Disruption of Brca2 increases the spontaneous mutation rate in vivo: synergism with ionizing radiation

We have described previously a mouse model to examine the in vivo phenotype of a truncating mutation (Brca2^Tr2014, hereafter referred to as Brca2^Tr; Connor et al., 1997), which mimics cancer‐predisposing mutations in humans. This mutation truncates the protein just before the final BRC repeat in exon 11 at a similar site to the common Ashkenazi 6174delT BRCA2 truncation. Mice homozygous for the Brca2^Tr mutation are prone to tumorigenesis. We used in vitro cell culture systems to show an increase in error‐prone DNA DSB repair after disruption of Brca2 (Tutt et al., 2001). To establish whether this defect seen in vitro is associated with increased somatic mutation in vivo, we have crossed mice heterozygous for the Brca2^Tr mutation with a LacZ‐plasmid transgenic mutation reporter mouse (Boerrigter et al., 1995; Figure 1). This system can detect spontaneous and IR‐induced somatic mutations including the large genomic deletions, insertions or translocations (Gossen et al., 1995) predicted to occur from DSB misrepair using non‐homologous end‐joining (NHEJ) or SSA (Jasin, 2000). We generated Brca2^Tr/Wt mice that were additionally homozygous for the pUR288 transgene mutation reporter concatamer at each of two chromosomal loci, then intercrossed these mice to examine the effect of the truncating mutation in Brca2 on spontaneous somatic mutation frequency using the LacZ reporter genes (Figure 1).

Most Brca2^Tr/Tr mice die at birth (Connor et al., 1997), so to generate sufficient samples for analysis we examined the mutation frequency in littermate Brca2^Tr/Wt, Brca2^Tr/Tr and wild‐type embryos at day 14.5 gestation (Figure 2A). We found that loss of both wild‐type Brca2 alleles causes a rapid and significant increase in spontaneous somatic mutation frequency at this embryonic stage: Brca2^Tr/Tr, 8.9 ± 0.4 × 10⁻⁵; Brca2^Wt/Wt, 3.9 ± 0.6 × 10⁻⁵ (P = 0.0018). This is equivalent to the somatic mutational load acquired by a 2‐year‐old wild‐type mouse (Dolle et al., 1997). Based on a LacZ reporter gene target size of 3 kb and a genome size of 6 × 10⁹ bp, this equates to an additional 100 mutations in every cell after only 2 weeks gestation. As unrepaired DNA DSBs within the reporter gene will not be recovered from genomic DNA by this assay, the observed increase in mutation frequency therefore represents completed DNA repair by an error‐prone mechanism. Heterozygosity for this Brca2 mutation has no effect on somatic mutation frequency: Brca2^Tr/Wt, 3.34 ± 0.3 × 10⁻⁵; Brca2^Wt/Wt, 3.9 ± 0.6 × 10⁻⁵ (P = 0.42). This suggests that at this embryonic stage, at least, there is no mutator phenotype associated with loss of one wild‐type copy of Brca2 and that loss of the second wild‐type allele drives the rapid acquisition of somatic mutation throughout the genome.

A high proportion of the infrequent spontaneous somatic mutations in wild‐type LacZ‐plasmid transgenic reporter mice arise from deletions and genomic rearrangements rather than point mutations (Dolle et al., 2000). This type of mutation can arise from DNA DSB misrepair events by error‐prone NHEJ and SSA repair pathways (Dolle et al., 2000). IR also increases this class of mutation (Gossen et al., 1995). We have shown that loss of wild‐type Brca2 results in increased use of error‐prone SSA to repair a chromosomal DSB in a cell culture system (Tutt et al., 2001). Therefore, we asked whether somatic mutations induced by loss of Brca2 in vivo are of the deletion/rearrangement type by analysing mutations present in the LacZ reporter plasmids recovered in the experiments described above (Figure 2A and B). Deletions, insertions and translocations, such as those induced by IR, cause restriction fragment size changes (Gossen et al., 1995), whereas point mutations, such as those induced by ethyl nitrosourea, do not (Dolle et al., 1996). We found no significant difference (P = 0.75) between the proportion of reporter plasmid mutants with detectable size changes recovered from Brca2^Tr/Tr embryos (76%; 51/67) compared with Brca2^Wt/Wt embryos (71%; 22/31) and Brca2^Wt/Tr embryos (78%; 43/55). We then used a PCR assay to distinguish size change mutations due to gross deletions/rearrangements involving flanking mouse genomic DNA from deletions internal to LacZ (Dolle et al., 2000). Interestingly, deletions/rearrangements involving the mouse genome are more common in Brca2^Tr/Tr recovered mutants (96%; 49/51) than in Brca2^Wt/Wt and Brca2^Wt/Tr control mutants (78%; 51/65) (P = 0.0065). The character of these mutations is compatible with use of error‐prone repair of DNA DSBs and shows that tissues lacking Brca2 acquire somatic mutation via this mechanism at a greater rate than in tissues with wild‐type Brca2 (Figure 2C).

