Spontaneous and Irradiation-Induced Tumor Susceptibility in Brca2 Germline Mutant Mice and Cooperative Effects with a p53 Germline Mutation

Abstract

Mutations in both p53 and BRCA2 are commonly seen together in human tumors suggesting that the loss of both genes enhances tumor development. To elucidate this interaction in an animal model, mice lacking the carboxy terminal domain of Brca2 were crossed with p53 heterozygous mice. Females from this intercross were then irradiated with an acute dose of 5 Gy ionizing radiation at 5 weeks of age and compared to nonirradiated controls. We found decreased survival and timing of tumor onsets, and significantly higher overall tumor incidences and prevalence of particular tumors, including stomach tumors and squamous cell carcinomas, associated with the homozygous loss of Brca2, independent of p53 status. The addition of a p53 mutation had a further impact on overall survival, incidence of osteosarcomas and stomach tumors, and tumor latency. The spectrum of tumors observed for this Brca2 germline mouse model suggest that it faithfully recapitulates some human disease phenotypes associated with BRCA2 loss. In addition, these findings include extensive in vivo data demonstrating that germline Brca2 and p53 mutations cooperatively affect animal survivals, tumor susceptibilities, and tumor onsets.

Introduction

Mutations in the breast cancer susceptibility gene BRCA2 are responsible for 32% of all hereditary breast cancers in women (Ford et al., 1998). Inherited mutations in BRCA2 also predispose individuals to a relatively high lifetime risk of ovarian cancer as well as increased risks for cancers of the stomach, pancreas, colon, prostate, and melanoma (Wooster et al., 1995; Ford et al., 1998; BCLC, 1999; King et al., 2003). Mutations in BRCA2 are also linked with Fanconi Anemia (FA), a disease characterized by increased susceptibility to multiple cancers, particularly squamous cell carcinomas (Howlett et al., 2002; D’Andrea, 2003; Hussain et al., 2004). It is now known that BRCA2 is 1 of the 8 complementation group proteins in the FA complex, and FANCD2 directly interacts with BRCA2 in DNA damage response pathways (Howlett et al., 2002; D’Andrea, 2003; Hussain et al., 2004). Germline Brca2 mutant models with extensive deletions have had limited viability (Connor et al., 1997; Friedman et al., 1998; Ludwig et al., 2001). We previously reported that mice with a germline homozygous Brca2 deletion of exon 27 alone, however, are viable but cancer-prone (McAllister et al., 2002).

BRCA2 has a direct role in DNA repair primarily through its interactions with RAD51, a protein critical for the recombinational repair of DNA double-strand breaks (Yuan et al., 1999; Howlett et al., 2002). The interaction of BRCA2 and RAD51 through the BRC repeats in the BRCA2 protein is necessary for homology-directed DNA repair (D’Andrea, 2003; Powell et al., 2003). Defects in this interaction are likely to contribute to an increased chromosomal and genetic instability, and a hypersensitivity to radiation and DNA cross-linking agents which is widely observed in FA patients, BRCA2-deficient cells, and Brca2 mutant mice (Morimatsu et al., 1998; Moynahan et al., 2001; Kraakman-van der Zwet et al., 2002; D’Andrea, 2003; Donoho et al., 2003; Hussain et al., 2004).

Brca2 mutations and p53 mutations commonly coexist in cancers, suggesting there is a possible interactive effect associated with these 2 deficiencies during tumorigenesis. BRCA2-associated cancers, particularly breast and ovarian cancer, frequently have mutations in p53 (Gretarsdottir et al., 1998; Rhei et al., 1998; Ramus et al., 1999). Some of the p53 mutations observed in BRCA2-linked tumors are unique to BRCA2-mutated tumors (Crook et al., 1998; Smith et al., 1999). An accelerated development of mammary tumors was observed in 2 independent mouse models combining a Brca2 conditional mutation with a p53 mutation (Jonkers et al., 2001; Cheung et al., 2002, 2004). Generally, the p53 protein controls cell-cycle arrest and apoptosis in response to DNA-damaging agents. p53 is also the cell cycle checkpoint for BRCA2-associated response to radiation (Powell and Kachnic, 2003; Tutt et al., 2003).

We sought to further examine the tumor spectrum of Brca2 deficient mice and the interaction of Brca2 and p53 mutations over time by crossing mice containing a germline homozygous Brca2 mutation with mice containing a germline p53 mutation that removes approximately 40% of the p53 coding region (Jacks et al., 1994). The Brca2 mutant mice have a mutation in exon 27, which deletes the final Rad51-DNA binding domain. This domain has been found to be essential for RAD51 foci formation during DNA damage and the stabilization of the RAD51 nucleoprotein filament (Yuan et al., 1999). This hypomorphic mutation is comparable to the C-terminal deletion of the human BRCA2 protein associated with Fanconi Anemia (Howlett et al., 2002). Several p53 germline knockouts, including the one we evaluated herein, have heterozygous tumor phenotypes, which generally resemble the hereditary Li-Fraumeni syndrome with significant predisposition to lymphoma, sarcoma, and carcinoma (Malkin et al., 1990; Donehower et al., 1992; Jacks et al., 1994; Kuperwasser et al., 2000).

