Cancer Genetics
-
(2014)
-
Investigation of a putative melanoma susceptibility locus at chromosome 3q29 €ran Jo €nsson b, Charlotta Enerba €ck c, Frida Appelqvist d, Rainer Tuominen a, Go b e a €iom a,* , Christian Ingvar , Johan Hansson , Veronica H o H akan Olsson a b
Department of Oncology and Pathology, Karolinska Institutet, Stockholm, Sweden; Department of Oncology, Clinical Sciences, Lund University, Lund, Sweden; c Division of Cell Biology and Dermatology, Department of Clinical and Experimental Medicine, Linko€ ping University, Linko€ ping, Sweden; d Department of Clinical Genetics, Sahlgrenska University Hospital, Go€ teborg, Sweden; e Department of Surgery, Lund University, Lund, Sweden Malignant melanoma, the most fatal form of skin cancer, is currently increasing in incidence in many populations. Approximately 10% of all cases occur in families with an inherited predisposition for melanoma. In Sweden, only a minor portion of such melanoma families carry a mutation in the known melanoma gene CDKN2A, and there is a need to identify additional melanoma susceptibility genes. In a recently performed genome-wide linkage screen, novel loci with suggestive evidence of linkage to melanoma were detected. In this study, we have further analyzed one region on chromosome 3q29. In all, 89 affected and 15 nonaffected family members from 42 melanoma-prone families were genotyped for 34 genetic markers. In a pooled linkage analysis of all 42 families, we detected significant evidence of linkage, with a maximum heterogeneity logarithm of odds (HLOD) score of 3.1 with 83% of the families contributing to the linkage score. The minimum critical region of linkage (defined by a 1LOD score support interval) maps to chromosome 3q29, spans 3.5 Mb of genomic sequence, and harbors 44 identified genes. Sequence variants within this region have previously been associated with cancer susceptibility. This study reports the presence of a putative novel melanoma susceptibility locus in the Swedish population, a finding that needs to be replicated in an independent study on other individuals with familial melanoma. Sequencing of genes in the region may identify novel melanoma-associated mutations. Keywords Candidate genes, familial cancer, linkage analysis, melanoma, susceptibility locus ª 2014 Elsevier Inc. All rights reserved.
Malignant melanoma is a malignancy showing a rapid increase in incidence among Caucasian populations. The annual increase is now over 5% and mortality is also rising (1). Approximately 10% of all melanoma cases belong to kindreds with an inherited susceptibility to melanoma. In the melanoma families from the StockholmeGotland region, less than 10% carry a mutation in the known high risk melanoma gene CDKN2A. There have also been a few reported mutations in CDK4 in other populations and some evidence of linkage to a susceptibility locus at 1p22, mainly in Australian families characterized by an early age of onset (2). Recently, a rare gene variant in the MITF gene was found to be
Received September 26, 2013; received in revised form January 14, 2014; accepted February 18, 2014. * Corresponding author. E-mail address: [email protected] 2210-7762/$ - see front matter ª 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.cancergen.2014.02.007
associated with increased risk of melanoma both in the familial setting and in the general population (3,4). This gene variant fits well with the criteria of a rare disease-related allele with intermediate penetrance. However, in the families originating from Stockholm, only 1% were found to carry this susceptibility variant (unpublished data). Thus, in the majority of families the genetic cause is still unknown. In addition to the high risk predisposing genes, there are low penetrance alleles of importance for melanoma risk. The most studied low penetrance gene is the pigmentation gene MC1R, variants of which have been associated with melanoma susceptibility and pigmentation traits in numerous populations (5,6). Moreover, sequence variants in genes such as ASIP, TYR, and IRF4 have been related to melanoma susceptibility in different populations (7e9). In order to identify novel melanoma genes of importance for melanoma susceptibility in the Swedish population, we conducted a genome-wide linkage screen on members of 32
2 melanoma-prone families using high density single nucleotide polymorphism (SNP) arrays. By using parametric linkage analysis, two chromosomal loci were detected with suggestive evidence of linkage (10). Here, we report linkage data from a fine-mapping study on the chromosome 3q28-q29 locus in Swedish melanoma-prone families.
