A transgenic mouse model for mammary carcinogenesis

Oncogene (1998) 16, 997 ± 1007  1998 Stockton Press All rights reserved 0950 ± 9232/98 $12.00

A transgenic mouse model for mammary carcinogenesis Baolin Li3,4, Kristen L Murphy1,4, Rodolfo Laucirica2, Frances Kittrell1, Daniel Medina1 and Je€rey M Rosen1 Departments of 1Cell Biology and 2Pathology, Baylor College of Medicine, Houston, Texas 77030-3498 and 3Hughes Institute, 2657 Patton Road, Roseville, Minnesota 55113, USA

Missense mutations in the p53 tumor suppressor occur frequently in human breast cancer and in¯uence both the prognosis and response to chemotherapy. Amino acid 175 (equivalent to murine 172) is the second most common site of missense mutations in p53 in human breast cancer. Over 95% of these mutations are arginine-to-histidine (R-H) substitutions resulting in a gain-of-function, and not merely a dominant-negative phenotype. Transgenic mice expressing a p53 172R ± H construct targeted to the mammary gland by means of a whey acidic protein (WAP) promoter were characterized as a model system in order to determine the speci®c e€ects of this mutation on mammary tumorigenesis. Although transgene expression alone had no apparent e€ect on normal mammary development, transgenic mice treated with the chemical carcinogen dimethylbenz(a)anthracene developed tumors with much shorter latency than did control littermates and had a greater tumor burden. Tumors arising in transgenic mice did not exhibit either decreased apoptosis or increased cell proliferation relative to tumors arising in nontransgenic littermates, but did display increased genomic instability. Large pleiomorphic nuclei were visible in many tumors from transgenic mice, and DNA ¯ow analysis con®rmed the presence of signi®cant aneuploid cell populations. Since these transgenic mice develop very few spontaneous tumors, while accelerating carcinogen-and oncogene-mediated tumorigenesis, this mouse model will, therefore, be useful in the investigation of early events in mammary tumorigenesis. It may also be used as a preclinical model to test newly developed chemotherapeutic strategies. Keywords: transgenic mouse model; mutant p53; carcinogenesis

Introduction The development of both p53 knockout and p53 mutant transgenic mice has greatly facilitated studies of the role of p53 in carcinogenesis and tumor progression (Donehower, 1996). Mice de®cient in p53 display a high frequency of spontaneous tumors the most frequent of which are lymphomas, osteosarcomas and soft tissue sarcomas (Donehower et al., 1992; Harvey et al., 1993). However, unlike Li ± Fraumeni syndrome patients, mammary adenocarcinomas are only infrequently observed in the p53 null mice. While the p53-de®cient mouse model has been useful for Correspondence: JM Rosen 4 These two authors contributed equally to this study Received 24 July 1997; revised 3 October 1997; accepted 3 October 1997

some in vivo chemical carcinogenesis assays (Harvey et al., 1993; Kemp et al., 1993), no di€erence in the induction of mammary adenocarcinomas has been observed between mice heterozygous or wild type for p53 following treatment with 7,12-dimethylbenz[a] anthracene (DMBA) (Jerry et al., 1994). Similar studies have not been performed in p53 null mice because of the early mortality due to spontaneous lymphomas and sarcomas. However, crossing p53 null mice with lines of transgenic mice overexpressing speci®c oncogenes provided new insights into the mechanisms by which di€erent signal transduction pathways interact with p53 to increase tumorigenesis (Donehower et al., 1995; Fukasawa and Vande Woude, 1997; Hundley et al., 1997). Unlike the deletion or nonsense mutations observed in other tumor suppressor genes, most p53 alterations are missense mutations resulting in the expression of a functionally altered protein (Hainaut et al., 1997; Hollstein et al., 1991). Such mutations in p53 may result not only in a loss-of-function, but also in the generation of molecules with dominant-negative as well as gain-of-function properties (Dittmer et al., 1993; Hsiao et al., 1994; Ko and Prives, 1996). Speci®c mutations in p53 have been associated not only with a poor prognosis in breast cancer patients, but also with primary resistance to chemotherapy (Aas et al., 1996). Colorectal tumors with point mutations within the conserved domains of p53 have been reported to be inherently more aggressive than tumors with point mutations outside these domains (Goh et al., 1995). Interestingly, mutations of codon 175, one of the ®ve `hotspot' codons present in the sequence-speci®c DNA binding domain of p53, were reported to be particularly aggressive. These represent *20% of all p53 mutations reported to date (Hainaut et al., 1997). The side chain of R175 is not involved in contacting DNA, but in loop stabilization (Cho et al., 1994). The nature of the substitutions found at the most frequently mutated codons, such as codon 175, may vary between tumor types and the properties of these mutants may be cell type-speci®c (Hainaut et al., 1997; Ory et al., 1994). The unique properties of certain p53 mutants may re¯ect their selective activation of speci®c DNA targets (Dittmer et al., 1993; Thukral et al., 1995) and their participation in novel protein-protein interactions (Chen et al., 1994). Using regulatory sequences from the rat whey acidic protein gene linked to a murine p53 minigene, we have established lines of transgenic mice that selectively overexpress a murine 172R ± L mutant p53 (analogous to the human 175R ± L mutant) in the mammary gland (Li et al., 1994). These mice exhibited impaired lobuloalveolar development, an increased sensitivity to girradiation-induced apoptosis, and a marked resistance

Transgenic mouse model for mammary carcinogenesis B Li et al

maximally (Figure 1c). Transgene expression did not appear to be copy-number dependent, and the highest expression levels of the WAP p53 172R ± H transgene were observed in line 8512.

a

b

1 2 3 4 5 6 7 8 9 10 M.W. (kb)

3.0

– – –

2.0

–

5.0 4.0

5.6 kb

2.3 kb

# 5071 # 8509 # 8512 # 8518 # 8522 # 8523 # 8531 # 8533 # 8535 # 8632

to DMBA-induced mammary tumorigenesis (Li et al., 1995). Furthermore, the p53 172R ± L mutant exhibited many properties characteristic of wild type p53. While the mutation of R to L at codon 175 has been detected previously, albeit infrequently, in human breast cancers, the R to H mutation is the second most frequent mutation observed in breast cancer, accounting for approximately 6% of all p53 mutations reported to date (Hainaut et al., 1997). In this study, we describe the generation and characterization of lines of transgenic mice that express the murine equivalent p53 172R ± H mutant in the mammary gland. These mice exhibited no detectable alterations in normal mammary gland development or alterations in the levels of proliferation or apoptosis. In contrast to the 172R ± L p53 transgenics, the expression of a 172R ± H p53 mutant in the mammary gland resulted in increased susceptibility to mammary carcinogenesis induced by DMBA. These carcinogen-induced mammary tumors displayed increased genomic instability, but no detectable alteration in the levels of apoptosis. Since a very low incidence of spontaneous mammary tumors was observed in mice expressing the p53 172R ± H transgene, these mice are an excellent model in which to study the mechanisms by which p53 172R ± H may cooperate with other oncogenes (Li et al., 1997) and tumor suppressor genes involved in mammary carcinogenesis. These mice should also aid in the identi®cation of novel genetic alterations leading to an increased predisposition to mammary cancer.

