jun N-terminal kinase 2 L31951 shb proto-oncogene X75342 Transport protein ADP/ATP carrier proteinb J02683 Down regulated Immune system FC-epsilon-receptor gamma subunitb M33195 Cytokine receptor Interleukin-1 receptor antagonist protein precursor M63099 Regulatory protein Protein kinase C inhibitor 1 U51004 GTP binding protein Rho GTPase activating protein 4b X78817 Guanine nucleotide-binding protein G(l) alpha subunit M17219 Genes specifially modulated in HCC developed on cirrhotic tissue Down regulated Growth factor Insulin-like growth factor binding protein 2 M35410 Immune system IgG, IgG Kb M63438+U72063 IgG3; IgG1 L; IgG1 K; IgG1 Fcb D78345+Y14737 IgC mu heavy chain constant regionb,c X57086; X57331

aGenes speci®cally modi®ed (w2 test) in more than 30% of the comparisons. bGenes modulated in more than 50% of the comparisons. cGenes previously reported in the literature

Table 5 Genes speci®cally regulated in well or moderately/poorly di€erentiated HCCsa Category Gene name GenBank accession

Genes specifically modulated in well differentiated HCCs Up regulated Transcription factor Zinc finger protein L08096; S69339 Cytoskeleton Integrin alpha 6 X53586; X59512 Growth factor Hepatoma-derived growth factorb,c D16431 Transforming growth factor, beta-inducedb M77349 Down regulated Cytokine Macrophage inhibitory cytokine 1b AF019770 Immune system Lymphocyte antigenb M81141 Genes specifically modulated in moderately/poorly differentiated HCCs Up regulated Transcription factor c-myc purine-binding transcription factor pufc L16785+M36981 Down regulated Metabolic pathway Aldehyde oxidase L1105 Dihydro-orotate dehydrogenase precursor M94065 Methylenetetrahydrofolate dehydrogenase J04031 Cytosolic superoxide dismutase 1 (SOD1) K00065; X02317 Liver specific Plasminogenb,c X05199 Haemoglobin alpha subunitb,c V00491 Cytoskeleton Vitronectin X03168

aGenes speci®cally modi®ed (w2 test) in more than 30% of the comparisons. bGenes modulated in more than 50% of the comparisons. cGenes previously reported in the literature

HBV- and HCV-related HCC. Di€erent studies have common genetic and epigenetic alterations, involve led to report that a number of genetic alteration are distinct pathways. Consistent with these results, other common to HCV- and HBV-related HCCs (Buendia, reports, focusing on patients with chronic hepatitis 2000; Honda et al., 2001; Kusano et al., 1999; Laurent- with or without HCC, have also suggested a distinct Puig et al., 2001; Tornillo et al., 2000). Our study adds pattern of gene expression association with HBV and further support to the hypothesis that liver carcinogen- HCV infections (Benvegnu and Alberti, 2001; Benveg- esis induced by HBV and HCV may, in addition to nu et al., 2001; Honda et al., 2001; Kusano et al., 1999;

