SEARCH

SEARCH BY CITATION

Keywords:

  • RAD50;
  • breast cancer;
  • familial;
  • cancer susceptibility

Abstract

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

The genes predisposing to familial breast cancer are largely unknown, but 5 of the 6 known genes are involved in DNA damage repair. RAD50 is part of a highly conserved complex important in recognising, signalling and repairing DNA double-strand breaks. Recently, a truncating mutation in the RAD50 gene, 687delT, was identified in 2 Finnish breast cancer families. To evaluate the contribution of RAD50 to familial breast cancer, we screened the whole coding region for mutations in 435 UK and 46 Finnish familial breast cancer cases. We identified one truncating mutation, Q350X, in one UK family. We screened a further 544 Finnish familial breast cancer cases and 560 controls for the 687delT mutation, which was present in 3 cases (0.5%) and 1 control (0.2%). Neither Q350X nor 687delT segregated with cancer in the families in which they were identified. Functional analyses suggested that RAD50 687delT is a null allele as there was no detectable expression of the mutant protein. However, the wild-type allele was retained and expressed in breast tumors from mutation carriers. The abundance of the full-length RAD50 protein was reduced in carrier lymphoblastoid cells, suggesting a possible haploinsufficiency mechanism. These data indicate that RAD50 mutations are rare in familial breast cancer and either carry no, or a very small, increased risk of cancer. Altogether, these results suggest RAD50 can only be making a very minor contribution to familial breast cancer predisposition in UK and Finland. © 2005 Wiley-Liss, Inc.

The genes predisposing to familial breast cancer are largely unknown. There are 2 major breast cancer susceptibility genes, BRCA1 and BRCA2, and 4 minor genes, TP53, CHEK2, ATM and PTEN; but collectively mutations in these genes only account for approximately 20–30% of the familial risk of breast cancer.1, 2, 3 Therefore, the susceptibility genes underlying most familial breast cancer remain to be identified.

Five of the six known breast cancer susceptibility genes, BRCA1, BRCA2, CHEK2, ATM and TP53, encode proteins implicated in genome integrity monitoring that have multiple interactions with each other, both directly and indirectly.4, 5 This suggests that mutations in some of the many other genes involved in the DNA damage response pathways may also be involved in breast cancer susceptibility.

The highly conserved MRE11/RAD50/NBS1 (MRN) complex has a central role in many cellular responses to DNA double-strand breaks, including homologous recombination, nonhomologous end-joining, telomere maintenance and DNA damage checkpoint activation.6 The complex consists of the large coiled-coil ATP-binding cassette ATPase RAD50, the nuclease MRE11 and the checkpoint mediator NBS1.

Hypomorphic biallelic NBS1 mutations cause the autosomal recessive condition, Nijmegen breakage syndrome, which is characterised by microcephaly, immunodeficiency and tumor predisposition, particularly lymphomas.7 Biallelic MRE11 mutations cause ataxia-telangiectasia-like disorder (ALTD), which is characterised by slowly progressive ataxia and ocular apraxia.8 Mutations in RAD50 have not been associated with a defined human phenotype. However, a hypomorphic mutation of RAD50 has been identified in a patient suffering from a syndrome broadly reminiscent of NBS.5 Recently, a 687delT truncating RAD50 mutation was reported in 2 breast cancer families from Northern Finland.9

RAD50 is a 1,312-amino acid protein that contains 2 ABC-ATPase domains separated by 2 coiled-coil regions required for intramolecular interactions. A Cys-X-X-Cys motif located in the middle of the coiled-coil domain functions as a dimerization domain between 2 RAD50 arms.10, 11 The 687delT mutation leads to a premature stop at codon 234 and the function of the mutant protein would be predicted to be severely compromised.

To evaluate the role of RAD50 in breast cancer predisposition, we have screened the coding region of the gene in 481 UK and Finnish breast cancer families and have specifically evaluated the Finnish 687delT mutation in 590 Finnish familial breast cancer cases and 560 controls.