c

1

2

3

4

5

6

7

8

9

Results

Expression of the 172R ± H p53 mutant in the mammary gland Nine transgenic FVB founders carrying the WAP-p53 172R ± H transgene were characterized by Southern blot analysis (Figure 1b, lanes 2 ± 10). Digestion of genomic DNA with BamHI generated a 5.6 kb endogenous mouse p53 fragment and the expected 2.3 kb transgene fragment. The approximate copy numbers of the transgene were assessed using appropriate loading and copy number standards (Li et al., 1994) Transgene expression was detected in ®ve of the nine lines examined by Northern blot analysis of total RNA isolated from the mammary gland at day 2 of lactation, a time when the transgene should be expressed

d

#8632

#8535

#8533

#8531

#8523

#8522

#8518

To facilitate the ecient expression of speci®c p53 mutants in the mammary gland, a genomic p53 minigene has been utilized rather than a cDNAderived construct (Li et al., 1994). A PCR mutagenesis and recombination protocol was employed to substitute histidine (CAC) for arginine (CGC) at codon 172 in the wild type p53 minigene by the method described previously for the leucine substitution (Li et al., 1994). The minigene bearing the correct mutation was subsequently placed under the control of whey acidic protein (WAP) regulatory sequences to target transgene expression to the mammary gland and to ensure high levels of expression (Figure 1a).

#8512

p53

Construction of a p53 minigene bearing the 172R ± H mutation

#8590

998

1 2 3 4 5 6 7 8 9 — 28 S — 18 S

Figure 1 (a) Construct used to generate murine p53 172R ± H transgenic mice (Li et al., 1994, 1995). (b) Quantitative Southern blot analysis of all founders of the 172R ± H mutant transgenic line. Mouse tail genomic DNA was digested with BamHI, electrophoresed on a 0.6% regular agarose gel, transferred to Zeta-probe GT (Bio-Rad) membrane and hybridized with a probe covering exons II, III, IV and introns II and III of the murine p53 gene. The 5.6 kb band is endogenous p53, and the 2.3 kb band is the transgene. Lane 1 (5071) is DNA from a p53 172R ± L transgenic mouse known to carry 45 copies of the transgene, as a copy number control. Lanes 2 ± 10 are DNA from nine lines of positive 172R ± H p53 mutant transgenic mice. Of these, lines 8518 and 8531 are C57BL mice, and all others are FVB. Founders 8509, 8512, and 8518 are female; all others are male. The ®lm was exposed for 22 h. (c) Demonstration of p53 172R ± H transgene expression in the mammary gland of transgenic mice by Northern blotting. Twenty mg of total RNA isolated from the mammary glands of transgenic mice at day 2 of lactation were analysed on a 1.2% denaturing RNA gel, transferred to Zeta-probe membrane, and hybridized with a 32P-labeled murine p53 cDNA probe. Five of the nine lines of 172R ± H transgenic mice express p53 in the mammary gland. This ®lm was exposed for 10 h. No speci®c p53 RNA transcripts were detected in lanes 1, 5, 6, or 9 even following longer exposure. (d) Shows the ethidium bromide stained gel demonstrating equivalent RNA loading

Transgenic mouse model for mammary carcinogenesis B Li et al

To determine if overexpression of p53 172R ± H in¯uenced the expression of genes known to be regulated by wild type p53 in the mammary gland, a comparative Northern blot analysis examined the di€erent transactivation properties of wild type p53 and the two p53 mutants (172R ± L and 172R ± H) at day 2 of lactation in several lines of transgenic and control mice. As illustrated in Figure 2, top panel, three of the four transgenic lines expressed high levels of mutant p53 (Figure 2, lanes 2, 4 and 5), and the expression in

1

2

3

4

5

p53

CIP 1

MDM 2

PCNA

MOUSE WAP

28 S

18 S

Figure 2 Northern analysis of gene expression in the mammary glands of wild type and p53 172R ± L and p53 172R ± H transgenic mice. Twenty mg of RNA isolated from the mammary gland at day 2 of lactation was resolved on a 1.2% denaturing RNA gel, transferred to Zeta-probe membrane, and hybridized with 32Plabeled probes of murine p53, CIP 1, MDM2, PCNA and mouse WAP, respectively. The bottom panel shows the ethidium bromide stained gel to demonstrate equivalent loading of total RNA. Lane 1 is RNA from FVB non-transgenic mice, lane 2 RNA from 172R ± L p53 mutant mice (line 5017), and lanes 3 ± 5 RNA from three independent lines of 172R ± H p53 mutant transgenic mice (in order, 8533, 8518 and 8512). The panels demonstrating p53, CIP 1, MDM2, PCNA, and mouse WAP expression were exposed for 20 h, 11 days, 9 days, 48 h and 1 h, respectively

