The p53 targeted allele generated by Donehower and coworkers 22 was introduced into the BALB/cHeA mouse at The Netherlands Cancer Institute (Amsterdam). The p53 heterozygous deficient mice (p53^+/-) were repeatedly backcrossed to BALB/cHeA mice more than 30 times, and maintained at the animal facility of Osaka Prefecture University. The Atm targeted mouse (129/SvEv-Atm^tm1Awb/+ mouse) was originally generated in the Jackson Laboratory 23. The Atm heterozygous deficient mice (Atm^+/-) were repeatedly backcrossed more than 10 times to MSM/Ms mice. The BALB/cHeA-p53^+/- mice were crossed with MSM/Ms-Atm^+/- mice, and females of the F₁ progeny ([BALB/cHeA × MSM/Ms]F₁) with four genotypes (i.e. p53^+/- Atm^+/-, p53^+/- Atm^+/+, p53^+/+ Atm^+/- and p53^+/+ Atm^+/+) were used in the experiments. The conditions for breeding were described previously 24.

Mice were exposed at 5 weeks of age to X-rays (5 Gy; 260 kV, 12.0 mA, 0.3 mm Cu + 0.5 mm Al filter; 0.5 Gy/min) from an X-ray generator (Radioflex 350; Rigaku Industrial Co., Takatsuki, Japan). All animal experiments were carried out in accordance with the standards relating to the care and management of experimental animals (Japan) and Osaka Prefecture University's guidelines for animal care and use.

Moribund mice were killed by cervical dislocation for autopsy. In cases of thymic lymphoma, the enlarged thymuses were examined as previously described 25. In tumour-bearing mice the tumours were fixed in 10% buffered formalin, processed histologically, and stained with haematoxylin and eosin. The processed tumour specimens were evaluated by medical and veterinary pathologists using the Annapolis guidelines established by Cardiff and coworkers 26.

Normal and mammary carcinoma tissues were removed. Isolation of DNA, PCR amplification, electrophoresis of PCR products and assessment of allelic losses were performed according to a procedure described previously 24. Genotypes for the wild-type and targeted alleles of p53 and Atm genes were determined by analyzing the PCR products for these alleles. Amplification for the p53 alleles was done as described elsewhere 27. The wild-type and the targeted alleles of the p53 gene were amplified by PCR using primers p53-4F (5'-CGACCTCCGTTCTCTCTCCTCTCTT-3') and p53-6R (5'-AGACGCACAAACCAAAACAAAATTACA-3'), and primers p53-NF (5'-GCCTTCTATCGCCTTCTTGACGAGT-3') and p53-6R, respectively. Similarly, amplification of the wild-type and the targeted allele of the Atm gene were performed by using primers IMR0640F (5'-GCTGCCATACTTGATCCATG-3') and IMR0641R (5'-TCCGAATTTGCAGGAGTTG-3'), and primers IMR0640F and AtmNeo410R (5'-CGGTGGATGTGGAATGTGTG-3'), respectively.

Statistical significance was evaluated for the incidence of mammary carcinoma and number of carcinomas per mouse by χ² analysis and Mann–Whitney U-test, respectively. Comparison of latency in mammary carcinoma development was examined by unpaired Student's t-test.

Mammary carcinomas occurred in BALB/c mice of the p53^+/- genotype, and tumours of similar macroscopic morphologies were also observed in the (BALB/c × MSM/Ms)F₁–p53^+/- mice. The histological features of these tumours in the mammary glands are shown in Fig. 1a,1b,1c,1d,1e,1f, which indicates that they are adenocarcinomas. The histopathology of the tumours in nonirradiated mice (Fig. 1a,1b) exhibited more hyperplastic lesions than did those in irradiated mice (Fig. 1c,1d,1e,1f), but basically there were no marked differences between the two groups. There was also no remarkable difference in histopathological features between Atm^+/- (Fig. 1c,1d) and Atm^+/+ mice (Fig. 1e,1f) in the irradiated groups. These tumours formed glands lined by highly pleomorphic cells exhibiting frequent mitosis, and were classified as glandular adenocarcinomas, high grade according to the Annapolis Pathology Classification 26.

