|
|
| |||||||||||||||||||||||||||||||||||||||||||||||
Background Germ-line mutations in the CDKN2A tumor-suppressor gene (also known as p16, p16 INK4a, and MTS1) have been linked to the development of melanoma in some families with inherited melanoma. Whether mutations in CDKN2A confer a predisposition to sporadic (nonfamilial) melanoma is not known. In some patients with sporadic melanoma, one or more additional primary lesions develop, suggesting that some of these patients have an underlying genetic susceptibility to the cancer. We hypothesized that this predisposition might be due to germ-line CDKN2A mutations.
Methods We used the polymerase chain reaction, single-strand conformation polymorphism analysis, and direct DNA sequencing to identify germ-line mutations in the CDKN2A gene in patients with multiple primary melanomas who did not have family histories of the disease. A quantitative yeast two-hybrid assay was used to evaluate the functional importance of the CDKN2A variants.
Results Of 33 patients with multiple primary melanomas, 5 (15 percent; 95 percent confidence interval, 4 percent to 27 percent) had germ-line CDKN2A mutations. These included a 24-bp insertion at the 5' end of the coding sequence, three missense mutations (Arg24Pro, Met53Ile, and Ser56Ile), and a 2-bp deletion at position 307 to 308 (resulting in a truncated CDKN2A protein). In three families, CDKN2A mutations identical to those in the probands were found in other family members. In two families with mutations, we uncovered previously unknown evidence of family histories of melanoma.
Conclusions Some patients with multiple primary melanomas but without family histories of the disease have germ-line mutations of the CDKN2A gene. The presence of multiple primary melanomas may signal a genetic susceptibility to melanoma not only in the index patient but also in family members, who may benefit from melanoma-surveillance programs.
For a white person living in North America, the estimated lifetime risk of melanoma is approximately 1 in 80.4,5 If the development of a single melanoma is considered to be an independent event, it follows that the probability of a second primary lesion is again 1 in 80, or 1.25 percent. In fact, about 5 percent of patients with melanoma have one or more additional primary lesions.6,7 This higher-than-expected prevalence of multiple primary tumors may be due to excessive exposure to an environmental factor, such as ultraviolet light. Alternatively, the phenomenon may have a genetic basis. In support of this latter possibility, it is noteworthy that 8 to 24 percent of patients with more than one melanoma have family histories of melanoma.8,9,10,11,12,13 Moreover, as with familial melanoma, the initial primary lesion in patients with multiple primary melanomas develops at a relatively early age.9,10
Approximately 10 percent of patients with melanoma have family histories of the disease,14,15,16 suggesting a genetic susceptibility. In fact, approximately 20 percent of families with a predisposition to melanoma have germ-line mutations in the CDKN2A tumor-suppressor gene.17 In some of these kindreds, there is also a predisposition to pancreatic cancer.18 CDKN2A (also known as p16, p16INK4a, and MTS1) encodes a cell-cycle regulator that inhibits the activities of cdk4 and cdk6, two protein kinases that in turn phosphorylate the retinoblastoma protein.19,20,21 Since unphosphorylated retinoblastoma protein inhibits the expression of genes involved in controlling the progression of the cell cycle, the impaired function of CDKN2A may result in dysregulated cellular proliferation.
We hypothesized that patients with more than one primary melanoma, even in the absence of a family history of the disease, might have germ-line mutations of the CDKN2A gene. We studied 33 patients with two or more primary melanomas and no known family histories of melanoma or pancreatic carcinoma to determine whether they had these mutations.
Methods
Patients
We reviewed a computerized data bank and medical records at the Toronto–Sunnybrook Regional Cancer Centre to identify patients with multiple primary melanomas and no family histories of the disease. We defined multiple primary disease as either two or more biopsy-confirmed invasive melanomas or one melanoma in situ and one or more invasive melanomas. A total of 28 patients met these criteria and were available to participate in the study. Five additional patients were referred to us: three from the Hamilton Regional Cancer Centre, Hamilton, Ontario, Canada, and two from the National Institutes of Health, Bethesda, Maryland. The 33 patients were unrelated to one another, all were white, and they ranged in age from 21 to 67 years. All but two of the patients resided in the greater Toronto area. The patients were questioned in detail about a possible family history of malignant melanoma or pancreatic carcinoma, and all 33 reported no knowledge of a family history of either cancer. Written informed consent was obtained from all the patients.
