Although screening for large deletions or duplications of the BRCA1 gene is becoming a routine component of the molecular diagnosis of familial breast cancer, little is known about the occurrence of such rearrangements in the BRCA2 gene. Because of the high frequency of BRCA2 mutations in breast cancer families with at least one case of male breast cancer, we selected a cohort of 39 such families, tested negative for mutations in the coding regions of BRCA1 and BRCA2, and developed an assay for BRCA2 rearrangements, based on quantitative multiplex PCR of short fluorescent fragments (QMPSF). We found three rearrangements: (1) a deletion of exons 12 and 13; (2) a duplication of exons 1 and 2; and (3) a complete deletion of BRCA2. We determined the boundaries of the deletion of exons 12 and 13, showing that it resulted from an unequal recombination between Alu sequences. We mapped the complete BRCA2 deletion, which extends over at least 298 kb and showed that it does not affect APRIN/AS3, previously characterized as a tumor suppressor gene, but it comprises several loci corresponding to proven or putative transcripts of unknown functional significance. These data suggest that screening for BRCA2 rearrangements should be done, especially in male breast cancer families tested negative for BRCA1 and BRCA2 mutations.
Several studies have described large genomic deletions or duplications of the BRCA1 gene in breast cancer or breast-ovarian cancer families (1, 2, 3, 4) , and sensitive assays designed to detect such rearrangements have been recently added to the routine molecular diagnosis of breast or breast-ovarian cancer predisposition (5 , 6) . Genomic rearrangements have been estimated to represent between 10 and 15% of the BRCA1 defects (2 , 7) , but founder effects resulting in higher frequencies of germline rearrangements have been described in some populations (1) . However, very little is known concerning the occurrence of germline rearrangements in the BRCA2 gene. Besides the report of an Alu element insertion into exon 22 (8) , only two examples of germline BRCA2 rearrangements have been published: a deletion that comprises a portion of exon 3 and most of intron 3, resulting into an inframe exon 3 skipping (9) and a deletion/insertion mutation removing exons 12 and 13 (10) . The former rearrangement was discovered at the cDNA level, whereas the latter was found at the genomic level, by Southern blot analysis. Therefore, we set out to develop for BRCA2 a sensitive method, capable of analyzing rapidly all of the exons for copy numbers. To this end, we adapted to BRCA2 the previously described quantitative multiplex PCR of short fluorescent fragments (QMPSF) method (5 , 11) . Breast cancer is rare in men, but BRCA2 defects are frequent in families with one or more cases of male breast cancer. Therefore, we applied the new QMPSF assay to the detection of BRCA2 rearrangements in families with multiple breast cancers, including one or more male breast cancers, and without detectable mutations within BRCA1 and BRCA2 coding sequences.
Materials and Methods
The inclusion criteria were as follows: at least one male breast cancer (any age) and at least one first- or second-degree relative with male breast cancer (any age) or female breast cancer (age of diagnosis below 60 years). In some cases, third-degree relatives were also considered if the intermediate was a male or an affected female. The index case was an affected male (any age) or female (diagnosed at age below 60 years) or, in a few cases, an obligate carrier. The diagnoses of the index case and of male breast cancer patients were verified by examining the histologic and/or the clinical records. All of the index cases had been found BRCA1 and BRCA2 negative on screening all of the exons for point mutations and microdeletions or -insertions by either denaturing high-performance liquid chromatography (dHPLC), denaturing gradient gel electrophoresis (DGGE), or by using different combinations of established methods [dHPLC, DGGE, fluorescence assisted mismatch analysis (FAMA), or the protein truncation test (PTT)].
Standard QMPSF Conditions.