the fourth line (Figure 2, lane 3) was greater than in the nontransgenic control (Figure 2, lane 1). Under these RNA loading conditions and using short autoradiographic exposure times, wild type p53 transcripts were not detected in control mammary glands at day 2 of lactation. In the absence of DNA damage, p53 levels are normally quite low in the mammary gland during lactation. However, as reported previously (Li et al., 1994), overexpression of the 172R ± L mutant, which retains some wild type transactivation properties, correlated with a substantial increase in CIP1/p21 expression (Figure 2, lane 2). In contrast, the levels of CIP1/p21 did not appear to be altered by the overexpression of the 172R ± H p53 mutant in the three independent transgenic lines analysed (Figure 2, lanes 3 ± 5). Similarly, mdm2 mRNA is known to be positively regulated by wild type p53 in an autoregulatory loop (Ko and Prives, 1996) and mice carrying the 172R ± L transgene express mdm2 mRNA at much higher levels. Here, it appears that lines of mice expressing the 172R ± H mutant and the control mice have approximately equal levels of mdm2 expression. In contrast, the expression of proliferating cell nuclear antigen (PCNA) mRNA was unchanged in the mice expressing the 172R ± H mutant and the control, but markedly reduced in the 172R ± L mutant, as reported previously (Li et al., 1994). Finally, levels of mouse WAP mRNA were assessed in order to determine whether milk protein gene expression was a€ected in these transgenic lines. Consistent with the observed failure of the 172R ± L mice to lactate (Li et al., 1994), no WAP expression was detected in these transgenics (Figure 2, lane 2). In contrast, mice bearing the 172R ± H p53 transgene lactate normally and expressed levels of WAP mRNA equivalent to those seen in controls. Transgene expression does not a€ect normal mammary gland development Overexpression of some transgenes in the mammary gland, including the WAP-172R ± L p53 construct (Li et al., 1995) has resulted in altered lobuloalveolar development, the result of markedly increased levels of apoptosis during pregnancy. Therefore, it was essential to determine whether or not the 172R ± H p53 mutant in¯uenced glandular morphology in the absence of carcinogen treatment. Female mice from the highest expressing line, #8512, bore normal numbers of pups that were well within the accepted weight range at each stage of development and comparable to FVB littermates. The pups survived to adulthood, suggesting that there was no transgenerelated lactational dysfunction. Several additional experiments were conducted in order to directly investigate if there were any detectable changes in mammary gland development that might have resulted from the transgene-induced alterations in cell cycle regulation or apoptosis. Mammary glands were removed from FVB mice and transgenic line #8512 at 17 days of pregnancy and at 2 days of lactation, times at which transgene expression was maximal. Cell proliferation levels were assessed using immunohistochemistry by counting the number of dividing cells that incorporated BrdU injected 2 h

999

Transgenic mouse model for mammary carcinogenesis B Li et al

1000

prior to sacri®ce. No signi®cant di€erence was observed in the percentage of BrdU positive cells in the mammary gland in the control or transgenic mice at either stage of development (Figure 3a). Moreover, the overall percentages of proliferating cells were quite low, as would be expected at these stages of mammary gland development. To determine if expression of p53 172R ± H a€ected apoptosis, the mammary glands from mice that underwent forced involution, i.e. whose litters were removed at day 10 of lactation and who were then left to involute for 5 days before sacri®ce, were paranembedded, sectioned, and stained for apoptotic cells using the TUNEL method (Gavrieli et al., 1992). The levels of apoptosis assessed by counting representative ®eld were as expected, i.e. approximately 4% (Li et al., 1996), and were comparable between the FVB and transgenic mice (Figure 3b). This lack of a signi®cant di€erence in either the level of proliferation or apoptosis was re¯ected in the mammary gland morphology assessed by comparative whole mount analysis. No striking di€erences were observed between glands of the transgenic and control mice, except for a slightly denser ductal network observed in the FVB as compared to the transgenic mice following forced involution (data not shown). Finally, it should be noted that no transgene-related pathologies were observed in any other tissues examined. No spontaneous tumors were observed in the ®rst year of life, through four lactations. Furthermore, only four spontaneous tumors have been observed in a population of at least 25 ± 30 mice over a year of age continuously expressing the transgene in the 5 yrs since this line was generated. Together, these results suggest that the presence of the transgene itself does not accelerate tumorigenesis.

a

Increased susceptibility to mammary carcinogenesis following DMBA administration Despite the rare incidence of spontaneous tumors in line #8512, it was of interest to determine the susceptibility of these mice to mammary tumorigenesis following exposure to a known mammary carcinogen. Starting 2 weeks after implantation of pituitary isografts to stimulate maximal transgene expression, groups of transgenic and nontransgenic mice were given the chemical carcinogen DMBA at 0.5 mg/wk for 2 weeks. Subsequently, the tumor incidence, time of appearance, total tumor number per animal, and tumor histopathology were assessed

Figure 4 Overexpression of p53 172R ± H results in the accelerated onset of DMBA-induced mammary tumors. The data are expressed as the percentages of tumor-free p53 172R ± H transgenic mice and FVB nontransgenic mice. The fractions at the bottom of the curves are the number of tumor-free animals divided by the total number of animals in each group

b

Figure 3 Transgene expression does not appear to a€ect either cell proliferation or rate of apoptosis in non carcinogen-treated mammary glands of transgenic mice as compared to FVB control mice. (a) Mean percentages of proliferating cells, measured by BrdU incorporation. Lanes 1 and 2 represent proliferation levels of FVB mammary glands removed at 17 days of pregnancy and 2 days of lactation, respectively, and lanes 3 and 4 are proliferation percentages in glands removed from 172R ± H p53 mutant mice at the same stages of pregnancy for comparison. Proliferation levels do not appear to vary signi®cantly between the FVB and transgenic mice at either stage. (b) Analysis of apoptosis levels by the TUNEL method. Glands analysed were from FVB and p53 172R ± H transgenic mice whose litters were removed at 10 days of lactation and then force-involuted for a further 5 days prior to sacri®ce. Lanes 1 and 2 depict mean apoptotic cell percentages in the number 3 and number 4 mammary glands respectively of the FVB mice, while lanes 3 and 4 depict levels of apoptosis in the corresponding mammary glands of the transgenic mice. Error bars re¯ect the standard error of the mean in both panels

Transgenic mouse model for mammary carcinogenesis B Li et al

(Figure 4, Table 1). By 28 weeks, 100% of the transgenic mice (n=8) had developed tumors, at an average of 1.9 tumors per animal. By comparison only 85% of the wild type mice (n=13) had developed tumors by week 45, and each of these mice developed only one tumor. Thus, it appeared that both the rate of tumorigenesis and the tumor burden induced by DMBA treatment were a€ected by the expression of the 172R ± H p53 mutant transgene in the mammary gland.