Polymerase Chain Reaction and Single-Strand Conformation Polymorphism Analysis
Leukocyte DNA was extracted from peripheral-blood samples with the use of standard methods. Exons 1
and 2 of the CDKN2A gene were amplified with the use of the primers listed in Table 1. The reaction mixture for the polymerase chain reaction (PCR) consisted of 0.5 U of DNA Taq polymerase (GIBCO BRL, Gaithersburg, Md.), 1x PCR buffer (20 mM TRIS–hydrochloric acid, pH 8.4, and 50 mM potassium chloride), 1.5 mM magnesium chloride, 0.2 mM deoxynucleoside triphosphate, 1 mM forward and reverse primers, 0.05 µCi of [-32P]deoxycytidine triphosphate, and 100 ng of DNA from the patient (the PCR reactions for exon 1 also contained 10 percent dimethylsulfoxide). The samples were denatured at 95°C for 5 minutes and then subjected to 35 cycles of denaturing for 1 minute at 95°C, annealing at 58 to 61°C for 1 minute, and extension at 72°C for 2 minutes, followed by a final period of extension at 72°C for 2.5 minutes. Since the sensitivity of single-strand conformation polymorphism (SSCP) analysis in detecting mutations tends to decrease as the size of the PCR product increases, fragments greater than 300 bp were digested with the appropriate restriction endonucleases. The reactions were stopped by the addition of an equal volume of stop buffer (95 percent formamide, 20 mM EDTA, 0.05 percent bromophenol blue, and 0.05 percent xylene cyanol), and the samples were denatured at either 98°C (exon 1) or 95°C (exon 2) for five minutes and then cooled on ice. The DNA fragments were separated electrophoretically at room temperature on a 6 percent nondenaturing polyacrylamide gel (supplemented with 10 percent glycerol) at 6 W for 12 to 18 hours. After electrophoresis, the gels were dried and exposed to autoradiographic film for 8 to 48 hours.
|
All DNA samples were also examined by direct sequencing. PCR was carried out as described above with the use of M13 tagged primers ME1-F and ME1-R for exon 1 and ME2-F and ME2-R for exon 2. The PCR products were separated electrophoretically on 2 percent agarose gels and purified with the use of the Qiaex II gel extraction kit (Qiagen, Chatsworth, Calif.). Both strands of the PCR products were sequenced with the use of the appropriate set of M13 primers and analyzed with the ABI Prism 310 Genetic Analyzer (Perkin–Elmer, Foster City, Calif.). To confirm the presence of allele-specific mutations, all DNA samples displaying variant SSCP bands or heterozygous sequences or both were reamplified with the use of specific primers for exon 1 and exon 2 (exon 1, E1-F and E1-R; exon 2, E2-F and E2-R). The resulting products were ligated into the pCR-Script Amp SK (+) cloning vector (Stratagene, La Jolla, Calif.) and sequenced with the use of the ABI Prism Dye Terminator kit (Perkin–Elmer).
Yeast Two-Hybrid Assay
A yeast two-hybrid assay22 was used to evaluate the interaction between the various CDKN2A variants and cdk4. As shown in Figure 1A and Figure 1B, this assay measures the reconstitution of a functional transcriptional activator from two separate fusion proteins: CDKN2A fused to an activation domain (derived from the herpes simplex virus VP16 protein), and cdk4 fused to a DNA-binding domain (derived from the Escherichia coli LexA protein).
|
The full-length CDKN2A allelic variants generated by reverse-transcription PCR were digested with Bgl II and Not I, purified, and subsequently ligated in frame into a pVP16 cDNA vector (a gift from S. Hollenberg, Fred Hutchinson Cancer Research Center, Seattle) to yield plasmid pVP16/p16.
Wild-type cdk4 cDNA (a gift from G. Hannon and D. Beach, Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y.) was ligated in frame with E. coli LexA DNA-binding domain encoded by the pBTM116 expression vector (a gift from S. Hollenberg) to yield the plasmid pBTM116/CDK4.
