Common Nonsense Mutations in RAD52

Introduction

RAD51, RAD52, and RAD54 encode the human homologues of yeast genes that are required for homologous recombination, an essential step in the repair of double-strand DNA breaks (reviewed in Ref. 1 ). This process involves the pairing of sister chromosomes and nucleolytic processing of double-strand breaks to produce single strands, which then pair with the complementary strand from the sister chromosome, leading to accurate repair and recombination. In Saccharomyces cerevisiae, Rad54 facilitates the pairing of chromosomes (2) , whereas Rad52 stimulates Rad51-mediated pairing of complementary single-stranded DNA and strand exchange (3, 4, 5) . Yeast strains lacking any of these genes, defined as belonging to the Rad52 epistasis group, demonstrate profound defects in recombination and increased sensitivity to DNA damage. The vertebrate homologues of the yeast RAD51, RAD52, and RAD54 genes demonstrate similar functional properties. Homozygous deletion of RAD51 and RAD54 in the mouse leads to an early embryonic lethal phenotype, and cultured embryonic stem cells display increased sensitivity to ionizing radiation, a reduction in homologous recombination, and progressive chromosome loss and genomic instability (6, 7, 8) . In contrast, RAD52-null mice are viable (9) , and inactivation of RAD52 in chicken B cells leads to a reduction in targeted chromosomal integration, without a detectable increase in radiation sensitivity (10) . The attenuated phenotype of RAD52 deletion in vertebrates is in striking contrast to its essential role in S. cerevisiae but consistent with that of its Schizosaccharomyces pombe homologue, RAD22, pointing to potentially divergent functional pathways (11) . In vertebrates, failure to repair double-strand breaks by RAD52-mediated homologous recombination may be compensated in part by Ku-dependent nonhomologous recombination (12) .

The discovery that RAD51 associates directly with the product of the breast cancer predisposition gene BRCA2 (7) and indirectly with the BRCA1 product (13 , 14) has implicated these tumor suppressor genes in this recombinational pathway for the repair of DNA damage. In addition to their physical association and colocalization with RAD51 within nuclear dots, the products of the BRCA genes appear to play a role in the repair of DNA damage induced by ionizing radiation. BRCA1- and BRCA2-null embryos are arrested in early embryonic development, and the latter display hypersensitivity to γ-irradiation (6 , 7) . Ionizing radiation leads to the hyperphosphorylation and physical relocalization of BRCA1, and mouse embryo fibroblasts expressing attenuated alleles of either BRCA1 or BRCA2 accumulate chromosomal breaks, characteristic of deficiencies in recombination-mediated repair (14, 15, 16) . Taken together, these observations raise the possibility that the BRCA genes are components of a functional pathway defined by RAD51, RAD52, and RAD54 and that mutations in these genes may themselves contribute to the development of breast cancer. Of note, RAD52 maps to chromosome locus 12p12.2-p13, a frequent site for allelic losses in breast and ovarian cancer (17, 18, 19) .

Germ-line mutations in either BRCA1 or BRCA2 account for 60–80% of breast cancer kindreds, suggesting the existence of additional breast cancer predisposition genes (20 , 21) . Population-based cohorts with an early age of onset of breast cancer have been particularly helpful in screening for genetic variations that may confer an increased risk of developing breast cancer, including highly penetrant mutations that are represented in cancer pedigrees and sequence variations in the population that may be associated with a more moderate increase in breast cancer risk. Epidemiological models have predicted that 36% of early-onset breast cancer cases (diagnosed under the age of 30 years) result from a single-gene defect (22) , whereas BRCA1 and BRCA2 mutations are observed in only ∼15% of these cases (23, 24, 25) . Given the observation that BRCA1 and BRCA2 mutations are present in the germ-line of women with genetic predisposition to breast cancer but not in sporadic breast tumors, analysis of candidate genes for their potential role in the development of breast cancer requires analysis of both tumor and germ-line specimens. We, therefore, analyzed the transcripts for RAD51, RAD52, and RAD54 in EBV-immortalized lymphoblasts from 100 women who had been identified from Boston area hospitals as having developed breast cancer under the age of 40 years and from an equal number of healthy controls from the same population. Nine of the 100 early-onset cases also had a strong family history of breast cancer but did not have a BRCA1 or BRCA2 mutation and, therefore, may carry a germ-line mutation in another breast cancer susceptibility gene. To search for somatic mutations in these genes, we analyzed 15 breast cancer cell lines.