1001

DMBA-induced tumor pathologies are variable and in¯uenced by p53 status The DMBA-induced tumors were excised, paranembedded and sectioned for staining as described in Materials and methods. A diverse spectrum of tumor pathologies were observed using Masson's trichrome staining as illustrated in Figure 5. The representative tumor histology from DMBA-treated FVB control mice is shown in Figure 5a and b. Figure 5a depicts a

Table 1 Summary of tumor pathology and ploidy analysis Tumor ID 848-12 (FVB) 850-15 (FVB) 853-7 (FVB) 855-1 (FVB) 861-10 (FVB) 864-6 (FVB) 864-9 (FVB) 864-10 (FVB) 845-10 (TGd) 845-12 (TGd) 847-16 (TGd) 850-1 (TGd) 850-2 (TGd) 850-4 (TGd)

Tumor pathologya

Ploidy

MAC/AAC, 50/50 epithelial/stromal MAC/AAC, very stromal MAC, extensive stroma MAC/squamous, 75% epithelial MAC, 90% epithelial MAC/squamous, lung metastasis MAC, lung metastasis MAC/squamous, lung metastasis adenoma, one focus of MAC, 90% epithelial adenoma, 480% epithelial MAC, extensive stroma MAC, undi€erentiated, 70% epithelial MAC, undi€erentiated, extensive stroma MAC, undi€erentiated, 490% epithelial

diploid aneuploid diploid aneuploid diploid diploid aneuploid diploid tetraploid aneuploid diploid aneuploid diploid aneuploid

DNA Indexb 1 1.7 1 1.2 1 1 1.7 1 2.0 2.4 1 1.2 1 1.8

% in S-Phase 9.30 17.86 (aneuploid 3.44 0.00 (aneuploid 6.27 3.76 13.01 (aneuploid 5.02 22.95 (tetraploid 29.68 (aneuploid 1.75 0.00 (aneuploid 3.15 14.25 (aneuploid

pop.) pop.) pop.) pop.) pop.) pop.) pop.)

p53 Stainingc + 7 + 7 7 + 7 + (one region only) ++ ++ +++ +++ +/++ +++

a

MAC, mammary adenocarcinoma; AAC, adenocanthoma. bDNA index: peak ¯uorescent intensity ratio of aneuploid G0/G1 population/ diploid G0/G1 population. cp53 staining is only seen in epithelial cells (see ®gure 8); scale values correspond to high (+++), medium (++), low (+) and negative (7) p53 expression. Used to con®rm p53 status of tumors. dTransgenic

Figure 5 DMBA-induced mammary tumors have widely varying morphologies. These panels were stained with Masson's trichrome. (a) and (b) illustrate tumors that arose in FVB mice, while (c) and (d) illustrate tumors that arose in transgenic p53 172R ± H mice. FVB tumors are often keratinized and contain extensive connective tissue stroma, while the tumors arising in transgenic mice are often more epithelial. Note the presence of more normal-appearing mammary tissue at the left border of d

Transgenic mouse model for mammary carcinogenesis B Li et al

1002

mammary adenocarcinoma (MAC) and adenocanthoma (AAC) with extensive stroma, whereas the tumor shown in Figure 5b is a pure MAC, but also contains a large stromal component. The representative tumor histology for the DMBA-treated transgenic mice is shown in Figure 5c and d. Figure 5c shows an adenoma with a greater than 80% epithelial component, and d is an undi€erentiated MAC with a greater than 90% epithelial content. The majority of tumors induced in the transgenic mice were classi®ed as mammary adenocarcinomas, type B; and the remainder were adenomas. In contrast, in the nontransgenic FVB mice, four tumors were carcinomas with extensive squamous metaplasia and seven were adenocarcinomas. Tumors with the greatest latency (weeks 43, 44 and 45) exhibited metastasis to lung (data not shown). Mechanisms responsible for p53 transgene-mediated tumorigenesis To study the mechanism by which expression of the p53 transgene might increase the rate of tumor appearance and tumor burden, it was ®rst determined that mutant p53 was in fact overexpressed in the DMBA-induced tumors from transgenic mice. Representative sections of tumors were stained with p53 using the CM5 antibody (the kind gift of Dr David Lane) to detect both wild type and mutant p53. As illustrated in Figure 6a, b and d DMBA-induced tumors from p53-transgenic animals

displayed intense nuclear staining for p53 in the epithelial cells. The p53 transgene was found to be expressed at levels several 100-fold higher than wild type p53 (Li et al., 1995). No staining was observed in the majority of the DMBA-induced tumors observed in nontransgenic FVB control mice (Figure 6c) because of the short half-life of wild type p53. However, mutations in p53 have been reported to occur infrequently in DMBA-induced mammary tumors in Balb/c and hemizygous p53 mice (Jerry et al., 1994), and accordingly, weak p53 staining was observed in only two of the DMBA-induced tumors in the nontransgenic FVB controls (data not shown). Based upon previous studies performed in several mouse model systems it was expected that loss of wild type p53 function resulting from the overexpression of a dominant-negative mutant p53 might decrease apoptosis leading to the increased tumorigenesis and tumor burden in transgenic mice. To test this hypothesis the levels of apoptosis were examined in carcinogen-induced tumors from both transgenic mice and nontransgenic controls using the TUNEL assay (Gavrieli et al., 1992). Representative TUNEL-stained sections of carcinogen-induced tumors in wild type mice, both a combination of MAC and AAC, are depicted in Figure 7a and b, while sections of tumors from transgenic mice are shown in Figure 7c and d (c is a very stromal MAC, while d is an undi€erentiated MAC). No signi®cant di€erence in the number of

Figure 6 Mutant p53 is expressed at much higher levels than endogenous p53. These panels depict tumor sections that were stained with the CM5 anti-p53 antibody, a gift from Dr David Lane. (a) and (b) Illustrate tumors that arose in mice carrying the transgene and are very heavily stained, indicating high levels of mutant protein expression. (c) Is a section of a tumor that arose in a nontransgenic control mouse, negative for p53 staining. (d) Is a higher magni®cation of part of the section shown in (a) and illustrates nuclear p53 localization

Transgenic mouse model for mammary carcinogenesis B Li et al

apoptotic cells (approximately 1.75% of cells displayed positive staining in each group) was observed between tumors arising in transgenic and nontransgenic mice. The 172R ± H p53 mutant appears to induce genomic instability In order to determine if the presence of the mutant p53 transgene exerted an e€ect on proliferation that might account for the increase in carcinogen-induced tumorigenesis, tumor sections were stained using a speci®c anti-PCNA antibody (Figure 8b and e). Equivalent numbers of PCNA-positively staining cells, ranging from 25 ± 35%, were detected in tumors arising in the nontransgenic and transgenic mice. However, marked di€erences in nuclear morphology were observed between tumor nuclei in the transgenic as compared to the nontransgenic mice as illustrated by both hematoxylin-and-eosin staining (Figure 8a and d) and PCNA staining (Figure 8b and e). The tumor nuclei in transgenic mice were pleiomorphic, usually larger and more irregular in shape, and demonstrated marked subnuclear structures. Flow cytometric analysis of paran-embedded cells from representative tumors induced by DMBA in transgenic and nontransgenic mice was employed to determine DNA content and the total number of cells in S-phase (Table 1). Representative histograms of a diploid tumor from a nontransgenic mouse, and an aneuploid tumor from a transgenic mouse are

