Material and methods

The index case was diagnosed with breast cancer at the age of 30 and ovarian cancer at the age of 49. She had one sister with breast cancer diagnosed at the age of 35, another sister with ovarian cancer diagnosed at the age of 35, and a paternal grandmother with breast cancer diagnosed at the age of 41. The index case was referred to Cedars-Sinai Medical Center (Los Angeles, USA) for a genetic consultation. She elected to participate inBRCA gene testing, as she hoped to characterise her apparent genetic susceptibility so that her daughter could know her own status with greater certainty. NoBRCA1 or BRCA2gene mutation was identified by direct DNA sequencing (BRACAnalysis^TM, Myriad Genetic Laboratories Inc, Salt Lake City, USA). Because the a priori likelihood of carrying aBRCA gene mutation was high, the case was referred to our laboratory to search for large rearrangements in theBRCA1 gene (family quoted IC2361).

The strategy for the detection of large rearrangements developed in our laboratory is based on a full four colour bar code of theBRCA1 region on combed DNA.16Combing relies on homogeneous stretching of DNA molecules at a constant rate of 2 kb/μm.17 Fluorescence in situ hybridisation (FISH) is then performed on combed DNA.18 The probes used include a PAC covering the whole BRCA1region and long range (LR) PCR products (6.5 to 10 kb long) covering a number of exons, therefore bar coding the PAC. We have optimised theBRCA1 bar code reported by Gadet al 16 by adding new LR products to allow for the detection of rearrangements as small as 2 kb. Finally, in addition to the PAC, a complex bar code of theBRCA1 region was designed with seven probes (fig 2). This approach allows for a panoramic view of theBRCA1 gene and its flanking regions.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

BRCA1 bar code of IC2361. (A) The BRCA1 region in 17q21. The BRCA1 gene is spread over 81 kb and has a common promoter with the NBR2 gene (Next to BRCA1 gene 2), which covers approximately 30 kb.15 ,26 The BRCA1 pseudogene (ΨBRCA1) lies next to NBR2 and corresponds to a partial duplication of BRCA1, from the promoter region to intron 2.27 ,28 The NBR1 gene (Next to BRCA1 gene 1) is located 5′ to ΨBRCA1 and may be phylogenetically at the origin of NBR2.26 ,29 The figure is not drawn to scale and is adapted from Xu et al26 and Brown et al.28 (B) Analysis of the BRCA1 region with our standard bar code. (1) This full signal corresponds to the normal allele. The PAC 103O14 insert covers the region between the first exons of NBR1 and the 3′ BRCA1 UTR (exon 24), which is approximately 120 kb long.15 ,26 ,28 ,30 The cosmid ICRFc105D06121 clone used in our experiment is rearranged, with an insert which is approximately 6 kb long and covers the region between NBR2 intron 1 and BRCA1 intron 2.16 ,26 ,28 The other probes are long range (LR) PCR products, covering the regions between the exons quoted, except LR24-3′, which covers BRCA1 exon 24 to 10 kb downstream (primers used are available on http://www.curie.net/genetique). (2) This full signal does not include the LR13-15 probe, which represents a first indication of a deletion of at least exons 13 to 15 in the BRCA1 gene for this patient. (C) Adaptation of the BRCA1 bar code to the detected deletion. The hybridisation of two additional long range products covering exons 12 to 13 and exons 15 to 18 on the normal allele (1) and the deleted allele (2) shows that the deletion comprises only exons 13 to 15.

Germline DNA was extracted from a lymphoblastoid cell line, with a step in agarose blocks in order to preserve its integrity. After combing of the DNA on silanised surfaces17 ,18 and FISH with the set of probes, microscope screening was then performed.

Results

With a few fields of microscopic view, a number of full signals without LR13-15 probes were observed, identifying the existence of a large deletion of the BRCA1 gene (fig 2B). In order to characterise the deleted exons better, the bar code was adapted by using two additional probes, covering exons 12-13 and exons 15-18. Taking the sizes of BRCA1 exons and introns into account,15 as shown in fig 2C, the deleted region was expected to comprise exons 13, 14, and 15.

To characterise the mutant mRNA, we performed RT-PCR using primers located at the 3′ end of exon 11 and in exon 16. A normal product was detected both in control and patient DNA corresponding to the 824 bp expected product (fig 3A). In the patient, a mutant product was also observed, corresponding to the 334 bp product expected in the absence of exons 13 to 15 (fig 3A). Sequencing of this shorter product showed that exons 12 and 16 were adjacent (fig 3B), leading to a premature stop codon at position 1437 and truncation of the BRCA1 protein. Thus, the deletion of exons 13 to 15 appears to be the origin of the cancer predisposition in this family.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Characterisation of the IC2361BRCA1 rearrangement. (A) RT-PCR products with primers “exon 11 forward” and “exon 16 reverse” on (lane 1) control RNA (824 bp expected) and on (lane 2) IC2361 RNA: a product between 300 and 400 bp is observed and corresponds to the deletion of exons 13 to 15 (334 bp). RNA extraction was performed according to the TRIzol protocol (Gibco, Gaithersburg, MD, USA). Long range PCR products with primers “ex12 forward” and “ex16 reverse” on (lane 3) genomic control DNA showing, as expected, no product and on (lane 4) IC2361 genomic DNA showing the 8.5 kb abnormal product comprising the breakpoints. PCR products with primers “Alu intron 12 forward” and “Alu intron 15 reverse” on (lane 5) genomic control DNA showing as expected no product and on (lane 6) IC2361 genomic DNA showing the 550 bp narrowing the breakpoint region. (B) Sequences of IC2361 RT-PCR products. The 824 and 334 bp PCR products were gel extracted and sequenced with primers “exon 11 forward” and “exon 16 reverse”. Sequence of the 824 bp product shows normal transition between exons 12 and 13, whereas sequence of the 334 bp product shows a junction of exons 12 to 16. (C) Box in which recombination has occurred, leading to a 11.6 kb deletion. The 550 bp was gel extracted and sequenced with primers “Alu intron 12 forward” and “Alu intron 15 reverse”. Alignment of this sequence with Alu Sx in intron 12 and Alu Sp in intron 15 shows the recombination box located between nucleotides 44 377 and 44 397 in intron 12 and between nucleotides 55 980 and 56 000 in intron 15 (accession number Genbank L78833 15), thus resulting in a 11 604 bp deletion.