Materials and Methods

Study Population.

EBV-immortalized lymphocytes from 100 women who were diagnosed with breast cancer at or before the age of 40 years were screened for mutations in the RAD51, RAD52, and RAD54 genes. These samples were part of a larger cohort of 408 women with early-onset breast cancer that was identified by survey of Boston area hospitals, and they represent a subset of the general population at high risk for genetic predisposition (23) . Nine of the 100 early-onset cases studied also have a strong family history of breast cancer, defined as more than three affected family members, in either paternal or maternal lineage, over at least two generations. Control EBV cell lines were derived from healthy blood donors at these hospitals. Fifteen human breast cancer cell lines were obtained from the American Type Culture Collection (Manassas, VA): MDA-MB-435, MDA-MB-468, MDA-MB-415, MDA-MB-231, MDA-MB-453, MDA-MB-175-VI, MDA-MB-436, MDA-MB-157, MCF-7, MCF-7/ADR, UACC-893, T-47D, BT-549, BT-483, and HS578.

nt 5 Sequencing Analysis.

Nested PCR was performed on a RAD51 primary PCR product using five sets of primers to generate overlapping PCR products for sequence analysis. To facilitate sequencing, we added the M13-21 primer sequence (TGTAAAACGACGGCCAGT) to the 5′ end of the forward primer, and the M13-28 primer sequence (AGGAAACAGCTATGACCAT) was added to the 5′ end of the reverse primer. Primers pairs used were: R51-ET-1F (AGTAGAGAAGTGGAGCGTAAGC) and R51-ET-1R (CCATCCATTGGAACTAATTTAGCTGCC); R51-ET-2F (TGCCTATGCGCCAAAGAAGGAGC) and R51-ET-2R (AGCAGCCGTTCTGGCCTAAAGG); R51-ET-3F (TCACCTGCCAGCTTCCCATTG) and R51-ET-3R (ACTGCTACACCAAACTCATCAGCG); R51-ET-4F (GCCCTTTACAGAACAGACTACTCG) and R51-ET-4R (CAATCTGGTTGTTGATGCATGG); and R51-ET-5Fa (AGTGGTAATCACTAATCAGGTGG) and R51-ET-5Ra (TTATCCTGGCAGTCACAACAGG). The coding region of RAD52 was amplified using the following primer sets: R52-ET-1F (ATCTGCCCATCTTGTTACTCC) and R52-ET-1R (GATTAATTACCCGATGACCCTC); R52-ET-2F (ATCCAGAAGGCCCTGAGGCAG) and R52-ET-2Ra (TGATATGAACCATCCTTCAGCTG); R52-ET-3Fa (GCGAGCCCTCAGGAGTTTTGG) and R52-ET-3R (GAGCTCAGGCTTCGGGAGCTGC); R52-ET-3F (CCAGCTGAAGGATGGTTCATATC) and R52-ET-2Rb (CCAAAACTCCTGAGGGCTCG); R52-ET-4F (GCAGCTCCCGAAGCCTGAGCTC) and R52-ET-4Ra (GTCTTCAAGAGTCTTGATTAATTC); R52-ET-5F (TGGAGAAAGACTTCCTTGCAGG) and R52-ET-5R (CAGTTTCCTGTTGTGCGTTGG); and R52-ET-6F (AGGACTCCACACAGCGTTTGCC) and R52-ET-6R (AAGTCCCTTTGTGACAGAGTCC). RAD54 cDNAs were sequenced in nine overlapping fragments generated using the following primer pair combinations: R54-ET-1F (CTTTGGCTCATGGGTACTTGACG) and R54-ET-1R (CAATGCTCGAGAGCCCAGAGGACC); R54-ET-2F (CTGGACAGCAGTCAGCATGAAGC) and R54-ET-2R (AGTGTCCACATCAATGTGATGC); R54-ET-3F (ATCCCTGGCAGCCATGGCTGCATC) and R54-ET-3R (GACTCCAACATGAAGGCGGAAGG); R54-ET-4F (CTGGAAGGATTCATGAACCAGCG) and R54-ET-4R (GCC TCA CTA GCA GCA GCG TCT CG); R54-ET-5F (CGGCATCCTAGGGACTGCCCATG) and R54-ET-5R (CACTTATCATAGATTAGAGCTGG); R54-ET-6F (GCAAGATGAGTGTGTCTTCCCTTTC) and R54-ET-6R (CAGGGCTCGATGGACTATTGAAGC); R54-ET-7F (CTTATACGTCCGCCTGGATGGCACG) and R54-ET-7R (AGCTCTGACGCTGGAAGATCTTC); R54-ET-8F (GAAGACTTGCTATATCTACCGC) and R54-ET-8R(CTGAAGTGCAGTCAGAACCATCAG); and R54-ET-9F (GACACACATGACAGGTTGC) and R54-ET-9R (CAGGGCCCTGTGGAGCAAGGAGC). For all primer sets, PCR was performed using the following conditions for amplification: 95°C for 1 min, followed by 40 cycles of 95°C for 30 s, 62°C for 30 s, and 72°C for 1 min.