Exons of the BRCA2 gene were analyzed by three QMPSF assays consisting of 8, 9, and 9 BRCA2 amplicons, respectively, and a control from the MLH1 gene. Amplicon sizes ranged between 172 bp and 300 bp. Reactions were done in 25 μL with 0.2 mmol/L of each deoxynucleoside triphosphate, 1.2 mmol/L MgCl2, 1.5 units of Thermoprime Plus DNA Polymerase (ABgene, Epson, United Kingdom), 7.5% DMSO, 75 to 100 ng of DNA, and between 0.2 and 2.4 μmol/L of each PCR primer, one of them carrying a 6-FAM label. After an initial step of denaturation at 94°C for 4 minutes, 24 cycles were done consisting of denaturation at 94°C for 10 seconds, annealing at 48°C for 15 seconds, and extension at 72°C for 20 seconds, followed by a final extension step at 72°C for 5 minutes. DNA fragments generated by QMPSF were separated on an ABI Prism 377 DNA sequencer, and the resulting fluorescence profiles were analyzed with the GeneScan 3.1 Software. Primer sequences are available on request.
Special QMPSF Designs.
The QMPSF assay for mapping the deletion of exons 12 and 13 comprised 7 amplicons as follows: an amplicon in exon 14 of MLH1 as a control, the standard amplicons for exons 11 (3′end), 12, 13, and 14, and two additional amplicons in introns 11 and 13, respectively. To map the boundaries of the complete BRCA2 deletion, amplicons were chosen in several loci in the BRCA2 region that correspond to shown or putative genes. Three of them were located in the 13CDNA73 gene as follows: one in exon 21, one in exon 44, and the other one in exon 61. One amplicon was located in exon 7 of the CG018 gene. Two amplicons were located in the PFAAP5 gene–one in exon 2 and the other one in exon 6. One amplicon was located 6 kb from the 5′end of the APRIN/AS3 gene, and the last one was located in exon 2 of the same gene. This QMPSF assay also contains the standard amplicon for BRCA2 exon 17 and a control amplicon located in the UFD1L gene (22q11).
Long-range PCR was carried out with the Expand Long Template PCR System (Roche, Meylan, France) as described previously (5) . The duplication of exons 1 and 2 of BRCA2 in one family was confirmed with the multiplex ligation-dependent probe amplification (MLPA) kit for BRCA2 from Microbiology Research Center-Holland (Amsterdam, The Netherlands).
Results and Discussion
BRCA2 mutations are frequent in families with multiple breast cancer cases including one or more male breast cancer cases. It has been estimated by linkage analysis that BRCA2 defects are present in 76% of the male breast cancer families of the Breast Cancer Linkage Consortium, which are characterized by at least 4 breast cancers (age of diagnosis <60 years for females, any age for males; ref. 12 ). Large deletions or duplications of BRCA2 are therefore likely in male breast cancer families tested negative after screening all exons and exon-intron junctions of BRCA2 and of BRCA1. Of the 39 French families that fulfilled our criteria (Table 1) ⇓ , 5 had two male breast cancer cases and 11, including 2 families with two male breast cancer cases, had four or more breast cancer cases. Moreover, other characteristic features of the BRCA2 phenotype (13) were found in several families of this cohort such as prostate cancer, pancreatic cancer, and melanomas (Table 1) ⇓ .
The QMPSF assay for BRCA2 that we have developed consists of three multiplex PCRs and was validated with DNA from patients with known microdeletions or microinsertions in different BRCA2 exons, which showed the expected size-shift of peaks because of the mutant amplicons and, in addition, a reduction to one half of the peaks of the corresponding wild-type allele (data not shown). We evaluated the sensitivity of this assay by using DNA from a chromosome 13 trisomy and found a ∼50% increase of the signals corresponding to all of the BRCA2 amplicons (data not shown). We then screened the index cases of the 39 families listed in Table 1 ⇓ and found three rearrangements: a deletion of exons 12 and 13, a duplication of exons 1 and 2, and a complete deletion of BRCA2 (families F1028, F1020, and F2832, respectively; shown in bold in Table 1 ⇓ ). All three families were among those with the highest probability of BRCA2-related cancer predisposition. In family F1028, a total of 11 cancers had been found, including 2 male breast cancer cases, 8 breast cancer cases, and one prostate cancer case. In family F1020, two brothers had breast cancer. Family F2832 showed a pancreatic cancer, a male breast cancer, and a female breast cancer in the same lineage in subsequent generations.
