Identification of a splice acceptor site mutation in p16INK4A/p14ARF within a breast cancer, melanoma, neurofibroma prone kindred
- A H Prowse1,*,
- D C Schultz1,3,†,
- S Guo1,
- L Vanderveer1,
- J Dangel1,
- B Bove1,
- P Cairns1,
- M Daly2,
- A K Godwin1
- 1Department of Medical Oncology, Fox Chase Cancer Center, Philadelphia, PA 19111, USA
- 2Department of Population Science, Fox Chase Cancer Center, Philadelphia, PA 19111, USA
- 3Department of Chemistry, Lehigh University, Bethlehem, PA, USA
- Correspondence to: Dr A H Prowse, Nuffield Department of Obstetrics and Gynaecology, University of Oxford, John Radcliffe Hospital, Oxford, OX3 9DU, UK;
Epidemiological studies estimate that at least 10% of all cancer cases, including breast cancer, can be attributed to inherited susceptibility. Two hereditary breast cancer genes, BRCA1 and BRCA2, have been identified and women who inherit a mutated copy of either gene have a raised lifetime risk of breast and ovarian cancer.1–4 More than 80% of families with multiple cases of breast cancer and ovarian cancer and most very large families with multiple cases of breast cancer carry mutations in either BRCA1 or BRCA2.5–7 However, only one third of families with only four or five cases of breast cancer and no cases of ovarian cancer carry mutations in either BRCA1 or BRCA2.6,8 Large scale screening studies for BRCA1 and BRCA2 mutations have shown that the percentage of high risk breast cancer families carrying a predisposing mutation in either gene is likely to be overestimated and depending on the criteria used to define the syndrome is likely to be greatly overestimated.6,9,10 Therefore, additional BRCA genes remain to be identified.
Inactivation of the INK4a/ARF locus on human chromosome 9p21 by point mutation, deletion, or hypermethylation is observed in many cancers.11–17 Accumulating evidence now suggests that the frequency of involvement of the INK4a/ARF locus in human cancers may be second only to that of TP53, underscoring its broad importance in tumorigenesis. The INK4a/ARF locus encodes for two distinct tumour suppressor genes, p16INK4a and p14ARF, which have alternative first exons (1α or 1β) and common exons 2 and 3; p16INK4a is encoded by three exons (designated 1α, 2, and 3), whereas p14ARF is encoded by a unique first exon (exon 1β) which splices into the INK4A exon 2, but is translated into an alternative reading frame (ARF).18 Germline mutations of p16INK4a have been identified in a proportion of familial melanomas.16 Many of these mutations are also predicted to inactivate p14ARF19,20 and, in addition, mutations in exon 1β of p14ARF have recently been identified in hereditary melanoma families.21–23 Interestingly, a high frequency of other carcinomas, particularly pancreatic tumours, nervous system tumours, head and neck tumours, and also breast carcinomas, has been reported in INK4a/ARF mutation positive melanoma families.21,22,24–29 In this study, we have evaluated members of 31 BRCA1 and BRCA2 mutation negative families presenting with multiple cases of early onset breast cancer, and in some cases malignant melanoma and pancreatic cancer, for germline mutations in p16INK4a and p14ARF.
MATERIALS AND METHODS
As part of a Fox Chase Cancer Center Institutional Review Board approved protocol, peripheral blood samples were obtained from consenting affected and unaffected high risk family members through the Margaret Dyson/Family Risk Assessment Program. DNA from these bloods was evaluated for germline mutations in the BRCA1 and BRCA2 genes through the Clinical Molecular Genetics Laboratory at Fox Chase Cancer Center. Subjects participating in the Family Risk Assessment Program have agreed to allow their samples to be used for a wide range of research purposes, including screening for mutations in other candidate predisposing genes. Determination of family history was as described previously.30,31 In addition, DNA samples from disease free controls were obtained from the Biosample Repository Core Facility (http://www.fccc.edu/clinicalresearch/BRCF). All participants in this study had signed a consent form to allow for genetic testing and the study was carried out under the approval of the Internal Review Board at Fox Chase.
Mutations in BRCA1 and BRCA2 are found in the majority of families with cases of both breast and ovarian cancer, but not all of breast cancer syndrome families, an indication that other breast cancer susceptibility genes exist. The INK4a/ARF locus on 9p21 is deleted or rearranged in a large number of human cancers and germline mutations in the gene have been shown to confer an inherited susceptibility to malignant melanoma and pancreatic carcinoma. The locus encodes two unrelated and independently acting negative cell cycle regulators, p16INK4a and p14ARF, arising from alternative first exons (1α or 1β) and common exons 2 and 3.