Irradiation of mice initiates the carcinogenic and DNA damaging process without the complications of altered metabolism associated with other carcinogenic agents (Ullrich et al., 1996; Ponnaiya et al., 1997; Ullrich and Ponnaiya, 1998). We therefore analyzed the interaction of Brca2 and p53 mutations in response to this environmental insult by exposing the Brca2 and p53-deficient mice to a single sublethal dose of 5 Gy irradiation and comparing them to nonirradiated controls. Four and 5 Gy irradiation doses had previously been used to induce substantial tumorigenesis in these p53-deficient animals (Kemp et al., 1994; Backlund et al., 2001). We irradiated the animals at 5 weeks of age, which is the time when the mammary glands of mice are undergoing ductal morphogenesis (Snedeker et al., 1991). Animals were sacrificed when morbid to fully evaluate the lifetime risk of developing tumors. Such an analysis also reveals whether there are differences in tumor latency and survival defined as time to natural death or morbid sacrifice.

Most viable Brca2 knockout models have displayed a substantial reduction in life span and some hallmarks of an aging phenotype, although an extensive analysis of tumor spectrum had not been possible due to limited viability (Connor et al., 1997; Friedman et al., 1998; Ludwig et al., 2001; McAllister et al., 2002; Donoho et al., 2003). We predicted that the tumor spectrum of these Brca2 mutant animals would model some of the tumor types observed in human diseases associated with BRCA2 or FA loss. We hypothesized that the loss of germline p53 and Brca2 deficiencies would cooperatively reduce survival and increase tumor susceptibility. We define cooperation as meaning that the combined effect of both of these mutations would be significantly greater than each individual effect alone. We predicted that the addition of radiation would further accelerate the effects of loss of these 2 genes that are closely associated with DNA repair and damage response pathways such that irradiated animals with combined p53 and/or Brca2 defects would have a further reduction in survival and increased tumor susceptibility.

Materials and Methods

Mice and Tissues

The targeting construct design and generation of germline mutant mice with an exon 27 mutation of Brca2 and genotyping of these mice were previously described (McAllister et al., 2002). The p53 wild-type and p53 mutant alleles were amplified using the following PCR primers: p53 MF, 5′-CTATCAGGACATAGCGTTGG-3′ and p53 WR, 5′-TATACTCAGAGCCGGCCT-3′ for the mutant allele and p53 WF, 5′-ACAGCGTGGTGGTACCTTAT-3′ and p53 WR for the wildtype allele. The thermal cycling conditions were as follows: 94′ for 2 minutes, followed by 35 cycles of 94′ for 1 minute, 55′ for 1 minute, 72′ for 1 minute followed by 72′ for 7 minutes.

The germline Brca2 ^{− / −} animals are now available at the NCI Mouse Models of Human Cancers Consortium (MMHCC) repository on the Balb/c (C.Cg-Brca2<tm1kamc>) and SWR (Cg-Brca2<tm1kamc>) strain. Mice heterozygous for a germline Brca2 exon 27 mutation on a C57Bl/6 background (backcross generation 6–8) were initially crossed with mice heterozygous for a p53 null mutation on a pure inbred (>10 backcross generations) BALB/c genetic background (Jacks et al., 1994) (see schematic below). Offspring from this cross that were heterozygous for both the Brca2 and p53 mutations were then intercrossed to generate the F2 offspring with all 9 possible genotypic classes represented. However, animals that were p53 homozygous null were generated in small numbers and died so early that they were not included in the survival or tumor analyses of this study.

A smaller number of F2 crosses were generated utilizing animals heterozygous for Brca2 alone (wild-type for p53 ) crossed with animals heterozygous for both the Brca2 and p53 mutations:

1st Cross: p53 ^{+ / −}(Balb/c) × Brca2 ^{+ / −} (C57BL/6)

2nd Cross: p53 ^{+ / −}/Brca2 ^{+ / −} x p53 ^{+ / −}/Brca2 ^{+ / −} or p53 ^{+ / +}/ Brca2 ^{+ / −} x p53 ^{+ / −}/Brca2 ^{+ / −}

F2 Offspring: 9 genotypic classes (p53 null animals not included in final analyses) irradiated at 0 or 5 Gy at 5 weeks of age.

Approximately 900 virgin females from these F2 intercrosses were either irradiated with a single acute dose of ionizing radiation of 5 Gy at 5 weeks of age or observed as untreated virgin female mice from the 6 representative genotypic classes as controls for the influence of irradiation on neoplastic development. Genomic DNA from pups was isolated from tail biopsies obtained at weaning for genotyping by PCR as previously described (McAllister et al., 2002). Virgin females were group housed in plastic cages on pressed wood-chip bedding. Animals had access to an NIH-31 diet (18% protein, 4% fat, and 5% fiber; Zeigler Bros., Gardeners, PA) and water ad libitum. Compliance with all NIH guidelines for the humane care and use of animals was followed in this study protocol.

This study followed NIEHS guidelines for “Endpoints for Solid Tumor Studies” which does not allow the actual death of the animal to be used as an endpoint. We therefore define survival for the purposes of this study with the following stated criteria that were used to determine when to euthanize the animal: (1) When the size or location of the tumor begins to interfere with the normal behavior of the animal in eating, drinking, freedom of movement, etc. (2) When the tumor ulcerates or develops necrotic areas. (Some very rapidly growing tumors outgrow their blood supply before the tumor is large enough to interfere with movement). (3) In tumors which are not palpable, clinical signs such as weight loss, lethargy, etc. were used. These endpoint determinations were made by trained and experienced personnel and were stringently and consistently applied to all the animals of this study.