PCR products were resolved by gel electrophoresis and treated with exonuclease I (Amersham Life Sciences, Cleveland, OH) and shrimp alkaline phosphatase (United States Biochemical, Cleveland OH) and diluted 1:6 prior to sequencing. Sequencing was performed using Energy Transfer Dye Primer sequencing (Amersham) according to the manufacturer’s instructions. Factura and Sequence Navigator (Perkin-Elmer Applied Biosystems, Foster City, CA) were used to mark potential heterozygous positions and display them for evaluation. Base positions at which the height of the secondary peak was ≥30% that of the primary peak were marked as heterozygous and were confirmed by analysis of both sense and antisense strands. Confirmation that the RAD52 Ser346ter mutation was present at the genomic DNA level required determination of the flanking intronic sequence, to avoid amplification of the known RAD52 pseudogene. Intronic sequence was PCR-amplified from genomic DNA and used to design a forward primer within the 5′ intronic sequence (CACTTTTCCTTGACCTCTGTAG), which was used in combination with a reverse primer (GTCACCATCTGGTTGTTCAAGG) within the RAD52 coding region to amplify codon 346 from genomic DNA. Both forward and reverse primers were compound primers that included the M13-21 and M13-28 5′ tails, respectively. PCR conditions were: 40 cycles of 95°C for 15 s, 62°C for 30 s, and 72°C for 1 min. PCR products were sequenced using dye primer chemistry, as described previously in “Materials and Methods.”

Yeast-based Protein Truncation Assay.

The coding regions of RAD51, RAD52, and RAD54 were analyzed for the presence of stop codons using a modification (26) of the SC yeast-based protein truncation assay (27) . Total cellular mRNA was isolated using STAT-60 (Tel-Test, Inc., Friendswood, TX), according to the manufacturer’s instructions. The entire coding regions of these genes were amplified by RT-PCR as follows: with primers RAD51F (CATCTTGGGTTGTGCGCAGAAGGC) and RAD51R (TTCCCGGAAGCTTTATCCTGGC) using 35 cycles of 95°C for 10 s, 60°C for 15 s, and 72°C for 3 min); with primers RAD52F (GAT CTG CCC ATC TTG TTA CTC C) and RAD52R (AAA GTC CCT TTG TGA CAG AGT C) using 35 cycles of 95°C for 30 s, 65°C for 30 s, and 72°C for 3 min; and with primers RAD54F (TCA ACA GGG TGT GGA GCA AGG) and RAD54R (CCG CTG CCT GCT TTT GAC CTT TGG) using 25 cycles of 95°C for 30 s, 56°C for 30 s, and 72°C for 3 min.

