Absence of rearrangements in the BRCA2 gene in human cancers

Mutations of BRCA2 in sporadic breast and ovarian carcinomas are exceedingly rare. This led to the suggestion that large genomic rearrangements could be involved. We performed Southern blots in genomic DNA from 130 primary breast cancers and 83 cancer cell lines (breast, ovarian, pancreatic and small cell lung carcinomas) and found no genomic rearrangements. These results suggest that a gene other than BRCA2 is the target of the frequent 13q12.3 allelic deletions in human cancers. © 2001 Cancer Research Campaign http://www.bjcancer.com

BRCA2, the second hereditary breast cancer gene, has been mapped to chromosome 13q12-13 (Wooster et al, 1994). BRCA2 is a very large gene spanning more than 70 kb of genomic DNA encoding 3495 amino acids (Wooster et al, 1995;Tavtigian et al, 1996). To date more than 300 distinct germline mutations of BRCA2 (http://www.nhgri.nih.gov/intramural_ research/lab_transfers/bic) have been identified that predispose carriers to breast cancer and to a lesser extent ovarian cancer (Rahman and Stratton, 1998). A small increase in risk for pancreatic and prostate cancer has also been reported in BRCA2 pedigrees (Wooster et al, 1995;Lancaster et al, 1996;Phelan et al, 1996). Frequent loss of heterozygosity (LOH) at the BRCA2 locus in a variety of sporadic cancers e.g. breast, ovarian , pancreatic, prostate (Cooney et al, 1996;Li et al, 1998), hepatocellular cancer (Kuroki et al, 1995), suggests this gene may behave as a tumour suppressor gene (Cleton-Jansen et al, 1995). However, no clear disease-causing somatic mutations have been described in BRCA2 in sporadic breast cancers, and somatic mutations in ovarian cancers are very rare (Foster et al, 1996;Lancaster et al, 1996;Miki et al, 1996;Takahashi et al, 1996;Teng et al, 1996). The lack of somatic BRCA2 mutations in sporadic breast and ovarian cancers could be due to the mutation detection assays used. Mutations may be missed if they are outside of the region of analysis and certain types of mutations (large deletions, insertions and duplications) may not be detected by PCR-based mutation detection assays. Southern blot analyses have identified 5 large Alu-mediated genomic deletions (Petrij-Bosch et al, 1997;Puget et al, 1997b;Swensen et al, 1997) and a 6 kb Alu-mediated duplication (Puget et al, 1997a) involving BRCA1 in breast cancer families that would have been missed by conventional PCR-based mutation screening methods such as SSCP, PTT or direct sequencing using genomic DNA as template. A recent study found one case of sporadic breast cancer out of 81 studied with BRCA1 genomic deletions (van der Looij et al, 2000). Similar large genomic deletions have also been described in other tumour suppressor genes e.g. p53 (Masuda et al, 1987;Ruggeri et al, 1992), hMLH1 (Nystrom-Lahti et al, 1995), hMSH2 (Wijnen et al, 1998) and Rb-1 (Ruggeri et al, 1992). Thus, we wanted to investigate whether similar genomic deletions occur in BRCA2 which may have escaped detection using PCR-based techniques.

MATERIALS AND METHODS
We undertook Southern blot analysis of genomic DNA in a large series of 130 invasive breast tumours comprising ductal, lobular, mucinous, tubular, cribriform and squamous cell metaplastic carcinomas. These tumours were snap frozen at the point of collection. Genomic DNA from these primary tumours was digested with EcoRI. In addition, genomic DNA from established cell lines derived from breast (39), ovarian (29), pancreatic (7) and small cell lung (SCLC) tumours (8) was digested with 4 different restriction endonucleases (BamHI, EcoRI, HindIII and PstI). Digested DNA was size fractionated by electrophoresis and transferred onto nylon membranes. Filters were hybridized separately with two clones containing BRCA2 cDNA. The first is a 5.8 kb fragment representing amino acids 1-1963 (BRCA2-front) and the second fragment is 4.6 kb from amino acids 1895-3495 (BRCA2-back).

RESULTS
Initially, we detected aberrant-sized fragments in two primary tumours; tumour 386 with both probes and tumour 64 NT with BRCA2-front probe (Figure 1). There were 8 breast cancer cell lines that showed restriction fragment size fragment abnormalities with BRCA2-front and 4 with BRCA2-back. In ovarian cancer cell lines, abnormalities were only observed in 3 cell lines using BRCA2-front. No abnormalities were detected in any of the SCLC or pancreatic cell lines. A representative sample of these Southern blot experiments is presented in Figure 1.

Absence of rearrangements in the BRCA2 gene in human cancers
To confirm the presence of genomic rearrangements in the abnormal samples, the experiments were repeated with longer incubation of the DNA with the respective restriction endonucleases to ensure complete digestion. No abnormal restriction fragments were detected in any of the samples suggesting that the initial abnormal bands were due to incomplete digestion (data not shown). While no large genomic deletions were observed in the BRCA2 gene, restriction fragment length polymorphisms were observed in both primary tumours and cell lines (Figure 2).

DISCUSSION
In summary, BRCA2 does not undergo large intragenic deletions in human tumours. Only 8 somatic BRCA2 mutations, 3 in breast tumours Miki et al, 1996;Weber et al, 1996), 4 in ovarian cancers (Foster et al, 1996;Takahashi et al, 1996) and one in a hepatocellular carcinoma , have been reported since the discovery of the gene using PCR-based mutation detection assays. Similar to BRCA1, the region containing BRCA2 undergoes loss of heterozygosity in a fraction of breast and ovarian tumours (Cleton-Jansen et al, 1995). In fact in another study using these cell lines, we found that 24/36 breast (67%), 6/30 ovarian (20%), 4/7 pancreatic (57%) and 5/8 SCLC (63%) used in this study had homozygosity for all markers tested in the region encompassing BRCA2 (data not shown) suggestive of allelic deletions. Partial mutational analyis of BRCA2 mutations was undertaken for some of the cell lines with LOH and to date, only one breast cancer cell line with LOH, MT-3, was found to have a 1 bp deletion in exon 23 of suggest either distinct genetic pathways or different mechanisms for inactivating gene function compared to the familial forms (Rahman and Stratton, 1998). The high frequency of LOH on chromosome 13q could be explained by the close proximity of other tumour suppressor genes e.g. retinoblastoma (Lee et al, 1988), Brush-1 (Schott et al, 1994) or other putative tumour suppressor gene(s) that might be targeted instead of BRCA2.