A small number of animals (approximately 2–3 of each available genotype) were sacrificed at 3 and 6 months of age to examine animals for preneoplastic lesions. All p53 heterozygous animals that were homozygous for the Brca2 mutation in all groups were dead by 18 months of age. Remaining p53 heterozygous mice in all groups were necropsied at 18 months of age while some of the p53 wild-type animals of all groups were observed for 24 months of age to determine tumor spectrum. The terminal sacrifice of all remaining animals occurred at 24 months of age. All tissue and tumor samples were removed postmortem and fixed in 10% neutral buffered formalin for histology and flash-frozen for future molecular studies. Specimens were processed for routine histology, embedded in paraffin, sectioned, and stained with H&E. Histological examination was performed by veterinary pathologists (C.H., J.S., and B.D.). When possible, the classification of mammary tumors followed the guidelines and recommendations established at the Annapolis meeting (Cardiff et al., 2000). The subtypes of mammary carcinomas were based on characterizing the predominant histological pattern in the tumor although if more than one pattern was found to predominate then the tumor was classified as mixed. Routine H&E staining was performed on the left #4 and #5 mammary gland of each animal, followed by histopathological analysis. An attempt was made to obtain tissues from all animals that were found dead. Animals that were found dead and too autolyzed to obtain good tissue samples for histopathology were included in the survival analysis and not in the tumor analysis.

Statistical Analyses

We defined survival time as time to sacrifice based on the criteria defined according to the NIEHS guidelines for tumor studies, regardless of cause. The effects of the different Brca2 and p53 genotypes and radiation on survival were analyzed using SAS Proc Lifetest with survival probabilities estimated using the Kaplan and Meier product-limit procedure (Allison, 1995). Because of the many comparisons and contrasts used in these analyses, the significance level was p < 0.001 for each individual comparison in order to insure an overall false positive rate of 0.05. The log-rank test was used to test for equality of survival curves between individual pair-wise Brca2, p53 , and exposure levels. Animals not dying from natural causes (such as all time point sacrifices) and animals reaching the end of the study without dying were treated as censored data for the statistical survival analysis. Animals used for time point sacrifices were incorporated in the survival analysis until the day of scheduled sacrifice.

All moribund sacrifices as well as the 3, 6, 18, and 24-month time point sacrifices were included in the tumor analysis. The tumor analysis was considered statistically significant at p ≦ 0.05. The tumor incidence was analyzed using a continuity-corrected Poly-3 test, a survival adjusted quantal-response procedure as used for National Toxicology Program reports (Bailer and Portier, 1988; Portier and Bailer, 1989; Bieler and Williams, 1993). Tests of significance included pairwise comparisons of each exposed group with controls.

Results

The following mice were included in this study based on genotypes as Brca2 wildtype (Brca2 ^{+ / +}), Brca2 heterozygous mutant (Brca2 ^{+ / −}), Brca2 homozygous mutant (Brca2 ^{− / −}), p53 wildtype (p53 ^{+ / +}), and p53 heterozygous mutant (p53 ^{+ / −}). For the nonirradiated animals, 56 p53 ^{+ / +}/Brca2 ^{+ / +}, 80 p53 ^{+ / −}/Brca2 ^{+ / +}, 104 p53 ^{+ / +}/Brca2 ^{+ / −}, 148 p53 ^{+ / −}/Brca2 ^{+ / −}, 36 p53 ^{+ / +}/Brca2-^{− / −}, and 49 p53 ^{+ / −}/Brca2 ^{− / −} animals were evaluated. For the animals irradiated at 5 Gy dose , 45 p53 ^{+ / +}/Brca2 ^{+ / +}, 45 p53 ^{+ / −}/Brca2 ^{+ / +}, 105 p53 ^{+ / +}/Brca2 ^{+ / −}, 92 p53 ^{+ / −}/-Brca2 ^{+ / −}, 49 p53 ^{+ / +}/Brca2 ^{− / −}, and 45 p53 ^{+ / −}/Brca2 ^{− / −}animals were evaluated. The following numbers of animals were sacrificed at scheduled timed necropsies: 27 animals at 3 months of age, 50 animals at 6 months of age, 36 animals at 18 months of age, and 82 animals at 24 months of age. Fifty-one animals were “found dead” animals in this study. All remaining animals were sacrificed according to the NIEHS guidelines as described in the Materials and Methods.

Survival

We were not allowed to use actual death as an endpoint in this tumor study in accordance with the NIEHS guidelines for tumor studies so some variability may have been introduced into the survival endpoints for this study. However, we were careful to apply the criteria for sacrifice of the mice very stringently and consistently to all animals in this study such that this data should be a true representation of the survival differences of the animals in the different genotypic and irradiated groups.

Individual and Combined Brca2 and p53 Mutations Decreased Survival in Nonirradiated Mice.

Mice with either a p53 or a Brca2 mutation had significantly reduced survival in the non-irradiated groups. Specifically, regardless of p53 genotype, all Brca2 ^{− / −} mice had a significantly reduced lifespan compared to Brca2 ^{+ / −} or Brca2 ^{+ / +} animals in the nonirradiated groups (p < 0.0001) (Figures 1 and 2 and Table 1—Comparisons 7, 9, 11, and 13). The heterozygous loss of p53 (p53 ^{+ / −}) also significantly shortened survival compared to the p53 ^{+ / +} animals for all three Brca2 genotypes in the non-irradiated groups (p < 0.0001) (Figure 1 and 2 and Table 1—Comparisons 1, 3, and 5). Additionally, the combined homozygous loss of Brca2 and heterozygous loss of p53 significantly affected the survival outcome compared to either gene mutation alone (p < 0.0001 for both comparisons) (Figure 3 and Table 1—Comparisons 5 and 9). For untreated p53 ^{+ / +} /Brca2 ^{+ / +}mice, 50% survival was not yet achieved at the time of the final terminal sacrifice (104 weeks). For the untreated p53 ^{+ / +} /Brca2 ^{− / −}, 50% survival occurred at 78 weeks of age. For the untreated p53 ^{+ / −}/Brca2 ^{+ / +} animals, 50% survival occurred at approximately 63 weeks of age. In contrast, 50% survival occurred at 39 weeks of age in animals having both genetic mutations (p53 ^{+ / −}/Brca2 ^{− / −}).

