A Region Within Murine Chromosome 7F4, Syntenic to the Human 11Q13 Amplicon, Is Frequently Ampli®Ed in 4NQO-Induced Oral Cavity Tumors*

Home , Amplicon

A region within murine chromosome 7F4, syntenic to the human 11q13 amplicon, is frequently ampli®ed in 4NQO-induced oral cavity tumors*

Bo Yuan1,2,3, Margaret N Oechsli2,3,4 and Fred J Hendler1,2,3,4,5

Departments of 1Biochemistry and Molecular Biology, and 2Medicine and the 3Henry Vogt Cancer Research Institute, J Graham Brown Cancer Center; 4Center for Genetics and Molecular Medicine; 5Center for Environmental Health Sciences; 6University of Louisville, and Louisville VA Medical Center, Louisville, Kentucky, USA

Our previous reports have shown that two thirds of 4- also occurs frequently in esophageal (Tsuda et al., nitroquinoline-1-oxide (4NQO)-induced murine oral 1989), breast (Dickson et al., 1995), and bladder squamous cell carcinomas (SCC) have Hras1 mutations. tumors (Bringuier et al., 1996). The region contains Loss of heterozygosity (LOH) involving the distal many growth regulatory genes implicated in tumor- portion of chromosome (Chr) 7 occurred in half of the igenesis, including CCND1 (cyclin D1) (Inaba et al., tumors with Hras1 mutations. Here, we demonstrate that 1992). CCND1, as well as other genes in this region can ®ve of six tumors with LOH have 4 ± 8-fold ampli®cation be, but are often not, overexpressed in tumors with involving the distal portion of Chr 7 (7F4). Ccnd1, Fgf4 11q13 ampli®cation (Gaey et al., 1995). and Fgf3, within the most telomeric region of Chr 7 The murine region syntenic to most of the 11q13 (70.5 cM), are co-ampli®ed. The region is syntenic to a amplicon is located in the telomeric portion of Chr 7 previously identi®ed human amplicon at 11q13. Only one (Peters et al., 1989). Mouse Chr 7 contains Fgf4 and out of eight tumors without LOH at Chr 7 had twofold Fgf3 loci that are separated by 17 kb with Fgf3 being ampli®cation; the other seven had no detectable more telomeric (Brookes et al., 1989). Ccnd1 is located ampli®cation. Signi®cant ampli®cation is restricted to in the same region 75 kb centromeric to Fgf4 (Lammie the chromosome with the Hras1 mutation. Gene et al., 1992; Bianchi et al., 1993a). The human ampli®cation occurred without overexpression since only amplicon is approximately 50 kb larger than the one of ®ve tumors with ampli®cation and one of six syntenic region on mouse Chr 7 with the genes having tumors without Ccnd1 ampli®cation expressed increased the same orientation (Lammie and Peters, 1991). A protein. Although ampli®cation of 11q13 occurs rather portion of the 11q13 amplicon is also syntenic with the frequently in human tumors, 4NQO-induced oral cavity centromeric region of murine Chr 19. Ccnd1 is tumors in inbred mice are the ®rst example of a murine ampli®ed in approximately 15% of 7,12-dimethyl- tumor with consistent ampli®cation. Our observations benz(a)anthracene) (DMBA)/12-O-tetradecanoylphor- are strikingly similar to human head and neck SCC bol 13-acetate (TPA)-induced murine squamous cell where overexpression of genes within the 11q13 amplicon carcinomas (SCC) (Bianchi et al., 1993a) while cyclin is inconsistently detected. The ampli®cation of genes D1 overexpression is detected in most tumors (Bianchi localized to human 11q13 and the syntenic region on et al., 1993a; Robles and Conti, 1993). murine Chr 7 during tumorigenesis suggests that similar In contrast to Ccnd1 ampli®cation, other mutations structural elements are present which predispose these initiated by chemical carcinogenesis in murine squa- regions to ampli®cation during malignant transforma- mous cells frequently occur on Chr 7. Point mutations tion. at Hras1 occur in 60 to 95% of skin tumors (Balmain and Brown, 1988; Kemp et al., 1993; Bianchi et al., Keywords: head and neck neoplasia; chemical carcino- 1993b). Loss of heterozygosity (LOH), trisomy and genesis; gene ampli®cation; cyclin D1; chromosome 7 chromosomal non-disjunction involving Chr 7 frequently are detected in the late phases of DMBA- induced tumorigenesis (Bremner and Balmain, 1990; Bianchi et al., 1990, 1991; Buchmann et al., 1991). Introduction Hras1 activation appears to be associated with increased cyclin D1 expression (Filmus et al., 1994; At least 20% (range 20 ± 60%) of human and neck Winston et al., 1996). However, the incidence of Hras1 squamous cell carcinomas (HNSCC) have amplifica- mutations and LOH on Chr 7 has not been correlated tion of genes that localize to 11q13 (Rubin et al., 1995; with Ccnd1 ampli®cation and cyclin D1 overexpression. Lammie and Peters, 1991). The presence of 11q13 In human HNSCC Hras1 mutations occur in no more ampli®cation portends a poor prognosis (Akervall et than 10% of tumors (Rumsby et al., 1990; Clark et al., al., 1995; Michalides et al., 1995). This ampli®cation 1993) and the incidence of Hras1 mutations has not been reported in tumors with either 11q13 amplification or cyclin D1 overexpression. Correspondence: F. Hendler We have shown that approximatley two thirds of 4- Received 20 February 1997; revised 13 May 1997; accepted 13 May nitroquinoline (4NQO) induced murine oral cavity 1997 SCCs have HRAS1 point mutations (Yuan et al., *Supported in part by the Department of Veterans Aairs, The 1994). Half of the tumors with Hras1 mutations Alliant Community Trust Foundation, NIEHS ULCHEHS PG94-3 & PG95-5, the University of Louisville Hospital Medical Foundation, undergo LOH involving the distal half of mouse Chr the Jewish Hospital Foundation, and the James Graham Brown 7 during 4NQO-induced squamous cell carcinomas Cancer Center Foundation. (SCC) (Yuan et al., 1997). This portion of the Gene amplification in 4NQO-induced oral SCCs BYuanet al 1162 chromosome contains oncogenes, tumor suppressor (Table 1). A crude estimate of tumor DNA genes, and growth regulatory genes including Hras1, concentrations was determined using serial dilutions H19, Igf2, Kip, as well as the region syntenic to a of liver DNA as external standards. In this analysis portion of 11q13. Since LOH occurs in the syntenic using Prp and Ngfg, the yield of each PCR region of the chromosome that undergoes amplifica- ampli®cation depends on the initial DNA concentration in HNSCC (Lammie and Peters, 1991) we posited tion when ampli®cation is exponential and was that the syntenic region might be similarly ampli®ed in con®rmed with 1 and 20 ng of liver DNA (Figure 4NQO-induced murine HNSCC, that ampli®cation 1a and b). The DNA concentration was estimated by was associated with LOH, and that ampli®cation comparison of the tumor DNA to known concentra- would lead to cyclin D1 overexpression. This report tions of mouse liver DNA ampli®ed in parallel shows that the syntenic region is ampli®ed and (Gilliland et al., 1990). Subsequently, all samples ampli®cation is associated with LOH. Cyclin D1 were diluted to an estimated concentration of 50 pg/ overexpression, however, can occur without gene ml. This approach will not discriminate dierences in ampli®cation. copy number of less than two (Gilliland et al., 1990). To more accurately determine the concentration of tumor DNA, segments within the Th loci were ampli®ed in a competitive PCR (Gilliland et al., Results 1990). DNA that contained the same primer sites as those of the endogenous targets, but with a dierent Quantifying DNA template length, was added to the reaction mixture as To determine if gene ampli®cation had occurred an internal standard and co-ampli®ed (Figure 1c). A