Products of initial PCRs were diluted 100-fold and subjected to nested PCR. The primers used in the nested PCR were complementary to the RAD cDNA, with a 5′ tail of 45 bp complementary to the sequence flanking an engineered BamHI site within the yeast URA3 gene (forward primer tail, TTCCTGACTATGCGGGCTATCCCTATGACGTCCCGGACTATGCAG; and reverse primer tail, GCAGCAACAGGACTAGGATGAGTAGCAGCACGTTCCTTATATGTG). Primers were designed to preserve the ORF of both URA3 and RAD51, RAD52, or RAD54 following insertion of these sequences by homologous recombination. PCR conditions for RAD51, which was analyzed in two overlapping fragments, were 35 cycles of 95°C for 10 s, 60°C for 15 s, and 72°C for 3 min, and primers were: fragment A, RAD51AF (GAGTAATGGCAATGCAGATGC) and RAD51AR (AGCAGCCGTTCTGGCCTAAAG); and fragment B, RAD51BF (ACATTGACACTGAGGGTACCT) and RAD51BR (CAGTCTTTGGCATCTCCCACT). RAD52 was analyzed in four overlapping fragments A2 (forward primer, GCC TGA GCT CAT CCGCCGTGGAG; and reverse primer, AATGTGTCAGAGGTCTGGGCTGGGTC), A3 (forward primer, AGCGAGCCCTCAGGAGTTTTGGGAATG; and reverse primer, GAGCTGCAGTCCTGGTCCGCCG), A6 (forward primer, CAACCAGAATCAAGATGTCTGGGACTG; and reverse primer, GCTCGCTTCAGCCCGTCTGTCACC), and B (forward primer, GCACTCCTGTAACTGTCTCAG; and reverse primer, TAAGATGGATCATATTTCCTTTTC). PCR conditions were: for fragment A2, 35 cycles of 95°C for 30 s, 62°C for 30 s, and 72°C for 45 s; for fragment A3, 40 cycles of 95°C for 30 s, 62°C for 30 s, and 72°C for 1 min; for fragment A6, 35 cycles of 95°C for 30 s, 62°C for 30 s, and 72°C for 1 min; and for fragment B, 19 cycles of 95°C for 1 min, 52°C for 1 min, and 72°C for 2 min, followed by 19 cycles of 95°C for 1 min, 65°C for 1 min, and 72°C for 2 min. RAD54 was analyzed in two overlapping fragments A and B, which were nested within the primary PCR product. The primary PCR product was amplified using the forward primer R54-Prim-F (TCAACAGGGCCCTGTGGAGCAAGG) and R54-Prim-R (CCGCTGCCTGCTTTTGACCTTTGG) and 25 cycles of 95°C for 1 min, 56°C for 30s, and 72°C for 3 min, by 35 cycles of 95°C for 30s, 60°C for 30s, and 72°C for 1 min 30s with primers RAD54AF, CTAGGCCCAGGATGAGGAGGAGCTTG; RAD54AR, GGTGTCAGCCTACAACAAACGACCTGC; RAD54BF, TAGGAGAGGAGCGGCTGCGGGAGC; and RAD54BR, CAGCGGAGGCCCCGCTGTTCCTC.

