A missense mutation in the BRCA2 gene in three siblings with ovarian cancer.

Inherited susceptibility to ovarian cancer has been associated with germline defects at several loci. The major known ovarian cancer susceptibility gene is BRCA1 on chromosome 17q, which confers a risk of approximately 60% by the age of 70 years. Truncating mutations in BRCA2 on chromosome 13q also predispose to ovarian cancer, although they confer a lower risk than mutations in BRCA1. We have studied the molecular basis of ovarian cancer predisposition in a Finnish family with three affected sisters. Analysis of polymorphic markers provided evidence against linkage to BRCA1, but the sibship was consistent with linkage to BRCA2. Conformation-sensitive gel electrophoresis was used to screen the entire coding sequence of BRCA2. A G to A transition at nucleotide 8702 was observed, which is predicted to convert glycine 2901 to aspartate in the encoded protein. This sequence variant was not detected in 220 cancer-free Finnish control individuals, or in several hundred cancer families of many nationalities previously screened for BRCA2 mutations. Taken together with the fact that this amino acid residue and the surrounding region of BRCA2 is identical in mouse and chicken, the data suggest that this alteration is a disease-causing BRCA2 missense mutation. Previously published data indicate that the risks of breast and ovarian cancer conferred by BRCA2-truncating mutations varies with the position of the mutation in the gene. The missense mutation reported here suggests that the BRCA2 domain including and surrounding glycine 2901 may be more important in preventing neoplastic transformation in ovarian epithelium than in breast epithelium. ImagesFigure 1

Summary Inherited susceptibility to ovarian cancer has been associated with germline defects at several loci. The major known ovarian cancer susceptibility gene is BRCA1 on chromosome 17q, which confers a risk of approximately 60% by the age of 70 years. Truncating mutations in BRCA2 on chromosome 1 3q also predispose to ovarian cancer, although they confer a lower risk than mutations in BRCA 1. We have studied the molecular basis of ovarian cancer predisposition in a Finnish family with three affected sisters. Analysis of polymorphic markers provided evidence against linkage to BRCA1, but the sibship was consistent with linkage to BRCA2. Conformation-sensitive gel electrophoresis was used to screen the entire coding sequence of BRCA2. A G to A transition at nucleotide 8702 was observed, which is predicted to convert glycine 2901 to aspartate in the encoded protein. This sequence variant was not detected in 220 cancer-free Finnish control individuals, or in several hundred cancer families of many nationalities previously screened for BRCA2 mutations. Taken together with the fact that this amino acid residue and the surrounding region of BRCA2 is identical in mouse and chicken, the data suggest that this alteration is a disease-causing BRCA2 missense mutation. Previously published data indicate that the risks of breast and ovarian cancer conferred by BRCA2-truncating mutations varies with the position of the mutation in the gene. The missense mutation reported here suggests that the BRCA2 domain including and surrounding glycine 2901 may be more important in preventing neoplastic transformation in ovarian epithelium than in breast epithelium.
Keywords: ovarian cancer; BRCA2 gene; missense mutation Ovarian cancer is the sixth most common cancer in women worldwide (Parkin et al, 1993). It is estimated that 5-10% of ovarian cancers are associated with the inheritance of a mutant allele conferring autosomal dominant predisposition with high penetrance (Lynch et al, 1987). Genetic linkage studies have indicated that the majority of hereditary breast-ovarian and site-specific ovarian cancer families appear to be linked to the BRCAJ gene on chromosome 17q (Hall et al, 1990;Miki et al, 1994). Mutations in BRCAJ confer an overall risk of ovarian cancer of approximately 60% by the age of 70 years in high-risk families (Easton et al, 1995), although some other studies have suggested a lower risk (Struewing et al, 1997;Whittemore et al, 1997).
Mutations in the BRCA2 gene on chromosome 13q (Wooster et al, 1994;1995) also confer susceptibility to ovarian cancer. The risk is lower than that conferred by BRCAJ and has recently been estimated at 27% (D Ford, DF Easton and the Breast Cancer Linkage Consortium, submitted for publication). The risks of breast and ovarian cancer conferred by BRCA2 mutations also vary according to the position of the mutation within the gene. In particular, the ratio of ovarian cancer to breast cancer cases observed in each familial cluster appears to be higher in a region of approximately 3000 bp within exon 11, which has been termed the ovarian cancer cluster region (OCCR) (Gayther et al, 1997). In principle, this pattern could be due to a lower breast cancer risk conferred by mutations in the OCCR compared with the rest of the gene; a higher ovarian cancer risk conferred by mutations in the OCCR compared with the rest of the gene; or both. Whichever option is correct, however, it appears likely that mutational inactivation of certain domains within BRCA2 differentially alters risks of breast and ovarian cancer. To characterize further such domains, we have analysed a Finnish family in which three sisters developed ovarian cancer.