within the region syntenic to 11q13 in 4NQO-induced (GA)n(GT)n microsatellite in Th had been shown to be murine tumors we had to quantify the DNA that was polymorphic in CBA/J liver (Yuan et al., 1996) and extracted from the tumor. Since little tumor remained tumor DNA (Yuan et al., 1997). The only two alleles after our previous studies (Hawkins et al., 1994; Yuan detected diered by approximately 20 bases (Yuan et et al., 1994, 1997), this was done using PCR based al., 1996) and their distribution in these tumors had analysis. DNA was isolated from cryopreserved tissue been reported (Yuan et al., 1997). When tumors were sections that contained tumor and morphologically homozygous for the shorter allele, plasmid DNA, uninvolved tissues (Yuan et al., 1994). To estimate containing the longer Th allele, was used as an DNA concentration known amounts of normal mouse internal standard and, vice versa for co-ampli®cation. liver DNA were simultaneously ampli®ed with The products of the exogenous and endogenous unknown amounts of DNA from tumor specimens. templates were readily distinguished by gel electro- Since LOH included the region of interest on Chr 7 phoresis and the relative ratio of ampli®ed products was detected in one third of the 4NQO-induced determined (Figure 1c). Since the product ratio of the tumors (Yuan et al., 1994, 1997), we were concerned internal standard to target DNA did not vary during that the process had signi®cantly altered the copy ampli®cation, PCR was carried out to saturation number of alleles since this region of the Chr 7 may maximizing the assay's sensitivity. Using a series of be hemizygous in a tumor. Therefore, segments within standard DNA concentrations in simultaneous co- Prp on Chr 6, a chromosome that did not undergo ampli®cations, the DNA concentration in the tumor LOH, Ngfg at 21 cM on Chr 7, outside the region of samples was extrapolated by determining the equiva- LOH, and Th on Chr 7, within the region of LOH lence point of the Th ampli®cation (Figure 1c). This but 7 cM outside the region syntenic to the 11q13 was veri®ed by repeating the ampli®cation using amplicon (Yuan et al., 1997), were ampli®ed to equivalent concentrations of tumor DNAs with the quantify the DNA present in the tissue specimens Prp and Ngfg primers (data not shown).

Table 1 Oligonucleotide primers ++l] Locus PCR Primer Sequences ID Chr (cM) Size (bp) RE Tm8C [Mg Ngfga CTCCACATGTGTATGTGTATG X01799 7 (21) 147 Mnl I 60 1.5 ATGGAGGCCGAAGAAAGAATC Thb TAGATCCCTGGGTCTTGCTGG X13781 7 (68) 424 Rsa I 62 1.5 GAGAAGCAGCAGGCATGGG Ccnd1c GCCATGGAACACCAGCTGGTGTG X78355 7 (75) 201 Hpa II 65 1.5 ACCTCCAGCATCCAGGTGGCCAC Fgf3b AGACCTTCCAGGGTTCTGTTGA M26284 7 (75) 261 Kpn I 60 1.5 CAAAGAGGTGCTGATGCCCA Fgf4d CAGAGCCAGGCTGGGTTTCTAG X14849 7 (75) 341 Rsa I 65 2.0 GCTTCGTAGGCGTTGTAGTTGTTG Prpb GTGGGCACTCTTACTGAGGCTTAGC M12099 6 (60) 479 Alu I 60 1.5 CTTTCGTCCTTTGTGTGACACTGG The top sequence is the GenBank or reported primer sequence and the bottom is a complementary primer sequence 3' to the top sequence. ID: sequence identi®cation numbers were obtained from GenBank. Chr (cM): chromosome number and distance in centimorgans from the centromere (Love et al., 1990; Rinchik et al., 1992; Buchberg et al., 1992). Size (bp): the length of the ampli®ed segment in base pairs. RE: the ++ ++ restriction endonuclease used to verify the ampli®ed product. TM8C: optimized annealing temperatures. [Mg ]: optimized Mg concentration. Ngfg: nerve growth factor-g; Th: tyrosine hydroxylase; Ccnd1: cyclin D1; Fgf3: ®broblast like growth factor-3; Fgf4: ®broblast like growth; Prp: mouse proline-rich protein-2. a: oligonucleotide primer sequences obtained from (Love et al. 1990). b: primer sequences obtained from (Yuan et al., 1997). c: a segment from Ccnd1 from murine cDNA (Wang et al., 1994) 95% homologous to exon I of human CCND1 (Motokura and Arnold, 1993). d: primer sequences developed for this study Gene amplification in 4NQO-induced oral SCCs BYuanet al 1163

a Tumor # Standard MW 5 11 29 3 9 21 15 4 30 20ng 1ng

Prp

Ngfg

c Tumor # 5 11 3 9 21 4 30 29 15

Figure 1 Estimation of tumor DNA concentrations using PCR. Genomic DNA was isolated from tumors (Yuan et al., 1994). DNA was ampli®ed using Prp (a) and Ngfg (b) primers for 35 cycles. Known concentrations of mouse liver DNA were ampli®ed simultaneously in separate reactions as external standards. The ampli®ed product was identi®ed by electrophoresed in a 3% agarose containing 1 mg/ml ethidium bromide. Molecular weight standards (MW), a HaeIII digest of pBR322. (c) Quantifying tumor DNA using competitive PCR at the Th locus. All DNA samples analysed were homozygous at the Th locus. The cloned longer allele of Th was added to the ampli®cation mixture as an internal standard for all tumor DNA except #29 and #15 where the cloned shorter allele was added. Plasmid DNA containing 20, 80, 140 and 200 copies were added to ampli®cation mixtures containing a constant amount of tumor DNA as estimated by PCR (b). 10 ml of each PCR product was subjected to electrophoresis in a 3% agarose gel containing 1 mg/ml ethidium bromide. The numbers identify the tumor specimens as previously reported (Yuan et al., 1994, 1997). Arrowheads identify the equivalence points