illustrated (Figure 8c and f). No correlation was observed between tumor genotype and the percentage of cells in S-phase, but a slightly greater number of tumors in the transgenic mice were tetraploid or aneuploid than those in the nontransgenic group. Ras mutations are not detected in DMBA-induced tumors Carcinogen-induced mammary tumorigenesis following DMBA treatment has been reported to result from the activation of the Ha-ras-1 oncogene by an A-T transversion at codon 61 (Dandekar et al., 1986; Kumar et al., 1990). In addition, chromosome instability induced by the loss of p53 is greatly enhanced by the constitutive activation of the MAPK pathway (Fukasawa and Vande Woude, 1997). Accordingly, tumors were screened for the presence of Ha-ras mutations using a nested PCR-RFLP protocol, as well as through direct sequencing of the PCR products (see Materials and methods for details). Surprisingly, carcinogen-induced tumors arising in both nontransgenic and transgenic mice appeared to lack the Ha-ras codon 61 A-T transversion commonly caused by DMBA administration (data not shown). Several carcinogen-induced tumors were used as positive controls. Preliminary results also suggested that at least two other common carcinogen-induced ras mutations (Ki-ras, codons 12 and 61) were not present in either group of the DMBA-induced tumors.

Figure 7 Apoptosis levels in tumors from the transgenic p53 172R ± H mice and the FVB mice are comparable and very low (approximately 1.75% for both groups). (a) and (b) depict TUNEL-stained sections arising in nontransgenic FVB mice, while (c) and (d) depict sections of tumors that arose in transgenic p53 172R ± H mice

1003

Transgenic mouse model for mammary carcinogenesis B Li et al

1004

Figure 8 Relative to tumors that arise in FVB nontransgenic mice, tumors that arise in mice carrying the p53 172R ± H transgene often have strikingly larger and more irregular nuclei (here, highlighted by both hematoxylin and eosin staining (a and d) and proliferating cell nuclear antigen (PCNA) staining (b and e). Further analysis by ¯ow cytometry (c and f) was performed to determine whether genomic irregularities were present in the aberrant nuclei. The presence of an aneuploid population (e.g. yellow peak, f) was noted more frequently in the tumors from mice expressing the p53 172R ± H transgene. (a and c) Are representative analyses of a nontransgenic FVB tumor, while (d and f) represent analyses of a tumor that arose in a p53 mutant transgenic mouse

Discussion Targeted overexpression of a 172R ± H p53 transgene did not appear to a€ect normal mammary development as determined by morphological examination of mammary gland wholemounts, analysis of the levels of apoptosis and proliferation at several stages of development, and ®nally, by the ability of the mice to nurse their young and synthesize normal levels of milk protein mRNAs. These observations are consistent with previous studies that indicated that mammary gland development, including both ductal morphogenesis (Humphreys et al., 1996), lobuloalveolar development and lactation (Donehower et al., 1992), and apoptosis and remodeling of the mammary gland during involution will occur through p53-independent pathways (Li et al., 1996). The loss of wild type p53 function may slightly alter the timing of these processes, but p53 is not essential for normal mammary gland development. Expression of the 172R ± H mutant p53 did not alter the expression levels of p21, MDM2, PCNA, or several other genes known to be regulated by wild type p53. Again this result was expected because the 175R ± H mutant human p53 protein was shown not to interact with a consensus p53 DNA-binding site, and failed to transactivate a p53-dependent reporter gene construct (Davido€ et al., 1991; Lavigueur et al., 1989; Ory et al., 1994; Prosser et al., 1990). Despite the lack of an essential role for p53 in normal mammary gland development, p53 does appear to play a critical role in mammary tumorigenesis. Mice expressing the p53 172R ± H transgene exhibited a very low level of spontaneous tumorigenesis, but were predisposed to the development of mammary tumors when treated with the chemical carcinogen, DMBA. Tumors developed much earlier in transgenic mice than

in controls, and the transgenic mice had a greater tumor burden. The absence of ras mutations in DMBA-induced tumors from both the nontransgenic and p53 mutant transgenic mice was unexpected based upon previous reports of ras mutations in carcinogeninduced tumors in Balb/c mice (Dandekar et al., 1986; Guzman et al., 1992; Kumar et al., 1990). Previous studies have indicated that FVB mice given four doses of DMBA of 1 mg each were quite refractory to mammary carcinogenesis as compared to Balb/c mice. However, the present studies were performed in FVB mice containing pituitary isografts that produce high serum prolactin and progesterone levels, facilitating the expression of milk protein gene-based transgenes and replacing the need for continuous breeding. These mice all developed tumors within 13 weeks following the last dose of carcinogen (Li et al., 1995). In the present study, one fourth the dose of DMBA (0.5 mg given twice) was employed, and tumors developed with a mean latency of 30 weeks for the nontransgenic FVB controls. Thus, the constitutive activation of the JAK/ Stat, Src and MAPK pathways by prolactin may have abrogated the need for activating ras mutations (Zhang et al., 1990). Mice carrying the 172R ± H p53 mutant transgene have also been crossed in separate experiments with mice overexpressing erb-B2 (Li et al., 1997) and desIGF-1 (Murphy KL, Hadsell D and Rosen JM, unpublished observations), both of which were speci®cally targeted to the mammary gland. Tumors appeared in these bigenic mice with a signi®cantly shorter latency than mice expressing either growth factor transgene alone. Thus, mice expressing the p53 172R ± H transgene, while developmentally normal, exhibited accelerated tumorigenesis in conjunction with the overexpression of certain growth factors or growth factor receptors, as well as following exposure to a chemical carcinogen.