R. Tuominen et al. between the closest SNP marker upstream and downstream with LOD scores more than one level below the peak LOD score. MERLIN and ProgenyLAB software were used to construct the most likely haplotypes shared by affected family members, using the informative microsatellite markers.
Results Materials and methods Families The study cohort consisted of 42 families from Stockholm (n €teborg (n Z 4), and Lund (n Z 10). The general Z 28), Go features regarding the families are described in Table 1. Among the Stockholm families, 24 families were included in the previously performed genome-wide linkage screen (10). Six of the families consisted of two second-degree relatives with melanoma, whereas the remaining families had two firstdegree relatives or three or more relatives with melanoma. Eight families from the genome-scan were excluded because they were considered low risk families, with only two thirddegree relatives affected with melanoma (n Z 5), multiple genotyping errors due to poor quality of DNA (n Z 2), or nonSwedish ethnicity (n Z 1). Of the genotyped family members, 89 had a melanoma diagnosis and 15 were classified as unaffected.
Genetic analysis Genomic DNA was extracted from the peripheral blood leukocytes by standard methods. In all, 29 single nucleotide variants (SNVs) and five microsatellite markers covering a region of 6.9 Mb of the chromosome 3q28-q29 region were analyzed. The region of interest was determined by including markers with LOD scores two levels below the peak LOD score in the original study. More information about the genetic markers can be found in Supplementary Table 1. The genotyping was performed at deCODE Genetics (Reykjavik, Iceland) using their in-house genotyping technique developed by Nanogen. Linkage format files were generated with the ProgenyLAB software (Progeny, South Bend, IN). Inconsistent genotypes were removed by MERLIN 1.1.2 (11), which also was used for calculating multipoint parametric linkage estimates. The linkage analyses were performed assuming an autosomal-dominant inheritance pattern using two different genetic models. One model assumed a rare disease allele with a population frequency of 0.0001, with complete penetrance for disease-gene carriers and 0.1% for noncarriers. The second model postulated a more common disease gene with an allele frequency of 0.003, with 90% penetrance for disease gene carriers and 2% for noncarriers. SNVs showing evidence of linkage disequilibrium (LD) (r2 > 0.16) were also removed, since high LD markers may inflate linkage data (12). Linkage data from the families were combined and analyzed together as well as in subgroups (divided into the original genome-scan families and the fine-mapping families, respectively), to calculate multipoint LOD scores allowing for heterogeneity (HLOD). In order to estimate the chromosomal region with the greatest probability of linkage, we defined the minimum critical region (MCR)
When analyzing the complete set of families, we detected significant evidence of linkage to 3q29 with a maximum multipoint HLOD score of 3.10. More than 80% of the families showed linkage to the locus, peaking at SNP rs1055161 (Figure 1). The size of the region of linkage was determined to be 3.5 Mb using a 1LOD support interval approach, bounded by SNPs rs3109183 and rs12487782. According to the GRCh37.3 human genome build, this region harbors 44 protein coding genes and 245 transcripts. All genes mapped to this region are listed in Supplementary Table 2. When the families included in the original genome-scan were analyzed separately, a maximum HLOD score of 2.53 at 3q29 was detected with all families linked to the region. The MCR covered 3.9 Mb, harboring some additional genes, in total 56 protein coding genes and 324 transcripts. In both analyses, the highest LOD score was obtained when we assumed a rare disease model (see Figure 1AeB). More details about the linked chromosomal regions are shown in Table 2. This table also shows that, whereas all original genome-scan families showed linkage to the identified region, only 54e62% of the additional fine-mapping families were linked, depending on the analysis model. It was possible to identify a shared haplotype among affected family members in 30 out of 42 families (Figure 1C). Parts of the haplotypes were also shared between families. Constructing three marker haplotypes (using D3S3663, D3S2748, and D3S1305), we found two common haplotypes (218/117/197 and 218/117/203) to be shared by 23% and 13% of the families, respectively.