Material and methods

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Patients and controls

Familial breast cancer cases and controls were collected from UK and Finland. The UK samples were collected through the Breast Cancer Susceptibility Collaboration (UK). Consent was obtained from all cases and the research was approved by the London Multiresearch Ethics Committee. Samples from individuals with breast cancer from 702 familial breast cancer pedigrees were analyzed. All samples were screened for mutations in the full coding sequence of BRCA1 and BRCA2, as previously described.12 The breast cancer history in relatives of the index case was quantified using a family history score, with 1 point for an affected first-degree relative or a second primary in the index case, 0.5 point for an affected second-degree relative and 0.25 for an affected third-degree relative. The majority of cases had a family history score of at least 2 (i.e. at least 2 first-degree relatives with breast cancer in addition to the affected index case, or equivalent) and no case had a family history score less than 1 (i.e. at least one affected first-degree relative in addition to the affected index case, or equivalent). Samples from 786 UK controls were obtained from Human Random Control DNA panels from the European Collection of Cell Cultures (Salisbury, UK).

The Finnish series consisted of 590 familial breast cancer patients collected at the Helsinki University Central Hospital, as previously described.13 Of these, 338 breast cancer patients had a strong family history (3 or more first- or second-degree relatives with breast or ovarian cancer in the family, including the index case), as verified through the Finnish Cancer Registry and hospital records, and 252 unrelated breast cancer cases had only a single affected first-degree relative. Among the 590 patients, 502 had a family history of breast cancer only and 88 had a relative with ovarian cancer. For 435 cases, BRCA1 and BRCA2 mutations had been excluded as previously described,14, 15 and for 155 the BRCA1 and BRCA2 mutation status was unknown. Controls were from 560 anonymous healthy blood donors from the Finnish Red Cross Blood Transfusion Service in Helsinki. Informed consent from the patients and permission for the study from the Ethics Committee of the Department of Obstetrics and Gynecology, HUCH and the Ministry of Social Affairs and Health in Finland were obtained.

RAD50 mutation analyses

The coding sequence and intron–exon boundaries of RAD50 were analyzed using conformation sensitive gel electrophoresis, CSGE,16 in 435 UK and 46 Finnish familial breast cancer cases. Primers and conditions are available on request. Genomic DNA from cases showing mobility shifts on CSGE was bidirectionally sequenced using the BigDyeTerminator cycle sequencing kit and an ABI 3100 automated sequencer (Applied Biosystems, Foster City, CA).

The exon 7, Q350X mutation was analysed in 267 additional UK familial breast cancer patient samples and 786 population controls by bidirectional sequencing of exon 7. The Finnish 687delT mutation and the Q350X mutation were screened in Finnish familial breast cancer patient samples and controls by minisequencing (primer extension).17 All positive minisequencing results were confirmed by reamplification from the original genomic DNA sample and direct sequencing (ABI BigDyeTerminator cycle sequencing kit (v3.0), ABI Prism 310 genetic analyzer (Applied Biosystems, Foster City, CA).

Protein expression analysis of RAD50 687delT

The RAD50 protein expression was evaluated by immunoblotting on cell lysates from 3 lymphoblastoid cell lines established from heterozygous carriers of the 687delT mutation, compared with a control lymphoblastoid cell line homozygous for wild-type RAD50. For immunoblotting analysis, proteins from total cell lysates were separated using denaturing polyacrylamide gel electrophoresis (10% SDS-PAGE) and blotted onto nitrocellulose membrane. The RAD50 protein was visualized on parallel blots after incubation with 3 distinct primary antibodies (ab3622 from Abcam, Cambridge, UK, clone 13 from Transduction Laboratories, Lexington, KY and 13B3 from GeneTex, San Antonio, TX), using the ECL visualization reagents (Amersham, Piscataway, NJ).