Oncogene Macroarray analyses of hepatocellular carcinoma O Delpuech et al 2935 Laurent-Puig et al., 2001; Okabe et al., 2001). Materials and methods Strikingly, HCV-associated HCC exhibits a more heterogeneous gene expression pro®le when compared Patients and tissue samples with HBV-related HCC. The present study has in particular highlighted the link between cell cycle- Tissues from the tumour and from surrounding non- tumorous liver were obtained from 15 patients undergoing controlling gene expression deregulation and HBV surgical hepatectomy or liver transplantation for hepatocel- infection. The expression of some of these genes lular carcinoma (Table 1). Cancerous and non-cancerous (GADD45, PIG7) has been reported to be p53- tissues were meticulously dissociated, snap-frozen in liquid dependent (Polyak et al., 1997; Zhan et al., 1994). nitrogen, and stored at 7808C until use. The tumoral or non- Their down-regulation implies a modulation of the p53 tumoral localizations of samples used for RNA puri®cation pathway in HBV-related HCC, as already suggested were veri®ed histologically. Eight patients were HCV positive, (Laurent-Puig et al., 2001; Teramoto et al., 1994). six were HBV-positive and one was co-infected with HCV However, we did not detect any alteration to p53 and HBV. Histological analysis of non-tumour tissues encoding gene expression, so that our observations showed that 11 HCCs had developed on cirrhotic and four may therefore re¯ect post transcriptional and/or on non-cirrhotic tissues. The tumour was classi®ed histolo- gically, using the Edmondson grading system; ®ve HCCs translational regulations, associated or not with p53 were de®ned as well di€erentiated, and 10 as moderately or gene mutations, undetectable using our cDNA-macro- poorly di€erentiated. array technique. A further potentially important result of our study is the over-expression of TGF-beta induced-encoding gene in HCV-related HCC. To our RNA extraction and macroarray hybridizations knowledge, a speci®c alteration to the TGF-beta Frozen tissues were pulverized in liquid nitrogen using a pathway in HCV-related HCC has only been reported pestle and mortar. Frozen tissue powder was immediately in one study, to date (Kim et al., 2000). placed in Trizol Bu€er (Life Technologies, BRL), and Our ®ndings also provide further support for the homogenized, and total RNA extraction was performed hypothesis of di€erent pathways being involved in liver according to the manufacturer protocol, before storage at 7808C until analysis. Probe synthesis (7 mg of total RNA carcinogenesis, associated or not with cirrhosis. A few from tumour and non-tumour tissues) and macroarray genetic alterations have been reported speci®cally in hybridization (Cancer 1.2 ± 1 186 genes ± CLONTECH) were HCC developing in a context of non-cirrhotic HCC, performed as indicated by the manufacturer (CLONTECH) such as 13q and 8p LOH (Laurent-Puig et al., 2001; using [a-P32]-dATP. Hybridized membranes were exposed to Terris et al., 1999). The results of the present study are a Phosphor Screen overnight and scanned using a Storm 840 consistent with the involvement of the PKC dependent (Molecular Dynamics, Inc, Watertown, MA, USA). For each pathway in such tumours. We also observed an up- patient, comparison of the gene expression pro®les of tumour regulation of the beta-catenin-encoding gene. Muta- and non-tumour tissue was performed in duplicate during tions in the beta-catenin sequence have been associated independent experiments (probe synthesis and macroarray with low LOH tumour pro®le and HBV negative status hybridization). The radioactive intensity of each spot was analysed with Atlas Image 1.5 software (CLONTECH). HCCs (Laurent-Puig et al., 2001). An over-expression Intensity which was twice as high as the background was of beta-catenin has been reported to have pathological de®ned as a positive signal. The ratio of tumour versus non- and prognostic signi®cance (Nhieu et al., 1999; Wong tumorous intensity was calculated and normalized using the et al., 2001). Our results therefore suggest a speci®c mean values for the nine housekeeping genes. Genes with a altered expression of the beta-catenin encoded gene in ratio 52or05ratio 40.5 were de®ned as up-regulated or tumours developing in the absence of cirrhosis. down-regulated, respectively. Finally, our investigation highlighted a distinct pro®les of TGF-beta-superfamily encoding genes in Semi-quantitative RT ± PCR analysis HCC with varying degrees of di€erentiation. TGF-beta has been extensively studied as regulator of liver cell To verify macroarray hybridization data, semi-quantitative RT ± PCR was performed. To achieve this, four paired proliferation and di€erentiation (Fausto, 1991); our tumour (T) and non-tumour (NT) tissue samples from study may therefore help to dissect these network patients with HCC were used, and the RNA expression level alterations during di€erent stages of liver carcinogen- of seven selected genes were evaluated in these paired tissues. esis. Random cDNA synthesis was performed using 15 mg of total cDNA microarray is still an emerging technology RNA, with ®nal concentrations of the following components with several experimental pitfalls (Knight, 2001). (Promega): 8 u of M-MLV, 2 mM dNTPs, 16Bu€er, 0.01 u Therefore, before launching large-scale studies, it is RNasine. Semi-quantitative PCR was performed on 2 ml important to validate its feasibility in the speci®c (0.4 mg of the starting total RNA) of di€erent cDNA experimental context of the tumours under analysis. dilutions (1, 1/10, 1/100) with 30 cycle numbers. PCR The number of patients in the present study is small, ampli®cations were achieved using the following components (Promega) and conditions: 0.4 mM dNTPs, 0.3 mM primers, but the ®ndings support the hypothesis that despite the 16bu€er, 0.1 u Taq; denaturation at 948C (5 min); 30 cycles HCC pattern heterogeneity, the large-scale screening of at 948C (1 min), primer hybridization temperature (1 min) gene expression may provide data signi®cant to our and 728C (1 min), then with a 728C extension (10 min). The understanding of the mechanism of liver carcinogenesis. sequences of the PCR primer pairs of the seven selected genes It should also encourage studies on larger patient series, with the MgCl2 concentration, the primer hybridization so as to validate the clinical impact of this approach. temperature and the expected size of the ampli®ed fragment,

