Oncogene (2002) 21, 1518 ± 1526 ã 2002 Nature Publishing Group All rights reserved 0950 ± 9232/02 $25.00 www.nature.com/onc

Recurrent allelic deletions at mouse chromosomes 4 and 14 in Myc-induced liver tumors

Yuanfei Wu1, Claire-Ange lique Renard1, FrancËoise Apiou3, Michel Huerre2, Pierre Tiollais1, Bernard Dutrillaux3 and Marie Annick Buendia*,1

1Unite de Recombinaison et Expression GeÂneÂtique (Inserm U163), Institut Pasteur, 28 rue du Dr. Roux, 75015 Paris, France; 2Unite d'Histopathologie, Institut Pasteur, 28 rue du Dr. Roux, 75015 Paris, France; 3CNRS UMR 147, Institut Curie, 26 rue d'Ulm, Paris, France

Transgenic mice expressing the c-Myc oncogene driven by Introduction woodchuck hepatitis virus (WHV) regulatory sequences develop hepatocellular carcinoma with a high frequency. To Hepatocellular carcinoma (HCC) is among the com- investigate genetic lesions that cooperate with Myc in liver monest cancers worldwide, with an increasing annual carcinogenesis, we conducted a genome-wide scan for loss of incidence in many countries (Bosch, 1997; El-Serag and heterozygosity (LOH) and mutational analysis of b-catenin Mason, 1999). In more than 80% of cases, HCC in 37 hepatocellular adenomas and carcinomas from development has been linked to chronic infection with C57BL/6 x castaneus F1 transgenic mice. In a subset of hepatitis B and C viruses. Other risk factors include these tumors, chromosome imbalances were examined by alcohol-related cirrhosis and dietary exposure to comparative genomic hybridization (CGH). Allelotyping a¯atoxin B1 (Schafer and Sorell, 1999). Allelotype with 99 microsatellite markers spanning all autosomes studies of human HCC have demonstrated recurrent revealed allelic imbalances at one or more chromosomes in loss of heterozygosity (LOH) at multiple chromosome 83.8% of cases. The overall fractional allelic loss was rather loci, mostly on chromosome arms 1p, 4q, 6q, 8p, 9p, low, with a mean index of 0.066. However, signi®cant LOH 13q, 16p, 16q and 17p (Boige et al., 1997; Nagai et al., rates involved chromosomes 4 (21.6% of tumors), 14, 9 and 1997). Comparative genomic hybridization (CGH) 1 (11 to 16%). Interstitial LOH on chromosome 4 was con®rmed these losses and revealed frequent chromo- mapped at band C4 ± C7 that contains the INK4a/ARF and some gains involving 1q, 8q, 6p and 17q arms INK4b loci, and on chromosome 14 at band B ± D including (Marchio et al., 1997). Among known tumor suppres- the RB locus. In man, the homologous chromosomal regions sor genes (TSGs), LOH has been reported for TP53 at 9p21, 13q14 and 8p21 ± 23 are frequently deleted in liver 17p13, RB at 13q14, axin at 16p13 and mannose-6- cancer. LOH at chromosomes 1 and 14, and b-catenin phosphate/IGF-II receptor at 6q27 (De Souza et al., mutations (12.5% of cases) were seen only in HCCs. All 1995; Murakami et al., 1991; Satoh et al., 2000). TSGs tumors examined were found to be aneuploid. CGH analysis localized in other frequently deleted regions remain of 10 representative cases revealed recurrent gains at presently unknown. Few oncogenes, namely c-Myc and chromosomes 16 and 19, but losses or deletions involving b-catenin, are known to be activated in HCC (reviewed mostly chromosomes 4 and 14 generally prevailed over in Buendia, 2000). Recent studies support the notion gains. Thus, Myc activation in the liver might select for that distinct pathways might be di€erentially activated inactivation of tumor suppressor genes on regions of in HCCs of various etiological origins (Laurent-Puig et chromosomes 4 and 14 in a context of low genomic al., 2001). A hot-spot mutation at codon 249 of TP53 instability. Myc transgenic mice provide a useful model is considered as a hallmark of a¯atoxin B1 contamina- for better de®ning crosstalks between oncogene and tumor tion (Bressac et al., 1991). Higher LOH and p53 suppressor pathways in liver tumorigenesis. mutation rates, and lower incidence of b-catenin Oncogene (2002) 21, 1518 ± 1526. DOI: 10.1038/sj/ mutations have been evidenced in hepatitis B virus- onc/1205208 related tumors (Hsu et al., 2000; Legoix et al., 1999; Marchio et al., 2000; Sheu et al., 1999). Keywords: hepatocellular carcinoma; c-Myc; LOH; Tumor-prone transgenic mice o€er powerful tools CGH; INK4A/ARF for investigating the molecular mechanisms of onco- genesis. Animals with uniform genetic background are maintained in controlled environment, providing a stable and theoretically unlimited resource for analys- *Correspondence: MA Buendia, Unite de Recombinaison et ing tumors induced by the same initial oncogenic event. Expression Ge ne tique, De partement des Retrovirus, Institut Transgenic models of hepatocarcinogenesis have de- Pasteur, 28 rue du Dr. Roux, 75724 Paris Cedex 15, France; monstrated the high oncogenic impact of c-Myc E-mail: [email protected] Received 10 October 2001; revised 12 November 2001; accepted 28 activation on liver cells. In previous studies, we have November 2001 shown that liver expression of a woodchuck c-myc Allelotype of Myc-induced liver tumors YWuet al 1519 transgene controlled