MATERIAL AND METHODS Patients
The method used for identifying a series of Finnish ovarian cancer families has been described in a previous study (Auranen et al, 1996). Briefly, patients diagnosed with epithelial ovarian cancer in Finland between 1980 and 1982 and their first-degree relatives were studied. Of the 559 ovarian cancer probands, 27 had one or more first-degree relatives affected with ovarian cancer. Paraffin-embedded tumour samples were obtained for 47 patients belonging to these 27 different families. Blood samples were obtained from members of a single family (family 19). This family includes three sisters with epithelial ovarian cancer. The mother of the sisters had died of tuberculosis at * These authors contributed equally to this study. . Patient 19-1 was diagnosed at the age of 59 years with ovarian cancer, patient 19-2 at the age of 58 years and patient 19-3 at the age of 55 years. All these three sibs are at present clinically cancer free and alive at the ages of 71, 68 and 63 years respectively. One sister and one brother have died without evidence of cancer and one brother is alive and cancer free. The germline allele data at the marker loci used are shown beneath each analysed subject. Below the allele data, electrophoresis analysis of Sfanl digests from patients and one healthy brother are presented. The normal sample (blood) is indicated with N and tumour samples with T. The upper fragment in the gel is PCR product without sequence variant, digestion of this 1 94-bp fragment resulted in 1 27-bp and 67-bp (not shown) fragments in patients. In patient 19-1 tumour sample showed loss of the wild-type allele age 41 years and the father died free of cancer at age 71 years. One sister and one brother have died without evidence of cancer and one brother is alive and cancer free (Figure 1).
Patient 19-1 has had eight pregnancies and was 55 years of age at menopause. In 1981, at the age of 59 years, she presented with post-menopausal bleeding and was diagnosed with a grade III transitional cell stage Ib ovarian cancer. Patient 19-2 has had one pregnancy and one miscarriage, and was 52 years at menopause. In 1983, at the age of 58 years, she presented with loss of weight and ascites, and was diagnosed with serous papillary grade I stage HIlc ovarian cancer and in September 1996 she was operated for an ovarian cancer metastasis in the colon. She received six courses of cytostatic treatment and is currently in follow-up without clinical evidence of disease. Patient 19-3 has had no pregnancies and was 51 years of age at menopause. In 1986, at the age of 55 years, she was hospitalized because of acute severe abdominal pains and was diagnosed with an endometrioid grade II stage Ia ovarian cancer. All three patients underwent hysterectomy, bilateral salpingooophorectomy and cytostatic treatments. They are currently clinically disease free.

Mutational screen of BRCA2
The entire coding sequence and intron/exon junctions of the BRCA2 gene were amplified by polymerase chain reaction (PCR) using previously published primer sequences (Gayther et al, 1997). Both primers were end-labelled with gamma-32P, and heteroduplexes were formed by heating the PCR products to 98°C for 10 min, holding at 60°C for 15 min and allowing them to return to room temperature. Samples were then analysed by conformation-sensitive gel electrophoresis (CSGE) (Ganguly et al, 1993).
Samples showing variant migration patterns were reamplified and directly sequenced using fluorescent dye terminators and analysed on an ABI 377 DNA sequencer.
Analysis of G8702A using a PCR restriction enzyme assay Screening of the G to A change at nucleotide 8702 was performed by restriction enzyme digestion. The PCR amplification was performed in a total volume of 50 gl, including 150 ng genomic DNA, 25 gM of dNTP (Pharmacia Biotech), 0.8 gM of each primer, 1.5 mM magnesium chloride and 10 x reaction buffer (Perkin Elmer). The denaturation and synthesis steps were: 1 min at 95°C for denaturation, 45 s at 56°C for annealing and 1 min at 72°C for synthesis. The three-step amplification cycle was repeated 40 times. The following primers were used for amplification: forward primer 5'-TGGTTCTTTAGTTTTlAGTTGCTlTTG; reverse primer 5'-TCACCTCAAGGTAAGCTGGG.
The digestion was performed with SfaNI restriction enzyme in 1 x NEBuffer 3 (New England BioLabs) at 37°C overnight. SfaNI enzyme cuts the PCR fragment (194 bp), which contains mutations in two fragments (127 bp and 67 bp by size), whereas the wild-type fragment lacks the restriction site and is not digested. After digestion, PCR products were electrophoresed through 3% agarose gels.