Radiation Decreased Overall Survival for all Genotypes

A single high dose of radiation (5 Gy) significantly decreased survival compared to the corresponding nonirradiated group for all p53 and Brca2 genotypes (Figures 1, 2, 4, 5, and Table 1—Comparisons 15–20) (p < 0.0001 except for comparison 20 which is p = 0.0002). Only 21% of the irradiated animals were still alive at 18 months of age. Only 3% of the animals were alive at 24 months of age in all irradiated groups. In contrast, approximately 15% of all nonirradiated animals were alive at 24 months.

Individual and Combined Brca2 and p53 Mutation Affected Survival in Irradiated Mice.

Survival of Brca2 ^{− / −}animals with p53 ^{+ / +} genotypes was significantly decreased compared to the corresponding Brca2 ^{+ / −} or Brca2 ^{+ / +} mice irradiated at 5 Gy (p = 0.0004 and p = 0.0001) (Figure 4 and Table 1—Comparisons 8 and 12). The Brca2 ^{− / −} mice were not significantly different from Brca2 ^{+ / −} or Brca2 ^{+ / +} mice on the p53 heterozygous background when irradiated at 5 Gy (Figure 5 and Table 1—Comparisons 10 and 14) (p = 0.0025 and p = 0.0047). The 50% survival occurred at approximately 26 weeks for p53 ^{+ / −}/Brca2 ^{− / −} animals that were irradiated. A significant difference in survival in the irradiated groups occurred between the p53 ^{+ / −} and p53 ^{+ / +} for all three Brca2 genotypes (Table 1—Comparisons 2, 4, and 6) (p < 0.0001). There was no significant difference in survival between animals having both genetic defects (p53 ^{+ / −}/Brca2 ^{− / −}) compared to animals having only the p53 genetic defect (p53 ^{+ / −}/Brca2 ^{+ / +}) in irradiated groups (Table 1—Comparison 10).

Effect of the Brca2 or p53 Mutation with 5 Gy Irradiation for Survival.

We examined whether the combined effect of a single high-dose radiation exposure with either the homozygous loss of Brca2 or the heterozygous loss of p53 would significantly affect survival compared to the individual survival effect of either the radiation alone or the Brca2 or p53 genetic defect alone. Animals irradiated at 5 Gy that were p53 ^{+ / +}/Brca2 ^{− / −} had significantly decreased survival compared to irradiated p53 ^{+ / +}/Brca2 ^{+ / +} animals or nonirradiated p53 ^{+ / +}/Brca2 ^{− / −} animals (Table 1—Comparisons 8 and 17) (p = 0.0004 and p < 0.0001, respectively). Similarly, animals irradiated at 5 Gy that were p53 ^{+ / −}/Brca2 ^{+ / +} had significantly decreased survival compared to irradiated p53 ^{+ / +}/Brca2 ^{+ / +} animals or non-irradiated p53 ^{+ / −}/Brca2 ^{+ / +} animals (Table 1—Comparisons 2 and 18) (p < 0.0001 in both cases).

Effect of the p53 Compared to Brca2 Mutation for Survival in Nonirradiated and Irradiated Mice.

To further understand the possible contributions of each genetic loss to the survival reduction in mice having both mutations, we also examined whether the individual p53 heterozygous mutation had a significantly greater effect alone on survival compared to the effect observed with a Brca2 homozygous mutation alone in either the nonirradiated or irradiated animals. In the nonirradiated animals, there was no significant difference between p53 ^{+ / −}/Brca2 ^{+ / +} animals compared to p53 ^{+ / +}/Brca2 ^{− / −}animals (p = 0.0024). In the irradiated animals, a statistically significant difference between p53 ^{+ / −}/Brca2 ^{+ / +} animals compared to p53 ^{+ / +}/Brca2 ^{− / −} animals was observed (p = 0.0003).

Survival Analysis Adjusted for Tumor Burden

Survival was also analyzed using tumor burden (defined as the number of different types of tumors by location) as a variable. The survival analysis results remained the same with tumor burden taken into account in all cases with the exception of the comparison of the irradiated Brca2 ^{− / −} and Brca2 ^{+ / +} animals when p53 was wild-type at the 5 Gy exposure level (the p-value was 0.0018 with a cutoff of 0.001).

Tumor Susceptibility and Spectrum

Tumor data were analyzed using the Poly 3 test instead of the standard Cochran-Armitage test (see Materials and Methods) because the Poly 3 test accounted for survival differences in the time-point and moribund sacrifice data and for the considerable survival differences between the distinct genotypes. A few results reached statistical significance with this test with small changes observed in tumor incidences and other results were not statistically significant even with substantial differences observed in tumor incidence.

Individual Brca2 Mutation Associated with Overall Increase in Tumors and Specific Tumor Types in Nonirradi-ated Animals.

