Xrcc2 Modulates Spontaneous and Radiation-Induced Tumorigenesis in Apc^min/+ Mice

Double-strand breaks (DSB) are processed by one of two pathways: homologous recombination (HR) or nonhomologous end joining. The formation of chromosomal rearrangements caused by misrepair of ionizing radiation–induced DSBs is a hallmark of radiation damage (e.g., ref. 1). The RecA/RAD51 gene family plays an essential role in HR repair in all organisms, promoting chromosome stability and protecting against DNA DSBs, cross-links, and other forms of damage. Whereas RAD51 has the central role of facilitating exchange between homologous DNA strands, RAD51-like proteins such as XRCC2 have important but as yet poorly defined roles in HR. For example, the complete knockout of Xrcc2 in mice is lethal (2) and the accumulation of Rad51 at sites of DNA damage is compromised in the absence of Xrcc2 (3).

Importantly, it has been found that the human breast cancer susceptibility genes BRCA1 and BRCA2 have a role in the HR pathway. The homozygous loss of either gene leads to a severe reduction in HR repair of DNA damage and to the accumulation of chromosomal aberrations (4, 5). Germline mutations of BRCA1 and BRCA2 account for less than 2% of human breast cancer cases (6), and it is speculated that familial risk is associated with a large number of genetic components, each contributing a small effect (7). Given the association of the BRCA genes and HR repair, it is possible that other HR genes are candidates for this role in breast cancer.

XRCC2 has been shown to interact with other proteins involved in the DNA damage response, such as the Bloom's syndrome protein (8). When mice heterozygous for disruption of the Blm gene were crossed to mice heterozygous for a mutant form of the Apc tumor suppressor gene (important for the suppression of colorectal and other cancers), a 2-fold increase in the average number of spontaneous intestinal tumors was found (9).

Here, we aimed to investigate the effect of Xrcc2 deficiency on cancer susceptibility by introducing Xrcc2 heterozygosity onto the cancer-susceptible Apc^min/+ background (10). Female Apc^min/+ mice also have a susceptibility to the formation of mammary tumors (11). The Apc^min/+ mouse model has proved valuable for the verification of genes as cancer risk modifiers (e.g., refs. 9, 12, 13). The main focus of this study was to examine the incidence of cancers, particularly breast cancer, in mice with one or two copies of Xrcc2, with and without 2-Gy X-irradiation.

A 129/Sv heterozygous Xrcc2 knockout mouse strain has been described previously (2). Xrcc2 deficiency was bred onto a C57BL/6J background for five successive backcross generations. Xrcc2^+/− mice were subsequently crossed with Apc^min(Mom3R)/+ (Mom3, modifier of min 3; R, resistant) mice (14), which have a variant C57BL/6 background. The F1 generation yielded four genotypes: Apc^min/+ Xrcc2^+/−, Apc^+/+ Xrcc2^+/−, Apc^min/+ Xrcc2^+/+, and Apc^+/+ Xrcc2^+/+. Mice were housed in conventional cages, with a standard maintenance diet and water provided ad libitum. All procedures involving animals were carried out in accordance with the Animals (Scientific Procedures) Act 1986 and with guidance from the local Ethical Review Committee on animal experiments.

Genotyping of mice to detect both mutant and wild-type alleles was carried out as previously described for Apc (15) and Xrcc2 (16). Groups of at least 25 female and 10 male mice per genotype were irradiated with a single whole-body dose of 2-Gy X-rays at 35 days of age. Concurrent sham-irradiated control groups were also examined from the same litter where possible to minimize genetic bias. Irradiations were done at the Medical Research Council (Harwell, Oxon, United Kingdom) using a Siemens Stabiliplan X-ray machine (Siemens AG) operating at 250-kV constant potential and 14 mA (half value layer, 1.23 mm and 3.65 mm Cu) at a dose rate equal to 0.48 Gy min⁻¹.

Mammary tumors and gastrointestinal tracts were routinely sampled from all mice. These were fixed in 10% buffered formalin for 24 hours and subsequently stored in 70% ethanol. Other tissues were sampled from mice where abnormalities presented. Preparation of intestinal tracts and quantification of adenomas were previously described (14).

PCR was performed using 1 μL of DNA prepared (17) from approximately 10 to 20 mg of tissue dissected from fixed mammary tumors or single intestinal adenomas, approximately 0.5 to 1 mm in diameter, dissected from the third segment of the small intestine (ileum). All PCR products were run on 3% agarose gels and some were labeled with fluorescent dyes and analyzed using a GeXP genetic analyzer (Beckman Coulter). A loss of heterozygosity (LOH) screen was done using a marker for Xrcc2 (16) and an RFLP marker for Apc (15). Four distal chromosome 18 markers, polymorphic between C57BL/6J and the Apc^min(Mom3R)/+ strain, were also used. Three of these were located within genes known to have a role in tumorigenesis: D18mit207, D18mit33 (Tcf4), D18mit186 (Dcc), and a sequence-tagged site marker for Smad4 (F 5′-tcacggatgtcttggatctgg, R 5′-tgagggctgttattcctggg). Primers and WellRED fluorescently labeled primers were obtained from Sigma-Aldrich Company Ltd.