Immunohistochemistry of RAD50 expression on 687delT carrier tumors

The RAD50 protein expression was studied by immunohistochemical staining in 3 archival breast tumors from the same patients as above, compared to a series of normal breast tissues (n = 15), familial breast carcinomas with apparently wild-type RAD50 (n = 25) and sporadic breast tumors (n = 27). For immunohistochemistry, the tumor sections were deparaffinized and processed for sensitive immunoperoxidase staining with the primary mouse monoclonal antibody against human RAD50 (clone 2C6, Abcam, 1:500 dilution), incubated overnight, followed by detection using the Vectastain elite kit (Vector Laboratories, Burlingame, CA), as previously described.18 For analyses of the MRE11 and NBS1 proteins on parallel sections, rabbit antibodies No. 4895 against MRE11 (Cell Signaling, diluted 1:1,500) and No. 3002 against NBS1 (Cell Signaling, diluted 1:100) were used, respectively. Given that around 95% of epithelial cells were reproducibly and strongly positive on sections of normal breast, we regarded as an aberrant decrease when the RAD50 protein was detectable in fewer than 70% of cancer cells in a particular lesion, and the staining intensity of the remaining positive cancer cells was reduced compared with normal cells present on the same section.

Bioinformatic analyses

SIFT-analysis19 (http://blocks.fhcrc.org/sift/SIFT) and PolyPhen20 (http://genetics.bwh.harvard.edu/pph/) were used to evaluate functional significance of the RAD50 missense variants found. SIFT program calculates tolerance scores for amino acid changes based on sequence alignment and conservation across protein family or across evolutionary history. PolyPhen (=Polymorphism Phenotyping) predicts possible impact of an amino acid substitution on the structure and function of a human protein, using straightforward physical and comparative considerations.

Results

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

RAD50 mutation analysis

We screened the coding region and exon–intron boundaries of the RAD50 gene in 435 UK and 46 Finnish familial breast cancer cases and identified a single protein truncating mutation 1048C>T, Q350X, in a woman who developed bilateral breast cancer at 43 years (Fig. 1a, Table I). The mutation was not present in her sibling, who developed breast cancer at 73 years. Her mother developed ovarian cancer at 55 years and a paternal aunt developed breast cancer at 50 years. Unfortunately, no further samples were available and therefore it is unknown whether the mutation arose de novo in the index case or was inherited from one of her parents, or whether the paternal aunt carried the mutation. To further evaluate the Q350X mutation, 267 additional UK familial breast cancer cases, 786 UK controls, 235 Finnish breast cancer cases and 319 Finnish controls were screened for the mutation by direct sequencing or minisequencing. No further case or control with Q350X was identified.

thumbnail image

Figure 1. Pedigrees from 4 families with RAD50 truncating mutations and 3 families with a possibly pathogenic missense mutation. (a) Family B1321, a UK family with a Q350X mutation. (b) UK families B682, B1339 and B1353 with missense mutations R327H, R224H and R193W, respectively. (c) Three Finnish families with the 687delT mutation.

Download figure to PowerPoint

Table I. RAD50 Variants Found in UK and Finnish Samples
Nucleotide changeAmino acid changeExonUKFinnish
CasesControlsCasesControls
Truncating variants
 1048 C>TQ350X71/7020/7860/2810/319
 687delT 50/435 3/5901/560
Missense variants
 280 A>CI94L36/435 0/46 
 373 G>AV127I41/435 0/46 
 577 C>TR193W51/435 0/46 
 671 G>AR224H51/435 0/46 
 695 C>AA232D51/435 0/46 
 943 G>TV315L72/435 0/46 
 980 G>AR327H71/435 0/46 
 2177 G>AR726H131/435 0/46 
 2525 T>CV842A161/435 0/46 
Synonymous variants
 204 C>TH68H21/435 0/46 
 3879 C>TI1293I55/435 0/46 
Intronic variants
 IVS1+12G>A IVS11/435 0/46 
 IVS9+73C>T IVS92/435 0/46 
 IVS20+35C>T IVS202/435 1/46 
 IVS20+49G>C IVS203/435 0/46 
 IVS22-15delTTC IVS221/435 0/46 
 IVS22+24A>G IVS2210/435 0/46 

We identified 11 coding RAD50 sequence changes of which 9 were nonsynonymous and 2 were synonymous, 7 changes were intronic (Table I). The exonic missense variants were evaluated for possible functional effect by SIFT and PolyPhen analysis, which suggested that R193W, which was found in one UK breast cancer case and alters a key residue in the MRE11 binding segment, likely affects RAD50 function. R224H and R327H may affect RAD50 function but the other variants appear well tolerated by SIFT and PolyPhen analysis. Pedigrees with tumor information of the families carrying the R193W, R224H and R327H missense mutations are shown in Figure 1b.