Transgenic mouse model for mammary carcinogenesis B Li et al

These mice, therefore, provide an excellent model in which to study early events in mammary tumorigenesis, and also the interaction of p53 with altered growth factor signaling pathways. This transgenic model in some cases also provides a distinct advantage over p53 null mice that exhibit high levels of spontaneous lymphomas and sarcomas (Donehower et al., 1992). Bigenic mice derived through crosses of mice expressing a targeted oncogene with p53 null mice may also develop tumors in organs not usually prone to tumorigenesis, e.g. the salivary gland tumors observed in the MMTV-ras/p53 null mice (Hundley et al., 1997). Thus, it may not be possible to utilize these mice for drug or carcinogen screens in mammary tumorigenesis studies. Loss of p53 function has been shown in the choroid plexus to in¯uence tumorigenesis primarily through the inhibition of apoptosis (Symonds et al., 1994). However, this did not appear to be the case for DMBA-induced mammary carcinomas in our mouse model. Apoptosis in tumors from transgenic and nontransgenic mice were compared, but no signi®cant di€erences were detected. Likewise, apoptosis did not appear to play a major role in p53 function for the development of mammary gland (Jones et al., 1997) or salivary gland (Hundley et al., 1997) tumors in Wnt-1/ p53 null and MMTV-ras/p53 bigenic mice, respectively. Thus, the e€ects of p53 loss or overexpression of a dominant-negative p53 mutant on apoptosis may be cell-type speci®c. Increased rates of cellular proliferation have also been observed in mammary tumors in the Wnt-1/p53 null and the salivary tumors in the MMTV-ras/p53 null bigenic mice as compared to those arising in the Wnt-1 and MMTV-ras monogenic mice (Jones et al., 1997). However, no signi®cant di€erences in proliferation were observed in the DMBA-induced tumors between the nontransgenic and mutant p53 transgenic backgrounds. Because of the marked heterogeneity in the histopathology of DMBA-induced tumors, small di€erences in proliferation rates may not have been detectable in these di€erent tumor types. However, increased genomic instability has been uniformly observed in these bigenic mice de®cient in p53 as well in the mutant p53 transgenic mice treated with DMBA or crossed with mice overexpressing erbB2 (Donehower et al., 1995; Hundley et al., 1997; Li et al. 1997). Larger pleiomorphic nuclei and aneuploidy, characteristic of these tumors, most likely resulted from spindle errors during cell division. These results are consistent with the observation that p53 is a component of the spindle checkpoint that ensures maintenance of diploidy (Cross et al., 1995; Hartwell and Kastan, 1994). Certain missense mutations of p53 create a unique conformation (Cho et al., 1994) that may facilitate interactions with other proteins by creating a novel interaction site or by disrupting interactions with other cellular proteins that normally serve to block aberrant p53-involved protein-protein interactions (Chen et al., 1994). Amino acid 175 is not localized within the regions of the p53 protein that directly contact DNA, as are most of the other commonly mutated residues. The arginine side chain is located in the space between loops 2 and 3 of the protein, where it participates in bonds bridging the two domains (Cho et al., 1994). This p53 mutant was found associated with the heat

shock protein hsc70, bound the antibody PAb240, and was highly sensitive to proteolytic digestion (unlike the wild type protein), all of which suggested that the protein was at least partially denatured (Cho et al., 1994). The 175R ± H mutation appeared to be a gain-offunction mutation, and not merely a functional null or dominant negative mutant (Dittmer et al., 1993; Hinds et al., 1990). It was capable of stimulating expression of MDRCAT (a human multidrug resistance [MDR]1 gene promoter-CAT construct) in p53-null cells, in a manner reversible by co-transfection of wild type p53 (Chin et al., 1992). It conferred a growth advantage in the absence of endogenous wild type p53 protein in transfection experiments with Saos-2 cells, and injection of a cell line expressing this mutant p53 protein into nude mice resulted in tumorigenesis, which was not seen with the parental p53-null cells (Dittmer et al., 1993). This mutant protein was also able to cooperate in co-transfection experiments with activated H-ras in the transformation of rat embryo ®broblasts (Hinds et al., 1990). Finally, as mentioned previously, p53 has been implicated in regulation of the G2/M checkpoint and mitosis, with particularly striking e€ects upon centrosomal duplication (Fukasawa et al., 1996). Carcinogeninduced primary skin tumors from mice bearing a skintargeted 172R ± H mutation displayed a much greater degree of aberrant centrosomal duplication than did tumors from p53-null mice, also supporting a gain-offunction e€ect of this mutation (Wang XJ and Roop DR, personal communication). As cited previously, the 175R ± H p53 mutant was unable to bind a consensus p53 DNA-binding site, so it is likely that it mediates its gain-of-function through protein-protein interactions. It has been hypothesized that mutant p53 proteins may gain additional functions by interacting with other proteins that supply a DNA-binding domain. The resulting dimer or complex could then use the p53 transactivation domain to modulate a novel set of genes, including MDR-1 (Lin et al., 1995), Studies are in progress to identify these genes and novel protein targets. Finally, since both the prognosis and the response to chemotherapy has been correlated with the presence of speci®c missense mutations in p53 (Aas et al., 1996), the WAP-p53R ± H mutant transgenic mouse model is of interest not only to study the mechanisms involved in mammary carcinogenesis, but also as a potential model for testing new therapies.

Materials and methods Materials ABC reagent and diaminobenzidine (DAB) were obtained from Vector Labs (Burlingame, CA). 16-dUTP biotin and proteinase K were obtained from Boehringer Mannheim (Germany). Terminal deoxynucleotidyl transferase (TdT) was obtained from Pharmacia (Milwaukee, WI). The kit for analysis of cell proliferation by means of bromodeoxyuridine (BrdU) labeling was obtained from Amersham (Buckinghamshire, England). Probe-on Plus slides, Omniset tissue cassettes, 30% hydrogen peroxide, and Permount slide mounting solution were obtained from Fisher Scienti®c (Pittsburgh, PA). The PC-10 anti-PCNA primary antibody was obtained from Santa Cruz Biotechnology (Santa Cruz, CA), and the CM5 anti-p53 primary antibody

1005

Transgenic mouse model for mammary carcinogenesis B Li et al

1006

was the gift of Dr David Lane, University of Dundee, Scotland. The 74-micron mesh used in ¯ow cytometry was obtained from Small Parts Inc., (Miami, FL), and the chicken erythrocyte nuclei (CEN) from Accurate Chemical and Scienti®c Co. (Westbury, NY). All other chemicals and reagents used were obtained from Sigma Chemical Company (St. Louis, MO). Mice were acquired from Charles River Labs (Wilmington, MA) or from a breeding colony at Baylor College of Medicine. Southern and Northern blot analyses The generation and screening of p53 transgenic mice, the isolation of DNA and RNA, and Southern and Northern blot analyses were performed as described previously (Li et al., 1994).

the absence of carcinogen treatment by removing a litter after 10 days of lactation and allowing a further 5 days for involution to occur. Both 3rd (thoracic) and 4th (inguinal) mammary glands were surgically removed from each mouse under anesthesia, spread in Omniset tissue cassettes, and ®xed in 10% NBF for 6 h. Glands were stored in 70% ethanol, and then either paran embedded and sectioned for TUNEL analysis or stained with hematoxylin for whole mount analysis. TUNEL was performed on 5 mm sections mounted on Probe-on-Plus slides. Whole mount staining Staining was carried out according to the protocol of Williams and Daniel (1983) except that the glands were stained for only 2 h in hematoxylin solution.