by hepatitis virus sequences in Genome-wide screen of LOH in liver tumors WHV/c-myc transgenic mice induces hepatocellular adenomas and HCCs within 8 ± 12 months, with a We initially genotyped 81 tumor samples collected from high (near 100%) penetrance (Etiemble et al., 1994). In 30 mouse livers by microsatellite marker analysis, using this model, c-Myc is transiently overexpressed in the 99 markers distributed along all 19 autosomes and X liver after birth, and it is re-expressed together with chromosome. Di€erent tumor samples from a given IGF-II at tumor onset (Liu et al., 1997). Activation of mouse that showed identical LOH patterns were the Wnt pathway has been implicated in a subset of considered as the same tumor, while they were regarded these tumors by the ®nding of frequent b-catenin as independent tumors when presenting di€erent LOH mutations (de La Coste et al., 1998), but other patterns. Finally, 37 independent tumors (27 HCCs and secondary genetic events that cooperate with Myc 10 adenomas) were included in this study. Among HCCs, remain poorly de®ned. 24 were well di€erentiated, and three were moderately or In this study, we performed a genome-wide screen of poorly di€erentiated. LOH was assessed with at least LOH in 37 tumors (27 HCCs and 10 adenomas) of three widely spaced markers per chromosome, spanning C57BL/6 x Mus Musculus Castaneus F1 hybrids de- a total distance of 1075 cM and covering 75% of the rived from WHV/c-myc transgenic mice. In 10 selected mouse genome. Because some regions of mouse chromo- tumors, we further analysed DNA ploidy by ¯ow somes 1, 4, 8 and 14 are homologous to frequently cytometry and chromosome imbalances by CGH. In altered chromosomal loci in human HCC (Boige et al., addition, we compared the pro®les of genetic altera- 1997; Nagai et al., 1997), these chromosomes were tions in tumors carrying mutated or wild-type b- analysed with a higher density of markers. We also catenin alleles. Our data, showing nonrandom losses positioned markers nearby 10 known TSGs, including on regions of chromosomes 4 and 14 that are the metastasis-suppressing gene KAI1/CD82, Wilms' homologous to frequently involved regions in human tumor (WT1), cyclin-dependent kinase inhibitors HCCs, imply that TSGs located on these chromosomes INK4a/ARF and INK4b, Modi®er of Min-1 (Mom-1), are selectively inactivated in Myc-related liver tumori- breast cancer susceptibility BRCA2, ataxiatelangiecta- genesis. sia-mutated (ATM), mutL homologue (MLH1), TP53, retinoblastoma (RB), adenomatous polyposis coli (APC), dual speci®city phosphatase PTEN (MMAC1/ TEP1) and Myc-antagonist (Mxi1) genes. Based on Results reconstruction experiments, LOH was judged by reduc- tion of allelic intensity of more than twofold, as Liver tumorigenesis in hybrid C57BL/6 x Cast/ei illustrated in Figure 1. All makers detected one or two transgenic mice alleles with the same size in tumor and matched non- We have shown previously that liver-speci®c expres- tumor samples. Therefore, in the absence of replication sion of c-Myc in WHV/c-myc transgenic mice is error on short repeat sequences, these tumors did not associated with the invariable development of liver exhibit microsatellite instability (MIN). tumors (hepatocellular adenomas and HCCs) at 8 ± LOH was detected at one or more chromosomal loci in 12 months of age (Etiemble et al., 1994). The 30/37 tumors (81%). The number of chromosomes with transgenic founder 93-7 (hybrid C57Bl/6 x SJL/J LOH varied from 0 ± 4 per tumor (average: 1.3) (Figure 2). background) and progeny were serially backcrossed The fractional allelic loss or FAL (fraction of chromo- to the C57BL/6 strain for 18 generations. The resulting animals developed liver tumors in all cases within 6 ± 24 months (mean latency period: 12 months). For the present study, male and female transgenic mice were crossed with Cast/ei mice to produce F1 o€spring harboring a high rate of polymorphism in microsatellite marker loci. The onset of liver tumors was markedly delayed in F1 animals compared to pure-strain C57BL/6 mice, with 80% of the F1 transgenic cohort harboring liver tumors between 12 ± 24 months (average: 16.5 months). At autopsy, most animals carried several tumor masses, with average three macroscopic nodules, often localized in the right liver lobe. At histopathological examination, the tumors consisted mainly of well-di€erentiated, trabecular-type HCCs Figure 1 Examples of microsatellite analysis of liver and tumor with frequent necrosis, or hepatocellular adenomas. DNAs. PCR products were separated on agarose gels and stained These tumors were indistinguishable from those of with ethidium bromide. (a) Liver DNA from a normal F1 hybrid the parental C57Bl/6 mice (Etiemble et al., 1994), (NF1) was ampli®ed with D4Mit52 primers and PCR products were compared to those from Cast/ei and C57B1/6 DNAs mixed and adjacent liver similarly carried dysplastic lesions in di€erent ratios. (b) Typical examples of allelic imbalances at of varying severity (data not shown). markers D4Mit52, D14Mit84 and D1Mit486 in liver tumors