Homozygous loss of Brca2 resulted in an increased incidence of specific cancers compared to Brca2 wild-type or heterozygous animals with the same p53 genotype. Stomach cancers were primarily observed in Brca2 ^{− / −}animals for the nonirradiated group, with the highest incidence occurring in the nonirradiated p53 ^{+ / −}/Brca2 ^{− / −} animals (17%) (Table 2). The most common gastric tumors observed in this study were squamous cell carcinomas of the forestomach (Figure 6A). A significantly higher incidence of squamous cell carcinomas of the stomach was observed in Brca2 ^{− / −} animals compared to Brca2 ^{+ / −} or Brca2 ^{+ / +} mice for both p53 ^{+ / −} and p53 ^{+ / +} genotypes (Table 2) (p = 0.001, p = 0.001, p = 0.004, and p = 0.017, respectively). A significantly higher incidence of stomach squamous cell hyperplasia (diffuse rather than focal) also occurred in the Brca2 ^{− / −} animals compared to wild-type or heterozygous animals for the same p53 genotype (Table 2) (p = 0.001 for all 4 comparisons).

Squamous cell carcinomas in other locations occurred predominantly in the Brca2 ^{− / −} animals and carcinomas at all sites were significantly higher in the Brca2 ^{− / −} mice compared to Brca2 ^{+ / −} or Brca2 ^{+ / +} mice in both the nonirradiated p53 ^{+ / −} and p53 ^{+ / +} groups (Table 2) (p = 0.007, p =0.001, p = 0.001, and p = 0.001, respectively). Osteosarcomas were significantly increased by 20% and 23% for Brca2 ^{− / −} animals with p53 ^{+ / −} genotypes compared to the corresponding Brca2 ^{+ / −} or Brca2 ^{+ / +} groups, respectively (p = 0.001 for both) (Figure 6B). Osteosarcomas were also significantly increased for Brca2 ^{− / −} animals compared to Brca2 ^{+ / −} animals with p53 ^{+ / +} genotypes as well (p =0.049). Finally, a significantly higher overall tumor incidence and incidence of animals with multiple tumors was also observed in the Brca2 ^{− / −} animals compared to Brca2 ^{+ / −} or Brca2 ^{+ / +} animals on the p53 wild-type background (Table 2) (p = 0.006 and p = 0.028, respectively).

Individual p53 Mutation Increased Tumors in Nonirradiated Animals.

p53 heterozygosity also affected the tumor spectrum in the non-irradiated groups. A significantly higher incidence of osteosarcomas, mammary carcinomas, and lymphomas (Figure 6C) occurred in the p53 heterozygotes compared to p53 wild-type animals for all three Brca2 genotypes (Table 2) (p = 0.001 in all cases except p = 0.009 for Brca2 wild-type for lymphoma). Few mammary tumors occurred on the p53 wild-type background. The predominantly p53 ^{+ / −} animals developed mammary carcinomas of a variety of different subtypes. The most common mammary carcinomas were the glandular types (Figure 6D). Additional subtypes observed were adenosquamous, solid, spindle cell, papillary, and cribiform patterns. Hemangiosarcomas, myoepitheliomas, lymphomas, and undifferentiated carcinomas were also found in the mammary gland. No correlations of the mammary histological subtypes with either the Brca2 or p53 genotype were seen. Heterozygosity of p53 also significantly increased tumor multiplicity (number of tumors per animal) compared to p53 ^{+ / +} for all three Brca2 genotypes (p = 0.021 forBrca2^{− / −} , p =0.001for both Brca2^{+ / −} and Brca2^{+ / +}).

Individual Brca2 Mutation Affected Tumor Spectrum in Irradiated Animals.

Some specific cancers were still significantly increased in irradiated Brca2 homozygous mice with a p53 ^{+ / +} background (compared to irradiated Brca2 ^{+ / −} or Brca2 ^{+ / +} mice on the p53 wild-type background) (Table 3). These included osteosarcomas (p = 0.010 for Brca2 ^{+ / −}comparison), and squamous cell carcinomas (p = 0.001 for Brca2 ^{+ / +} comparison). Lymphomas were also significantly increased in the Brca2 ^{− / −} animals (p = 0.002 for Brca2 ^{+ / −}and p = 0.046 for Brca2 ^{+ / +} comparison).

Individual p53 Mutation Affected Tumor Spectrum in Irradiated Animals.

A significantly higher incidence of lymphomas was still observed in the p53 heterozygous compared to p53 wild-type groups for all 3 Brca2 genotypes with irradiation (Table 3) (p = 0.001 for Brca2 ^{+ / +}, p = 0.001 for Brca2 ^{+ / −}, and p = 0.040 for Brca2 ^{− / −}). A significantly higher incidence of mammary carcinomas occurred in the p53 ^{+ / −} compared to p53 ^{+ / +} animals with the Brca2 ^{− / −} mutation after 5 Gy exposure (Table 3) (p = 0.001). There was no longer a significant difference in osteosarcoma incidence between irradiated p53 ^{+ / −} and p53 ^{+ / +} mice as occurred between the non-irradiated p53 ^{+ / −} and p53 ^{+ / +} mice with all 3 Brca2 genotypes (p = 0.996 for Brca2 ^{+ / +}, p = 0.221 for Brca2 ^{+ / −}, and p = 0.589 for Brca2 ^{− / −}).