The pBTM116/CDK4 plasmid was introduced into L40 Saccharomyces cerevisiae through lithium acetate transformation. Colonies were selected on Trp(-) medium, transformed with pVP16/p16 plasmids containing either the wild-type or variant CDKN2A alleles, and selected on Trp(-)/Leu(-) medium. The remaining colonies were then cultured in Trp(-)/Leu(-) medium overnight at 37°C, and a colorimetric 
-galactosidase assay22 was used to quantitate CDKN2A binding with CDK4. The assays were performed at least three times for all CDKN2A variants and in all cases were normalized to the binding activity of wild-type CDKN2A.
Results
Study Population
Table 2 shows the clinical characteristics of the patients. The mean age at the time of the first diagnosis of melanoma was 43 years. In less than half the cases (39 percent), the second melanoma was diagnosed within one year after the first (defined as synchronous). Seventy-six percent of the patients had only two melanomas removed, and 24 percent had three or more excised. In addition to multiple primary melanomas, most of the patients (91 percent) also had clinically or histologically diagnosed dysplastic nevi. The relatively young age at the time of the first diagnosis of melanoma and the presence of dysplastic nevi are typical of multiple primary melanomas.10,12
|
The primers used for PCR–SSCP analysis were selected so that the amplified DNA fragment encompassed the entire coding sequence and splice junctions of exons 1 and 2 of CDKN2A. We did not analyze exon 3, since no mutations have been identified within this 11-bp exon.23 Using SSCP analysis, we detected bands of aberrant mobility in 4 of the 33 patients; two bands were located in exon 1, and two in exon 2.
We also sequenced the genomic DNA of exons 1 and 2 of the CDKN2A gene in all 33 patients. This analysis showed that five patients (15 percent; 95 percent confidence interval, 4 percent to 27 percent) had germ-line mutations in the CDKN2A gene. Four of these patients had abnormal SSCP banding patterns (as described above), whereas the pattern in the fifth patient was indistinguishable from that of the wild-type gene. The five mutations (Figure 2A, Figure 2B, Figure 2C, Figure 2D, Figure 2E, and Figure 2F) included a 24-bp insertion at the beginning of exon 1 that resulted in the addition of eight amino acids to the amino terminal end of the CDKN2A protein, a 2-bp deletion of nucleotides 307 to 308 that resulted in a premature termination signal at codon 117, and three missense mutations: the substitution of C for G at nucleotide 71 (Arg24Pro), the substitution of C for G at nucleotide 159 (Met53Ile), and the substitution of T for G at nucleotide 167 (Ser56Ile). To confirm these findings, a second blood sample from each of the five patients was analyzed. The CDKN2A exon bearing the relevant mutation was amplified by PCR, subcloned, and sequenced. Although SSCP screening is generally considered to be a less sensitive means of detecting mutations than DNA sequencing, it is technically straightforward and has been used in a number of studies to screen for CDKN2A mutations. Although not unexpected, our observations indicate that direct DNA-sequence analysis of the CDKN2A gene gives fewer false negative results than SSCP analysis.
|
Results of the Yeast Two-Hybrid Assay
An alteration in the germ-line sequence of the CDKN2A gene may not necessarily alter the function of the corresponding protein. To help discriminate clinically relevant germ-line mutations from polymorphic variants, which do not usually affect the function of the protein, we used a yeast two-hybrid system to measure the binding of CDKN2A to one of its targets: cdk4. The yeast two-hybrid system exploits the fact that many transcriptional activators have two distinct domains: a site-specific DNA-binding domain and an activation domain. When the two domains are separated, the transcriptional activator is inactive. To evaluate the interaction between two proteins, each domain is separated and fused to one of the two proteins. If the two proteins interact, the transcriptional activator is reconstituted, resulting in the expression of a reporter gene. In our assay, wild-type or mutant forms of CDKN2A were fused to an activation domain derived from the herpes simplex virus VP16 protein, and wild-type cdk4 was fused to an E. coli LexA DNA-binding domain (Figure 1A and Figure 1B). Binding of CDKN2A to cdk4 resulted in the expression of the reporter gene, and its product was quantitated with the use of a colorimetric assay.