Oncogene Allelotype of Myc-induced liver tumors YWuet al 1520 somes with LOH over the total chromosomes tested) detected in HCCs from WHV/c-myc transgenic mice (de varied from 0 to 0.108 with mean value of 0.066 (+0,032, La Coste et al., 1998). In the present work, we found b- 99% con®dence interval). Average FAL indexes were catenin mutations in three HCCs from F1 hybrids, slightly elevated in HCCs compared to adenomas including one GAT to GTT transversion at codon 32 (average: 0.070 versus 0.055) without reaching statistical (Asp to Gly), one GGA to GAA transition at codon 34 signi®cance. The incidence of LOH at each of the 20 (Gly to Glu), and one ACC to TGC two-base transversion chromosomes tested ranged from 0 to 21.6% (Figure 3). at codon 41 (Thr to Cys). These residues are frequently While chromosomes 13, 15 and X remained una€ected in mutated in mouse and human HCCs (de La Coste et al., all tumors, signi®cant LOH was seen at chromosomes 1, 1998). In contrast with our previous report, interstitial 4, 9 and 14 (P50.01), with higher frequency at deletions were not observed in this panel. This marked chromosome 4 (Table 1). Interestingly, allelic imbalances di€erence might re¯ect either mouse strain speci®city or a at both chromosomes 4 and 14 were not seen in the same bias due to the limited number of tumors studied. tumors. LOH at chromosome 4 was observed in both Comparison of the LOH patterns between tumors adenomas (3/10 cases) and HCCs (5/27 cases), while carrying mutated and wild-type b-catenin alleles revealed chromosomes 1 and 14 were a€ected only in HCCs. In that b-catenin mutation tendered to cluster with LOH at 40% of cases, allelic loss was demonstrated at all markers chromosome 4, but this association did not reach along a chromosome, suggesting a high incidence of non- statistical signi®cance. disjunction, but we observed more frequently interstitial LOH on di€erent chromosomes. A critical region was Flow cytometry analysis of DNA ploidy mapped on chromosome 4, at markers D4MP15 and D4Mit117 located on each side of the INK4A/ARF locus DNA ploidy was analysed by ¯ow cytometry in a subset of (Figure 4a). Regional LOH was also observed on 10 representative liver tumors (eight HCCs and two chromosome 14 at markers D14Mit84 to D14Mit224, adenomas). Only one tumor (C690) had no LOH detected an area spanning both the RB locus and a region at any marker. Control analysis of normal F1 livers homologous to human chromosome 8p21 ± 23 (Figure showed about 80% of diploid and 20% of tetraploid cells, 4b). Other TSG-associated markers revealed no LOH rate but aneuploid hepatocytes were not found (data not above average on the corresponding chromosome; shown). In most tumors, aneuploid cell populations were notably, allelic imbalance at the TP53 locus on chromo- predominant (Table 3). In four hyperdiploid tumors, the some 11 was found only in a unique tumor (Table 2). diploid population represented less than 18%, likely Although tumor onset in F1 mice was markedly delayed re¯ecting contamination by normal cells. All other cases compared to that of C57Bl/6 mice, preferential loss of the showed a higher percentage of diploid cells (32.5 ± Cast/ei allele was seen only for chromosomes 1 and 19, 61.5%), suggesting that a signi®cant fraction of tumor and the Bl/6 allele of chromosome 14 was predominantly cells remained diploid. Aneuploid populations consisted deleted. Finally, parental-speci®c allelic deletion suggest- either of hyperdiploid or hypertetraploid cells, and three ing inactivation of an imprinted TSG was not found in tumors demonstrated mixed aneuploid populations with these tumors. two di€erent DIs (Table 3).

b-catenin mutations in HCCs CGH analysis of chromosome imbalance In previous studies, frequent somatic mutations or We next analysed these 10 tumors with known ploidy deletions in the N-terminal region of b-catenin have been by CGH. Typical examples of the experimental data

Figure 2 LOH frequency in 37 WHV/c-myc liver tumors. Only chromosomes harboring LOH on at least two contiguous markers are shown. The medium number of chromosomes with LOH per tumor is 1 (average, 1.3)

Oncogene Allelotype of Myc-induced liver tumors YWuet al 1521 are presented in Figure 5. Chromosomal imbalances Complete or regional losses were frequent at chromo- were found in all cases, including a tumor showing no somes 14 (eight cases) and 4 (seven cases), with abnormality by microsatellite analysis. Most CGH common regional losses at bands 4 C3 ± C6, 4 D1 ± pro®les exhibited both DNA copy gains and losses at E1 and 14 D (Figure 6). Recurrent losses occurred also di€erent chromosomal regions, but one tumor had at chromosomes 10 (three cases), 7 and 8 (two cases), only gains and three had only losses (Table 3). DNA but curiously, chromosome 9 remained apparently losses prevailed over gains in most tumors (average unaltered in all tumors. Complex abnormalities were values: 3.1 losses versus 1.7 gains per tumor). not common and only seen on chromosome 3 in a Interestingly, tumors carrying a b-catenin mutation unique case (Table 3). had more gains and fewer losses than the others The frequency of numerical chromosome imbalances (P50.01). DNA gains, usually on the entire chromo- revealed by CGH was signi®cantly higher than LOH some, were seen at chromosomes 16 (four tumors), 19 frequency in the tumors analysed (P50.01), in (three tumors), and at chromosomes 1, 3, 15 and 18 in agreement with previous observations in human HCC two cases (Figure 6). Overrepresentation of chromo- (Marchio et al., 1997). Tumor C690 is an extreme case some 1 was observed in tumors carrying a b-catenin in which no LOH was detected, but losses involved ®ve mutation (Table 3). High level ampli®cation of a chromosomes in CGH analysis. These di€erences may limited chromosomal region was never observed. re¯ect di€erent types of abnormalities measured by

Figure 3 LOH incidence at each chromosome in WHV/c-myc liver tumors. The indicated values are to those of the marker showing the highest LOH frequency along a given chromosome. The cut-o€ ratio for background LOH is 9.87% (99% con®dence interval). LOH rates at chromosomes 1, 4, 9 and 14 are signi®cant (P50.01)