IR induces, amongst other lesions, DNA DSBs and is recognized as being carcinogenic to breast tissue (Goss and Sierra, 1998). An increase in error‐prone repair of IR‐induced DNA DSBs due to abnormal function of BRCA2 would be predicted to increase radiation‐induced mutagenesis. This is of clinical interest, as carriers of BRCA2 are exposed to IR during mammography and adjuvant radiation therapy associated with breast conserving surgery for ductal carcinoma in situ and invasive breast cancer. To examine the effect of Brca2 mutation on IR‐induced somatic mutations, we compared littermate Brca2^Wt/Wt, Brca2^Wt/Tr and Brca2^Tr/Tr embryo sets irradiated with 4 Gy X‐rays 24 h prior to embryo harvest on day 14.5 gestation. We found that 4 Gy induces a significant increase over the spontaneous mutation frequency win ild‐type (10.6 ± 1 × 10⁻⁵ versus 3.9 ± 0.6 × 10⁻⁵; P = 0.0013), Brca2^Wt/Tr (10.6 ± 0.6 × 10⁻⁵ versus 3.34 ± 0.3 × 10⁻⁵; P <0.0001) and Brca2^Tr/Tr embryos (31.6 ± 2.9 × 10⁻⁵ versus 8.93 ± 0.4 × 10⁻⁵; P = 0.0004; Figure 3A). The induction of mutations (in absolute numbers) above the spontaneous background is therefore ∼3‐fold greater in Brca2^Tr/Tr embryos than in the wild type (22.7 ± 2.5 × 10⁻⁵ versus 6.7 ± 1 × 10⁻⁵; P <0.0001; Figure 3B). We estimate that loss of Brca2 in vivo results in an extra 320 somatic mutations in every cell as a result of misrepair of 4 Gy IR‐induced damage. We also asked whether any subtle mutator phenotype associated with haploinsufficiency due to mutation in one Brca2 allele, not revealed in untreated Brca2^Wt/Tr embryos, might become apparent under the stress of DNA damage induced by 4 Gy of IR. In fact, we found no increase in radiation‐induced mutagenesis in Brca2^Wt/Tr embryos compared with the wild type (7.2 ± 0.8 × 10⁻⁵ versus 6.7 ± 1 × 10⁻⁵, P = 0.68). This further supports the contention that mutation or loss of both copies of Brca2 is required before a mutator phenotype is acquired, after which Brca2‐deficient tissues then suffer a greater number of somatic mutations per unit dose of IR.

Characterization of the size of recovered mutant plasmids reveals that increased mutation induction by IR in Brca2^Tr/Tr embryos is predominantly via a deletion/rearrangement mechanism (86%; 32/37). We found that although the mutation spectrum is similar to that in wild‐type embryos (76%; 16/21) and Brca2^Wt/Tr embryos (74%; 32/43; P = 0.39), the frequency at which these somatic mutations occur is lower than in Brca2^Tr/Tr embryos (Figure 3C).

C57BL/6‐TgN(LacZpl)60vij mice were obtained from the Jackson Laboratory and were crossed with C57BL/10 Brca2^Wt/Tr2014 mice (Connor et al., 1997). These mice were then back‐crossed to the C57BL/6‐TgN(LacZpl)60vij mutation reporter strain and homozygosity for both pUR288 concatamer integration sites was confirmed by test breeding to wild type and genotyping with primer sets specific for the two pUR288 cloned integration sites. For generation of embryos, Brca2^Wt/Tr2014 C57BL/6‐TgN(LacZpl)60vij mice were intercrossed and the females killed on day E14.5 gestation measured from the ‘plug’ date.

For irradiation experiments, female mice received 4 Gy whole‐body irradiation on day E13.5 gestation from a Pantak 250 kV X‐ray set at 90 cm focus skin distance at a dose rate of 0.75 Gy/min. Embryos were harvested 24 h later. For all experiments, embryos were harvested on day E14.5, washed in phosphate‐buffered saline at 4°C and were snap frozen on dry ice and stored at −80°C. Genomic DNA was prepared by phenol–chloroform extraction.

Complete protocols for this system have been described elsewhere (Vijg, 1996). Briefly, between 10 and 20 μg of genomic DNA were digested with HindIII for 1 h in the presence of magnetic beads (Dynal) pre‐coated with lacI–lacZ fusion protein. The beads were washed three times to remove the unbound mouse genomic DNA. Plasmids were subsequently eluted from the beads with isopropyl‐D‐thiogalactoside. After circularization with T4 DNA ligase, plasmids were used to electrotransform Escherichia coli C (ΔLacZ, galE‐) cells by electroporation. Transformations were plated on titration plates containing X‐gal and LacZ mutant selection plates containing P‐gal. Mutant frequencies were determined as the ratio of the number of colonies on the selective plates versus titration plates. Each determination point is based on at least 100 000 recovered plasmids. The in vitro background mutant frequency of this system is ∼1 × 10⁻⁵ (Dolle et al., 1999). Identification of size‐change mutations was performed by digestion of rescued LacZ mutant pUR288 plasmids with AvaI and PstI. To screen for size‐change mutants with a breakpoint in the mouse genome, a PCR amplification was performed by using a forward primer specific for the last 18 bp of lacZ before the HindIII site (pUR3285‐F) in combination with a reverse primer after the HindIII site (pUR3421‐R; Dolle et al., 2000). This procedure results in a 137 bp product only when both breakpoints are in the lacZ gene. As a positive control, a primer set (pUR4071‐F and pUR4181‐R) specific for the Amp^r gene was included in the same reaction.

Mutation frequencies were compared using a two‐tailed unpaired Student's t‐test. Proportions of recovered reporter plasmids containing restriction fragment size changes or with mouse genome rearrangements were compared using two‐sided analysis of contingency tables by Fisher's exact test or the Chi squared test. All statistical comparisons were performed using Graphpad Prism software.