We evaluated members of 31 BRCA1 and BRCA2 mutation negative families presenting with multiple cases of early onset breast cancer, and in some cases malignant melanoma and pancreatic cancer, for germline mutations in p16INK4a and p14ARF. We identified a mutation in one family which in addition to breast cancer had multiple other tumours, including early onset melanomas, neurofibromas, and pancreatic carcinomas. The mutation, a splice site mutation at intron 1 (α and β/exon 2 boundary) of p16INK4a and p14ARF, was present in patients diagnosed with malignant melanoma, dysplastic naevi, neurofibroma, osteochondroma, and/or breast cancer. This mutation appears to lead to aberrantly spliced transcripts of p16INK4a and p14ARF, both of which lack exon 2, and is predicted to result in the expression of two severely defective proteins.
Based on these findings and other studies, we suggest that mutations affecting p16INK4a, p14ARF, or both may predispose to a portion of inherited breast cancer in melanoma prone kindreds but are not a common event in other families in which breast cancer is predominant.
Candidates for genetic evaluation had previously tested negative for a mutation in BRCA1 and BRCA2, and fell within one of the following categories: (1) affected with breast cancer, with a family history of at least one first degree relative with breast and/or ovarian cancer (any age); (2) affected with ovarian cancer, with a family history of at least one first degree relative with breast and/or ovarian cancer (any age); (3) affected with breast and ovarian cancer, with a family history of at least one first degree relatives with breast and/or ovarian cancer (any age); and (4) affected with breast and malignant melanoma. Diagnoses of disease were confirmed with pathology reports or death certificates whenever possible.
Preparation of genomic DNA
Single strand conformation polymorphism analysis of INK4a/ARF
SSCP analysis was performed as previously described.32 The DNA sequences of the primers used for INK4a, ARF, and p15 amplification and sequencing were as previously described.34–37 Any SSCP variant detected was verified by sequencing. Briefly, a fresh aliquot of DNA was amplified by PCR and the product separated from primers using Wizard resin (Promega) according to the manufacturer’s specifications. The purified DNA was subjected to cycle sequencing using an automated fluorescence based cycle sequencer (Model 377A Automated Sequencer, Applied Biosystems) and taq dye terminator chemistry. Sequencing primers were the same as those used to amplify the template.
Transformation of cryopreserved lymphocytes
Cryopreserved leucocytes from the proband in family 543 (11860) and her daughter (11079) were immortalised with Epstein-Barr virus using standard methods. The lymphoblastoid lines were maintained in RPMI 1640 medium supplemented with 15% heat inactivated FBS. Cultures were re-fed twice each week by withdrawing approximately one half of the medium aseptically and replacing it with warm (37°C) medium. HIO-118 cells, a previously established lymphoblastoid cell line from a disease free person, were used as the wild type control in the RT-PCR studies.
RT-PCR analysis of INK4a/ARF expression
RNA was isolated from EBV immortalised lymphoblastoid lines by standard methods.38 RT-PCR was performed using Qiagen OneStep RT-PCR kit (Qiagen Inc, Valencia, CA); 500 ng of total RNA was reverse transcribed at 50°C for 35 minutes. INK4a and ARF transcripts were amplified with the following primers: 5′-CCC AAC GCA CCG AAT AGT-3′ (p16INK4a sense) and 5′-GGT AGT GGG GGA AGG CTT AT-3′ (p16INK4a antisense) or 5′-GGA GGC GGC GAG AAC AT-3′ (p14ARF sense) and 5′-CGA AAG CGG GGT GGG TTG T-3′ (p14ARF antisense) using the following conditions: 15 minutes at 95°C, followed by 40 cycles of denaturation at 94°C for 45 seconds, annealing at 58°C (for p16INK4a) or 60°C (for p14ARF) for 45 seconds, and an extension step of 72°C for 45 seconds, and a final extension at 72°C for 10 minutes. RT-PCR products were resolved on a 1% agarose gel and the bands were excised and purified using the QIAquick gel extraction kit (Qiagen Inc) before sequencing.
Southern blot analysis
Southern blot analysis was performed using standard methods. Five μg of genomic DNA were digested with EcoRI and separated and blotted as we have previously described.39 The p16INK4a/p14ARF probe was derived by PCR and included exons 1α, 1β, 2, 3, and flanking intronic sequences.