Figure 4 LOH mapping on chromosomes 4 (a) and 14 (b). Microsatellite markers were positioned according to linkage data, and distances (in cM) of the marker loci to the centromere are shown at the left of chromosome ideograms. Only tumors showing partial or total LOH at chromosomes 4 and 14 are shown, and the LOH frequency of each marker in 37 tumors is indicated at the right side. *, loss of the parental Cast/ei allele; &, loss of the C57B1/6 allele; O or, no LOH. Tumors also analysed by CGH are indicated by a (see Table 3)

Oncogene Allelotype of Myc-induced liver tumors YWuet al 1522 Table 1 Microsatellite loci with signi®cant LOH frequency in 37 Table 2 LOH frequency in 10 tumor suppressor gene loci liver tumors Gene Position Markers LOH (%) Chra Markers Position (cM) Number of tumors LOH (%)b KAI1 Chr 2 (49.6 cM) D2Mit476 (49.2 cM) 5.4 1 D1Mit294 1.1 4 10.8 D2Mit15 (50.0 cM) 5.4 D1Mit126 33.9 4 10.8 WT1 Chr 2 (58.0 cM) D2Mit398 (57.9 cM) 5.4 D1Mit486 52.5 4 10.8 INK4/ARF Chr 4 (42.7 cM) D4Mit15 (42.6 cM) 18.9 4 D4Mit38 19.8 5 13.5 D4MP15 (42.7 cM) 21.6 D4Mit178 35.5 5 13.5 D4Mit117 (44.05 cM) 21.6 D4Mit26 42.5 6 16.2 MOM-1 Chr 4 (66.0 cM) D4Mit170 (66.6 cM) 13.5 D4Mit27 42.5 6 16.2 BRCA2 Chr 5 (84.0 cM) D5Mit409 (84.0 cM) 2.7 D4Mit15 42.6 7 18.9 ATM Chr 9 (30.0 cM) D9Mit162 (30.0 cM) 10.8 D4MP15 42.7 8 21.6* MLH1 Chr 9 (62.0 cM) D9Mit17 (62.0 cM) 8.2 D4Mit117 44.05 8 21.6* TP53 Chr 11 (39.0 cM) D11Mit90 (38.3 cM) 2.7 D4Mit230 44.5 7 18.9 RB Chr14 (41.0 cM) D14Mit160 (40.0 cM) 13.5 D4Mit52 54.9 5 13.5 D14Mit224 (42.0 cM) 13.5 D4Mit170 66.6 5 13.5 APC Chr18 (15.0 cM) D18Mit34 (12.0 cM) 5.4 D4Mit259 76.0 4 10.8 D18Mit200 (16.0 cM) 5.4 D4Mit208 81.5 4 10.8 MMAC1(PTEN) Chr 19 (24.5 cM) D19Mit19 (26.0 cM) 8.2 9 D9Mit162 30.0 4 10.8 MXI1 Chr 19 (49.5 cM) D19Mit4 (48.0 cM) 8.2 D9Mit36 53.6 4 10.8 D9Mit18 68.9 4 10.8 14 D14Mit253 12.0 5 13.5 D14Mit84 30.6 5 13.5 tumors were nicely con®rmed as losses by CGH (see D14Mit160 40.0 5 13.5 Figure 4). In contrast, allelic imbalance at chromo- D14Mit224 42.0 5 13.5 D14Mit71 52.5 4 10.8 somes 1 and 19 correlated with gain by CGH. Similarly D14Mit105 59.0 4 10.8 in human HCC, LOH at di€erent chromosome arms was generally correlated with DNA copy number losses aChr, chromosome, bOnly markers with LOH values above 9.87% by CGH, with few exceptions including 1q, which is (99% con®dence interval of loci ampli®ed) are shown. The highest LOH rates are denoted by an asterisk (P50.01) homologous to mouse chromosome 1 (Buendia, 2000; Laurent-Puig et al., 2001). Thus, despite frank diverg- ence in tumors C471 and C690, there was a general agreement between data generated by the two tech- niques. In addition, our CGH analysis revealed recur- rent gains on chromosomes 16, which had not been reported so far, while chromosomal gains on chromo- somes 15 and 19 have been detected by conventional cytogenetic analysis in carcinogen-induced liver tumors (Ogawa et al., 1999).

Discussion

We have shown here that murine liver tumors induced by targeted expression of c-Myc display major, nonrandom allelic losses on chromosomes 4 and 14. Myc activity in cell transformation has been related to its ability to trigger proliferation and genetic instability (Felsher and Bishop, Figure 5 Examples of CGH aberrations on chromosomes 4 and 1999; Pelengaris et al., 2000). Accordingly, all WHV/ 14. A CGH image of hybridized chromosomes is shown with the c-myc tumors analysed by CGH displayed numerical corresponding ¯uorescence ratio pro®le plotted alongside the chromosome ideogram. The red bars indicate regions of loss. The chromosomal changes, and most of them harbored mean ratio pro®le of 19 chromosomes is depicted with 95% complex diploid/aneuploid patterns indicating mosaic con®dence interval. Red and green lines represent thresholds for cell populations with di€erent DNA content. In the DNA copy number losses (0.85) and gains (1.05). Normal absence of microsatellite instability (MIN) revealed by chromosomes 4 and 14 are shown in the left panels polymorphic marker analysis, our CGH data provide evidence for moderate chromosomal instability (CIN) in Myc-induced liver tumors. LOH is considered as a rate- CGH and microsatellite analysis. In aneuploid tumors, limiting event in tumorigenesis, and in some cases, it CGH may reveal gains or losses that restore the ratios might represent the ®rst hit in the biallelic inactivation of of paternal to maternal alleles on the corresponding tumor suppressor genes (Luttges et al., 2001). However, chromosome, and LOH cannot be detected in these we found a low incidence of LOH throughout the regions. However, LOH at chromosomes 3, 4, 8, 14 genome, a feature also reported in a variety of murine and 17, including total or regional loss was seen as cancers (Davis et al., 1994; Dietrich et al., 1994; Hegi et chromosome loss by CGH in all cases (Table 3). In al., 1994; Lander and Fan, 1997). Because LOH studies particular, LOHs at chromosomes 4 and 14 in ®ve can be carried out only in F1 hybrids of evolutionary