Fluorescently labeled PCR products were analyzed using peak height ratios to determine allelic imbalance, LOH, or homozygous loss in tumor tissue. Contamination with normal tissue is difficult to eliminate in solid tumors, and therefore, a product of less than 20% of the signal seen for normal tissue was considered to represent allelic loss. Allelic imbalance was scored when the sample peak was less than 80% of normal and homozygous loss was scored when the sample peak height was less than 20% of a control marker in the same sample. Verification of homozygous loss was done in two ways: by either multiplexing two markers and running products on agarose gels or combining fluorescently labeled sample and control marker products for a single tumor (pool-plex) and analyzing as a single sample by capillary electrophoresis compared with normal genomic DNA.

The difference between the percentage mammary tumor incidence for sham-irradiated Apc^min/+ Xrcc2^+/− and Apc^min/+ Xrcc2^+/+ mice was determined using the χ² test. The sham-irradiated mammary tumor incidence was subtracted from the incidence following 2-Gy irradiation in Apc^min/+ Xrcc2^+/− and Apc^min/+ Xrcc2^+/+ mice, yielding the excess incidence due to irradiation. The binomial distribution of mammary tumor data was used to obtain SEs. The difference between the two Xrcc2 genotypes was calculated using Fisher's exact test. A one-way ANOVA was also done on the percentage incidence of mammary tumors in the sham and irradiated groups of Apc^min/+ mice for the two Xrcc2 genotypes using Minitab 15 software (Minitab, Inc.).

Differences between mammary and intestinal tumor multiplicities and fold changes in Apc^min/+ and Apc^+/+ mice, for both Xrcc2 genotypes, were calculated using Minitab 15 software (Minitab).

The average life span of 2 Gy–irradiated Apc^min/+ mice was 116.5 days and that of sham-irradiated controls was 144.5 days. Apc^+/+ mice were postmortem examined at 285 days old, which was in excess of the maximum life span of Apc^min/+ mice of 196 days.

Mammary tumors presenting in these mice were classified as adenocarcinoma with squamous cell differentiation. For female sham-irradiated Apc^min/+ mice, the incidence of mammary tumors was lower in Apc^min/+ Xrcc2^+/− mice (14%; n = 28) than in Apc^min/+ Xrcc2^+/+ mice (35%; n = 26; Table 1). This difference only approached significance due to relatively small numbers of mice and, therefore, small statistical power (χ², P = 0.08). The addition of 13 Apc^min/+ Xrcc2^+/+ mice from a parallel experiment increased the statistical power (n = 39; difference between the two Xrcc2^+/+ groups indistinguishable; χ² test, P = 0.739) and showed a significantly reduced incidence of mammary tumors in the Xrcc2^+/− mice (14% versus 38%; χ² test, P = 0.03). The sham-irradiated mammary tumor incidence was subtracted from the incidence following 2-Gy X-irradiation, giving a background corrected incidence for X-irradiation–induced tumors in the Apc^min/+ Xrcc2^+/− group of 57 ± 11% compared with 17 ± 13% in Apc^min/+ Xrcc2^+/+ mice (excluding additional mice described above) and this difference was significant (Fisher's exact test, P = 0.001; data not shown). Inclusion of the larger Apc^min/+ Xrcc2^+/+ group gave a greater significance. The increase in incidence between the sham and irradiated groups of Apc^min/+ Xrcc2^+/+ mice was 1.50-fold, which was not significantly different, but the increase following X-irradiation in Apc^min/+ Xrcc2^+/− mice was 4.99-fold (ANOVA, P < 0.001; Table 1).

Both Apc^min/+ Xrcc2 genotypes spontaneously presented with single mammary tumors and multiplicity increased following X-irradiation. The difference in sham-irradiated multiplicity for the two Apc^min/+ Xrcc2 genotypes approached statistical significance at P = 0.08, and addition of 13 Apc^min/+ Xrcc2^+/+ mice, described above, gave a significant difference (t test, P = 0.05). For female Apc^min/+ Xrcc2^+/+ mice, the increase in multiplicity following 2-Gy X-ray exposure was significant, increasing by a factor of 2.57 (Table 1). The corresponding change in the Apc^min/+ Xrcc2^+/− groups was greater with an increase of 7.25-fold (Table 1). The fold increase in multiplicity following X-irradiation was significantly different between the two genotypes (t test, P < 0.001; Table 1).