Analysis of the INK4a/ARF gene locus in BRCA1/BRCA2 negative early onset breast carcinoma kindreds
Subjects belonging to early onset breast kindreds were collected through the Dyson Foundation/Family Risk Assessment Program at the Fox Chase Cancer Center in Philadelphia. Thirty-one subjects affected with breast and/or ovarian cancer, with a family history of breast cancer, were analysed for mutations in p16INK4a and p14ARF by SSCP and sequencing analyses. All participants had previously tested negative for deleterious mutations in BRCA1 and BRCA2 by full gene analysis through the Clinical Molecular Genetics Laboratory at Fox Chase, yet all reported a significant family history of disease (table 1A). Six PCR primer pairs covering the 5′UTR and the coding regions of 1α, 1β, exon 2, and exon 3 including adjacent splice junction sites34,35,37 were used to screen DNA for germline mutations in p16INK4a and p14ARF by SSCP analysis. As a result of our screen, we identified one novel sequence variant and several recurrent changes. The recurrent changes had previously been reported to be polymorphisms.16 The other alteration, a single nucleotide transversion, G to C, at position −1 of intron 1 (fig 1) is predicted to interfere with the splice site and thus affect the coding region of both p16INK4a and p14ARF. In addition to the analysis of the regions listed above, exon 2 of p15 was evaluated and no sequence alterations were detected (data not shown).
We were struck by the observation that several breast cancer prone families reported a family history of other diseases commonly associated with p16INK4a mutations such as melanoma and pancreatic cancer (table 1B). Eight of these families lacking mutations in p16INK4a/p14ARF and p15 as determined by sequence analysis were evaluated for large germline deletions encompassing these genes by Southern blotting. No additional mutations were found (data not shown).
INK4a/ARF gene mutation in a melanoma-neurofibroma-breast carcinoma kindred
A single mutation (IVS1-1G>C) was detected from our screen of 31 kindreds. The proband of family 543 reported a paternal family history of cancer that included six breast cancers all diagnosed before the age of 50, three melanomas, and one pancreatic cancer (fig 2). In addition, several benign tumours (neurofibroma and osteochondroma) and preneoplastic lesions (dysplastic naevi) were reported. Further testing showed that the mutation was present in the proband’s daughter (diagnosed with osteochondroma and dysplastic naevi at the age of 22), her two sons, one with malignant melanoma (aged 21) and the other with neurofibroma (age unknown), and her sister with bilateral breast cancer (aged 26), but was not present in her unaffected husband (fig 2).
Population frequency of INK4a/ARF substitution
To show that this missense mutation was unlikely to be a benign polymorphism, a population frequency analysis was performed using unrelated subjects. All DNA samples were from people of similar race to members of family 543. Constitutional DNA from these subjects was genotyped by rapidly sequencing the site of the mutation using pyrosequencing (Pyrosequencing Inc). The IVS1-1G>C mutation was not identified in the 200 unrelated subjects tested. Furthermore, this alteration has not been reported previously in any other mutational analysis of the p16INK4a/p14ARF locus.16 Mutational analysis of the p16INK4a/p14ARF locus has been extensive; for p16INK4a at least 280 sequence variants have been detected (19 nonsense and 152 missense alterations, of which at least 11 are found extensively in the general population and are deemed polymorphisms, and 109 deletions, insertions or frameshift alterations). For p14ARF at least 101 sequence variants have been detected (76 missense, two nonsense alterations, and 23 deletions, insertions, or frameshift alterations). These data indicate that the G to C sequence alteration is rare in the normal population and, therefore, it is likely to be either an uncommon polymorphism or a cancer susceptibility germline mutation.