The rearrangement in family F1028 is illustrated in Fig. 1 ⇓ . On superimposition of the fluorescence profiles of control DNA and of DNA from the index case of this family, we found a reduction of the peak intensities of amplicons corresponding to exons 12 and 13, shown in Fig. 1, B and C ⇓ , respectively. To define the boundaries of this deletion, we designed a specific QMPSF assay for this region (not shown), which included additional amplicons in introns. This specific assay allowed us to reduce the target size for long-range PCR and to amplify from the mutant allele a fragment containing the deletion boundaries. Sequencing of the junction showed that the deletion had occurred by unequal recombination between an AluYa5 sequence in intron 11 and an AluSx sequence in intron 13 and that its size was 7947 bp (Fig. 1D) ⇓ . This deletion therefore differs from the one described by Wang et al. (10) , which was shorter (6.2 kb) and could not be explained by a simple unequal recombination event between Alu repeats.
In family F1020 (Fig. 2) ⇓ , an ∼50% increase of the intensity of the peaks corresponding to exons 1 and 2 was observed (Fig. 2, A and B) ⇓ , and this difference in the QMPSF profile was not seen in any of the >100 individuals tested in this study. Moreover, this profile was identical, concerning exons 1 and 2, to that of our duplication control obtained with DNA from a chromosome 13 trisomy (data not shown). However, we were unable to amplify the duplicated segment by PCR and to characterize it at the nucleotide level. We therefore confirmed our QMPSF result using a commercially available multiplex ligation-dependent probe amplification (MLPA) assay for BRCA2, which is based on single-stranded probes that hybridize at adjacent positions to exonic targets and can then be joined covalently by ligase and amplified by PCR (6) . The relevant portion of the MLPA profile is shown in Fig. 2C ⇓ and confirms the duplication of exons 1 and 2. Our failure to obtain duplication-specific PCR fragments, despite the numerous combinations of primers used in long-range PCR experiments, may indicate that this rearrangement is more complex than a tandem duplication of exons 1 and 2.
A complete BRCA2 deletion was found in family F2832. Fig. 3 ⇓ shows our analysis of the boundaries of this large deletion, particularly with regard to the known genes or open reading frames surrounding BRCA2. The 13q12.3 region (schematically shown in Fig. 3B ⇓ ) contains (in the centromer to telomer direction) the following: the large gene 13CDNA73 encoding the hypothetical protein CG003; the BRCA2 gene; the PRO0297 gene, inferred from a human fetal liver cDNA (data not shown); the hypothetical gene CG018, with opposite transcriptional orientation; a locus with unknown significance (LOC88523; data not shown), also in opposite orientation; the PFAAP5 gene, predicted to encode phosphonoformate immunoassociated protein 5, with unknown function; and the APRIN gene, also known as AS3 gene (henceforth named APRIN/AS3), which is transcribed from the same strand as BRCA2, and whose 5′ end lies 187 kb downstream of the 3′ end of BRCA2. It is especially noteworthy that the AS3 protein (androgen-induced prostate proliferative shutoff associated protein) is a mediator of the proliferative arrest of prostate epithelial cells (14) and that the gene coding for this protein had previously been found affected by large somatic deletions of the BRCA2 region, described initially in a pancreatic carcinoma (ref. 15 and the references therein). It was therefore of great interest to define the boundaries of the germline deletion found in family F2832, and particularly the telomeric one. To this end, we designed additional QMPSF assays containing amplicons distributed over the ∼500-kb region shown in Fig. 3 ⇓ (an example of these assays is shown in Fig. 3A ⇓ ). These analyses showed that the germline deletion extended over at least 298 kb and did not affect the APRIN/AS3 gene (Fig. 3B) ⇓ . Because of the undefined function of genes surrounding BRCA2 (except for APRIN/AS3), one cannot predict which phenotypes could have been expected for this family, because of the large deletion. It is, however, noteworthy that the cancer characteristics observed in this family (a pancreatic cancer, a male breast cancer, and a female breast cancer in the same lineage in subsequent generations) are typical of the BRCA2 phenotype (13) .