Oncogene Allelotype of Myc-induced liver tumors YWuet al 1523 distant strains, low LOH rates might re¯ect the that inactivation of tumor suppressor genes on inhibitory e€ect of divergent chromosomal sequences chromosomes 4 and 14 might play a crucial role in on mitotic recombination (Shao et al., 2001). Such a Myc-induced tumorigenesis. mechanism could be involved in increased tumor A critical region showing frequent LOH was located resistance of F1 hybrids compared to C57Bl/6 inbred on chromosome 4 between 42.6 cM and 44.5 cM (band strain, although involvement of modi®er loci cannot be C4 ± C7). Allelic losses in the homologous region on excluded. In this context, our allelotype study indicates human chromosome 9p21 have been found in about

Figure 6 Summary of chromosomal gains and losses detected by CGH analysis of 10 liver tumors. Gains are shown on the right side of the chromosome ideogram and losses on the left. Each vertical line represents the a€ected chromosomal region seen in a single tumor sample

Table 3 DNA ploidy and genetic alterations in 10 liver tumors Tumor DNA indexes (%)a Chromosome copy number changes by CGH LOH

A313 1.0 (44%) +2, +11, +19, 71, 78, 710, 714 8, 19 2.45 (26%) 2.78 (30%) A801 1.0 (48%) 72 (C1-F1), 74(A3-D3) 4 (p) 3.18 (52%) C471 1.0 (13%) +16, 74, 710, 714 (D) 1, 6, 11 1.19 (34%) 1.32 (53%) C261 1.0 (48%) +15, 74 (C3-C6), 77(D2-ter), 714 14 2.5 (52%) C770 1.0 (11.5%) +16, 74 (D1-E1), 75 (E3-ter), 77 (F2-ter), 714 (E3-ter) 14 (p) 1.28 (39%) 1.38 (49.5%) C690 1.0 (35%) 74 (C6-E1), 711, 714 (D), 715 (D3-ter) NO 2.02 (65%) C940 1.0 (16.5%) +3 (E1-F2), +18, 73 (F3-G3), 710, 712, 714, 717 17, 3 (p) 1.47 (83.5%) C400b 1 (61.5%) +16 (B5-C2), 74 (D1-E1), 76 (D1-ter), 78 (B3-C1), 714 (D) 4 (p), 8 (p) 1.73 (38.5%) C220b 1.0 (32.5%) +1, +8, +19, 74 (C4-E2), 714 (D1-E3) 4 (p), 5 (p) 2 (67.5%) C390b 1.0 (18%) +1, +3, +15, +16, +18, +19 1 (p), 9, 18, 19 1.28 (82%)

A, hepatocellular adenoma; C, hepatocellular carcinoma; (p) partial LOH; a DNA indexes (DI) measured by ¯ow cytometry and population sizes (% of cells in the total tumor cell population). DI=1.0 denotes diploid cells, DI=2.0 tetraploid cells, and intermediate values aneuploid cell populations. b Tumors with b-catenin mutation