TdT dUTP nick end labeling (TUNEL) analysis

Immunohistochemical analysis

TUNEL was performed as described (Gavrieli et al., 1992) with the following modi®cations. Proteinase K digestions were carried out at 0.5 mg/ml for 10 min at 248C. Tissue sections were labeled with one unit of TdT and 1 nmol of 16-dUTP-biotin in 150 ul of TdT bu€er. These sections were then incubated with ABC reagent following the manufacturer's protocol and subsequently with DABnickel substrate for 5 ± 6 min. Sections were counterstained with 0.1% (w/v) methyl green for 30 s to 1 min, rinsed in distilled water, dehydrated through graded alcohols and xylene, and then mounted with Permount according to standard protocols.

Tumor sections collected as described were deparanized and rehydrated according to standard protocols. For PCNA detection, sections were ®rst blocked with 10% normal horse serum (NHS) in 16TBS for 20 min at 378C, and then incubated with primary antibody at a 1 : 50 (v/v) dilution in 10% NHS at 150 ml/slide. The slides were coverslipped and held in a humidi®ed chamber for 25 min at 378C. A 1 min TBS wash preceded application of goat anti-mouse secondary antibody at 1 : 1500 dilution in 10% NHS and incubation at 378C for 15 min in the chamber. For p53 deletion, slides were blocked in 5% normal goat serum (NGS) in 16TBS for 5 min at 378C. Primary antibody (rabbit anti-mouse) was applied at a dilution of 1 : 200 in 10% NGS, and slides were incubated at 378C for 15 min. After a 2 min TBS wash, slides were incubated with biotinylated goat anti-rabbit secondary antibody at a dilution of 1 : 500 in 10% NGS at 378C for 15 min. Following incubation with secondary antibody in both cases, slides were incubated with ABC reagent according to the manufacturer's protocol (Vector Labs) and visualized with diaminobenzidine for 5 ± 7 min. Sections were counterstained with 0.1% (w/v) methyl green for approximately 1 min, rehydrated through graded alcohols and xylene and mounted with Permount according to standard protocols. Bromodeoxyuridine labeling was performed in accordance with the manufacturer's protocol.

Quantitation of apoptosis and cell proliferation Images from TUNEL and BrdU/PCNA-analysed tumor sections and control glands were captured with a Sony 3CCD color video camera attached to a BX-50 Olympus microscope with Adobe Photoshop software. The number of apoptotic or proliferating cells and the total number of cells in three representative ®elds (1000 ± 1500 cells total per section) were counted, and the percentage of apoptotic or proliferating cells was used in the statistical analysis. This was performed using the JMP Statistical Visualization software from SAS Institute, Inc. (Cary, NC). At least ®ve tumors from each group were included in the comparisons reported, and several sections of each tumor were counted. Clusters of TUNEL-labeled fragments that occurred within the same cell diameter were counted as one apoptotic cell. Pituitary isografting and carcinogen treatment In order to ensure continuous hormonal stimulation, which results in WAP-p53 transgene expression, all mice treated with the carcinogen were ®rst given a pituitary isograft under anesthesia. In this procedure, one donor syngeneic pituitary gland was implanted under the kidney capsule of each experimental mouse at 3 to 5 weeks of age. Carcinogen treatment started 2 weeks later, and consisted of two doses of DMBA (0.5 mg each) administered by gavage, seven days apart. Isolation of tumors and mammary glands Tumors were surgically excised from p53172R-H mice under anesthesia when they reached approximately 161 cm in size, and were ®xed for 6 ± 8 h in 10% neutral bu€ered formalin (NBF). They were stored in 70% ethanol, and then paran embedded and sectioned (5micron sections, Probe-On Plus slides) for immunohistochemical and histopathological analysis. Control mice (p53172R-H and FVB, n=1 of each) were assessed for e€ect of the transgene in the mammary gland in

Flow cytometric analysis of paran-embedded tumor sections Flow cytometric analysis of all 13 paran-®xed tumor samples was performed. Nuclear suspensions were prepared from tissue blocks using a modi®cation of Hedley's protocol (Hedley et al., 1985). Brie¯y, three 50-micron sections were cut from each sample. These were ®rst deparanized in xylene for two 50 min intervals at 258C, and then rehydrated through graded alcohols (100%, 95%, 70%, and 50%), ending with deionized water. There was a 50 min incubation at 258C at each alcohol concentration change. The tissue was subsequently treated with a 0.5% pepsin solution for 30 min at 378C to create a single cell suspension. The resulting solution was ®ltered through a 74-micron mesh and then the nuclear suspension was stained using the Vindelov technique (Vindelov et al., 1983). Fresh chicken erythrocyte nuclei (CEN) was used as a control in order to detect problems with ®xation, pepsinization, etc. A 5 ml solution of a 20 000 000 CEN/ ml suspension was added prior to DNA staining. These nuclei ¯uoresence at or near channel 100. The sample for DNA ¯ow analysis was acquired on a FACScan ¯ow cytometer (Becton Dickinson, San Jose, CA). Twenty thousand events were collected, and the parameters used for gating were FL2-W versus FL2-A. Aneuploid tumors met the following criteria: two individual G0/G1 peaks or

Transgenic mouse model for mammary carcinogenesis B Li et al

an asymmetric G0/G1 peak with two well-de®ned G2/M peaks. As suggested by Shankey et al. (1993), the ®rst peak, appearing near channel 200, is assumed to be the diploid G0/G1 peak. Histograms were analysed using the ModFit LT software (Verity Software House, Topsham, ME). H-ras codon 61 A-T mutation analysis PCR ampli®cation of the codon 61 region of the H-ras gene was carried out as in Gimenez-Conti et al. (1992). Subsequent detection of A-T transversions by RFLP was performed as in Nakazawa et al. (1992). DNA sequence analysis was performed according to manufacturer's

instructions (Exo-Pfu Cyclist Stratagene, La Jolla, CA).

DNA

Sequencing

Kit,

Acknowledgements These studies were supported by grant CA16303 from the National Institutes of Health. KLM was supported by training grant GM08231 from the National Institutes of Health. We would like to thank Liz Hopkins for help with the histology, Jason Gay, Joyce Pike, Lester Patton and Shu-wen Sun for help with animal husbandry and surgery, and Dr Norman Greenberg for a critical review of this manuscript.