Effect of Radiation on Tumor Spectrum

Specific effects due to radiation on tumor spectrum and distribution were also observed. Reproductive tumors, particularly ovarian tumors, were the most obvious tumor type induced with radiation alone. Specifically, there was a significantly higher incidence of ovarian adenomas in the irradiated compared to the non-irradiated group for the p53 ^{+ / +}/Brca2 ^{+ / −} mice (data not shown) (p = 0.009). Approximately half of the irradiated animals that were heterozygous for the p53 mutation and Brca2 mutation developed mammary tumors (Table 3). A significant increase of 29% in overall mammary tumor incidence was observed in the p53 ^{+ / −}/Brca2 ^{+ / −} animals when comparing the 5 Gy level to the 0 Gy level (Tables 2 and 3) (p = 0.045). A significantly higher incidence of mammary carcinomas was also observed between the 5 Gy exposure compared to the nonirradiated group for p53 ^{+ / +}/Brca2 ^{+ / −} animals (p = 0.014). A significantly higher incidence of lymphomas was observed for the 5 Gy level compared to the 0 Gy level for all genotypic classes except the p53 ^{+ / −}/Brca2 ^{− / −} genotypic class (Tables 2 and 3) (p = 0.001 for all significant comparisons except p = 0.013 for p53 ^{+ / +}/Brca2 ^{+ / +} and p = 0.052 for p53 ^{+ / −}/Brca2 ^{− / −}). The overall incidence of animals with tumors and the incidence of animals with multiple tumors were both increased with radiation compared to the corresponding nonirradiated groups for p53 ^{+ / +}/ Brca2 ^{+ / +}, p53 ^{+ / +}/Brca2 ^{+ / −}, and p53 ^{+ / −}/Brca2 ^{+ / −} mice (Tables 2 and 3) (p = 0.015, p = 0.001, and p = 0.001, respectively, for total tumor comparisons and p = 0.001, p = 0.001, and p = 0.003, respectively, for multiple tumor comparisons).

Combined Effect of p53 and Brca2 Mutations for Tumor Incidence in Control and Irradiated Mice.

In order to examine whether the combined effect of the p53 and Brca2 mutations acted cooperatively in cancer development, we compared the incidence of specific tumor types that occurred in each mutant genotype to the incidence that occurred in the double mutants. In the nonirradiated groups, the incidence of stomach tumors and osteosarcomas in animals with both genetic mutations was significantly greater than the incidence observed for these tumor types with either mutation alone (Table 2) (p = 0.003 and p = 0.006). Specifically, for gastric carcinomas, none occurred in the p53 heterozygous animals that were wild-type for Brca2 while 11% occurred in Brca2 ^{− / −} that were wild-type for p53 and 17% occurred in p53 ^{+ / −}/Brca2 ^{− / −} mice. For osteosarcomas, a 19% incidence was observed in the p53 heterozygous animals that were wild-type for Brca2 while 11% incidence occurred in Brca2 ^{− / −} mice that were wild-type for p53 and 42% of the p53 ^{+ / −}/Brca2 ^{− / −} animals had osteosarcomas. These combined effects of the Brca2 and p53 mutations were not observed in the irradiated groups.

Impact of Genotypes on Survival of Animals with Distinct Tumors

We further examined the potential interactions of these mutations on cancer development by considering that shorter survival times of mice with malignant cancers would represent shorter latency for that malignancy. Because many animals in this study had more than 1 distinct tumor type and we could not determine which tumor led to morbid sacrifice, we only analyzed mice with a single malignant tumor type. Subsetting the data resulted in reduced sample sizes and power. However, we suggest that the following comparisons provided evidence of genuine biological survival differences even though some did not actually meet the stringent 0.001 cutoff value.

Individual Brca2 Mutation Effect on Time of Tumor Onset.

A decreased survival for animals with lymphomas, osteosarcomas, or mammary tumors alone was associated with the presence of 2 mutant Brca2 alleles. Specifically, survival of animals with lymphomas alone was decreased in the Brca2 ^{− / −} mice compared to Brca2 ^{+ / −} or Brca2 ^{+ / +} animals with the p53 ^{+ / +} genotype in the nonirradiated group (data not shown) (p = 0.007 and 0.001, respectively). The p53 ^{+ / −}/Brca2 ^{− / −} mice with osteosarcomas alone had a significant decrease in survival compared to p53 ^{+ / −}/Brca2 ^{+ / −}mice and an almost significant decrease in survival compared to p53 ^{+ / −}/Brca2 ^{+ / +} mice with osteosarcomas alone in the non irradiated group (p < 0.0001 and p = 0.0015, respectively). The p53 ^{+ / −}/Brca2 ^{− / −} mice with mammary tumors alone had a significant decrease in survival compared to p53 ^{+ / −}/Brca2 ^{+ / +} or p53 ^{+ / −}/Brca2 ^{+ / −} mice with mammary tumors alone in the nonirradiated group as well (p = 0.0007 and p < 0.0001, respectively). Finally, Brca2 ^{− / −} mice with tumors of any type had a decrease in survival compared to tumor-bearing Brca2 ^{+ / +} or Brca2 ^{+ / −} mice for both p53 genotypes in both the irradiated and nonirradiated animals (p ≦ 0.005 for all four cases).

Combined Brca2 and p53 Mutation Effect on Timing of Tumor Onsets

The combination of a single p53 and homozygous Brca2 mutation decreased the survival of mice with a single distinct tumor compared to the individual genetic defect alone for some tumor types as well. The combination of a single p53 mutation and a homozygous Brca2 mutation decreased the survival of animals with lymphomas alone compared to the survival observed for p53 ^{+ / −}/Brca2 ^{+ / +} mice or p53 ^{+ / +}/Brca2 ^{− / −} animals with lymphomas alone at the 5 Gy dose (data not shown) (p = 0.003 and p < 0.0001, respectively). As previously described, there was a significant decrease in survival in animals with both genetic defects with osteosarcomas or mammary tumors alone compared to the survival observed for p53 ^{+ / −}/Brca2 ^{+ / +}or p53 ^{+ / −}/Brca2 ^{+ / −}mice with this single tumor type in the nonirradiated group. (Limited nonirradiated p53 ^{+ / +}/Brca2 ^{− / −} mice displayed these tumor types alone).