Oncogene Macroarray analyses of hepatocellular carcinoma O Delpuech et al 2936 were as follows: 40S, 5'-AGC GGA CGC AAC ATG CCA dendograms, the pattern and length of branches re¯ect the GTG G-3',5'-TCG CCG ATC AGC TTC AGC TCT T-3' relatedness of samples (patients or genes). (3 mM,608C, 103 bp); p-21, 5'-GGA AGA CCA TGT GGA So as to assess the signi®cance of gene expression pro®les CCT GT-3',5'-CCC AGC ACT CTT AGG ACC CTC-3' related to clinical factors (HBV/HCV, cirrhotic/non-cirrhotic, (1.5 mM,608C, 400 bp), c-myc,5'-CCA GCA GCG ACT well/moderately-poorly di€erentiated) statistical analyses CTG AGG-3',5'-GTT GTG TGT TCG CCT CTT GA-3' were performed on genes modulated in at least 30% of the (1.5 mM,608C, 337 bp), SHP1, 5'-GAC CCC ATG GTC experiments. The relative gene expression pro®les were GGG CCA G-3',5'-CAC CCG AGA GGT GGA GAA arranged in two groups and a w2-test was performed. In the AGGC-3' (1.5 mM,588C, 218 bp), beta-actin, 5'-CAA AGA ®rst group up-regulated genes (ratio 52) were compared with CCT GTA CGC CAA CAC A-3',5'-AAC CGA CTG CTG non-up-regulated genes, while in the second group, down- TCA CCT TCAC-3' (2 mM,608C, 434 bp), Egr1, CLON- regulated genes (05ratio40.5) were compared with non- TECH primers (2 mM,568C, 281 bp), ISPK1, CLONTECH down-regulated genes. The co-infected patient was excluded primers (2 mM,538C, 200 bp). Reaction of the 5 mlofRT± from statistical analysis of HBV/HCV comparisons. PCR products from each cDNA dilution (1, 1/10, 1/100) were analysed by electrophoresis on 1.5% agarose gel and visualized by ethidium bromide staining. Gel electrophoresis images were saved using Fisher Bioblock Scienti®c software, and the original intensity of each speci®c band was quanti®ed Acknowledgements with NIH 1.62 software. The data on paired T and NT We would like to thank the following for their invaluable tissues from each patients were compared after normalization help in collecting human liver samples: Prof Dominique by the intensity of the 40S ribosomal encoding gene, used as Franco, Dr Cyrille Feray and all the Pathological Anatomy an internal control. Department at the Antoine Beclere and Paul Brousse Hospitals. We would like to thank Yves Chretien for advice on statistical analysis, Dominique Daudry and Cluster and statistical analysis Fanny Baran-Marszak for providing primer sequences Cluster analysis was performed using Cluster software and and PCR conditions. This work was supported by grants visualized using TreeView software (http://rana.Standord.E- from Institut National de la Recherche Me dicale (IN- DU/Software) (Eisen et al., 1998). Ratio values were log SERM), from the Ligue contre le Cancer (LNC), from transferred, ®ltered with a 60% ®lter, and applied to average Association pour la Recherche contre le Cancer (ARC) and uncentered linkage clustering. The hierarchical clustering from the European Community. Oona Delpuech was algorithm organizes experimental samples on the basis of supported by a fellowship from Agence Nationale de la overall similarities in their gene expression pro®le patterns. In Recherche sur le Sida (ANRS).