Discussion We report here a novel locus at chromosome 3q29 with significant evidence of linkage to melanoma susceptibility in 42 melanoma-prone families originating from Sweden. In approximately 10e40% of melanoma families worldwide, germline mutations are present in the tumor suppressor gene CDKN2A. In Sweden, a founder mutation (dupArg112) accounts for most of the known melanoma-associated mutations. This mutation has been found in various parts of Sweden, but so far not elsewhere, suggesting that the Swedish population is relatively genetically homogenous. Segregating CDKN2A mutations are far more common in families with many affected family members. However, the majority of the Stockholm families consist of three or fewer affected family members, among which only 5% (R. Tuominen, personal communication) carry a mutation in CDKN2A (including low risk families with two affected relatives irrespective of their relationship). Thus, in a majority of the Stockholm families, the genetic background is still unknown. Following more stringent criteria established for familial melanoma (at least two first- or second-degree relatives with melanoma or three or more cases of melanoma), we
3q29 melanoma susceptibility locus Table 1
3
Description of the Swedish melanoma families included in the fine-mapping study
All families Gene scan families Fine map families Lund €teborg Go Stockholm
No. of genotyped unaffected family members
No. of families with DN
Mean age at diagnosis, y
No. of families
No. of affected individuals (mean/family)
No. of genotyped affected (mean/family)
42 24
103 (2.45) 55 (2.3)
89 (2.1) 50 (2.1)
15 13
NA 22
NA 46.5
10 4 4
27 (2.7) 10 (2.5) 11 (2.75)
22 (2.2) 9 (2.3) 8 (2.0)
1 0 1
NA 3 4
NA 48.9 67.1
Abbreviations: NA, not available; DN, dysplastic nevi.
included 75% of the genome-scan families in the finemapping analysis. One limitation with the study is that the families consist mainly of two case families, which are less informative for linkage analyses. However, simulation
analyses indicated that detection of a putative susceptibility gene was possible, if enough families share a common susceptibility gene (data not shown). Indeed, parametric multipoint linkage analysis revealed significant evidence of
Figure 1 Linkage and haplotype analyses of the 3q28-q29 chromosomal region. (A) Multipoint linkage analyses assuming a rare disease gene frequency of 0.0001 with full penetrance for carriers and 0.1% for noncarriers for all families combined (n Z 42), and for the genome-scan families (n Z 24) and the fine-mapping families (n Z 18) analyzed separately. (B) The same analyses as in part A but using the more common disease gene frequency of 0.003 with 90% penetrance for disease gene carriers and 2% for noncarriers. (C) The most likely haplotypes, estimated by MERLIN and ProgenyLAB, that were shared between affected individuals in each family. The same color represents the same allele.Abbreviation: NA, not available.
4
R. Tuominen et al.
Table 2 Summary of the linkage results with multipoint parametric LOD scores for all families and the genome-scan and finemapping families separately 1LOD support interval SNPs rsID, bp Study cohort All families n Z 42
Disease modela
Rare Common Genome-scan families Rare n Z 24 Common Fine-mapping families Rare n Z 18 Common
Maximum HLOD score a valueb
Peak SNP Position, bp Left-flanking SNP
Right-flanking SNP
rs1055161 rs1055161 rs3732474 rs724767 rs2550240 rs753815
rs12487782 (195600227) 3.10 rs1045075 (195968643) 2.0 rs1045075 (195968643) 2.53 3qter 1.19 3qter 0.85 3qter 0.78
194310731 194310731 194063153 193722330 195495916 193359205
rs3109183 rs1024946 rs3109183 rs2193880 D3S1314 D3S1314
(192102141) (190347523) (192102141) (190262799) (190092160) (190092160)
0.83 0.82 1.0 1.0 0.54 0.62
a
The rare disease model assumes a disease gene frequency of 0.0001, with complete penetrance for carriers and 0.1% noncarriers, whereas the common disease model uses a disease gene frequency of 0.003, with 90% penetrance for carriers and 2% for noncarriers. b The a value indicates the proportion of linked families.