Discussion

This study provides in vivo evidence that mutations in both p53 and Brca2 cooperatively enhance deleterious effects on cancer susceptibility, cancer latency, and overall survival compared to mutations in either gene alone. We found that homozygous loss of Brca2 alone causes a specific cancer spectrum, decreases tumor latency, and increases tumor multiplicity and consistently replicates some of the tumor spectrums observed for FA and BRCA2 carriers. Gastric squamous cell carcinomas and hyperplasias and nongastric squamous cell carcinomas are primarily associated with the homozygous loss of Brca2, and a decreased latency time for mammary cancers, lymphomas, and osteosarcomas is observed in the Brca2 ^{− / −} animals. The homozygous loss of Brca2 also significantly decreases overall survival which is consistent with that of Donoho et al. (2003). As expected based on previous studies, mice that are heterozygous for a p53 mutation also have decreased survival and have significantly higher incidences of lymphomas and osteosarcomas.

Several specific results from this study support our hypothesis that the loss or reduction of both gene products may cooperate in both tumorigenesis and survival reduction: (1) When both genetic defects are combined in nonirradiated animals, survival is reduced to a greater extent than with either genetic defect alone. (However, we were unable to contribute to a further understanding of the individual contributions of the p53 heterozygous mutation compared to Brca2 homozygous mutation because there was no statistically significant difference in the survival curves for this comparison alone in the nonirradiated animals); (2) The presence of both mutations significantly increases the incidence of gastric tumors and osteosarcomas compared to the effect with either gene mutation alone; (3) The combination of mutations also decreases the average age of death for animals presenting with some individual tumor types alone compared to a single genetic defect in Brca2 or p53 having this single tumor type, suggesting that inactivation of these 2 genes can decrease the latency time of tumors in some tissues as well.

Our hypothesis was supported in predicting that several tissue or tumor-specific phenotypes observed prominently in human BRCA2 or FA mutation carriers are recapitulated in this murine knockout model, including increased carcinomas of many kinds (Ford et al., 1998; BCLC, 1999; Johannsson et al., 1999; Figuer et al., 2001; Jakubowska et al., 2002). Gastric tumors (predominantly squamous cell carcinomas) specifically occurred almost exclusively in the Brca2 ^{− / −} animals, and recent studies show an overrepresentation of gastric cancer in BRCA2-linked families (Johannsson et al., 1999; Figuer et al., 2001; Jakubowska et al., 2002). A C-terminal mutation of the human BRCA2 gene comparable to the C-terminal deletion in our Brca2 ^{− / −} mice has recently been found to be causative for a form of Fanconi Anemia (Howlett et al., 2002). Patients with Fanconi Anemia develop squamous cell carcinomas and epithelial tumors that are very similar to those observed in these Brca2 ^{− / −} mice and in BRCA2-linked families (Howlett et al., 2002; D’Andrea, 2003). We also find Brca2 deficiency associated with a significantly decreased latency for mammary tumors that is of interest given the role of BRCA2 in hereditary breast cancer.

The consequences of radiation exposure in animal models carrying defective Brca2 and/or p53 alleles is important to evaluate given the concern of the possible additional sensitivity to radiation with women who inherit mutations in the BRCA2 or p53 genes. The combined survival effect of a single high dose of radiation exposure with either the homozygous loss of Brca2 or the heterozygous loss of p53 was significantly greater than with either the radiation alone or the Brca2 or p53 genetic defect alone, suggesting a possible interactive effect between radiation and these individual genetic losses. This interactive effect observed between the Brca2 ^{− / −} mutation and 5 Gy irradiation for survival in this study suggests that this gene, along with p53 , may play a role in protection from radiation-induced damage. Protection from radiation by Brca2 has been studied and appears to be confined to actively proliferating cells (Tutt et al., 2003). This may explain some of the tissue-specificity of tumor development in the homozygous Brca2 irradiated animals. It is interesting that gastric squamous cell carcinomas are one of the cell types that is significantly increased in Brca2 ^{− / −} mice compared to Brca2 ^{+ / −} or Brca2 ^{+ / +} mice with the additional insult of radiation since this tumor originates from a highly proliferating cell type. Surprisingly, many tumor incidences in irradiated Brca2 homozygous mutants, irradiated p53 mutants, or irradiated double mutants were not significantly increased compared to the corresponding nonirradiated group in this study. This may be because the loss of p53 or Brca2 substantially impacts the overall survival and incidence of tumors in the nonirradiated groups such that the additional insult of radiation did not make a significant difference. Alternatively, some tumor types such as carcinomas may not have had time to develop in irradiated mice with one or both genetic mutations due to the significantly decreased overall survival of these mice and the possible earlier onset of other rapidly developing tumor types, such as lymphomas.