Primers used to amplify a RAD51 fragment encoding a frameshift were: RAD51AFmut, TTCCTGACTATGCGGGCTATCCCTATGACGTCCCGGACTATGCAGAGTAATGGCAATGCAGATGC; and RAD51AR. Wild-type (RAD51AF/RAD51AR products) and mutant (RAD51AF-mut/RAD51AR products) were mixed together in variable ratios and transformed into YPH499.

Nested PCR products (10 μl) and BamHI-digested yeast expression vector, pCI-HA(ura3)-2(-leu) (10 ng), were cotransformed into competent yeast (strain YPH499; Stratagene, La Jolla, CA). Vector alone was transformed into yeast as a negative control. Transformants were selected by plating on SC medium lacking leucine. For each sample, 30 transformants were replica-streaked unto SC plates lacking either leucine or leucine and uracil. Transformants expressing a chimeric RAD51 (ORF)-URA3 fusion gene exhibit uracil-independent growth, whereas those containing a stop codon in RAD51 are unable to grow in the absence of uracil. As expected, virtually 100% of transformants derived from specimens with wild-type RAD inserts display the ura3+ phenotype, whereas ∼50% of transformants from specimens with a heterozygous truncating mutation are ura3+ (see “Results”). To sequence the inserts of ura3− yeast transformants, the cDNA inserts were amplified from yeast with the ura 3− phenotype by PCR amplification using primers directed against the flanking pCI-HA(ura3)-2(-leu) sequence: forward primer, CCC ATA CGA TGT TCC TGA CTA TGC G; and reverse primer, TTG GCA GCA ACA GGA CTA GGA TGA G. PCR products were subjected to cycle sequencing using an automated fluorescence-based cycle sequencer (model 373; PE Applied Biosystems) and Taq dye terminator chemistry. Internal primers used in sequencing RAD52 clones were R52B-S2F (AGACCTCTGACACATTAGCCTTG) and R52B-S1R (TTTGCT TGTGGTTTCTGGTGG). The sense and antisense strands of the sequenced products were aligned with that of the wild-type sequence using the Sequence Navigator program (PE Applied Biosystems).

Results and Discussion

Analysis of RAD51.

To screen for the presence of definitive RAD51 mutations in the germ-line of women at high risk for genetic predisposition to breast cancer, we first analyzed EBV-immortalized lymphocytes from a cohort of women with early-onset breast cancer (diagnosed before the age of 40 years), using a high-throughput, yeast-based protein truncation assay outlined in Fig. 1A ⇓ (26 , 27) . To validate this truncation assay for analysis of RAD51 mutations, we mixed cDNAs with synthetic stop codons at various ratios. As predicted, specimens with reconstituted heterozygous mutations yielded 50% ura3+ and 50% ura3− colonies, whereas those with wild-type RAD51 sequence produced virtually 100% ura3+ colonies (Fig. 1B) ⇓ .

Analysis of germ-line specimens from 120 women with early-onset breast cancer using this protein truncation assay did not identify any premature stop codons in RAD51 (Table 1) ⇓ . Whereas such truncating mutations might have provided definitive evidence for inactivation of RAD51 (see below), our analysis could not exclude the presence of missense mutations that might be associated with an increased risk for breast cancer. We, therefore, screened the highest-risk subset of the early-onset breast cancer cohort (27 women diagnosed before the age of 30 years) by direct automated nt sequencing of uncloned RT-PCR products, combined with analysis using software for detection of heterozygous mutations (see “Materials and Methods”). No sequence variations in RAD51 were detected in any of these cases, leading us to conclude that this gene is unlikely to contribute to genetic predisposition to breast cancer.