Lymphomas (Fig. 1g), mainly thymic lymphomas, were efficiently induced by exposure to X-rays, regardless of p53 and Atm genotype, although only a few lymphomas were observed in nonirradiated mice. Ovarian carcinomas, osteosarcomas (Fig. 1h) and hepatomas developed spontaneously. Squamous cell carcinomas, basal cell carcinomas, histiocytic sarcomas and granulocytic leukaemias were also observed in irradiated mice.

Twenty-eight p53^+/- Atm^+/-, 22 p53^+/- Atm^+/+, 11 p53^+/+ Atm^+/- and 25 p53^+/+ Atm^+/+ mice were examined for spontaneous development of tumours until age 26 months (113 weeks). Fourteen out of 28 p53^+/- Atm^+/- mice (50%) and seven out of 22 p53^+/- Atm^+/+ mice (32%) developed mammary carcinomas during the period of observation (Table 1, Fig. 2). The incidence of mammary carcinomas in p53^+/- Atm^+/- mice appeared to be higher than that in p53^+/- Atm^+/+ mice, but the incidences did not differ significantly between Atm^+/- and Atm^+/+ genotypes (P = 0.32, by Fisher's exact probability test). Among the tumours that developed in the p53^+/- mice, mammary carcinoma were the most common. These mammary carcinomas were mainly observed at 41–77 weeks after birth (Fig. 2). No significant difference in latency in the nonirradiated groups was observed between doubly heterozygous mice and p53 singly heterozygous mice (P = 0.47, by unpaired Student's t-test). None of 11 p53^+/+ Atm^+/- mice and 25 p53^+/+ Atm^+/+ mice developed mammary carcinomas. Several lymphomas and a few other tumours developed in the four genotypes (Table 1). Thus, mammary carcinoma development depended strongly on p53 heterozygous deficiency in (BALB/c × MSM/Ms)F₁ mice, and p53^+/- mice of both Atm^+/+ and Atm^+/- genotypes developed mammary carcinoma.

To test the effect of haploinsufficiency of the Atm gene on mammary carcinogenesis after X-irradiation in p53^+/- mice, 55 p53^+/- Atm^+/-, 61 p53^+/- Atm^+/+, 47 p53^+/+ Atm^+/- and 53 p53^+/+ Atm^+/+ mice (for a total of 216 mice) were exposed to X-rays (5 Gy) at age 5 weeks. Only one out of 53 p53^+/+ Atm^+/+ mice and none of 47 p53^+/+ Atm^+/- mice developed mammary carcinoma, indicating that almost all p53^+/+ mice fail to develop mammary carcinomas, despite X-irradiation and irrespective of Atm gene status. In contrast, 32 out of 55 p53^+/- Atm^+/- mice (58%) and 19 out of 61 p53^+/- Atm^+/+ mice (31%) developed mammary carcinomas (Table 2, Fig. 2). The proportion of mice developing mammary carcinomas in the p53^+/- Atm^+/- group was significantly greater than that in the p53^+/- Atm^+/+ group (P = 0.0034, by χ² test). A total of 52 mammary carcinomas developed in 55 p53^+/- Atm^+/- mice (average number of mammary carcinomas/mouse = 0.95), whereas 28 mammary carcinoma developed in 61 p53^+/- Atm^+/+ mice (average number of mammary carcinomas/mouse = 0.46; Table 2). The average number of mammary carcinomas per mouse in the p53^+/- Atm^+/- group was significantly greater than that in p53^+/- Atm^+/+ mice (P = 0.0052, by Mann–Whitney U-test). Thus, Atm heterozygous deficiency enhanced development of mammary carcinoma in irradiated p53 heterozygous knockout mice. Spring and coworkers 14 observed no tumours in Atm knockout (Atm^+/-) heterozygous mice. Mice bearing a knockout allele of Atm and humans carrying a mutant allele of truncated type in the ATM gene have been shown not to have obviously elevated susceptibility to mammary carcinogenesis. Our findings show that heterozygosity for a null knockout allele of Atm enhances mammary carcinogenesis under p53^+/- status, although the Atm mutation is not a dominant-negative type. Heterozygous deficiency of p53 might make clear the effect on mammary carcinogenesis of haploinsufficiency in the Atm gene.