In order to define the boundaries of the deletion, long range PCR was performed on genomic DNA using primers located in exons 12 and 16. From the patient's DNA, an abnormal product was observed at 8.5 kb, whereas the 20.2 kb expected product was not observed in control DNA (fig 3A). The 8.5 kb product was gel extracted. Its restriction map was determined (data not shown) and compared to the restriction map of the normal region of exons 12 and 16 (accession number GenbankL78833 15). From this comparison and by taking the location of the Alu sequences in this region into account, we hypothesised that a recombination event had occurred between two Alu sequences, in the 3′ end of intron 12 and the 5′ end of intron 15, respectively. Primers were designed to amplify a DNA fragment comprising the putative breakpoints. From the patient's DNA, a 550 bp fragment was obtained, gel extracted, and sequenced (fig 3A). It showed that unequal recombination had occurred between an Alu Sx in intron 12 and an Alu Sp in intron 15. These two Alu sequences share 86% homology (data not shown). Recombination breakpoints were located between nucleotides 44 377 and 44 397 in Alu Sx and between nucleotides 55 980 and 56 000 in Alu Sp (accession number Genbank L78833 15), resulting in a 11 604 bp deletion (fig 3C).

In order to examine the frequency of this rearrangement among breast/ovarian cancer families, we screened a series of 90 women affected with breast or ovarian cancer and ascertained at the cancer genetics clinic of the Institut Curie (Paris, France) according to family criteria previously reported.3 Most of patients had French ancestors. These women, screened negative forBRCA1 and BRCA2point mutations, were then tested for the presence of the 550 bp PCR product by using primers “Alu intron 12 forward” and “Alu intron 15 reverse”. No 550 bp fragment was observed, suggesting that this rearrangement is not frequent in the population studied (data not shown).

Discussion

Our report of a previously undescribed 11.6 kb deletion encompassing exons 13 to 15 of the BRCA1gene illustrates the diversity of large rearrangements and their contribution to the molecular pathology of theBRCA1 gene. Few series of breast/ovarian cancer families have been systematically screened for large rearrangements of the BRCA1 gene. The reported frequencies of BRCA1 rearrangements range between 12% and 36%.4 ,6 ,7 ,14 Even with a conservative estimate of 10%, it would be advisable to include a search for large rearrangements in BRCA1when analysing high risk breast/ovarian cancer families. The family reported here serves as a prime example of a case in which additional testing was warranted in the absence of a detectable point mutation with standard PCR methods. The prior probability of the index case being a BRCA1/2 mutation carrier has been estimated at 95%. This value was obtained by using the MLINK program of the LINKAGE package, with the parameters of the Claus segregation model modified by Easton and the estimated contributions ofBRCA1 and BRCA2mutations to breast/ovarian cancer predisposition.1 ,19-21 In the absence of an identifiable mutation, closely related family members would have to be considered to be at high risk and would have to make decisions regarding cancer prevention on the basis of empirical data. With the identification of the familial BRCA1 deletion, at risk family members can now consider testing for the identified familial mutation and can learn their mutation status with certainty.

The broad diversity of rearrangements, ranging from 0.5 to 23.8 kb and spread over the 81 kb of the BRCA1region, requires methods that allow for complete analysis of the gene. In this respect, colour bar coding on combed DNA appears useful. It allows for a panoramic view of the BRCA1region and for the detection of a rearrangement of about 6 kb (the size of a probe deleted or duplicated) at a glance. In addition, deletions and duplications as small as 2 kb can be detected with measurement of the probe signals.16 Finally, more complex rearrangements involving inversions can also be detected. We think that software allowing for the automatic capture and analysis of signals would streamline the approach and, therefore, favour the use of colour bar coding on combed DNA. Searching for large gene rearrangements is a recurrent challenge for molecular geneticists. In addition to Southern blotting, other promising PCR based methods have recently been reported, including a long range PCR strategy and quantitative PCR.11 ,22-24 Haploid conversion of human lymphocytes via a cell fusion strategy may be another alternative to these methods, as it allows for suppression of the normal allele, facilitating the detection of large rearrangements by standard PCR.25Comparative analysis of the different methods listed above, taking both sensitivity and cost into consideration, are now needed to improve genetic testing for breast and ovarian cancer predisposition.