Oncogene Allelotype of Myc-induced liver tumors YWuet al 1524 20% of human HCCs, and they were correlated with mice, demonstrating recurrent cytogenetic alterations poor prognosis (Laurent-Puig et al., 2001). Our CGH of chromosome 14, with frequent involvement of band data, showing loss of chromosome 4 in 7/10 cases, are 14E1 (Sargent et al., 1999). Evidence for disruption of in agreement with recent cytogenetic studies of c-myc/ the RB/E2F pathway in these tumors has been TGF-a tumors, in which partial or complete loss of reported (Santoni-Rugiu et al., 1998). Further genetic this chromosome was evidenced in all HCCs, with non- and expression studies of INK4A, INK4B, ARF and random breakage in band C2 (Sargent et al., 1999). In RB in these tumors will provide critical information on other murine cancers, including lung and mammary the importance of these genes in Myc-induced tumors, frequent allelic losses have been detected in hepatocarcinogenesis. By contrast, we did not ®nd this chromosomal region (Hegi et al., 1994; Herzog et any interstitial deletion in other TSG loci investigated. al., 1994; Ritland et al., 1997). In contrast, LOH at Although recent data have demonstrated a synergistic chromosome 4 was not recorded in spontaneous and cooperation between Myc and p53 in liver tumorigen- chemically induced liver tumors of B6C3 mice, in esis (Renard et al., 2000), LOH at the TP53 locus was which the overall allelotype pattern was clearly seen here in a unique case, and numerical alterations di€erent, suggesting distinct pathways of hepatocarci- were not detected on chromosome 11 by CGH, in nogenesis (Davis et al., 1994). The candidate region on agreement with cytogenetic analysis of c-myc/TGF-a chromosome 4 contains three tumor suppressor genes, tumors (Sargent et al., 1999). p15INK4B, p16INK4A and p19ARF. The INK4 proteins In conclusion, comprehensive allelotyping combined bind and inhibit cyclin-dependent kinases CDK4 and with molecular cytogenetic technology has led to de®ne CDK6, thereby inhibiting their interaction with D-type major genetic changes in murine liver tumors. This cyclins and consequent phosphorylation of RB and study represents a ®rst step towards the identi®cation related proteins. P16INK4A (CDKN2A/MST1) is fre- of tumor suppressor networks that cooperate with quently inactivated by homozygous deletions or c-Myc in liver cell transformation. It has been shown hypermethylation of the promoter region in a variety that loss of INK4A/ARF promotes lymphoma harbor- of neoplasms (Serrano, 2000). In about one half of ing apoptotic defects and increased chemoresistance human HCCs, extensive CpG methylation of the p16 (Schmitt et al., 1999). The WHV/c-myc transgenic promoter was correlated with downregulated expres- model might therefore be useful for testing therapeutic sion of the protein (Liew et al., 1999; Matsuda et al., strategies to counteract the selective disruption of 1999). The product of the alternative reading frame of tumor suppressor pathways in liver tumorigenesis. the INK4a locus, p19ARF (p14ARF in humans) activates the p53 pathway by inhibiting functions of Mdm2, and it might also induce growth arrest independently of p53 (Duro et al., 1995; Eymin et al., 2001; Quelle et al., Materials and methods 1995; Weber et al., 1999, 2000). P19ARF is activated by c-Myc, and disruption of the ARF/ p53 pathway plays Transgenic mice and tumor samples a crucial role in Myc-induced lymphomagenesis The WHV/c-myc transgenic mice have been described (Eischen et al., 1999; Sherr and Weber, 2000). More- previously (Etiemble et al., 1994). Male and female transgenic over, engineered alterations of the INK4a/ARF locus mice in the C57BL/6 strain were crossed with M. musculus have suggested that p19ARF might be more relevant castaneus Ei (Cast/ei) mice. Transgenic (B16 x Cast/ei) F1 than p16INK4A for tumor suppression in mice (Serrano, o€spring were identi®ed by hybridization of tail DNA with a 2000). A second locus harboring signi®cant LOH rates WHV DNA probe. Mice were examined at weekly interval and regional losses by CGH is located on the distal for up to 24 months and euthanized when presenting with part of chromosome 4. This region has been implicated abdominal broadening. All animals were maintained in a in intestinal tumor resistance in mice carrying inactive pathogen-free facility and handled according to the guidelines Min-1 alleles. It is homologous with human chromo- of the MinisteÁ re de l'Agriculture et de la PeÃche (France). Tumor samples and adjacent liver tissues were ®xed in 4% some 1p32 ± 36 that is frequently altered in human paraformaldehyde and embedded in paran for histopatho- HCCs. logical examination. For nucleic acid analysis, tumor samples The second chromosome showing recurrent LOH were dissected free of adjacent liver tissues, capsule and and DNA losses in WHV/c-myc mice is chromosome necrosis, snap-frozen in liquid nitrogen and kept at 7808C. 14, with interstitial deletions between 30.6 and 42.0 cM Livers from nontransgenic C57BL/6, Cast/ei, and F1 hybrid (band D ± E1). Homologous sequences on human mice were used as normal controls. DNA was isolated by chromosome arms 8p21 ± 23 and 13q14 are frequently standard proteinase K digestion and phenol extraction. deleted in liver tumors (Marchio et al., 1997; Nagai et al., 1997). While no candidate gene for HCC has been Microsatellite marker selection and PCR amplification localized so far on 8p, the RB gene at 13q14 is The selection of microsatellite markers for genome-wide scan frequently downregulated in human liver cancer, via was based on their linkage position and on size di€erence INK4A allelic deletions, silencing of p16 or direct 420 bp between the BL/6 and Cast/ei alleles, as described in interaction with gankyrin leading to degradation by the MIT database and the Mouse Genome Database of the proteasome (Higashitsuji et al., 2000; reviewed in Jackson Laboratory. The complete list of markers used in Buendia, 2000). Our CGH data are in agreement with this study is available upon request. Genomic DNA (20 ng) previous studies of liver tumors from c-myc/TGF-a was ampli®ed by PCR as described (Nagai et al., 1997), and