Table 2 Comparison of cyclin D1 expression with mutation analysis Gene ampli®cation in 4NQO-induced tumors To determine if ampli®cation had occurred in the region 7F4 Ccnd1 syntenic to 11q13, segments within Fgf3 were similarly Tumor# Hras1 LOH Ampli®cation Expression ampli®ed using competitive PCR. For this analysis D25, 4 */* + 4 + an Fgf3 ampli®ed segment with a 25 bp deletion between 10 */* + 7 7 two EarI sites, was synthesized and cloned. Standard 15 */* + 4 7 concentrations of cloned plasmid DNA containing the 24 */* + none nd 26 */* + 8 7 deletion (pFgf3CA2D25) were added to a constant 30 */* + 8 ++++ amount of tumor DNA as determined in the previous 9 */+ 7 2 7 experiments and co-ampli®ed with Fgf3 oligonucleotide 12 */+ 7 none 7 primers. Four- to eightfold gene ampli®cation at Fgf3 11 +/+ 7 none ++ 23 +/+ 7 none 7 was observed in ®ve tumors which had Hras1 mutations 27 +/+ 7 none 7 and had undergone LOH on Chr 7 (Figure 2a and b; 29 +/+ 7 none 7 Table 2). Tumor #21, which had neither an Hras1 Hras1 mutations (Yuan et al., 1994) and LOH on Chr 7 (Yuan et al., mutation or LOH on Chr 7, appeared to have twofold 1997) as previously in 4NQO-induced tumors. *:allele with Hras1 ampli®cation (Figure 2c). The other seven tumors mutation; +: wild type Hras1 allele; +: the presence of LOH. Gene without LOH had no evidence of ampli®cation. ampli®cation was quanti®ed from analysis presented in Figure 2. Cyclin D1 expression (Figure 4) was semi-quantitative and graded. 7: The DNA from tumors that were heterozygous at not detected; +: detected but not overexpressed; ++ to ++++: the Th locus could not be analysed using the increased expressed; nd: not done competitive PCR assay. These were analysed using semi-quantitative PCR and the ampli®cations were compared with simultaneous ampli®cation of ®ve were analysed. Since the DNA that remained was tumor DNAs that had been quanti®ed using limited, these loci were analysed using a semi- competitive PCR. The semi-quantitative assay was quantitative assay (Figure 3b and c). At least two consistent with the competitive assay when either DNA samples from each tumor were ampli®ed ampli®cation was not detected or when the amplifica- independently, in duplicate, and all samples were tion exceeded twofold. However, the semi-quantitative ampli®ed in parallel with liver DNA standards of assay did not detect the apparent twofold amplifica- known concentrations. No signi®cant variation was tion in tumor #21 observed by competitive PCR. observed among duplicate specimens in multiple None of the tumors assayed using only the semi- reactions. The level of ampli®cation observed with quantitative method had detectable Fgf3 ampli®cation semi-quantitative PCR at Fgf3, Ccnd1 and Fgf4 was (Figure 3a). identical to that observed with Fgf3 using the To determine if other genes within the region competitive PCR assay (compare Figures 2 and 3). syntenic to 11q13 were ampli®ed, Ccnd1 and Fgf4 That the presence or absence of gene ampli®cation was Gene amplification in 4NQO-induced oral SCCs BYuanet al 1164 a

MW 24 10 26 30

4 15 3 5

11 12 MW 27 MW 29 9 MW 21 Tumor # Figure 2 Detection of Fgf3 ampli®cation using competitive PCR. (a and b) An estimated 100 copies of DNA from tumors #3, 4, 5, 10, 15, 24, 26 and 30 were mixed with 100, 200, 400, 800 and 1000 copies of pFgf3CA2D25; (c) DNA from tumors #9, 11, 12, 21, 27 and 29, estimated to be 100 copies (Figure 1), were mixed with 100, 200 and 400 copies of pFgf3CA2D25. The mixtures were ampli®ed for 40 cycles. Electrophoresis was carried out in a 3% agarose gel containing 1 mg/ml ethidium bromide. For each assay, the concentrations of internal standard increase from right to left except for the tumors #4 and 15. Arrowheads indicate the equivalence points. MW is a HaeIII digest of pBR322