References

Abou-Shady M, Baer HU, Friess H, Berberat P, Zimmer- Eisen MB, Spellman PT, Brown PO and Botstein D. (1998). mann A, Graber H, Gold LI, Korc M and Buchler MW. Proc. Natl. Acad. Sci. USA, 95, 14863 ± 14868. (1999). Am.J.Surg.,177, 209 ± 215. Endo K, Ueda T, Ueyama J, Ohta T and Terada T. (2000). Alizadeh AA, Eisen MB, Davis RE, Ma C, Lossos IS, Hum. Pathol., 31, 558 ± 565. RosenwaldA,BoldrickJC,SabetH,TranT,YuX,Powell Fairlie WD, Moore AG, Bauskin AR, Russell PK, Zhang HP JI,YangL,MartiGE,MooreT,HudsonJrJ,LuL,Lewis and Breit SN. (1999). J. Leukoc. Biol., 65, 2±5. DB,TibshiraniR,SherlockG,ChangWC,GreinerTC, Fausto N. (1991). Prog. Growth Factor Res., 3, 219 ± 234. Weisenburger DD, Armitage JO, Warnke R, Levy R, Feige JJ, Negoescu A, Keramidas M, Souchelnitskiy S and Wilson W, Grever MR, Byrd JC, Botstein D, Brown PO Chambaz EM. (1996). Horm. Res., 45, 227 ± 232. and Staudt LM. (2000). Nature, 403, 503 ± 511. Feo F, Pascale RM, Simile MM, De Miglio MR, Muroni Benvegnu L and Alberti A. (2001). Antiviral Res., 52, 199 ± MR and Calvisi D. (2000). Crit.Rev.Oncog.,11, 19 ± 62. 207. Francki A, Bradshaw AD, Bassuk JA, Howe CC, Couser Benvegnu L, Noventa F, Bernardinello E, Pontisso P, Gatta WG and Sage EH. (1999). J. Biol. Chem., 274, 32145 ± A and Alberti A. (2001). Gut, 48, 110 ± 115. 32152. Bertucci F, Houlgatte R, Benziane A, Granjeaud S, Adelaide Garrett SH, Sens MA, Shukla D, Nestor S, Somji S, Todd JH J, Tagett R, Loriod B, Jacquemier J, Viens P, Jordan B, and Sens DA. (1999). Prostate, 41, 196 ± 202. Birnbaum D and Nguyen C. (2000). Hum. Mol. Genet., 9, Gilles C, Bassuk JA, Pulyaeva H, Sage EH, Foidart JM and 2981 ± 2991. Thompson EW. (1998). Cancer Res., 58, 5529 ± 5536. Bradbury LE, Kansas GS, Levy S, Evans RL and Tedder TF. Golub TR, Slonim DK, Tamayo P, Huard C, Gaasenbeek (1992). J. Immunol., 149, 2841 ± 2850. M, Mesirov JP, Coller H, Loh ML, Downing JR, Caligiuri Brechot C, Gozuacik D, Murakami Y and Paterlini-Brechot MA, Bloom®eld CD and Lander ES. (1999). Science, 286, P. (2000). Semin. Cancer Biol., 10, 211 ± 231. 531 ± 537. Buendia MA. (2000). Semin. Cancer Biol., 10, 185 ± 200. Gozuacik DK, Murakami Y, Saigo K, Chami M, Mugnier C, Clark EA, Golub TR, Lander ES and Hynes RO. (2000). Lagorce D, Okanoue T, Urashima T, Brechot C and Nature, 406, 532 ± 535. Paterlini-Brechot P. (2001). Oncogene, 20, 6233 ± 6240. Delhem N, Sabile A, Gajardo R, Podevin P, Abadie A, Hayashi J, Aoki H, Arakawa Y and Hino O. (1999). Blaton MA, Kremsdorf D, Beretta L and Brechot C. Intervirology, 42, 205 ± 210. (2001). Oncogene, 20, 5836 ± 5845. Hegde RS, Voigt S, Rapoport TA and Lingappa VR. (1998). Demetrick DJ, Zhang H and Beach D. (1996). Cytogenet. Cell, 92, 621 ± 631. Cell Genet., 73, 250 ± 254. Hidalgo J and Carrasco J. (1998). Neurotoxicology, 19, 661 ± Dotto GP. (2000). Biochim. Biophys. Acta, 1471, M43 ± M56. 666.