References Aas T, Bùrresen A-L, Geisler S, Smith-Sùrensen B, Johnsen H, Varhaug JE, Akslen LA and Lùnning PE. (1996). Nature Med., 2, 811 ± 814. Chen Y, Chen PL and Lee WH. (1994). Mol. Cell Biol., 14, 6764 ± 6772. Chin K, Ueda K, Pastan I and Gottesman MM. (1992). Science, 255, 459 ± 462. Cho Y, Gorina S, Je€rey PD and Pavletich NP. (1994). Science, 265, 346 ± 355. Cross SM, Sanchez CA, Morgan CA, Schimke MK, Ramel S, Idzerda RL, Raskind W and Reid BJ. (1995). Science, 267, 1363 ± 1356. Dandekar S, Sukumar S, Zarbl H, Young LJT and Cardi€ RD. (1986). Mol. Cell Biol., 6, 4104 ± 4108. Davido€ AM, Humphrey PA, Iglehart JD and Marks JR. (1991). Proc. Natl. Acad. Sci. USA, 88, 5006 ± 5010. Dittmer D, Pati S, Zambetti G, Chu S, Teresky AK, Moore M, Finlay C and Levine AJ. (1993). Nature Genetics, 4, 42 ± 46. Donehower L. (1996). Biochim. Biophys. Acta., 1242, 171 ± 176. Donehower LA, Godley LA, Aldaz CM, Pyle R, Shi Y-P, Pinkel D, Gray J, Bradley A, Medina D and Varmus HE. (1995). Genes & Dev., 9, 882 ± 895. Donehower LA, Harvey M, Slagle BL, MacArthur MJ, Montgomery Jr CA, Butel JS and Bradley A. (1992). Nature, 356, 215 ± 221. Fukasawa K, Choi T, Kuriyama R, Rulong S and Vande Woude GF. (1996). Science, 271, 1744 ± 1747. Fukasawa K and Vande Woude GF. (1997). Mol. Cell Biol., 17, 506 ± 518. Gavrieli Y, Sherman Y and Ben-Sasson SA. (1992). J. Cell Biol., 119, 493 ± 501. Gimenez-Conti IB, Bianchi AB, Stockman SL, Conti CJ and Slaga TJ. (1992). Mol. Carcinogenesis, 5, 259 ± 263. Goh H-S, Yao J and Smith DR. (1995). Cancer Res., 55, 5217 ± 5221. Guzman RC, Osborn RC, Swanson SM, Sakthivel R, Hwang S-I, Miyamoto S and Nandi S. (1992). Cancer Res., 52, 5732 ± 5737. Hainaut P, Soussi T, Shomer B, Hollstein M, Greenblatt M, Hovig E, Harris CC and Montesano R. (1997). Nucleic Acids Res., 25, 151 ± 157. Hartwell LH and Kastan MB. (1994). Science, 266, 1821 ± 1828. Harvey M, McArthur MJ, Montgomery CA Jr, Brutel JS, Bradley A and Donehower LA. (1993). Nature Genetics, 5, 225 ± 229. Hedley DW, Friedlander ML and Taylor IW. (1985). Cytometry, 4, 327 ± 333. Hinds PW, Finlay CA, Quartin RS, Baker SJ, Fearon ER, Vogelstein B and Levine AJ. (1990). Cell Growth & Di€erentiation, 1, 571 ± 580.

Hollstein M, Sidransky D, Vogestein B and Harris CC. (1991). Science, 253, 49 ± 53. Hsiao M, Low J, Dorn E, Ku D, Pattengale P, Yeargin J and Haas M. (1994). Am. J. Pathol., 145, 702 ± 714. Humphreys RC, Krajewska M, Krnacik S, Jaeger R, Weiher H, Krajewski S, Reed JC and Rosen JM. (1996). Development, 122, 4013 ± 4022. Hundley JE, Koester SK, Troyer DA, Hilsenbeck SG, Subler MA and Windle JJ. (1997). Mol. Cell Biol., 17, 723 ± 731. Jerry DJ, Butel JS, Donehower LA, Paulson EJ, Cochran C, Wiseman RW and Medina D. (1994). Mol. Carcinogenesis, 9, 175 ± 183. Jones JM, Attardi L, Godley LA, Laucirica R, Medina D, Jacks T, Varmus HE and Donehower LA. (1997). Cell Growth & Di€erentiation, 8, 829 ± 838. Kemp CJ, Donehower L, Bradley A and Balmain A. (1993). Cell, 74, 812 ± 822. Ko LJ and Prives C. (1996). Genes & Dev., 10, 1054 ± 1072. Kumar R, Medina D and Sukumar S. (1990). Oncogene, 5, 1271 ± 1277. Lavigueur A, Maltby V, Mock D, Rossant J, Pawson T and Bernstein A. (1989). Mol. Cell Biol., 9, 3982 ± 3991. Li B, Greenberg N, Stephens LC, Meyn R, Medina D and Rosen JM. (1994). Cell Growth & Di€erentiation, 5, 711 ± 721. Li B, Kittrell FS, Medina D and Rosen JM. (1995). Mol. Carcinogenesis, 14, 75 ± 83. Li B, Rosen JM, McMenamin-Belano J, Muller WJ and Perkins AS. (1997). Mol. Cell Biol., 17, 3155 ± 3163. Li M, Hu J, Heermeier K, Hennighausen L and Furth PA. (1996). Cell Growth & Di€erentiation, 7, 13 ± 20. Lin J, Teresky AK and Levine AJ. (1995). Oncogene, 10, 2387 ± 2390. Nakazawa H, Aguelon A-M and Yamasaki H. (1992). Oncogene, 7, 2295 ± 2301. Ory K, Legros Y, Auguin C and Soussi T. (1994). EMBO J., 13, 3496 ± 3504. Prosser J, Thompson AM, Cranston G and Evans HJ. (1990). Oncogene, 5, 1573 ± 1579. Shankey TV, Rabinovitch PS, Bagwell B, Bauer KD, Duque RE and Hedley DW. (1993). Cytometry, 14, 472 ± 477. Symonds H, Krall L, Remington L, Saenz-Robies M, Lowe S, Jacks T and Van Dyke T. (1994). Cell, 78, 703 ± 713. Thukral SK, Lu Y, Blain GC, Harvey TS and Jacobsen VL. (1995). Mol. Cell Biol., 15, 5196 ± 5202. Vindelov LL, Christensen IJ and Nissen NI. (1983). Cytometry, 3, 323 ± 327. Williams JM and Daniel CW. (1983). Dev. Biol., 97, 274 ± 290. Zhang R, Haag JD and Gould MN. (1990). Cancer Res., 50, 4286 ± 4290.

1007