consistent in the 18 specimens analysed indicates that a Liver DNA Tumor # ampli®cation involved the entire region between Fgf3 MW 0.25 0.5 1 2 (ng) 5 14* 23* 21 22* 28* 10 26 24 and Ccnd1. Four- to eightfold gene ampli®cation was observed in ®ve of six tumors with Hras1 mutations and LOH at Fgf3 Chr 7 (Figure 2). The degree of ampli®cation was b determined by comparison with the number of Th Liver DNA (ng) Tumor # copies present. If the mechanism of LOH were deletion, then only a single copy of Th was present 0.2 0.5 1 27 14 23 12 22 28 10 26 24 MW and the amount of ampli®cation detected in tumors with LOH would be doubled. The range of amplification would be 8 ± 16-fold. Gene ampli®cation was not detected in tumor #24 which had LOH on Chr 7. Only Liver DNA (ng) Tumor # one tumor (tumor #21) that remained heterozygous at 0.2 0.5 1 MW 5 11 29 3 9 21 15 4 30 Hras1 and none of the tumors with wild type Hras1 had evidence of ampli®caiton. The ampli®cation in tumor #21 was not detected in the semi-quantitative assay using the same oligonucleotide primers speci®c Ccnd1 for segments in Fgf3 or with primers which ampli®ed segments in the Ccnd1 and Fgf4. As this twofold c ampli®cation was detected only using the competitive, Control DNA (ng) Tumor # more quantitative assay, this minimal ampli®cation 0.2 0.5 1 MW 27 14 23 12 22 28 10 26 24 may represent the accumulation of errors in the multiple reactions required for the analysis. Alterna- tively, the results obtained with the competitive assay were correct and the semi-quantitative assay lacked the sensitivity to discriminate twofold dierences. The Control DNA (ng) Tumor # tumor DNA available was insucient to repeat the 0.2 0.5 1 MW 5 11 29 3 9 21 15 4 30 quantitative analysis using Ccnd1 and Fgf4 primers. The incidence of ampli®cation in tumors with Hras1 and LOH when compared to those that remained heterozygous at Hras1 or had no Hras1 mutations was Fgf4 highly signi®cant (P=0.001; X2=13.8462; d.f.=2) (Systat 6.0, Evanston, IL 1994). Figure 3 Detection of 7F4 ampli®cation by semi-quantitative PCR. Tumor #3, 4, 5, 9, 10, 15, 21, 24, 26, 27 and 30 DNA estimated by either competitive PCR (Figure 1) or tumor #14, 22, 23 and 28 DNA estimated by semi-quantitative PCR (indicated Expression of cyclin D1 by *) (Figure 1) were ampli®ed with Fgf3 (a), Ccnd1 (b), and Fgf4 Conti's group observed that two of 13 DMBA- (c) speci®c primers (Table 1) for 35 cycles. Liver DNA (0.25, 0.5, 1.0 and 2.0 ng) was simultaneously ampli®ed as external induced skin tumors had Ccnd1 ampli®cation standards. MW is HaeIII digest of pBR322 (Bianchi et al., 1993a), while most tumors over- Gene amplification in 4NQO-induced oral SCCs BYuanet al 1165 expressed cyclin D1 (Bianchi et al., 1993a; Robles and be required for overexpression in 4NQO-induced Conti, 1993). To compare cyclin D1 expression with tumors. Cyclin D1 was not signi®cantly overexpressed ampli®cation, cyclin D1 was assayed using a in either normal tongue or metaplastic tissue polyclonal rabbit antibody (Robles and Conti, 1993) surrounding either of the tumors with increased which speci®cally recognizes murine cyclin D1 (Figure cyclin D1 (Figure 4). 4). Similar analysis with HD-11 and 72-13G, rat anti- murine cyclin D1 monoclonal antibodies (Santa Cruz Biotechnology, Santa Cruz CA) had signi®cant back- Discussion ground binding which could not be eectively blocked (data not shown). The ®ve tumors with gene 7F4 Gene ampli®cation frequently occurs in inbred CBA/ ampli®cation at Ccnd1 as well as six tumors without J mice gene ampli®cation were analysed (Table 2). Tumor #21, the tumor with twofold ampli®cation and no Ccnd1 had been previously shown to be occasionally LOH, could not be evaluated because the entire tumor ampli®ed in DMBA/TPA-induced murine skin SCC had been used in the earlier assays. Tumor #30, with (Bianchi et al., 1993a). The present report demonstrates eightfold ampli®cation and tumor #9, which was that in 4NQO-induced oral cavity squamous carcino- heterozygous at Hras1 and with no ampli®cation, genesis, not only is Ccnd1 consistently ampli®ed but had increased cyclin D1 expression (Figure 4). Cyclin the region extends from Ccnd1 to Fgf3. These results D1 was not overexpressed in all other tumors suggest that 11q13 and the syntenic murine region on analysed. Thus, cyclin D1 expression was independent Chr 7 (7F4 amplicon) are both amplicons. This is the of the level of gene ampli®cation and was not ®rst report of syntenic regions in human and mouse associated with LOH. Although most tumors with chromosomes undergoing ampli®cation during tumor- Hras1 mutations did not overexpress cyclin D1, both igenesis. Furthermore, in 4NQO-induced SCC the tumors with cyclin D1 overexpression were mutated at ampli®cation frequency is similar to that observed in Hras1. Thus, it is possible that Hras1 mutations may human neoplasms and in particular, HNSCC.