To extend our analysis to the potential contribution of RAD51 mutations to sporadic breast cancer, we screened breast cancer cell lines for the presence of truncating mutations. RAD51 has been localized to chromosome band 15q15.1, a region that exhibits frequent loss of heterozygosity in sporadic breast tumors (28) . Loss of heterozygosity was primarily observed in specimens of metastatic breast cancer (7 of 10 cases), with less frequent allelic losses in primary breast tumors (2 of 19 cases). Analysis of 15 breast cancer cell lines derived from sporadic tumors using the yeast truncation assay identified neither heterozygous nor homozygous mutations in any cases (Table 1) ⇓ . Thus, definitive, truncating mutations in RAD51 are unlikely to contribute to the development of sporadic breast cancer. It is noteworthy that we detected an apparent alternative splicing variant within the 3′ end of the RAD51 transcript. This variant mRNA was detected by RT-PCR analysis in the majority of germ-line specimens and tumor cell lines. Sequencing analysis demonstrated a deletion of 122 nt (inclusive of positions 1007–1128), resulting in a frameshift and generation of a premature stop codon at nt 1195. Whereas this transcript removes the COOH-terminal region of RAD51 implicated in its interaction with the BRCT motifs of BRCA2 (29) , its physiological significance is uncertain because it represents a very small proportion of the total cellular RAD51 mRNA (estimated at <1% by RT-PCR analysis, in both germ-line and cancer cell lines). However, this truncated transcript is preferentially amplified following nested RT-PCR analysis, and it should not be misconstrued as a genuine mutation.

Analysis of RAD52.

In contrast to RAD51, analysis of 100 early-onset breast cancer cases, using the yeast-based protein truncation assay, revealed two recurrent nonsense mutations in RAD52: Ser346ter (TCG→TAG) and Tyr415ter (TAT→TAG; Table 1 ⇓ ). These heterozygous germ-line mutations were confirmed by nt sequencing of uncloned PCR products derived from both cDNA and genomic DNA (Fig. 2A) ⇓ . The Ser346ter mutation was present in 3 of 100 cases with early onset of breast cancer, whereas the Tyr415ter mutation was detected in 2 of 100 cases. To determine whether these mutations were present within the general population, we analyzed 100 specimens from healthy blood donors. These controls are well matched with our cohort of early-onset breast cancer cases, as demonstrated by the equal prevalence of known p53 polymorphisms (data not shown). Direct sequencing analysis revealed the Ser346ter mutation in 5 of 102 controls and the Tyr415ter mutation in 3 of 102 controls (comparison with cases of early-onset breast cancer, P was not significant; Table 2 ⇓ ). Hence, these two nonsense mutations do not appear to confer genetic susceptibility to early-onset breast cancer. Remarkably, however, these truncating mutations together are present in 8% of a diverse population. They were not restricted to a specific ethnic group, and no mutations were found in a cohort of 40 women of Ashkenazi Jewish descent, a population with an increased frequency of founder mutations in BRCA1 and BRCA2. No missense mutations in RAD52 were detected by direct nt sequencing of 27 cases with early-onset breast cancer.

RAD52 encodes a protein of 421 amino acids; the Ser346ter mutation results in loss of 75 COOH-terminal residues, whereas the Tyr415ter mutation only removes the terminal 7 amino acids. The functional properties of RAD52 remain to be defined, but known domains, including the sites of RAD51 association (30) and RAD52 self-association (31) , and domains that bind DNA and interact with RPA (32) are NH₂-terminal to Tyr415. To determine whether carriers of this heterozygous mutation might be more susceptible to DNA damage induced by ionizing radiation, we treated EBV-immortalized lymphoblasts from carriers with X-ray (1 Gy) and determined cell viability by vital dye staining. No significant increase in radiosensitivity was observed in the RAD52 ^Ser346ter/+ cells compared with RAD52 ^+/+ cells (data not shown). Within the limitations of these assays, carriers of these common truncating mutations in RAD52, therefore, do not appear to have increased sensitivity to ionizing radiation, although we cannot exclude more subtle abnormalities in homologous recombination and the repair of DNA damage.