Finally, we screened for BRCA2 rearrangements our collection of 91 breast-ovarian cancer families that had been tested BRCA1 and BRCA2 negative, except for five distinct rearrangements detected in BRCA1 (5 , 16) . We found no BRCA2 rearrangement in this large breast-ovarian cancer cohort, a result consistent with the notion that in breast-ovarian cancer families BRCA1 defects are predominant, although BRCA2 mutations are also found, particularly in the ovarian cancer cluster region (17) .
This work shows that large constitutive rearrangements are not uncommon in the BRCA2 gene. Two studies had previously addressed this question by Southern blot analysis, but they were not based on families selected for particularly high probability of BRCA2-related cancer predisposition. In a study of 82 Finnish breast and/or ovarian cancer families, no germline rearrangement was found in BRCA1 or BRCA2 (18) . Moreover, no BRCA2 rearrangement was found in 81 Dutch breast and/or ovarian cancer families (19) . These negative results therefore indicate the absence, in these populations, of BRCA2 rearrangements with a founder effect, as described for BRCA1 (1) , but leave open the possibility of a variety of BRCA2 rearrangements such as those that we have described.
The ability of both QMPSF and MLPA to detect the duplication of exons 1 and 2 (Fig. 2) ⇓ illustrates the high sensitivity of these methods and indicates that the vast majority of rearrangements should be detected, if one uses target DNAs of good and homogeneous quality. Nevertheless, particular types of rearrangements could be missed if they involve only portions of exons, because only short sequence stretches are targeted in each exon. The cost per sample of screening for BRCA2 rearrangements is the same as that of screening for BRCA1 large germline deletions or duplications, which is already being implemented routinely in many laboratories, but screening for BRCA2 rearrangements is presently limited to selected families with high probability of a BRCA2 defect, tested negative for BRCA1 and BRCA2 mutations.
Our cohort of BRCA1- and BRCA2-negative male breast cancer families was composed of families studied in several French laboratories and did not result from consecutive cases. Therefore, it would be highly troublesome to evaluate directly the contribution of the three rearrangements found to the total number of BRCA2 defects. However, the criteria for inclusion of the families analyzed here are similar to those of 29 of the 33 families analyzed by Evans et al. (20) . They found BRCA2 point mutations or microdeletions/insertions in 41% of these families (12 of 29). Assuming a similar proportion in the families from which our cohort of 39 BRCA1- and BRCA2-negative families was derived, one can estimate that BRCA2 point mutations or microdeletions/insertions accounted for not more than 30 defects. Although these estimates have been extrapolated and should be taken with caution, they suggest that the 3 BRCA2 rearrangements found represented an additional 10% of BRCA2 defects detected in our families. Moreover, considering that the BRCA2 rearrangements described here have different predicted consequences at the RNA level and that no genotype-phenotype correlation is known between BRCA2 mutations and male breast cancer, it is very likely that germline BRCA2 rearrangements are also present in a fraction of families with large numbers of breast cancers, regardless of the presence of male breast cancer cases.
Consequently, it seems appropriate to add the search of large deletions/duplications of BRCA2 to the molecular diagnosis of male breast cancer families and possibly also of breast cancer families with large numbers of cases, also considering the sensitivity and low cost of both assays (i.e., QMPSF and MLPA) that are now available.
We wish to thank all of the members of the different laboratories participating in this study who prepared and provided DNA samples.
Grant support: Association pour la Recherche sur le Cancer, Ligue Contre le Cancer de Haute Normandie. I. Tournier is supported by a fellowship from Ministère de l’Education Nationale, de la Recherche et de la Technologie (MENRT).
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.
Requests for reprints: Mario Tosi, INSERM U614, Faculté de Médecine, 22 boulevard Gambetta, 76183 Rouen, France. Phone: 00332-35148311; E-mail:
- Received July 12, 2004.
- Revision received September 15, 2004.
- Accepted September 28, 2004.
- ©2004 American Association for Cancer Research.