Figure 4 Cyclin D1 expression in tumor specimens. Cryosections (6 to 8 mm) from 4NQO-induced tumors were reacted with a polyclonal rabbit antibody according to the Materials and methods. The cryosections were lightly counterstained with hematoxylin. The red staining indicates cyclin D1 expression. Magni®cation 4006.(aand b) Two microscopic ®elds of tumor #30 which was mutated at Hras1, had undergone LOH, and had an eightfold 7F4 ampli®cation. (c) Tumor #15 which was mutated at Hras1, had undergone LOH, and had a sixfold 7F4 ampli®cation. (d) Tumor #9 which was mutated at Hras1, had not undergone LOH, and did not have 7F4 ampli®cation. (e) Tumor #11 which was mutated at Hras1, had not undergone LOH, and did not have 7F4 ampli®cation. (f) Tumor #23 which was not mutated at Hras1, had not undergone LOH, and did not have 7F4 ampli®cation Gene amplification in 4NQO-induced oral SCCs BYuanet al 1166 However, since DMBA-induced SCC are infrequently Gene ampli®cation in inbred CBA/J mice appears to be associated with 7F4 ampli®cation, LOH is not the associated with LOH and Hras1 mutations critical event. Ampli®cation could have also developed Carcinogen-induced murine SCCs frequently develop as a direct consequence of gross chromosomal events, Hras1 mutations which undergo LOH involving the such as deletion, mitotic recombination or chromoso- distal portion of Chr 7 regardless of the carcinogen, the mal non-disjunction that result in LOH (a cis eect). site of the tumor, or the genetic background of the Since the incidence of ampli®cation was signi®cantly mouse. The relationship of these events with amplifica- higher in inbred mice, the lack of genetic diversity in tion has not been previously addressed. The present the CBA/J mice may have aected the incidence of study of CBA/J mice demonstrated a highly signi®cant ampli®cation. coexistence of these molecular events with 7F4 ampli®cation. Extrapolating from the reported 90% Possible mechanisms by which LOH and ampli®cation incidence of Hras1 mutations induced by DMBA/TPA occur (Bianchi et al., 1990), the 70% incidence of LOH (Bianchi et al., 1991) and the less than 20% incidence of 4NQO-induced tumors with Hras1 mutations and Ccnd1 ampli®cation (Bianchi et al., 1993a) in tumors LOH have either undergone deletion or mitotic suggests that these events are independent in outbred recombination. The gene copy number detected with SENCAR mice. Gross chromosomal events including the probes for Ngfg and Th, loci on Chr 7 that lay LOH on Chr 7 are restricted to tumors that have Hras1 outside and within the region of LOH, appear to be mutations (Bremner and Balmain, 1990; Bianchi et al., equivalent. This suggests that LOH involves mitotic 1990; Buchmann et al., 1991; Bremner et al., 1994; Yuan recombination and not deletion. However, these data et al., 1994, 1997). Since gene ampli®cation was only are far from conclusive since the analysis probably can detected in both models of squamous carcinogenesis not accurately discriminate between one and two after carcinogen exposure has ceased, it is our copies of an allele. presumption that the ampli®cation is not due to Since 7F4 represents the most telomeric region on dierences in the carcinogen application or action. In Chr 7, an attractive model for ampli®cation during DMBA-treated outbred SENCAR and F1 mice that LOH would be the addition of the 7F4 amplicon from develop skin SCC, the intermediate events can include the lost chromosomal region to the Chr 7 that is deletion, mitotic recombination, trisomy and chromo- retained. LOH on Chr 7 appears to include loss of all somal non-disjunction (Bianchi et al., 1990; Buchmann alleles from a region telomeric to Tyr to the telomere et al., 1991). Whereas in 4NQO-induced oral cavity including the 7F4 amplicon (Yuan et al., 1997). Since SCC, LOH occurs only through either deletion or LOH occurred at Fgf3, ampli®cation must have mitotic recombination (Yuan et al., 1997). However, involved only the Fgf3 allele that was retained during trisomy of Chr 7 was detected in all of the SENCAR LOH; i.e., ampli®cation involved only the Chr 7 with mice in which Ccnd1 was not ampli®ed (Bianchi et al., the Hras1 mutation. Therefore, either chromosomal or 1993a). Since 7F4 amplicon ampli®cation occurs more episomal ampli®cation developed using the remaining frequently in 4NQO than in DMBA-induced skin 7F4 amplicon as template. The lone human study tumors (P50.0001), it is our speculation that 4NQO- addressing the relationship of LOH and 11q13 induced tumors develop ampli®cation as a result of the ampli®cation found the opposite relationship (Foulkes gross chromosomal event which lead to LOH involving et al., 1993). Four of four tumors with ampli®cation at Hras1 by means of deletion or mitotic recombination. 11q13 maintained heterozygosity at FGF3 while Loss of the wild type Hras1 allele results in apparently undergoing LOH at 11q22-ter. The region unopposed mutant activated p21 that could poten- of LOH did not apparently include 11q13. Alterna- tially lead to gene ampli®cation. Stambrook's group tively, LOH at region centromeric to 11q22 would have has demonstrated that 20-fold overexpression of a been missed if the region containing FGF3 were added transfected mutant Hras1 can rapidly induce gross to the `normal' Chr 11 and ampli®ed. chromosomal degeneration and gene ampli®cation (Denko et al., 1994; Wani et al., 1994). Since the Cyclin D1 expression is independent of Ccnd1 introduction of an activating point mutation into ampli®cation and Hras1 activation Hras1 is not sucient to transform rat cells (Finney and Bishop, 1993), it is unlikely that the loss of the Cyclin D1 expression does not correlate with Ccnd1 wild type Hras1 allele, in itself can result in a dosage gene ampli®cation in 4NQO or DMBA-treated mice. eect which leads to ampli®cation. The incidence of cyclin D1 overexpression is very low in LOH on Chr 7 could have resulted in the loss or CBA/J when compared to SENCAR mice. The decreased expression of a tumor suppressor gene(s) observations in 4NQO-induced tumors are similar to which might have aected DNA repair, progression that observed in human tumors where typically no more through the cell cycle, or apoptosis. These regulators than 50% of tumors with ampli®cation in this region either acting in concert or independently might allow have increased cyclin D1 expression (Gillett et al., 1994; ampli®cation and/or result in a mutator phenotype Gaey et al., 1995). Hras1 mutations are relatively rare leading to genomic instability. This speculation is in human tumors, no correlation exists with cyclin D1 supported by our preliminary observations: tumors overexpression. Cyclin D1 overexpression can occur in with 7F4 ampli®cation also have deletion or mitotic human tumors with the absence of ampli®cation as well recombination on Chr 11 (Lentsch et al., manuscript in (Gillett et al., 1994). Although Conti's group analysed preparation) and only in tumors with gene amplifica- Ccnd1 DNA and PCNA expression in one set of tumors tion have multiple new segments identi®ed by DNA (Bianchi 1993a) and cyclin D1 protein expression in a ®ngerprint and sequence analysis (Hu et al., 1995). second set (Robles and Conti, 1993), the implication is Gene amplification in 4NQO-induced oral SCCs BYuanet al 1167 that overexpression had occurred in tumors with and frequent 11q13 ampli®cation. Oral cavity lesions without Ccnd1 ampli®cation. developing among tobacco chewers are the one Hras1 activation appears to be associated with cyclin exception where the incidence may approach 60% D1 expression in cultured rodent cells (Filmus et al., and half of these tumors are associated with 11p15 1994; Winston et al., 1996). Since most DMBA-induced LOH (Saranath et al., 1991b; Huber et al., 1996). skin tumors have Hras1 mutations (Bianchi et al., 1990), 11q13 ampli®cation has not as yet been studied in these this association has probably been observed in tumors as tumors. Hras1 point mutations and LOH at 11p15 well (Bianchi et al., 1993a). However, in the 4NQO- have been reported in many human tumors (Fearon et induced tumors no association between Hras1 and cyclin al., 1984, 1985; Ali et al., 1987; Srivatsan et al., 1991; D1 expression was detected and cyclin D1 overexpres- Howell et al., 1989; Saranath et al., 1991a; Riou et al., sion was rare. Conti's group has attributed the increased 1988; Viel et al., 1991; Gustafson et al., 1994; Huber et cyclin D1 expression as a response to exposure to the al., 1996), but again 11q13 ampli®cation was not promoter, TPA (Bianchi et al., 1993a). Since the 4NQO- evaluated in these tumors. treated mice were not exposed to a promoter, this may Gene ampli®cation at 11q13 frequently occurs (Tsuda explain the observed absence of cyclin D1 expression. et al., 1989; Somers et al., 1990; Lammie and Peters, 1991; Hinds et al., 1994) and may be as frequent as 60% in HNSCC (Rubin et al., 1995). Since Hras1 point The temporal relationship of Hras1 mutations, LOH and mutations are rarely detected in HNSCC among gene ampli®cation during tumorigenesis western populations (Rumsby et al., 1990; Clark et al., Chemical carcinogens cause mutations through DNA 1993; Michalides et al., 1995), 11q13 ampli®cation must adduct formation resulting in Hras1 mutations occur in the absence of Hras1 mutations. These (Quintanilla et al., 1986; Bremner and Balmain, 1990; observations do not preclude the possibility that Bianchi et al., 1991; Yuan et al., 1994). 