RESULTS
Linkage analysis using markers D17S250, DJ7S579, D17S588 and D17S855 for the BRCAJ locus provided evidence against linkage to BRCAI, whereas marker D13S310 at the BRCA2 locus was W1I 0 Cancer Research Campaign 1998 BRCA2 mutation in three siblings with ovarian cancer 1201 compatible with linkage ( Figure 1). CSGE was used to screen the entire coding sequence of the BRCA2 gene. The only variant detected was a G to A transition at nucleotide 8702, which is predicted to convert glycine to aspartate at amino acid 2901. This variant was present in all three sisters with ovarian cancer. Analysis of tumour material from one case showed loss of heterozygosity of the allele not carrying the sequence variant (Figure 1). To evaluate further the significance of this finding, we analysed 220 Finnish cancer-free controls and detected it in none. We also analysed 190 Finnish patients with primary ovarian cancer. A total of 44 of these ovarian cancer cases were familial and derived from 26 different families (Auranen et al, 1996). Among the other 146 cases there was a series of 15 patients who were selected by age (younger than 40 years at the time of diagnosis). The remaining 131 cases were unselected. The G8702A change was also searched for in index patient DNA samples for 100 Finnish breast cancer families (with three or more cases of breast or ovarian cancer in firstor second-degree relatives, all probands affected with breast cancer) previously analysed for BRCA1 and BRCA2 mutations. Ten of these patients were BRCA1 mutation carriers and 11 carried BRCA2 mutations, whereas 65 breast cancer only families and 15 breast-ovarian families were negative for both (Vehmanen et al, 1997). In addition, 21 breast cancer families not scrutinized for the above criteria (not previously analysed for BRCA1 or BRCA2) were studied for G8702A variant. None of these 311 patients or 220 cancer-free controls showed the change.

DISCUSSION
We have detected a BRCA2 sequence variant that is predicted to generate a missense amino acid change in a family with three sisters affected by ovarian cancer. Until this variant is evaluated in a functional assay for BRCA2, it will not be possible to determine unambiguously if it is a disease associated mutation or a rare polymorphism. However, the variant has not been detected in a large series of Finnish controls and has not been reported previously in several hundred individuals screened. It is present in all three sisters with ovarian cancer, and the one tumour sample examined showed loss of the other allele (as predicted for a tumoursuppressor gene and previously demonstrated for cancers in BRCA2 mutation carriers). Moreover, in a protein that shows only 59% sequence identity between human and mouse, Gly-2901 is conserved in human, mouse and chicken, and is situated within a domain of the protein that shows strong sequence conservation in these three species (Table 1). Taken together, the evidence indicates that this is likely to be a disease-associated BRCA2 mutation.
Truncating BRCA2 mutations associated with a high ratio of ovarian to breast cancer have previously been mapped to the OCCR within exon 11. This is the first BRCA2 family reported with more than two cases of ovarian cancer and a mutation outside this region. Unfortunately, it has not been possible to evaluate further the cancer risks associated with this mutation because we have failed to detect additional examples of the variant in other Finnish ovarian or breast cancer families, or in a consecutive series of ovarian cancer cases. Nevertheless, given the rarity of such ovarian cancer clusters associated with BRCA2 mutations, we propose that this variant is associated with different breast/ovarian risks compared with mutations elsewhere in the gene. Whether this is a result of an increased ovarian cancer risk, decreased breast cancer risk or both is impoEFible to evaluate at present. However, The target of the mutation, amino acid Gly-2901 (*), is conserved in human, mouse and chicken, and is located within a domain of the protein that shows strong sequence conservation in these three species (amino acid residues that are identical in all three species are shown in bold).
the results suggest that inactivation of the BRCA2 domain that includes glycine 2901 may have different effects on breast and ovarian epithelium from inactivation of other domains of the protein.