These mammary carcinomas were observed significantly earlier (at 18–38 weeks after irradiation; i.e. 23–43 weeks of age) than in the nonirradiated group (age 41–75 weeks; Fig. 2). In particular, mammary carcinomas frequently developed 23–28 weeks after irradiation. The mean (± standard deviation) latency periods were 32.6 ± 4.8 and 29.8 ± 3.6 weeks in p53^+/- Atm^+/- and p53^+/- Atm^+/+ mice, respectively. Thus, X-irradiation at 5 Gy at age 5 weeks considerably shortened the latency period of mammary carcinoma development in these two groups with different genotypes. As shown in Tables 1 and 2, the incidences of mammary carcinoma in p53^+/- Atm^+/- mice were 58% (32 out of 55) and 50% (14 out of 28) in irradiated and nonirradiated groups, respectively; in p53^+/- Atm^+/+ mice the incidence in the irradiated group was 31% (19 out of 61) and that in the nonirradiated group was 32% (7 out of 22). The incidence of mammary carcinoma for each genotype was similar between irradiated and non-irradiated groups. Thus, irradiation may not elevate the incidence of the tumours.

Altogether, irradiation markedly hastened mammary carcinoma development in the p53^+/- mice, in which mammary carcinomas developed spontaneously. Furthermore, irradiation also induced lymphomas, mainly thymic lymphomas, in all four genotypes of mice. The incidence of the lymphomas did not differ significantly among the four groups, with different genotypes for p53 and Atm genes. A high incidence of thymic lymphoma was observed in previous studies performed using p53 heterozygous deficient F₁ mice 1728. In the present study an extremely high incidence of tumours, most of which were mammary carcinomas and thymic lymphomas, was observed in irradiated p53^+/- Atm^+/- mice.

The wild-type alleles of the p53 and Atm genes were examined in mammary carcinoma tissue from heterozygous mice. The wild-type p53 allele was lost in 25 (96%) out of 26 mammary carcinomas from irradiated p53^+/- mice and in all of 15 mammary carcinomas from nonirradiated p53^+/- mice, regardless of Atm gene status. Only one mammary carcinoma in irradiated p53^+/+ Atm^+/+ mice (Table 2) was found to retain the p53 wild-type allele. On the other hand, wild-type Atm allele was preserved in all of 17 mammary carcinomas from irradiated Atm^+/- mice and in all of 10 mammary carcinoma

s

Haploinsufficiency for tumour development was reported for the Nbn knockout mice, the mouse homologue of NBS1. Heterozygosity for the Nbn knockout allele rendered the mice susceptible to tumour development, yet the Nbn wild-type allele was fully retained in all 12 tumours examined 29. p27 heterozygotes are also predisposed to tumours in multiple tissues when challenged with irradiation or a chemical carcinogen, and in the developed tumours the remaining wild-type allele is neither mutated nor silenced, indicating that p27 is haploinsufficient for tumour suppression 30. In that study, heterozygous Atm knockout enhanced mammary carcinoma development in p53-heterozygous deficient mice, but the effect of Atm deficiency was not as profound. A previous study showed that heterozygosity for the Atm knockout allele did not enhance tumour development but that dominant-negative type missense mutations in the Atm gene did 14. In humans, heterozygosity of the truncation-type mutations, which represent the majority of Atm mutations that occur in humans, had no effect on carcinogenesis, suggesting that enhancement in tumourigenesis depends strongly on mutation type. However, in the present study we demonstrated that haploinsufficiency does occur for the Atm null allele in combination with heterozygous deficiency in the p53 gene. The p53 heterozygous deficient BALB/c mice, which developed mammary carcinomas early and efficiently, may represent a useful model for the study of effects of genes other than Atm on mammary carcinogenesis. F₁ mice between different subspecies may also provide an experimental system for precise genome-wide allelotype analysis of genes that cooperate with p53.