Functional analysis of a novel INK4a/ARF germline mutation
Because the nucleotide substitution occurs at a highly conserved base in the 3′ splice junction site of intron 1 (both α and β), it was likely that aberrantly spliced p16INK4a and p14ARF transcripts would be expressed. Owing to the fact that we had very limited amounts of leucocytes from members of family 543, we derived EBV immortalised cell lines from the leucocytes of the proband (11860) and her daughter (11079), in order to provide adequate cell numbers for extraction of RNA. Reverse transcriptase-PCR (RT-PCR) of RNA isolated from the immortalised cell lines derived from the leucocytes of the proband (11860), her daughter (11079), and a control (HIO-118) identified unique p16INK4a and p14ARF transcripts in the mutant carriers 11860 and 11079 that were not present in the control (fig 3A). Mutant and wild type bands were gel purified and sequenced. Sequence analysis of the novel 231 bp p16INK4a transcript and the 325 bp p14ARF transcript from both 11860 and 11079 showed that the entire coding region of exon 2 is removed as a result of this mutation (fig 3B and data not shown). This mutation is predicted to result in a frameshift in p16INK4a protein translation and the expression of a novel protein of 89 amino acids, extending exon 3 by 39 amino acids. The mutation detected should also lead to the expression of a truncated p14ARF protein. This mutation is predicted to result in a frameshift in p14ARF protein translation and the expression of a novel protein of 68 amino acids, including the four amino acids encoded by exon 3 in wild type p16INK4a (that is, DIPD) (fig 3C, D). Surprisingly, we were repeatedly unable to detect the wild type p14ARF transcript from the proband (11860) (fig 3A), suggesting that the wild type allele may have become inactivated during culturing and immortalisation. Therefore, we isolated DNA from the cell line and analysed it for aberrant promoter region hypermethylation and LOH. However, using methylation specific PCR analysis we found that the promoter region was unmethylated, and sequence analysis showed that the IVS1-1G>C mutation was still heterozygous (data not shown). It is still possible that the wild type allele is inactivated by other mechanisms such as decreased mRNA stability through an acquired mutation. Sequencing of an additional 538 bp band (fig 3A) showed that this was unrelated to p16INK4a and p14ARF and must be the result of a PCR artefact (data not shown). Overall, our results suggest that mutations affecting p16INK4a, p14ARF, or both may predispose to a portion of inherited breast cancer in melanoma prone kindreds but are not a common event in other families in which breast cancer is predominant.
We have analysed the tumour suppressor genes, p16INK4a/p14ARF, as candidates for involvement in familial breast cancer syndrome by evaluating 31 families with familial breast cancer which have tested negative for mutations in BRCA1 or BRCA2. Only one mutation in our analyses of 31 breast and breast/melanoma prone families was detected, thus excluding p16INK4a/p14ARF as significantly important in the development of the majority of breast cancer syndrome families not attributed to BRCA1 or BRCA2.
We did identify a mutation in one family, which was of particular interest because, in addition to the six breast cancers all diagnosed before the age of 50, there were three cases of melanomas and one case of pancreatic cancer. Several benign tumours (neurofibroma and osteochondroma) and pre-neoplastic lesions (dysplastic naevi) were also reported. This mutation, a splice site mutation, affects both p16INK4a and p14ARF, and appears to result in transcripts which lack exon 2 of p16INK4a and p14ARF. Therefore both p16INK4a and p14ARF would be functionally inactive, and it is not clear whether the mutation is selected because of the inactivation of p16INK4a or p14ARF or both. The mutation was identified in the proband, who was afflicted with breast cancer aged 38, melanoma aged 43, and neurofibroma, the proband’s daughter (diagnosed with osteochondroma and dysplastic naevi at the age of 22), her two sons, one with malignant melanoma (aged 21) and the other with neurofibroma (aged unknown), and the proband’s sister with bilateral breast cancer (aged 26), but was not present in her unaffected husband. Although these findings suggest that the mutation may be involved in the predisposition to breast cancer in addition to melanoma, it is still possible that the breast and melanoma are unlinked in this family, especially as only the proband presented with both melanoma and breast cancer. Ideally, our studies would benefit from the study of tumour tissue from family 543 for loss of heterozygosity at the INK4A/ARF locus; however, no archived tissue could be located. Mutations in other genes, as discussed below, may therefore still be important in the development of breast carcinomas in this family.
The mutation detected in family 543 is a novel single nucleotide transversion, G to C, at the −1 position of intron 1, which is predicted to interfere with the splice site and thus affect the coding region of both p16INK4a and p14ARF. RT-PCR analysis showed novel p16INK4a and p14ARF transcripts from mutation positive members of the family, and sequence analysis showed that the entire coding region of exon 2 appears to be removed as a result of this mutation. However, we were unable to show definitively that the skipped exon fragment came from the primary transcript carrying the mutation, owing to the lack of a suitable single allele marker. RT-PCR analysis has been known to detect minor, normally occurring skipped exon splice products in an irreproducible manner. Thus, it is possible that the failure to see the aberrantly spliced p16INK4a and p14ARF transcripts in the control samples could be a coincidental artefact. Although we would predict that the mutation would result in an aberrant splice product, this product may not be stable, or a cryptic splice acceptor could be used rather than skipping exon 2.