4NQO- tumors with Hras1 mutations and/or LOH at 11p or treatment of the oral cavity was relatively short 11q are associated with 11q13 ampli®cation. Since the (Hawkins et al., 1994) and LOH and ampli®cation Hras1 focus is at 11p15.5 and the amplicon is at 11q13, were only detected in late tumors with Hras1 separation of these loci by the centromere should mutations. In murine skin carcinogenesis LOH at prevent overlapping LOH from occurring. In human Hbb and Fgf3 (Bremner and Balmain, 1990; Bianchi breast carcinomas, presumably without Hras1 muta- et al., 1991) occurs during the conversion step from tions, LOH at 11p15 and l1q22-23 are frequent but late papillomas to SCCs. Cyclin D1 overexpression as apparently independent events (Carter et al., 1994). In well as gene ampli®cation has been associated with the one study of ovarian cancer probably all of the tumors conversion of papillomas to SCC after the cessation of with 11q13 ampli®cation (four of 28) had LOH at exposure to the carcinogen (Bianchi et al., 1993a). 11q22-23, but an equal number of tumors with LOH did Thus, ampli®cation did not develop as a direct not have 11q13 ampli®cation (Foulkes et al., 1993). consequence of exposure. The presumption is that The 7F4 amplicon is in the most telomeric region of gene ampli®cation occurs subsequent to LOH because mouse Chr 7. LOH in murine carcinogenesis universally one 4NQO-induced tumor had LOH without gene involves the telomeric half of Chr 7. The linkage with ampli®cation. Our preliminary data suggest that other ampli®cation in inbred mice would suggest that the gross chromosomal events have occurred as well on process by which LOH occurs either allows or causes Chr 11 and presumably result from genomic instability. the ampli®cation. Since ampli®cation occurs in 11q13, These tumors may be developing a mutator phenotype. in the middle of the 11q arm, and apparently in the It is important to emphasize that only two-thirds of absence of LOH and Hras1 mutations, this would the 4NQO-induced SCC have Hras1 mutations, and suggest that the LOH may have little to do with the only half of these undergo LOH and gene amplifica- ampli®cation. 11q13 ampli®cation may be due to fragile tion. Therefore, only one third have Hras1 mutations, sites which ¯ank the amplicon (Coquelle et al., 1997). It LOH, and gene ampli®cation. Neoplastic transforma- is our speculation that the 7F4 amplicon contains tion and tumorigenesis must occur by alternative homologous ampli®cation prone elements. pathways which may or may not involve Hras1 mutations and in which LOH and gene ampli®cation The role of the 11q13/7F4 amplicon in tumorigenesis on Chr 7 do not occur. Mutations of growth regulatory genes may lead to multiple subsequent gross genetic events (Hartwell and Comparison of the ampli®cation in murine and human Kastan, 1994; Hinds et al., 1994; Denko et al., 1994). malignancies Overexpression of cyclin D1 dramatically shortened the 4NQO-induced tumors had 4 ± 8-fold which is modest G1/S transition, presumably preventing apoptosis and in comparison with what has been observed in human enabling cells with damaged DNA to enter S phase tumors. However, the murine tumors were no larger more rapidly and accumulate genetic defects (Quelle et than 5 mm with no nodal metastasis (Hawkins et al., al., 1993). However, ampli®cation of CCND1/Ccnd1,in 1994), whereas most human tumors with ampli®cation human and murine tumors does not always lead to are very large lesions and associated with lymph nodal cyclin D1 overexpression (Gillett et al., 1994; Gaey et metastasis. al., 1995). Ampli®cation of 11q13 could result in In contrast to murine carcinogenesis, there is little overexpression of other regulatory genes in this reason to suspect in human tumors an association of region. Although this has been considered since Hras1 mutations, LOH at Chr 11p15 or 11q22-ter, and 11q13 ampli®cation was ®rst identi®ed (Winqvist et 11q13 ampli®cation. Hras1 mutations occur in no more al., 1993), overexpression of other genes within the than 10% of HNSCC and other malignancies with region has been inconsistently observed. Gene amplification in 4NQO-induced oral SCCs BYuanet al 1168 In tumors with 7F4 ampli®cation but without cyclin tumors #15 and 29, homozygous for the longer allele, and D1 overexpression, cyclin D1 plays no role in tumors #10 and 26 which had lost the shorter allele during tumorigenesis. The murine 7F4 amplicon contains tumorigenesis (Yuan et al., 1997), the plasmid containing the other important regulatory genes, including Fgf3 and shorter Th allele was added as the internal standard. Since Fgf4, and possibly Men1 and Ems1 for which the tumors #14, 22, 23 and 28 remained heterozygous during tumorigenesis, the DNA concentration could not be murine loci have not as yet been identi®ed. Until determined using competitive PCR with Th alleles. Five expression or loss of expression of other important microliters of tumor DNA, containing either 20 or 50 pg regulatory genes are demonstrated, the role of the (approximatley 100 or 250 copies) (Cha et al., 1992), were amplicon in transformation remains an enigma. Thus, pipetted into 4 tubes that contained 30 ml of a PCR master ampli®cation without cyclin D1 expression may be a solution (1.256Taq buer, 250 mM dNTP, 1.25 mM of each tumor associated event not aecting tumorigenesis, i.e., primers) and 5 ml of plasmid DNA with 20, 80, 140 and 200 a marker of genomic instability. copies, respectively. Mineral oil (30 ml) was added and a `hot start' was carried out (958C for 1.5 min with the subsequent addition of 10 mlof16Taq buer with 1.25 U of Taq polymerase). Ampli®cation continued for 45 cycles. The PCR Materials and methods products was analysed on 2% agarose/0.56TBE gels containing 1 mg/ml ethidium bromide. The concentration of 4NQO-Induced oral cavity tumors DNA was extrapolated from the equivalence point, i.e. where the ampli®ed product from the tumor DNA and the plasmid CBA/J mice were exposed to 4NQO for 4 to 16 weeks DNA were equal. The concentration of DNA was veri®ed by (Hawkins et al., 1994). SCC developed in the oral cavity at a second independent ampli®cation of tumor DNA using least 6 months after exposure to the carcinogen had ceased. equivalent concentrations. The concentration of DNA These tumors have been previously evaluated for mutations extracted from tumors that were not polymorphic at Th and LOH at Hras1 (Yuan et al., 1994), and the region on was estimated by comparing the ampli®ed products generated Chr 7 at which LOH occurred (Yuan et al., 1997). For this with oligonucleotide primers for Prp and Ngfg with that report 18 specimens were selected from tumors with no obtained with standard concentrations of liver DNA. Hras1 mutations (n=6), tumors heterozygous for mutant Hras1 at codon 12 (n=6), and tumors with LOH at Hras1 (n=6). The tumors as well as their normal livers had been Construction of an Fgf3 DNA segment with a deletion frozen in liquid N and stored at 7708C. 2 A 261 bp region of the murine Fgf3 gene was ampli®ed by PCR using speci®c oligonucleotide primers (Table 1) (Yuan TM Determination of DNA concentration et al., 1996), ligated, and cloned into pCR II (pFgf3CA2)