Oncogene Macroarray analyses of hepatocellular carcinoma O Delpuech et al 2937 HolFA,vanderPutNM,GeurdsMP,HeilSG,TrijbelsFJ, Polette M and Birembaut P. (1998). Int. J. Biochem. Cell Hamel BC, Mariman EC and Blom HJ. (1998). Clin. Biol., 30, 1195 ± 1202. Genet., 53, 119 ± 125. Polyak K, Xia Y, Zweier JL, Kinzler KW and Vogelstein B. Honda M, Kaneko S, Kawai H, Shirota Y and Kobayashi K. (1997). Nature, 389, 300 ± 305. (2001). Gastroenterology, 120, 955 ± 966. Porte H, Triboulet JP, Kotelevets L, Carrat F, Prevot S, Iredale JP, Murphy G, Hembry RM, Friedman SL and Nordlinger B, DiGioia Y, Wurtz A, Comoglio P, Gespach Arthur MJ. (1992). J. Clin. Invest., 90, 282 ± 287. C and Chastre E. (1998). Clin. Cancer Res., 4, 1375 ± 1382. ItoY,MatsuuraN,SakonM,MiyoshiE,NodaK,Takeda Ray RB and Ray R. (2001). FEMS Microbiol. Lett., 202, T, Umeshita K, Nagano H, Nakamori S, Dono K, 149 ± 156. Tsujimoto M, Nakahara M, Nakao K, Taniguchi N and Rouault JP, Falette N, Guehenneux F, Guillot C, Rimokh R, Monden M. (1999). Hepatology, 30, 90 ± 99. Wang Q, Berthet C, Moyret-Lalle C, Savatier P, Pain B, Juengel JL, Melner MH, Clapper JA Turzillo AM, Moss GE, Shaw P, Berger R, Samarut J, Magaud JP, Ozturk M, Nett TM and Niswender GD. (1998). J. Reprod. Fertil., Samarut C and Puisieux A. (1996). Nat. Genet., 14, 482 ± 113, 299 ± 305. 486. Jungmann RA, Huang D and Tian D. (1998). J. Exp. Zool., Sato Y, Itoh F, Hareyama M, Satoh M, Hinoda Y, Seto M, 282, 188 ± 195. Ueda R and Imai K. (1999). J. Gastroenterol., 34, 486 ± Kamai T, Arai K, Tsujii T, Honda M and Yoshida K. (2001). 493. B. J. U. Int., 87, 227 ± 231. Schauber C, Chen L, Tongaonkar P, Vega I, Lambertson D, Kasai Y, Takeda S and Takagi H. (1996). Semin. Surg. Potts W and Madura K. (1998). Nature, 391, 715 ± 718. Oncol., 12, 155 ± 159. Schulze K and Schultheiss HP. (1995). Eur. Heart J., 16 Kim HG, Chung YH, Song BC, Kim J, Yang SH, Lee YS and (Suppl O), 64 ± 67. Suh DJ. (2000). Korean J. Intern. Med., 15, 165 ± 170. Shen L, Dean NM and Glazer RI. (1999). Mol. Pharmacol., Knight J. (2001). Nature, 410, 860 ± 861. 55, 396 ± 402. Kusano N, Shiraishi K, Kubo K, Oga A, Okita K and Sasaki Shimotohno K. (2000). Semin. Cancer Biol., 10, 233 ± 240. K. (1999). Hepatology, 29, 1858 ± 1862. Shirota Y, Kaneko S, Honda M, Kawai HF and Kobayashi Lau WY, Lai PB, Leung MF, Leung BC, Wong N, Chen G, K. (2001). Hepatology, 33, 832 ± 840. Leung TW and Liew CT. (2000). Oncol. Res., 12, 59 ± 69. Stuver SO. (1998). Semin. Cancer Biol., 8, 299 ± 306. Laurent-Puig P, Legoix P, Bluteau O, Belghiti J, Franco D, Takano T, Hasegawa Y, Miyauchi A, Matsuzuka F, Yoshida Binot F, Monges G, Thomas G, Bioulac-Sage P and H, Kuma K and Amino N. (2001). Cancer Lett., 168, 51 ± Zucman-Rossi J. (2001). Gastroenterology, 120, 1763 ± 55. 