Four of the five CDKN2A variants had decreased binding to cdk4, with binding activity ranging from 2 percent to 11 percent, as compared with the wild-type form of CDKN2A (Figure 3). In contrast, the mutant CDKN2A protein with the insertion of eight amino acids showed normal binding to cdk4. This result is not entirely unexpected, since the eight-amino-acid insertion is situated outside the ankyrin motifs that are believed to play a part in cdk4 binding. Parry and Peters25 analyzed the binding of this same mutant CDKN2A to cdk4 and also failed to observe a reduction in binding as compared with the wild-type form of CDKN2A. Although the functional importance of this insertional mutation is unclear, it is noteworthy that the corresponding 24-bp insertion has been observed in a number of melanoma-prone kindreds throughout the world.
|
Although all the patients in our study reported that they did not have family histories of melanoma or pancreatic carcinoma, three of the mutations we detected — the eight-amino-acid insertion, the Arg24Pro mutation, and the Met53Ile mutation — have been reported in melanoma-prone families and have been found to cosegregate with the cases of melanoma.18,26,27 For our five patients with CDKN2A mutations, we therefore made special efforts to uncover evidence of family histories of melanoma and other cancers by reviewing a computerized data base of sporadic and familial cases of melanoma at the Toronto–Sunnybrook Regional Cancer Centre and by contacting family members. We found that the father of the patient with the Met53Ile mutation had died of a "melanotic sarcoma" and that additional family members had received diagnoses of melanoma, one of whom was found to have the same mutation as the index patient (Figure 4, Pedigree 1). The patient with the Ser56Ile mutation was also found to have a family history of melanoma (Figure 4, Pedigree 2). This patient had an invasive melanoma removed at the age of 36 years and the following year had a melanoma in situ resected. Although the patient was unaware of any family history of melanoma, a data-base search revealed that one of his daughters (who was 4 years old at the time he received the second diagnosis of melanoma) had had a melanoma in situ excised at the age of 18 years. She also had the Ser56Ile mutation. The patient's other two children had clinically apparent dysplastic nevi, but their CDKN2A status is unknown.
|
Discussion
Several lines of evidence demonstrate that germ-line mutations of the CDKN2A gene confer a predisposition to the development of melanoma in some kindreds with familial malignant melanoma. First, mutational analyses of the CDKN2A gene in such kindreds have revealed a large number of germ-line CDKN2A gene mutations that cosegregate with the cases of melanoma. Second, the CDKN2A germ-line mutations in these melanoma-prone families encode CDKN2A proteins that fail to bind to or inhibit the activity of a downstream target of CDKN2A, cdk4.24,28,29,30 Third, if loss of function of the CDKN2A protein confers a predisposition to malignant melanoma, it follows that tumor specimens from affected members of melanoma-prone kindreds should contain one CDKN2A allele with the original germ-line mutation, with the second allele either lost or bearing a mutation that results in the production of a dysfunctional protein. Loss of the wild-type allele has, in fact, been detected both in cell lines and in tumor specimens from patients with familial melanoma.24,31 In addition to mutations of the CDKN2A gene, alterations of other components of the signaling pathway regulated by the CDKN2A gene product may also confer a predisposition to melanoma, although the frequency or penetrance of mutations in such genes may be lower than that of mutations of CDKN2A. For example, two melanoma-prone families have been described with an identical germ-line mutation in the CDK4 gene,32 and patients with germ-line mutations of the retinoblastoma gene, who have retinoblastoma as children, have an increased incidence of melanoma as adults.33,34,35,36
Several clinical observations suggest a genetic predisposition in some patients with multiple primary melanomas. The probability that more than one melanoma will develop simply by chance is far lower than the observed frequency of multiple melanomas. Moreover, the first tumor in patients with multiple primary melanomas tends to develop at a younger age than in patients with a solitary melanoma. Many patients with multiple primary melanomas also have dysplastic nevi, which are associated with an increased risk of melanoma. The genetic abnormalities conferring a predisposition to multiple primary melanomas may parallel the genetic abnormalities in the familial form of this cancer. The frequency of germ-line mutations of CDKN2A in our patients (5 in 33 patients, or 15 percent; 95 percent confidence interval, 4 to 27 percent) was similar to that observed among kindreds in southern Ontario with two or more first- or second-degree relatives who had diagnoses of melanoma (11 in 39 patients, or 28 percent; 95 percent confidence interval, 14 to 43 percent; data not shown). Thus, a patient with multiple primary melanomas has about the same probability of carrying a germ-line mutation in the CDKN2A gene as an affected member of a melanoma-prone family.