(Figure 4). pFGF3CA2 DNA (10 mg) was digested with ®ve DNA was isolated from mouse liver using standard units of EarIin16buer M (USB, Clevland OH) at 378C techniques (Yuan et al., 1994). The concentration was for 2 h to delete a 25 bp segment. The 25 bp fragment was determined by u.v. spectroscopy. The DNA was diluted removed by centrifugation through a Chroma Spin-100 gel (20 pg/ml, 200 pg/ml, 2 ng/ml and 20 ng/ml)andusedas matrix (Clonetech, Palo Alto CA). The resulting linear external standards in subsequent ampli®cation reactions. DNA was ligated and transformed. The cloned plasmid Cryopreserved tissues containing SCC were sectioned (6 ± DNA (pFGf3CA D25) was ampli®ed by PCR and the 8 mm) and tumor tissues were dissectd (Yuan et al., 1994). 2 deletion (D25) veri®ed by EarI digestion. The pFgf3CA2D25 Tumor DNA from ®ve to 10 sections was solubilized with DNA concentration was determined by u.v. spectroscopy 100 mlof(0.5%NP-40 w/v (Sigma, St. Louis MO), 10 mM and used as an internal standard for competitive PCR. Tris-HCl (pH 8.0), 10 mM NaCl, 3 mM MgCl2,0.5% sodium dodecyl sulfate). Since the amount of DNA isolated from the tissue sections could not be determined Quantifying gene ampli®cation at the Fgf3 amplicon by by u.v. spectroscopy, the DNA concentration was crudely competitive PCR estimated by comparing the ampli®cation of mouse liver DNA with 5 ml of tumor DNA using oligonucleotide An estimated 100 copies of tumor DNA in 5 mlwas primers for segments within Prp and Ngfg (Table 1) ampli®ed with 30 ml of the PCR master solution, and (Yuan et al., 1996). Each ampli®cation was carried out mixed with 5 mlofpFgf3CA2D25 DNA. Sucient DNA independently in duplicate and in parallel using the same from tumors #9, 11, 12, 21, 27 and 29 was available for master solutions for 35 cycles. The ampli®ed products three PCR ampli®cations, using 100, 200 and 400 copies of (10 ml) were characterized by electrophoresis on 2% pFgf3CA2D25 as internal standards. Tumors #3, 5, 4, 15, agarose/TBE gels. Using serial dilutions of liver DNA, 24, 10, 26 and 30 were co-ampli®ed in ®ve reaction the range of ampli®cation cycles during which exponential mixtures with 100, 200, 400, 800 and 1000 copies of accumulation occurred was determined. The ampli®ed pFgf3CA2D25. After a hot start, the templates were products generated from each tumor specimen were ampli®ed for 45 cycles. The ampli®ed products (10 ml) compared to the ampli®ed products of the mouse liver were evaluated on 3% agarose gels containing 10 ng/ml DNA and the DNA concentration of the tumor specimen ethidium bromide. The tumor DNA concentration was extrapolated. The tumor DNA concentration was adjusted interpolated from the equivalence point where the to 50 pg/ml. ampli®cation from the DNA and the plasmid DNA were To more accurately estimate the DNA concentration in equal. the solutions from tumor specimens, competitive PCR was carried out (Gilliland et al., 1990). Two allelic variants of a Detection of co-ampli®cation at the Fgf3 amplicon by semi- microsatellite repeat within the tyrosine hydroxylase (Th) quantitative PCR gene of CBA/J mice which had been previously identi®ed (Yuan et al., 1996) were used as internal controls. The allelic To determine if other nearby genes were similarly variants had been isolated and cloned into the pCRTMII ampli®ed, the DNA copy number for Fgf3, Fgf4.and plasmid (Invitrogen, Irvine CA). Since tumors #3, 4, 5, 9, 11, Ccnd1 were compared using semi-quantitative PCR. Tumor 12, 21, 24, 27 and 30 were homozygous for the shorter allele, DNA (250 pg) was ampli®ed with Fgf3, Fgf4,andCcnd1 the plasmid containing the longer Th allele was added to the oligonucleotide primers (Table 1) at 948C, 658C, and 728C reaction mixture as an internal standard. In reactions with for 1 min for 35 cycles. Mouse liver DNA (0.2, 0.5 and Gene amplification in 4NQO-induced oral SCCs BYuanet al 1169 1ng/5ml) was ampli®ed simultaneously as external was detected with biotinylated conjugated second antibody standards. Ten microliters of the ampli®ed products were and strepavidin-horseradish peroxidase complex according subjected to electrophoresis on 2% agarose gels. The to the recommendations of the provider (Histomouse-SP amount of product present was estimated by comparisons kit, Zymed Laboratories, South San Francisco, CA). with that ampli®ed from liver DNA standards. To verify Sections were counterstained with hematoxylin. the assay's internal consistency, the semi-quantitative PCR ampli®cation was repeated with duplicate samples under identical conditions. Acknowledgements The 4NQO-induced tumors were identi®ed by Drs Brian Immunohistochemical analysis of cyclin D1 expression Hawkins and Briana Heniford, during their Otolaryngol- Cryopreserved tissue sections (6 to 8 mm) from 12 tumors, ogy research fellowships. We are indebted to the support normal mouse tongue, and CH72 cells (a gift of Dr that we have received from the Divison of Otolaryngology Claudio Conti, Science Park, UT MD Anderson Cancer and its director, Dr Ganzel and the previous director, Dr Center, Smithville TX) were ®xed with 50% acetone- Martinez. This report was submitted as partial ful®lment of phosphate buered saline (PBS) for 30 min, washed with Dr Yuan's doctoral thesis in The Department of Biochem- PBS, and incubated with a polyclonal rabbit anti-mouse istry and Molecular Biology at the University of Louisville. cyclin D1 antibody (UBI, Saranac NY) for 18 h at 48C. We appreciate the careful review of these data by his The sections were washed three times with PBS. Antibody committee.