Oncogene Allelotype of Myc-induced liver tumors YWuet al 1525 PCR products were analysed by electrophoresis through mouse Cot-I DNA (Gibco Life Technologies, Gaithersburg, 3.5% or 4% Neusive 3 : 1 agarose gels followed by ethidium MD, USA), denatured at 708C for 10 min, and incubated at bromide staining. Several markers for which BL/6 and Cast/ 378C for 2 h. Metaphase cell preparations obtained from ei alleles di€ered by less than 20 bp were analysed by PCR neonatal ®broblasts were denatured in 70% formamide/2x ampli®cation with a-32P dCTP (3000 Ci/mol) and electro- SSC at 708C for 2 min, and incubated with the probes in a phoresis through 6 or 7% denaturing polyacrylamide gels as humid chamber at 378C for 3 days. Slides were washed and described (Nagai et al., 1997). counterstained with 4,6-diamino-2-phenylindole (DAPI) in anti-fade solution. Digital images of the three ¯uorochromes were acquired LOH analysis using a computer-driven cooled CCD color camera mounted LOH in tumor DNA was evaluated by comparison with the on a Leitz ¯uorescence microscope (Leica, Wetzlar, Ger- mean of 10 DNAs from normal F1 mice. LOH was judged by many). Mouse chromosomes were stained with Dapi and a reduction 450% in signal intensity of one allele. The data identi®ed by computer-generated reverse Dapi banding. were con®rmed by repeated analysis of several DNA samples Quips Software (Vysis Inc, Downers Grove, IL, USA) was from the same tumor in two or three independent PCR used for analysis of digital images and determination of ratio experiments. Signal intensities were quanti®ed by Syngene pro®les of ¯uorescence intensity along chromosomes. Aver- Photo Image System (Syngene, Cambridge, UK). age ratio pro®les were calculated after analysis of 10 ± 15 representative metaphase spreads. As preliminary test for evaluating the quality of hybridization, we veri®ed correct Mutational analysis of b-catenin 2 : 1 labeling ratio of X chromosome. This chromosome was The N-terminal region of the b-catenin gene (exon 1 to exon 3) therefore excluded from CGH analysis. All juxtacentromeric was ampli®ed by PCR from tumor and paired non-tumor DNA loci and regions rich in heterochromatin in chromosomes 1 as described (Tran van Nhieu et al., 1999). The PCR products (band D, E1 ± 2), 4 (band E2), 10 (band A4), 12 (band B), 13 were either directly sequenced or cloned in the pCR2.1 TOPO (band B, C1) and 14 (band A3, B) that could not be vector (Invitrogen, Groningen, The Netherlands). Exon 2 was completely blocked by Cot-1 DNA were excluded from data sequenced with the forward primer 5'-TGATGGAGTTGGA- interpretation. Pro®les were then normalized to an average CATGGCCATG-3' and reverse primer 5'-CCCACTCATA- value of 0.95 for the entire genome, according to control CAGGACTTTGGGAGG-3', by using T7 Sequenase sequenc- experiments. The normal pro®le ranges were de®ned as mean ing kit (USB-Amersham, Cleveland, OH, USA). ratio deviation (95% con®dence interval) between 0.82 and 1.08 for pseudo diploid hyperdiploid tumors, and 0.85 ± 1.05 for tetraploid or polyploid tumors. Losses and gains were DNA flow cytometry considered when observed in at least 50% of pro®les. Frozen tumor tissues (1 ± 2 mm3) were suspended in 300 ml PBS-Tween and stained with propidium iodide. Samples were Statistical analysis analysed on a FAC SORT ¯ow cytometer (Becton Dickinson, San Jose, CA, USA) equipped with a 15 MW air-cooled 488- LOH frequencies at individual chromosomes and background nm argon-ion laser. Tumor cell populations with DNA index rate of LOH were compared by two-tailed Student's t-test (DI)=1 were de®ned as diploid and DI=2 as tetraploid. analysis. The di€erence was considered signi®cant at P value Tumors with intermediate DI values in 410% cells were 50.01. Association of strain bias and parental imprinting with considered as aneuploid. LOH and between group comparisons for LOH frequency was analysed by w2-test. CGH and digital image analysis CGH experiments were performed according to previously described protocols with some modi®cations (Kallioniemi et Acknowledgments al., 1994; Shi et al., 1997). Tumor DNA was labeled with We thank Danny Rouillard for assistance with ¯ow ¯uorescein-12-dUTP and normal reference DNA with Texas cytometry and Martine Lombard for technical assistance Red-5-dUTP (NEN Life Sciences, Boston, MA, USA) by in cytogenetic preparations. This work was supported in nick translation. Each tumor DNA was competitively part by grants from the Association pour la Recherche sur hybridized with normal reference DNA prepared from mice le Cancer (ARC contract 5236) and from the Association of the other sex. Equal amounts (500 ng) of labeled tumor Franco-Chinoise pour la Recherche Scienti®que et Techni- and reference DNAs were co-precipitated with 10 mgof que (Y Wu).

References

Boige V, Laurent-Puig P, Fouchet P, Flejou JF, Monges G, Davis LM, Caspary WJ, Sakallah SA, Maronpot R, Wise- Bedossa P, Bioulac-Sage P, Capron F, Schmitz A, manR,BarrettJC,ElliottRandHozierJC.(1994). Olschwang S and Thomas G. (1997). Cancer Res., 57, Carcinogenesis, 15, 1637 ± 1645. 1986 ± 1990. de La Coste A, Romagnolo B, Billuart P, Renard CA, Bosch FX. (1997). Liver cancer. Okuda K and Tabor E (eds). Buendia MA, Soubrane O, Fabre M, Chelly J, Beldjord C, New York: Churchill Livingstone, pp. 13 ± 28. Kahn A and Perret C. (1998). Proc. Natl. Acad. Sci. USA, Bressac B, Kew M, Wands J and Ozturk M. (1991). Nature, 95, 8847 ± 8851. 350, 429 ± 431. De Souza AT, Hankins GR, Washington MK, Orton TC and Buendia MA. (2000). Semin Cancer Biol., 10, 185 ± 200. Jirtle RL. (1995). Nat. Genet., 11, 447 ± 449.