No mammary tumors were found in sham-irradiated Apc^min/+ Xrcc2^+/+ male mice, and one presented in an Apc^min/+ Xrcc2^+/− male mouse. For X-irradiated mice, the majority of mammary tumors were seen in females; only one Apc^min/+ Xrcc2^+/+ male mouse and two Apc^min/+ Xrcc2^+/− males presented with a mammary tumor (data not shown). No Apc^+/+ mice presented with mammary tumors.

Following sham irradiation, the mean intestinal adenoma multiplicity was 105.8 for Apc^min/+ Xrcc2^+/− mice and 116.9 for Apc^min/+ Xrcc2^+/+ mice and this difference was not statistically significant. Some mice were killed prematurely due to other causes, such as development of a mammary tumor, and omission of these mice gave significantly lower intestinal adenoma multiplicities in Apc^min/+ Xrcc2^+/− mice than in Apc^min/+ Xrcc2^+/+ mice (t test, P = 0.005; Table 2). For mice given 2-Gy X-rays, there was a large variation in the intestinal adenoma multiplicity and the differences between Apc^min/+ Xrcc2^+/− and Apc^min/+ Xrcc2^+/+ mice were not statistically significant. However, the increase in adenoma multiplicity following X-irradiation was 4.14 ± 0.05-fold in Apc^min/+ Xrcc2^+/− mice and 3.30 ± 0.05-fold in Apc^min/+ Xrcc2^+/+ mice and this difference was statistically significant (t-test P < 0.001; Table 2).

Three irradiated Apc^+/+ Xrcc2^+/− female mice were killed at or before 285 days (152-250 days) of age due to similar signs of ill health. A full pathologic analysis on one of these mice confirmed a diagnosis of systemic lymphoma. Two irradiated Apc^+/+ Xrcc2^+/− female mice presented with ovarian tumors at postmortem as did one Apc^+/+ Xrcc2^+/+ mouse. Another irradiated female Apc^+/+ Xrcc2^+/+ mouse presented with a liver tumor. Two sham-irradiated Apc^+/+ Xrcc2^+/+ mice were postmortem examined before 285 days but no neoplastic changes were seen.

To assess the effect of Xrcc2 deficiency on mutational mechanisms operating in tumorigenesis, LOH analysis was done on mammary and intestinal tumors from Apc^min/+ mice. Results for mammary tumors were complex, with some tumors in sham- and X-irradiated mice for both Apc^min/+ Xrcc2 genotypes showing homozygous losses, and these, together with allelic imbalances, were more common events than LOH (Fig. 1A). For intestinal tumors, patterns of LOH were less complex and allelic imbalance was only seen in sham-irradiated tumors (Fig. 1B).

LOH of the Xrcc2 marker was seen in 25% of sham-irradiated mammary tumors from Apc^min/+ Xrcc2^+/− mice, and 23% of X-irradiated mammary tumors showed LOH or homozygous loss (Fig. 1A). Xrcc2 was retained in all intestinal tumors from Apc^min/+ Xrcc2^+/− mice (Fig. 1B).

Homozygous loss of Apc was identified in 14% of mammary tumors from X-irradiated Apc^min/+ Xrcc2^+/− mice that had also lost one or both copies of Xrcc2. Apc was retained in all other mammary tumors (Fig. 1A). LOH of Apc in intestinal tumors was similar for both Apc^min/+ Xrcc2 genotypes (Fig. 1B).

The marker for Dcc (D18mit186) was the most commonly lost chromosome 18 marker in mammary and intestinal tumors (Fig. 1A and B). Loss of all chromosome 18 markers was seen in approximately 23% of intestinal tumors from sham- and X-irradiated Apc^min/+ Xrcc2^+/− mice, in only one intestinal tumor (4%) from sham-irradiated Apc^min/+ Xrcc2^+/+ mice (Fig. 1B), and in none of the mammary tumors screened (Fig. 1A).

We have noted in Introduction that the breast cancer susceptibility genes BRCA1 and BRCA2 are involved in the HR damage response pathway, which led to the anticipation that other HR genes might play a role in the suppression of certain types of cancer. Very recently, analyses of germline mutations of RAD51C in breast and ovarian cancer pedigrees have established this HR repair gene as a cancer susceptibility gene (18). A marginally significant association of breast cancer susceptibility has also been found with a rare variant of XRCC2 (19, 20), as well as a potential link to ionizing radiation exposure (21). Earlier studies using Brca1 heterozygous knockout mice showed small increases in incidence of mammary tumors when Trp53 was also compromised, and this was further increased by irradiation (22). Similarly, we found that Apc^min/+ Xrcc2^+/− mice showed an increased incidence of mammary tumors after irradiation when compared with Apc^min/+ Xrcc2^+/+ mice, although the spontaneous frequency in the two genotypes also varied (see below). The main type of mammary tumor found in the Apc^min/+ Xrcc2^+/− mice was adenocarcinoma with squamous cell differentiation; similarly, this type of tumor was found in Brca1- or Brca2-compromised mice (23, 24). These data suggest that Xrcc2 may act as a tumor suppressor following radiation damage, but in parallel to studies of mouse mutants of the Brca genes (25, 26), further development of conditional and tissue-specific mouse Xrcc2 mutants would be required to support this finding.