References

Akervall JA, Jin Y, Wennerberg JP, Zatterstrom UK, Foulkes WD, Campbell IG, Stamp GW and Trowsdale J. Kjellen E, Mertens F, Willen R, Mandahl N, Heim S (1993). Br. J. Cancer, 67, 268 ± 273. and Mitelman F. (1995). Cancer, 76, 853 ± 859. Gaey MJ, Iezzoni JC, Meredith SD, Boyd JC, Stoler MH, Ali IU, Lidereau R, Theillet C and Callahan R. (1987). Weiss LM, Zukerberg LR, Levine PA, Arnold A and Science, 238, 185 ± 188. Williams ME. (1995). Hum. Pathol., 26, 1221 ± 1226. Balmain A and Brown K. (1988). Adv. Cancer Res., 51, 147 ± Gillett C, Fanti V, Smith R, Fisher C, Bartek J, Dickson C, 182. Barnes D and Peters G. (1994). Cancer Res., 54, 1812 ± Bianchi AB, Aldaz CM and Conti CJ. (1990). Proc. Natl. 1817. Acad. Sci. USA, 87, 6902 ± 6906. Gilliland G, Perrin S, Blanchard K and Bunn HF. (1990). Bianchi AB, Navone NM, Aldaz CM and Conti CJ. (1991). Proc.Natl.Acad.Sci.USA,87, 2725 ± 2729. Proc. Natl. Acad. Sci. USA, 88, 7590 ± 7594. Gustafson CE, Young J, Leggett B, Searle J and Chenevix Bianchi AB, Fischer SM, Robles AI, Rinchik EM and Conti Trench G. (1994). Br. J. Cancer, 70, 395 ± 397. CJ. (1993a). Oncogene, 8, 1127 ± 1133. Hartwell LH and Kastan MB. (1994). Science, 266, 1821 ± Bianchi AB, Rinchik EM and Conti CJ. (1993b). Mamm. 1828. Genome, 4, 220 ± 222. Hawkins BL, Heniford BW, Ackermann DM, Leonberger Bremner R, Kemp CJ and Balmain A. (1994). Mol. M, Martinez SA and Hendler FJ. (1994). Head & Neck, 16, Carcinog., 11, 90 ± 97. 424 ± 432. Bremner R and Balmain A. (1990). Cell, 61, 407 ± 417. Hinds PW, Dowdy SF, Eaton EN, Arnold A and Weinberg Bringuier PP, Tamimi Y, Schuuring E and Schalken J. RA. (1994). Proc. Natl. Acad. Sci. USA, 91, 709 ± 713. (1996). Oncogene, 12, 1747 ± 1753. Howell RE, Wong FS and Fenwick RG. (1989). J. Oral Brookes S, Smith R, Thurlow J, Dickson C and Peters G. Pathol. Med., 18, 79 ± 83. (1989). Nucleic Acids Res., 17, 4037 ± 4045. Hu LH, Yuan B and Hendler FJ. (1995). AACR, 36, 542. Buchberg AM, Buckwalter MS and Camper SA. (1992). Huber JL, Yuan B, Alagar R, Bumpous JM, Ackermann Mamm. Genome, 3, S162 ± S181. DM, Martin AW, Korfhage LG and Hendler FJ. (1996). Buchmann A, Ruggeri B, Klein Szanto AJ and Balmain A. Head and Neck Cancer ± Advances in Basic Research, (1991). Cancer Res., 51, 4097 ± 4101. Werner JA, Lippert BM and Rudert RR. (eds.) Elsevier Carter SL, Negrini M, Baa R, Gillum DR, Rosenberg AL, Science, Amsterdam. pp 89 ± 95. Schwartz GF and Croce CM. (1994). Cancer Res., 54, Inaba T, Matsushime H, Valentine M, Roussel MF, Sherr CJ 6270 ± 6274. and Look AT. (1992). Genomics, 13, 565 ± 574. Cha RS, Zarbl H, Keohavong P and Thilly WG. (1992). Kemp CJ, Bremner R and Balmain A. (1993). Mol Carcinog., PCR. Methods Appl., 2, 14 ± 20. 7, 147 ± 150. Clark LJ, Edington K, Swan IR, McLay KA, Newlands WJ, Lammie GA, Smith R, Silver J, Brookes S, Dickson C and Wills LC, Young HA, Johnston PW, Mitchell R and Peters G. (1992). Oncogene, 7, 2381 ± 2387. Robertson G. (1993). Br. J. Cancer, 68, 617 ± 620. Lammie GA and Peters G. (1991). Cancer Cells, 3, 413 ± 420. Coquelle A, Pipiras E, Toledo F, Buttin G and Debatisse M. Love JM, Knight AM, McAleer MA and Todd JA. (1990). (1997). Cell, 89, 215 ± 225. Nucleic Acids Res., 18, 4123 ± 4130. Denko NC, Giaccia AJ, Stringer JR and Stambrook PJ. Michalides R, van Veelen N, Hart A, Loftus B, Wientjens E (1994). Proc. Natl. Acad. Sci. USA, 91, 5124 ± 5128. and Balm A. (1995). Cancer Res., 55, 975 ± 978. Dickson C, Fanti V, Gillett C, Brookes S, Bartek J, Smith R, Motokura T and Arnold A. (1993). Genes Chromsom. Fisher C, Barnes D and Peters G. (1995). Cancer Lett., 90, Cancer, 7, 89 ± 95. 45 ± 50. Peters G, Brookes S, Smith R, Placzek M and Dickson C. Fearon ER, Vogelstein B and Feinberg AP. (1984). Nature, (1989). Proc. Natl. Acad. Sci. USA, 86, 5678 ± 5682. 309, 176 ± 178. Quelle DE, Ashmun RA, Shurtle SA, Kato JY, Bar Sagi D Fearon ER, Feinberg AP, Hamilton SH and Vogelstein B. and Sherr CJ. (1993). Genes Dev., 7, 1559 ± 1571. (1985). Nature, 318, 377 ± 380. Quintanilla M, Brown K, Ramsden M and Balmain A. Filmus J, Robles AI, Shi W, Wong MJ, Colombo LL and (1986). Nature, 322, 78 ± 80. Conti CJ. (1994). Oncogene, 9, 3627 ± 3633. Finney RE and Bishop JM. (1993). Science, 260, 1524 ± 1527. Gene amplification in 4NQO-induced oral SCCs BYuanet al 1170 Rinchik EM, Magnuson T, Holdener Kenny B, Kelsey G, Tsuda T, Tahara E, Kajiyama G, Sakamoto H and Terada Conti CJ, Chartier F, Brown KA, Brown SD and Peters J. M. (1989). Cancer Res., 49, 5505 ± 5508. (1992). Mamm. Genome, 3 Spec No, S104 ± S120. Viel A, De Pascale L, Tooli G, Tumiotto L and Miotto E. Riou G, Barrois M, Sheng ZM, Duvillard P and Lhomme C. (1991). Tumori., 77, 16 ± 20. (1988). Oncogene, 3, 329 ± 333. Wang TC, Cardi RD, Zukerberg L, Lees E, Arnold A and Robles AI and Conti CJ. (1993). Carcinogenesis, 16, 781 ± Schmidt EV. (1994). Nature, 369, 669 ± 671. 786. Wani MA, Xu X and Stambrook PJ. (1994). Cancer Res., 54, Rubin JS, Qiu L and Etkind P. (1995). J. Laryngol. Otol., 2504 ± 2508. 109, 72 ± 76. Winqvist R, Mannermaa A, Alavaikko M, Blanco G, Rumsby G, Carter RL and Gusterson BA. (1990). Br. J. Taskinen PJ, Newsham I and Cavenee W. (1993). Cancer Cancer, 61, 365 ± 368. Res., 53, 4486 ± 4488. Saranath D, Bhoite LT, Mehta AR, Sanghavi V and Deo Winston JT, Coats SR, Wang YZ and Pledger WJ. (1996). MG. (1991a). J. Cancer Res. Clin. Oncol., 117, 484 ± 488. Oncogene, 12, 127 ± 134. Saranath D, Chang SE, Bhoite LT, Panchal RG, Kerr IB, Yuan B, Heniford BW, Ackermann DM, Hawkins BL and Johnson NW and Deo MG. (1991b). Br. J. Cancer, 63, Hendler FJ. (1994). Cancer Res., 54, 5310 ± 5317. 573 ± 578. Yuan B, Shum-Siu A, Lentsch EM, Hu LH and Hendler FJ. Somers KD, Cartwright SL and Schechter GL. (1990). (1996). Genomics, 38, 58 ± 71. Oncogene, 5, 915 ± 920. Yuan B, Hu LH, Lentsch EM, Shum-Siu A and Hendler FJ. Srivatsan ES, Misra BC, Venugopalan M and Wilczynski SP. (1997). Mol. Carcinog., 19, 8 ± 16. (1991). Am. J. Hum. Genet., 49, 868 ± 877.