Oncogene Allelotype of Myc-induced liver tumors YWuet al 1526 Dietrich WF, Radany EH, Smith JS, Bishop JM, Hanahan D MatsudaY,IchidaT,MatsuzawaJ,SugimuraKand and Lander ES. (1994). Proc. Natl. Acad. Sci. USA, 91, Asakura H. (1999). Gastroenterology, 116, 394 ± 400. 9451 ± 9455. Murakami Y, Hayashi K, Hirohashi S and Sekiya T. (1991). Duro D, Bernard O, Della Valle V, Berger R and Larsen CJ. Cancer Res., 51, 5520 ± 5525. (1995). Oncogene, 11, 21 ± 29. Nagai H, Pineau P, Tiollais P, Buendia MA and Dejean A. Eischen CM, Weber JD, Roussel MF, Sherr CJ and (1997). Oncogene, 14, 2927 ± 2933. Cleveland JL. (1999). Genes Dev., 13, 2658 ± 2669. Ogawa K, Osanai M, Obata M, Ishizaki K and Kamiya K. El-Serag HB and Mason AC. (1999). N. Engl. J Med., 340, (1999). Carcinogenesis, 20, 2083 ± 2088. 745 ± 750. Pelengaris S, Rudolph B and Littlewood T. (2000). Curr. Etiemble J, Degott C, Renard CA, Fourel G, Shamoon B, Opin.Genet.Dev.,10, 100 ± 105. Vitvitski-Tre po L, Hsu TY, Tiollais P, Babinet C and Quelle DE, Zindy F, Ashmun RA and Sherr CJ. (1995). Cell, Buendia MA. (1994). Oncogene, 9, 727 ± 737. 83, 993 ± 1000. Eymin B, Karayan L, Seite P, Brambilla C, Brambilla E, Renard CA, Fourel G, Bralet MP, Degott C, De La Coste A, Larsen CJ and Gazzeri S. (2001). Oncogene, 20, 1033 ± Perret C, Tiollais P and Buendia MA. (2000). Oncogene, 1041. 19, 2678 ± 2686. Felsher DW and Bishop JM. (1999). Proc. Natl. Acad. Sci. Ritland SR, Rowse GJ, Chang Y and Gendler SJ. (1997). USA, 96, 3940 ± 3944. Cancer Res., 57, 3520 ± 3525. Hegi ME, Devereux TR, Dietrich WF, Cochran CJ, Lander Santoni-Rugiu E, Jensen MR and Thorgeirsson SS. (1998). ES, Foley JF, Maronpot RR, Anderson MW and Wise- Cancer Res., 58, 123 ± 134. man RW. (1994). Cancer Res., 54, 6257 ± 6264. Sargent LM, Zhou X, Keck CL, Sanderson ND, Zimonjic Herzog CR, Wiseman RW and You M. (1994). Cancer Res., DB, Popescu NC and Thorgeirsson SS. (1999). Am. J. 54, 4007 ± 4010. Pathol., 154, 1047 ± 1055. Higashitsuji H, Itoh K, Nagao T, Dawson S, Nonoguchi K, Satoh S, Daigo Y, Furukawa Y, Kato T, Miwa N, Nishiwaki Kido T, Mayer RJ, Arii S and Fujita J. (2000). Nat. Med., T,KawasoeT,IshiguroH,FujitaM,TokinoT,SasakiY, 6, 96 ± 99. Imaoka S, Murata M, Shimano T, Yamaoka Y and HsuHC,JengYM,MaoTL,ChuJS,LaiPLandPengSY. Nakamura Y. (2000). Nat. Genet., 24, 245 ± 250. (2000). Am.J.Pathol.,157, 763 ± 770. Schafer DF and Sorrell MF. (1999). Lancet, 353, 1253 ± 1257. Kallioniemi OP, Kallioniemi A, Piper J, Isola J, Waldman Schmitt CA, McCurrach ME, de Stanchina E, Wallace- FM, Gray JW and Pinkel D. (1994). Genes Chromosomes Brodeur RR and Lowe SW. (1999). Genes Dev., 13, 2670 ± Cancer, 10, 231 ± 243. 2677. Lander JK and Fan H. (1997). J Virol., 71, 3940 ± 3952. Serrano M. (2000). Carcinogenesis, 21, 865 ± 869. Laurent-Puig P, Legoix P, Bluteau O, Belghiti J, Franco D, Shao C, Stambrook PJ and Tisch®eld JA. (2001). Nature Binot F, Monges G, Thomas G, Bioulac-Sage P and Genet., 28, 169 ± 172. Zucman-Rossi J. (2001). Gastroenterology, 120, 1763 ± Sherr CJ and Weber JD. (2000). Curr. Opin. Genet. Dev., 10, 1773. 94 ± 99. Legoix P, Bluteau O, Bayer J, Perret C, Balabaud C, Belghiti Sheu JC, Lin YW, Chou HC, Huang GT, Lee HS, Lin YH, J, Franco D, Thomas G, Laurent-Puig P and Zucman- Huang SY, Chen CH, Wang JT, Lee PH, Lin JT, Lu FJ Rossi J. (1999). Oncogene, 18, 4044 ± 4046. and Chen DS. (1999). Br. J Cancer, 80, 468 ± 476. LiewCT,LiHM,LoKW,LeowCK,ChanJY,HinLY,Lau Shi YP, Naik P, Dietrich WF, Gray JW, Hanahan D and WY,LaiPB,LimBK,HuangJ,LeungWT,WuSandLee Pinkel D. (1997). Genes Chromosomes Cancer, 19, 104 ± JC. (1999). Oncogene, 18, 789 ± 795. 111. Liu P, Terradillos O, Renard CA, Feldmann G, Buendia MA Tran van Nhieu J, Renard CA, Wei Y, Cherqui D, Zafrani and Bernuau D. (1997). Hepatology, 25, 874 ± 883. ES and Buendia MA. (1999). Am. J Pathol., 155, 703 ± 710. Luttges J, Galehdari H, Brocker V, Schwarte-Waldho€ I, Weber JD, Je€ers JR, Rehg JE, Randle DH, Lozano G, Henne-Bruns D, Kloppel G, Schmiegel W and Hahn SA. Roussel MF, Sherr CJ and Zambetti GP. (2000). Genes (2001). Am. J Pathol., 158, 1677 ± 1683. Dev., 14, 2358 ± 2365. Marchio A, Meddeb M, Pineau P, Danglot G, Tiollais P, Weber JD, Taylor LJ, Roussel MF, Sherr CJ and Bar-Sagi Bernheim A and Dejean A. (1997). Genes Chromosomes D. (1999). Nat. Cell Biol., 1, 20 ± 26. Cancer, 18, 59 ± 65. MarchioA,PineauP,MeddebM,TerrisB,TiollaisP, Bernheim A and Dejean A. (2000). Oncogene, 19, 3733 ± 3738.

Oncogene