Other studies, in addition to our own, have suggested that there is likely to be an increased risk of breast cancer in melanoma prone families, and that thorough mutational analyses of both p16INK4A and p14ARF are important in these families.25,26 Epidemiological studies have suggested that there is an interaction between breast and melanoma genotypes.40,41 Borg et al25 and Ghiorzo et al26 have reported statistically significant increases in the incidence of breast cancer occurring in CDKN2A mutation positive melanoma families. Interestingly, it has been suggested that there is an increased risk of other cancers in melanoma families with mutations that affect both p16INK4A and p14ARF, rather than in families with p16INK4A specific mutations.20 It was found that 3/5 melanoma families with p16INK4A/p14ARF mutations had additional cancers, including breast and cervical cancers, whereas only 1/8 families with p16INK4A specific mutations had other cancers.20,42 The presence of breast cancer has been reported in several other melanoma families with CDKN2A mutations, although the significance of these cases has not been further investigated.21,27,43 When combined, these studies give a strong indication that there is an increased risk of breast cancer in melanoma prone families.
Although we did not detect mutations in other families who presented with both breast cancer and melanoma, only around 50% of melanoma kindreds show linkage to 9p21 and even then mutations in p16INK4 have not been detected in all these families.16 Family 543, afflicted with four cases of early onset melanoma, and cases of dysplastic naevi, is probably the best example of a FAMMM syndrome family in our study. The criteria for FAMMM includes the presence of melanoma, and increased number of naevi and atypical naevi. However, the naevi status in the other families has not been documented. Although family 2208 presented with as many cases of melanoma as family 543, the ages of diagnosis are unknown.
We did not detect any mutations in p16INK4A and p14ARF in our other early onset breast cancer families, and indeed it is possible that in family 543 the p16INK4a/p14ARF alteration is not linked to the breast cancer cases. Therefore, there is still a need to identify an additional BRCA gene(s). TP53, PTEN, and ATM are mutated in a proportion of familial breast cancers, as part of the hereditary syndromes Li-Fraumeni syndrome, Cowden disease, and ataxia telangiectasia (AT).44 In addition, studies suggest that p16INK4a/p14ARF may be inactivated in a proportion of breast cancers, as part of a FAMMM-breast cancer syndrome. However, the involvement of these genes in breast cancer only families and also in sporadic breast cancers is still unclear. Germline mutations in ATM have recently been identified in early onset multiple breast cancer families, not associated with AT; however, the true importance of this gene in site specific breast cancer cases remains to be fully elucidated.45 Putative tumour suppressor genes and oncogenes that are implicated in the development of sporadic breast cancer, and proteins which function in the same pathways as BRCA1 and BRCA2, continue to be isolated, but require further analysis to determine their importance in both sporadic and familial forms of breast cancer.46–49 It is also possible that we have missed mutations in BRCA1 or BRCA2, for example, large deletions and insertions which are not detected by the PCR based methods used in our analysis. The possibility that some of our probands had sporadic breast cancer also exists; many were diagnosed with breast cancer after the age of 40 and had just one relative with breast cancer.
In conclusion, there appears to be an increased risk of breast cancer in melanoma prone kindreds, owing to the inactivation of p16INK4a, p14ARF or both genes. However, we have found that germline mutations in the p16INK4a/p14ARF locus are not likely to contribute to a significant percentage of familial breast cancer in the absence of proclivity to melanoma. Therefore, there continues to be a need to identify additional breast cancer predisposing genes, which may also be important in the aetiology of sporadic breast cancer.
We would like to acknowledge Anna Naumer and Josephine Wagner-Costalas, genetic counsellors at Fox Chase, for their invaluable help with consenting subjects to participate in the Family Risk Assessment Program. We would also like to thank Christie Abbott and Kim Cattie for their technical assistance in evaluating subjects for germline mutations in the BRCA1 and BRCA2 genes as part of the services offered through the Clinical Molecular Genetic Laboratory at Fox Chase. These studies also relied on the expertise of the staff in the Biosample Repository, the DNA Sequencing, and the Tissue Culture Core Facilities at Fox Chase. This work was supported in part by the Ovarian Cancer Research Fund, Gustavus and Louise Pfeiffer Research Foundation, the Martha Rogers Charitable Trust, the Eileen Stein-Jacoby Fund, a grant from the National Institutes of Health Grants (Ovarian Cancer SPORE P50 CA83638), a grant from the United States Army Medical Research (DAMD17-01-0521), and by an appropriation from the Commonwealth of Pennsylvania.