The 687delT mutation found in the Finnish population previously was evaluated among index cases from 590 Finnish breast cancer families and 560 healthy population controls. The mutation was found in 3 familial patients (0.5%) and one control sample (0.2%). The characteristics of the breast cancer families with the 687delT mutation are shown in Figure 1c. Complete mutation screening of the whole RAD50 gene, among 46 Finnish breast cancer families, did not reveal any other coding RAD50 sequence variants present in breast cancer families from Southern Finland. The 687delT mutation was not found among the 435 UK families studied.

Protein expression analysis of RAD50 687delT

Immunoblotting analysis of RAD50 protein in 687delT carrier lymphoblastoid cell lines showed a decreased overall abundance of the full-length protein product, compared with a normal control lymphoblastoid cell line. In addition, the truncated protein form was not detected in lysates from the cell lines with heterozygous RAD50 687delT or with any of the 3 antibodies, including an antibody prepared specifically against the N-terminus of RAD50, predicted to be preserved in the mutant protein (Fig. 2a).

thumbnail image

Figure 2. (a) Immunoblot showing a lower level of the full-length RAD50 protein in the lymphoblastoid cell line heterozygous for 687delT (right), as compared to that in the noncarrier cell line (left). CDK7 was used as a loading control (the double band signal below). (b) Immunohistochemical detection of RAD50 in human breast tissues and breast carcinomas. (1) Strong nuclear positivity for RAD50 in epithelial and stromal cells of normal human breast tissue. Insert: higher magnification shows that both luminal (inner layer) and myoepithelial (outer layer) cells are positive. (2) Infiltrating ductal carcinoma showing the typical pattern of almost homogeneous RAD50 expression. (3)–(5). Examples of representative staining patterns for RAD50 in breast carcinomas from carriers of the RAD50 687delT mutation. (6) Aberrant, grossly reduced or lost RAD50 staining in a sporadic breast carcinoma contrasts with strong nuclear signal in the surrounding stromal cells.

Download figure to PowerPoint

Immunohistochemical analysis of 3 breast tumors from 687delT mutation carriers showed normal subcellular localisation of the RAD50 protein. The overall staining pattern was nuclear and the signal detected by this semiquantitative immunohistochemistry approach seemed comparable in the 3 mutation carrier tumors, in a set of additional 25 familial breast carcinomas with apparently wild-type RAD50, in the control normal breast tissues (n = 15, with both luminal epithelial and myoepithelial cells positive) and most of the 27 sporadic breast tumors examined (Fig. 2b). On the other hand, a small subset of sporadic carcinomas (3 out of 27) showed grossly reduced abundance of the RAD50 protein (Fig. 2b). No obvious defects of either the MRE11 or NBS1 protein were detected among the 28 familial breast tumors examined for RAD50, in contrast to the 3 sporadic carcinomas that showed concomitant reduction in staining of all 3 proteins of the MRN complex.

Discussion

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

The possible role of RAD50 in familial breast cancer predisposition is intriguing as RAD50 participates in critical cellular functions of DNA double-strand break repair, functionally interacting with other breast cancer predisposition genes.11, 21 In this study, we have investigated the role of RAD50 in familial breast cancer by mutation analysis of the full gene in 481 breast cancer families from UK and Finland, and have specifically evaluated a RAD50 truncating mutation previously identified in 2 Finnish breast cancer families.9

RAD50 mutation analysis revealed only a single protein truncating mutation, Q350X, in one of 481 breast cancer families studied. The mutation leads to truncation of 962 amino acids of the RAD50 protein, including most of the coiled-coil domain, the zinc-hook structure and the carboxyterminal ATP-binding domain.10, 22, 23 The mutation was identified in a young-onset bilateral breast cancer case, but was not present in her affected sister, although it is noteworthy that the sister did not develop breast cancer until 73 years. There was a history of breast cancer on the paternal side and ovarian cancer in the mother, but we were not able to evaluate whether or not either of these cases carried the mutation. No other case or control with the mutation was identified despite analysis of 1,488 UK samples and 717 Finnish samples. These data indicate that RAD50 Q350X is rare. Therefore, even if it is contributing to cancer predisposition in family B1321, it cannot be making a major contribution to familial breast cancer overall.