Our hypothesis of a cooperative effect of the combined genetic defects with radiation for tumor development or survival loss was not supported. The incidences of particular tumors in animals with both genetic mutations were not significantly greater than the incidences observed for these tumor types with either mutation alone in the irradiated groups. We did not observe an additional effect on survival with respect to the Brca2 mutation in animals with the combined p53 mutation and irradiation, perhaps due to the substantially reduced survival of all mice in this group. We also observed a statistically greater impact on survival for irradiated p53 ^{+ / −}/Brca2 ^{+ / +} compared to irradiated p53 ^{+ / +}/Brca2 ^{− / −} animals. These results suggest that the p53 defect may play a more substantial role in survival reduction than the Brca2 defect when combined with radiation. Based on the substantial survival reduction of animals in most genotypic classes with the 5 Gy irradiation dose, we believe that a lower dose of radiation may better clarify the combined contributions of irradiation with p53 and Brca2 defects on tumor spectrum, latency, and survival. Future studies will be necessary to elucidate the role of irradiation-induced damage as an environmental risk factor for BRCA2 or p53 mutant carriers.

Previous studies have shown a cooperative effect of p53 deficiency and other genetic defects on the timing and tissue-specificity of tumor formation (Williams et al., 1994; Donehower et al., 1995; Attardi and Jacks, 1999; Xu et al., 2001; Brodie et al., 2001; Blackburn and Jerry, 2002; Freie et al., 2003). However, the combined effect of germline p53 and Brca2 mutations in a mouse model was not previously evaluated due to the limited viability of most germline Brca2 knockouts (Connor et al., 1997; Friedman et al., 1998; Ludwig et al., 2001; McAllister et al., 2002). Conditional Brca2 mutant mice with a down-regulation in p53 expression have exhibited accelerated mammary tumorigenesis (Cheung et al., 2004). Jonkers et al.(2001) was also able to use a conditional Brca2 mutation combined with a p53 mutation to illustrate a synergy in the latency of mammary tumors. In that analysis, Brca2 conditional knockouts developed no epithelial tumors by 900 days. However, the combined conditional disruption of p53 and Brca2 resulted in all mice developing skin or mammary carcinomas by 300 days of age. Our studies similarly showed no mammary tumor development in p53 ^{+ / +}/Brca2 ^{− / −} mice and a significantly decreased latency for mammary tumor onset in animals with both a single p53 and homozygous Brca2 mutation compared to p53 ^{+ / −}/Brca2 ^{+ / +} mice. The fact that mammary tumor development was observed primarily in the p53 heterozygous animals alone in this study confirms previous findings of the importance of p53 loss for substantial mammary tumor formation (Blackburn and Jerry, 2002).

The segregation of Balb/c and C57Bl/6 alleles in the F2 intercross utilized in this study may account for some of the variation we observe in survival, tumor spectrum, and response to radiation both between and within genotypic groups. A substantial BALB/c and C57BL/6 strain difference in susceptibility for both spontaneous and radiation-induced tumorigenesis in mice has been established (Ullrich et al., 1996; Ponnaiya et al., 1997). Tumor samples from this study are therefore ideal for use in future studies to identify modifier loci responsible for strain-specific differences in spontaneous and radiation-induced tumor susceptibilities.

In addition, animals with this p53 heterozygous mutation display different tumor spectrums and susceptibilities depending on genetic background. Animals with this mutation on a C57BL/6 background develop primarily sarcomas (particularly osteosarcomas) and lymphomas (similar to Li-Fraumeni patients who inherit 1 defective p53 allele) but no mammary tumors (Jacks et al., 1994; Ponnaiya et al., 1997; Kuperwasser et al., 2000). Mice with this mutation on a pure BALB/c background develop predominantly mammary tumors (Jacks et al., 1994; Ponnaiya et al., 1997; Kuperwasser et al., 2000). Interestingly, in this study, nonirradiated p53 heterozygous animals (wild-type for Brca2) are approximately 50% C57BL/6 and 50% BALB/c and display some intermediate tumor spectrums with a 24% incidence of mammary tumors and a 23% incidence of sarcomas.

This in vivo study provides further support and is consistent with previous studies that suggest the existence of cooperation of Brca2 and p53 mutations in promoting tumorigenesis in specific tissues. Results from other animal models suggest that p53 might alter tumor specificity, timing, and progression by different mechanisms in different cell types when combined with another genetic alteration (Symonds et al., 1994; Jones et al., 1997; Blackburn and Jerry, 2002). In addition, there is evidence that inactivating mutations in other mitotic checkpoint genes besides p53 likely cooperate with a deficiency in BRCA2 in the progression of distinct tissue-specific tumors (Lee et al., 1999). The examination of allelic loss in tumors that are heterozygous for the p53 and/or Brca2 mutation might therefore allow us to further characterize tumor-specific molecular pathways in future studies.

This study further links the FA and BRCA2 pathway more clearly with epithelial cancer progression as observed in human FA as well as in these Brca2-deficient mice and the families with a comparable C-terminal deletion of the BRCA2 gene (McAllister et al., 2002; Oksuzoglu and Yalcin, 2002; D’Andrea, 2003; Houghtaling et al., 2003; Hu et al., 2003). A recently described mouse model with a targeted deletion in one of the downstream Fanconi Anemia genes (Fancd2) was found to have a similar predisposition for epithelial tumors (D’Andrea, 2003; Houghtaling et al., 2003). The similar phenotypes between these Brca2- and Fancd2-deficient mice suggest that Brca2 and Fancd2 function in the same biological pathway. Recently, these Brca2-deficient mice have been evaluated and proposed as a valuable model for Fanconi Anemia and the development of new therapies for this genetic disease (Navarro et al., 2005). Further research with this model may lead to a better understanding of human epithelial cancer progression.