About 5–10% of breast cancer cases are hereditary and attributable to mutations in several susceptibility genes from which to date only 2 have been identified: BRCA1 (17q21) and BRCA2 (13q12). However, the BRCA genes only explain about 30% of familial breast/ovarian cancer, suggesting the existence of other susceptibility genes.1 Familial cases not associated with mutations in the BRCA genes represent the best starting point for searching for these new genes using different approaches. One such approach is linkage study. Several loci on chromosomes 8p12, 10q23 and more recently 13q21 have been identified by linkage analysis, but the study of a large number of BRCA-negative families did not support the contribution of putative genes in these loci.2, 3, 4
Another approach is focused on the study of genes involved in specific pathways. Given that BRCA1/2 are directly involved in DNA double-strand break repair (DSB) by homologous recombination (HR), several genes participating in the same pathway have been considered good candidates for breast cancer susceptibility. The HR reaction is catalyzed by RAD51, which interacts directly with BRCA2, and some related proteins, XRCC2, XRCC3, RAD51C and RAD51D, are also involved in the process.5 Another protein, BACH1, a member of the DEAH helicase family, is known to interfere with DSB in a BRCA1-dependent manner.6 Some of the genes encoding for these proteins have been analyzed for putative mutations involved in breast cancer susceptibility. Thus, BACH1 has been suggested as a candidate high-penetrance gene because 2 germline missense mutations were found in a study of 65 individuals who were BRCA1-negative;6 however, 197 families without BRCA1/2 mutations were recently analyzed for mutations in BACH1, and none were found.7 CHEK2 is another protein involved in DNA damage signaling that was recently identified as a candidate by linkage analysis; a specific variant of CHEK2 could be acting as a low-susceptibility allele in about 4% of familial cases not associated with BRCA.8
In our study, we selected the XRCC2 gene as a new candidate for high penetrance because it is a parologue of RAD51 and because it was recently noted that the variant R118H in exon 3 of this gene increases sporadic breast cancer risk.9 To test the possible involvement of XRCC2 in hereditary breast cancer, we selected as index cases 105 high-risk breast/ovarian cancer families that were not associated with either BRCA1 or BRCA2. The families came from the Spanish National Cancer Center (CNIO) and the Hospital Clínico San Carlos in Madrid and were selected by choosing at least 3 women affected with breast and/or ovarian cancer (1 of them diagnosed before 50 years of age). All the women had given informed consent to participate in the study. The 105 index cases had an average age at diagnosis of 50.6 years, and 7 of these families had breast-ovarian cancer. Peripheral blood (PB) samples of the index cases were analyzed by single-stranded conformational polymorphism (SSCP) and denaturing gradient gel electrophoresis (DGGE) for mutations in the BRCA1/2 genes; it was confirmed that they did not carry mutations in the BRCA1/2 genes.10, 11 Two hundred PB samples from the National Blood Transfusion Center were selected as a control population; they were paired with the cases by sex and age.
The 3 coding exons of the XRCC2 gene were analyzed in DNA samples from PB of the 105 index cases by CSGE and direct sequencing when necessary; oligonucleotide primers and PCR conditions are available upon request. Two distinct germline alterations in exon 2 were found to be heterozygous in 3 cases. The IVS-16pb change was present in 2 patients and was considered a polymorphism because of its intronic location and its 1.5 % frequency in the control population of 200 individuals. The second variant consisted of an amino acid change, Leucine for valine in codon 31 (Leu31Val); this change was identified in 1 patient but not in the control population, suggesting that it could be acting as a deleterious mutation in the family in which it was detected.
To test this hypothesis, we generated a theoretical structural model of the XRCC2 protein using the Insight II Version 98.0 Molecular Modeling System. The model showed that the substitution Leu31Val is out of the 2 known domains, 1 interacting with RAD51D and the other with ATP.12 However, the Leu31Val variant was located in the hydrophobic core domain of the protein, which is a highly conserved region whose external surface is directly implicated in polymerization. In addition, substitution of the amino acid leucine for a smaller amino acid like valine could lead to an increase in the cavity from 10 to 3713, 14 destabilizing the protein. Thus, the meaning of this change remained uncertain.
We were able to collect samples from 6 more members of the family, 5 healthy female relatives (I-1, I-3, II-2, III-1 and III-2) and 1 affected female relative (I-2). Association study between the disease and the variant showed that it was not present in any of the relatives, including the second breast cancer-affected relative. This finding suggested that this variant was not the cause of the disease; however, given the late age of onset of the tumor in this patient (71 years), sporadic tumour can not be discarded.
To obtain further evidence on the role of the Leu31Val change, we obtained a tumoral sample from the proband and performed a loss of heterozygosity (LOH) analysis. We confirmed the existence of allelic loss, but sequencing analysis of the fragment revealed that the lost allele was the one harboring the variant, ruling out a possible mechanism of gene inactivation by mutation of 1 allele and LOH of the other (normal allele). This result convinced us of the lack of pathogenicity of this variant in the family, which was consistent with the variant being a rare polymorphism (<1%) without detectable risk associated with breast cancer.
In conclusion, we have found that the XRCC2 gene is unlikely to be a high-penetrance gene whose mutations could explain a proportion of Spanish high-risk breast/ovarian cancer families not associated with BRCA1 and BRCA2. In addition, in 105 high-risk individuals, we have not found the low-susceptibility allele R118H, which has recently been described in 2 sporadic breast cancer populations as conferring a risk of about 1.3.9, 15 New studies in other HR genes are necessary to illuminate the role and contribution of this pathway to hereditary breast cancer.