The absence of a family history in most patients with multiple primary melanomas suggests that additional genetic or environmental factors, or both, affect the phenotypic expression of CDKN2A mutations. It is often difficult to obtain an accurate family history of cancer, even among patients who are highly motivated to provide accurate information, but we uncovered evidence of family histories in two of our patients with multiple primary melanomas. A daughter of the patient with the Ser56Ile mutation was less than five years old at the time her father received the diagnosis of a second melanoma; 14 years later, unbeknownst to her father, she received a diagnosis of melanoma in situ (Figure 4, Pedigree 2). Similarly, although the patient with the Met53Ile mutation had no knowledge of a family history of melanoma, we ultimately identified other family members with melanoma (Figure 4, Pedigree 1).
Four of the five mutations we detected encoded dysfunctional CDKN2A protein that had reduced binding to one of its natural targets, cdk4. The abnormalities detected included the Arg24Pro, Met53Ile, and Ser56Ile missense mutations, as well as a 2-bp deletion at nucleotides 307 through 308 that resulted in a frame shift and a premature termination signal at codon 117. The functional consequence of the fifth mutation (the eight-amino-acid insertion) remains obscure. Despite numerous attempts, we were unable to demonstrate that this mutant form of CDKN2A bound to cdk4 less avidly than the wild-type form. Our results are similar to those of Parry and Peters,25 who evaluated the same mutant form using a different in vitro binding assay and also found no evidence of reduced binding to cdk4. These data, coupled with the fact that the insertion lies outside the ankyrin-repeat region of CDKN2A, which is believed to mediate binding to cdk4, suggest that this mutant form is functionally normal with respect to cdk4 binding. Four melanoma-prone families with the same insertion have been reported (two in Australia, one in England, and one in the United States). In all these cases, the mutant CDKN2A allele cosegregated with the cases of melanoma. There is also evidence that in two of these families, the mutations did not arise from a common founder (Hayward N: personal communication). These observations suggest that CDKN2A mutations outside the ankyrin motifs confer a predisposition to melanoma through a mechanism that has not yet been identified. It is possible that some mutant CDKN2A proteins fail to bind to other downstream targets, such as cdk6.
We conclude that patients with multiple primary melanomas have a genetic predisposition to this cancer that resembles the genetically determined susceptibility in melanoma-prone families. Our data reinforce the importance of obtaining a detailed family history from any patient with more than one melanoma and from sources other than the patient. Furthermore, we suggest that even in the absence of a family history, other family members may benefit from ongoing clinical surveillance. Although the precise role of genetic screening in the management of melanoma is currently unclear, close relatives of patients with multiple primary melanomas may benefit from such analyses.
Supported by grants from the National Cancer Institute of Canada and the Medical Research Council of Canada.
We are indebted to Drs. A. Schuh and E. Warner for critically reviewing the manuscript, to Drs. J. Rusthoven and M. McKenzie for their help in referring patients for the study, and to the patients and their families who participated in the study.
Source Information
From the Institute of Medical Sciences (J. Monzon, D.H., N.J.L.), the Department of Medical Biophysics (L.L., H.B., D.H., N.J.L.), and the Departments of Dermatology (L.F.) and Preventive Medicine and Biostatistics (J. McLaughlin), University of Toronto, Toronto; the Genetic Epidemiology Branch, National Cancer Institute, Bethesda, Md. (A.M.G., M.A.T.); and the Division of Medical Oncology, Toronto–Sunnybrook Regional Cancer Centre, Toronto (N.J.L.).
Address reprint requests to Dr. Lassam at Toronto–Sunnybrook Regional Cancer Centre, 2075 Bayview Ave., Toronto, ON M4N 3M5, Canada.
References
| |||||||||||||||||||||||||||||||||||||||||||||||
Related Letters:
CDKN2A Mutations in Multiple Primary Melanomas
Bystryn J.-C., Hughes-Davies T.H., Lassam N. J., Hogg D., McLaughlin J.
Extract |
Full Text
N Engl J Med 1998;
339:347-348, Jul 30, 1998.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
Previous