The reduction in spontaneous incidence of mammary cancers in the Apc^min/+ Xrcc2^+/− mice relative to the Apc^min/+ Xrcc2^+/+ mice, which was not significantly different unless supplementary mice were included in the analysis, was surprising given the role of Xrcc2 in the repair of DNA damage and the increase in spontaneous adenomas in Apc^min/+ mice carrying the Blm mutation (9). However, haploinsufficiency of Xrcc2 showed no life shortening in Xrcc2^+/− p53^−/− mice compared with Xrcc2^+/+ p53^−/− littermates, indicating that the tumor incidence was similar (27). Although chance or genetic background may have played a role in our result, the small reduction in the spontaneous incidence of intestinal adenomas is consistent with it (Table 2). Why should the spontaneous incidence of cancer reduce while the X-ray–induced incidence increases with Xrcc2 hemizygosity? It seems unlikely that this difference is simply due to an overall reduction in mitotic recombination (including interhomologue recombination potentially leading to LOH) because this should similarly affect both spontaneous and X-ray–induced events. However, it is likely that the misrepair of a subset of spontaneous DNA damage contributes to cancer formation, and this damage type may differ from the damage induced by radiation. Obviously, the level of DNA damage will be much higher in irradiated than in unirradiated mice, and whereas it is expected that damage repair in general will be reduced in mice with compromised HR repair functions, little is known about the misrepair of different damage types in these circumstances. It is therefore possible that in the Apc^min/+ Xrcc2^+/− mice, a decreased incidence of misrepair of a subset of spontaneous damage may occur while increased misrepair of radiation damage arises, due partly to the utilization of the nonhomologous end joining pathway.

The data for mammary and intestinal tumor multiplicity in Apc^min/+ Xrcc2^+/− mice give an indication that Xrcc2 deficiency may differentially modify mechanisms of tumor initiation and/or progression following X-irradiation in different tissues. The fold increase in mammary tumor multiplicity following X-irradiation was significantly greater than the fold increase in intestinal tumor multiplicity for the same genotype (t test, P < 0.001). This suggests that HR has a greater role in the repair of X-ray–induced DSB in mammary than in intestinal tissue. Tissue-, region-, or age-specific effects following ionizing irradiation have been shown for different loci in mouse models previously (28–31).

In addition to tissue differences in tumor multiplicity determined by Xrcc2 deficiency, the LOH data suggest differences in the molecular mechanisms of tumorigenesis between tissue types. LOH analysis identified loss of the functional copy of Xrcc2 in mammary tumors, but not in intestinal tumors, from Apc^min/+ Xrcc2^+/− mice, indicating that the role of Xrcc2 may differ in the two tissue types. Furthermore, homozygous loss of Apc was only seen in mammary tumors from X-irradiated Apc^min/+ Xrcc2^+/− mice that had also lost Xrcc2 function, suggesting that lack of Xrcc2 function influences efficiency of DNA repair on chromosome 18 following X-irradiation in mammary tissue. Intestinal tumors, however, showed a higher frequency of loss of all chromosome 18 markers in Apc^min/+ Xrcc2^+/− compared with Apc^min/+ Xrcc2^+/+ mice, indicating that Xrcc2 heterozygosity leads to a difference in HR-mediated LOH or possible nondisjunction (32), and further suggesting that there are tissue preferences for the molecular mechanisms that operate in Xrcc2-deficient mice.

To summarize, Xrcc2 hemizygosity on an Apc^min/+ background has a spontaneous tumor risk–modifying effect, reducing mammary and intestinal tumorigenesis and increasing radiation risk compared with the Apc^min/+ Xrcc2^+/+ genotype. Additionally, Xrcc2 hemizygosity seems to show some tissue preference–modifying effect, shown by a significantly greater increase in radiation-induced mammary tumorigenesis.

No potential conflicts of interest were disclosed.

We thank Kevin Whitehill for animal husbandry and Dr. Rosie Finnon for assistance with GM regulations and risk assessment.

Grant Support: EC RISC RAD contract FI6R-CT-2003-508842.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.