1773. Takayasu H, Horie H, Hiyama E, Matsunaga T, Hayashi Y, LeBailB,FaouziS,BoussarieL,BalabaudC,Bioulac-Sage Watanabe Y, Suita S, Kaneko M, Sasaki F, Hashizume K, P and Rosenbaum J. (1997). J. Pathol., 181, 330 ± 337. OzakiT,FuruuchiK,TadaM,OhnumaNand Liu C, Rangnekar VM, Adamson E and Mercola D. (1998). Nakagawara A. (2001). Clin. Cancer Res., 7, 901 ± 908. Cancer Gene Ther., 5, 3±28. Teramoto T, Satonaka K, Kitazawa S, Fujimori T, Hayashi LowensteinEJ,DalyRJ,BatzerAG,LiW,MargolisB, K and Maeda S. (1994). Cancer Res., 54, 231 ± 235. Lammers R, Ullrich A, Skolnik EY, Bar-Sagi D and Terris B, Pineau P, Bregeaud L, Valla D, Belghiti J, Tiollais Schlessinger J. (1992). Cell, 70, 431 ± 442. P, Degott C and Dejean A. (1999). Oncogene, 18, 6583 ± Miller YE, Minna JD and Gazdar AF. (1989). J. Clin. 6588. Invest., 83, 2120 ± 2124. Thorgeirsson SS, Teramoto T and Factor VM. (1998). Millian NS and Garrow TA. (1998). Arch. Biochem. Semin. Liver Dis., 18, 115 ± 122. Biophys., 356, 93 ± 98. Tornillo L, Carafa V, Richter J, Sauter G, Moch H, Minola Mira L, Maia L, Barreira L and Manso CF. (1995). Arch. E, Gambacorta M, Bianchi L, Vecchione R and Biochem. Biophys., 318, 53 ± 58. Terracciano LM. (2000). J. Pathol., 192, 307 ± 312. Murakami S. (1999). Intervirology, 42, 81 ± 99. Veneziani BM, Villone G, Romano R, Di Carlo A, Garbi C Nakamura H, Kambe H, Egawa T, Kimura Y, Ito H, and Tramontano D. (1990). J. Cell Biol., 111, 2703 ± 2711. Hayashi E, Yamamoto H, Sato J and Kishimoto S. (1989). Vousden KH. (2000). Cell, 103, 691 ± 694. Clin. Chim. Acta, 183, 273 ± 284. Weissman AM. (2001). Nat. Rev. Mol. Cell Biol., 2, 169 ± NhieuJT,RenardCA,WeiY,CherquiD,ZafraniESand 178. Buendia MA. (1999). Am. J. Pathol., 155, 703 ± 710. Welsh JB, Zarrinkar PP, Sapinoso LM, Kern SG, Behling Okabe H, Satoh S, Kato T, Kitahara O, Yanagawa R, CA, Monk BJ, Lockhart DJ, Burger RA and Hampton Yamaoka Y, Tsunoda T, Furukawa Y and Nakamura Y. GM. (2001). Proc. Natl. Acad. Sci. USA, 98, 1176 ± 1181. (2001). Cancer Res., 61, 2129 ± 2137. Wong CM, Fan ST and Ng IO. (2001). Cancer, 92, 136 ± 145. Okajima A, Miyazawa K, Naitoh Y, Inoue K and Kitamura Xu L, Hui L, Wang S, Gong J, Jin Y, Wang Y, Ji Y, Wu X, N. (1997). Hepatology, 25, 97 ± 102. Han Z and Hu G. (2001). Cancer Res., 61, 3176 ± 3181. Pateron D, Ganne N, Trinchet JC, Aurousseau MH, Mal F, Yamashita T, Kaneko S, Hashimoto S, Sato T, Nagai S, Meicler C, Coderc E, Reboullet P and Beaugrand M. Toyoda N, Suzuki T, Kobayashi K and Matsushima K. (1994). J. Hepatol., 20, 65 ± 71. (2001). Biochem. Biophys. Res. Commun., 282, 647 ± 654. PerouCM,SorlieT,EisenMB,vandeRijnM,Je€reySS, YuspaSH,DlugoszAA,ChengCK,DenningMF, ReesCA,PollackJR,RossDT,JohnsenH,AkslenLA, Tennenbaum T, Glick AB and Weinberg WC. (1994). J. Fluge O, Pergamenschikov A, Williams C, Zhu SX, Invest. Dermatol., 103, 90S ± 95S. Lonning PE, Borresen-Dale AL, Brown PO and Botstein Zhan Q, Bae I, Kastan MB and Fornace Jr AJ. (1994). D. (2000). Nature, 406, 747 ± 752. Cancer Res., 54, 2755 ± 2760. Podevin P, Sabile A, Gajardo R, Delhem N, Abadie A, Lozach PY, Beretta L and Brechot C. (2001). Hepatology, 33, 1503 ± 1511.

Oncogene