Although the chromosome 12p12–13 locus encompassing the RAD52 gene is commonly targeted by allelic losses in breast cancer (18) , analysis of 15 breast cancer cell lines did not demonstrate truncating mutations within RAD52. Our observations, therefore, do not indicate that inactivating mutations in RAD52 are commonly associated with sporadic breast cancer. Of note, we observed frequent aberrant RAD52 pre-mRNA splicing products, following RT-PCR amplification of the transcript from either breast cancer cell lines or germ-line of both patients and controls (Fig. 2B) ⇓ . However, RT-PCR of RAD52 from human thymus, a tissue in which the transcript is expressed at high levels physiologically, showed no evidence of these splicing products (Fig. 2C) ⇓ . As for RAD52, the aberrant RT-PCR products involved insertions of intronic sequences at nt positions 617, 736, 932, and 1236, leading to premature stop codons (Fig. 2B) ⇓ . Although they were present in a minority of the products amplified from either cases or controls, nested PCR amplification led to increased abundance of these aberrantly processed transcripts, leading to their potential misinterpretation as resulting from genuine mutations. In addition to preventing amplification of these aberrant PCR products, optimization of PCR conditions and the judicious placement of PCR primers was also required to avoid amplification of a conserved known RAD52 pseudogene (see “Materials and Methods”).

Analysis of RAD54.

Analysis of the RAD54 transcript using the yeast-based protein truncation assay did not reveal any premature stop codons in 100 cases of early-onset breast cancer. Direct nt sequencing of the 27 cases diagnosed before the age of 30 years identified a single missense mutation, Cys657Ser (Table 1) ⇓ . However, screening 70 additional cases of early-onset breast cancer for this specific mutation did not identify another case, and the mutation was not present in 100 controls. We, therefore, cannot distinguish a disease-associated missense mutation from a rare polymorphism. No truncating mutations in RAD54 were detected in 15 breast cancer cases.

Concluding Remarks.

In summary, analysis of RAD51, RAD52, and RAD54 in both breast cancer cells and in the germ-line of women at high risk for genetic predisposition to breast cancer did not reveal disease-associated mutations. Our germ-line analysis was aimed at a population-based cohort of 100 early-onset cases, but it included 9 women who also have a strong family history of breast cancer and do not have a germ-line mutation in either BRCA1 or BRCA2 and, hence, are likely to harbor a mutation in another breast cancer predisposition gene. Despite their implication in a functional pathway shared by BRCA1 and BRCA2, these RAD genes, involved in homologous recombination and the repair of double-strand breaks in DNA, therefore, do not appear to be targeted by mutations in the development of breast cancer. A striking observation from these mutational studies is that ∼8% of the United States population carries either one of two alleles encoding premature stop codons in RAD52. Truncating mutations have typically been taken as evidence of definitive gene inactivation, and to our knowledge, a similar mutation in a gene associated with human disease has only been reported in BRCA2, where an extreme COOH-terminal truncating mutation, Lys3326ter, is present in 2.2% of the British population (33) . Like the RAD52 mutations described here, the prevalence of this nonsense mutation at codon 3326 of BRCA2 is not increased among women with early-onset breast cancer, suggesting that it may be a functionally silent polymorphism. In our own early-onset breast cancer cohort, we also detected a rare truncating mutation at codon 3308 of BRCA2, upstream of Lys3326ter (26) . In the absence of functional assays for BRCA2 or RAD52, it is unclear at what point such progressive COOH-terminal truncations begin to affect protein function and whether they may be associated with attenuated phenotypes. As such, the study of common genetic variations within the population, ranging from subtle missense mutations to premature stop codons, is of considerable importance because even a modest increase in disease risk associated with a frequent genetic alteration may have significant consequences. This has been demonstrated most convincingly for the I1307K mutation in the APC gene, which is common in the Ashkenazi Jewish population and is associated with a moderately increased risk of colon cancer (34) . Further work will be required to determine whether the truncated RAD52 products described here display any deficiency in mediating homologous recombination and whether carriers in the population are at increased risk for clinical syndromes resulting from inadequate repair of DNA damage.