We identified 9 missense variants, 2 synonymous changes and 7 intronic RAD50 variants (Table I). Two of the missense variants (I94L and R224H) and one of the synonymous changes (H68H) have been reported previously.9 There was no strong evidence to suggest that any of these variants are breast cancer susceptibility alleles. Three variants, R913W, R224H and R327H, may affect RAD50 function, but all are rare, each identified in only 1/481 cases, limiting any potential contribution to breast cancer susceptibility, even if they are pathogenic.

The protein truncating mutation 687delT was detected in 3/590 (0.5%) Finnish familial breast cancer cases and 1/560 controls (0.2%), but was not present in the 435 UK families studied. The mutation showed incomplete segregation with cancer in the families (Fig. 1c). This is consistent with previous data showing that RAD50 687delT did not segregate with cancer in one of the 2 families in which it was identified; the family also carried a pathogenic BRCA1 mutation.9 The combined 687delT frequency in families not tested positive for a BRCA1/2 mutation from Finland does not differ significantly from the frequency in controls (4/730, 0.55% vs. 7/1,560, 0.40%, OR 1.22, 95% CI 0.36–4.19, p = 0.75). The presence of the mutation at similar frequency in cases and controls and the lack of segregation with cancer suggest that it either carries no increased risk of breast cancer, or is a very low penetrance breast cancer susceptibility allele. The absence of the mutation in UK cases limited our ability to further evaluate the role of RAD50 687delT in breast cancer predisposition.

To investigate the functional effect of RAD50 687delT, we studied RAD50 expression in lymphoblastoid cell lines from mutation carriers. This showed that the overall abundance of the full-length protein product was decreased and the truncated protein was not detected on immunoblotting. RAD50 687delT thus appears to be a null allele with no detectable expression of the mutant protein, suggesting it is unlikely to have a dominant-negative effect on RAD50 function. In addition, RAD50 protein expression and nuclear localization appeared normal in breast tumors from 687delT mutation carriers, suggesting that the wild-type allele was retained and expressed. Similarly to RAD50, the abundance and localisation of the MRE11 and NBS1 proteins appeared normal in all the 28 familial breast tumors examined by immunohistochemistry. The fact that we identified a small subset of sporadic breast tumors with a simultaneous gross reduction of all 3 proteins of the MRN complex is consistent with a previous study of sporadic carcinomas.24 Furthermore, the latter cases also document that the antibodies and our assay conditions were appropriate to detect such aberrations at the protein level when they existed. The subset of MRN-deficient carcinomas (ref.24 and the present study) may reflect the recently reported constitutive activation of DNA damage checkpoints in preinvasive human breast and other lesions, and the subsequent selection for defects of various components of the DNA damage response network as a way to overcome the anticancer barrier of such activated checkpoints during tumor progression.25

Our data on tumors and lymphoblastoid cell lines with the 687delT variant of RAD50 are consistent with loss of heterozygosity analyses both in our study (data not shown) and in breast tumors from previously reported 687delT carriers, which demonstrated that the wild-type allele was not lost.9 However, the overall amount of RAD50 in lymphoblastoid cells heterozygous for 687delT was reduced compared to that in noncarrier cells. These data suggest that RAD50 is not acting as a classical tumor suppressor gene in familial breast cancer, although it is possible that RAD50 haploinsufficiency is contributing to cancer. Such a mechanism has been suggested by mouse models for important DNA damage/repair proteins whose defects contribute to tumorigenicity via loss or decrease, including for example the tumor suppressor Chk1 kinase.26

In conclusion, our data demonstrate that germ line variants in RAD50 occur at very low frequency in breast cancer families from UK and Finland. Rare truncating variants may be associated with a small increased cancer risk, but do not appear to be acting as high penetrance mutations for breast cancer. RAD50 can only be making a very small contribution to familial breast cancer in the UK and Finnish populations.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

We thank all the families that participated in this study. We thank A. Hall and E. Mackie for sample ascertainment and pedigree entry and H. Eerola and N. Puolakka for patient contacts. The Finnish Cancer Registry is gratefully acknowledged for cancer data.

References

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References