linkage when all families were analyzed (83% of the families were contributing to the linkage signal). The linkage was mainly based on the families included in the original genomewide linkage scan, and therefore, this locus would need to be replicated in an independent cohort before being considered a confirmed melanoma susceptibility locus. The additional fine-mapping set of families contributed to the linkage signal but was considerably more heterogeneous than the original family set, since only 54% of the families were linked. One explanation for the increased heterogeneity among the finemapping families may be that they originate from various parts of Sweden (families were recruited from three different centers), whereas the genome-scan families were solely recruited from Stockholm. There was also a difference in age of onset between the Stockholm families, with a higher age in the fine-mapping set compared with those included in the genome scan. Whether this is just random or if this difference in age of onset could be consistent with a genetic background different from that of the original genome-scan families is difficult to elucidate given the low number of families in the fine-mapping cohort. We did not observe any difference in other clinical variables between these family groups. The 3q29 region has not previously been implicated as a susceptibility locus for hereditary melanoma. However, sequence variants near the TFRC and C3orf21 genes that lie within or in close proximity to this locus have previously been associated to pancreatic carcinoma and nonesmall-cell lung carcinoma, respectively (13,14). Furthermore, six of the 42 protein coding genes that mapped within the region of linkage showed somatic mutations in melanoma cells (three in tumor samples and three in cultured samples) according to the Catalogue of Somatic Mutations in Cancer (COSMIC) database (www.sanger.ac.uk/genetics/CGP/cosmic/). Interestingly, there is also some biological plausibility that this region may harbor a novel risk gene for hereditary melanoma, since there are several possible candidate genes in this 3.5-Mbp region. Candidate genes include the HRAS-like suppressor (HRASLS ), the hairy enhancer of split 1 (HES1), the ArfGAP with coiled-coil, ankyrin repeat and PH domains (ACAP2) and the tyrosine kinase, nonreceptor 2 (TNK2). The TNK2 gene, for example, exhibits three somatic mutations in melanoma, according to the COSMIC database. TNK2 regulates the activity of proteins involved in cell survival and proliferation by tyrosine phosphorylation, and has also been shown to enhance invasion in breast cancer cells (15). The other suggested candidates have been reported to be
mutated, not in melanoma but in other cancer types. HRASLS is somatically mutated in squamous cell carcinoma. ACAP2, which is involved in membrane trafficking by functioning as a GTPase-activating protein, shows somatic mutations in several tumor types, for example, ovarian carcinoma. HES1 carries mutations in lung and ovarian cancer tissue and encodes a protein functioning as a transcriptional repressor. Interestingly, increased NOTCH1 signaling through Hes1 has been found to play an important role in melanocyte survival by preventing initiation of apoptosis (16). The 3q29 locus and other candidate loci are currently being investigated further by massive parallel sequencing of candidate genes mapped within these chromosomal loci in order to detect novel disease-associated mutations in melanoma patients. In summary, we have fine-mapped a region on chromosome 3q29 that is likely to harbor a novel risk gene for melanoma development. Genomic variants within this region may explain some of the melanoma-susceptibility segregation in Swedish melanoma-prone families. However, an independent replication of this finding is crucial to confirm this region as a novel melanoma risk locus. Identifying the causative gene, or genes, would be invaluable for the identification of healthy family members at increased risk for disease.
Acknowledgment We are grateful to the family members belonging to the melanoma families for their participation in this project. We n for help with the clinical data. This project thank Diana Linde was supported by grants from the Swedish Research Council, the Swedish Cancer Society, the Karolinska Institutets research funds, and the Magnus Bergvall foundation.
Supplementary data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.cancergen.2014.02.007.
References 1. Swedish Cancer Registry. Cancer incidence in Sweden 2011. Stockholm, Sweden: Board of Health and Welfare. Available at,
3q29 melanoma susceptibility locus
2.
3.
4.
5.
6.
7.
8.
9.
http://www.socialstyrelsen.se/lists/artikelkatalog/attachments/ 18919/2012-12-19.pdf; 2012. Accessed: December 7, 2012. Gillanders E, Juo SH, Holland EA, et al. Localization of a novel melanoma susceptibility locus to 1p22. Am J Hum Genet 2003; 73:301e313. Bertolotto C, Lesueur F, Giuliano S, et al. A SUMOylationdefective MITF germline mutation predisposes to melanoma and renal carcinoma. Nature 2011;480:94e98. Yokoyama S, Woods SL, Boyle GM, et al. A novel recurrent mutation in MITF predisposes to familial and sporadic melanoma. Nature 2011;480:99e103. Hoiom V, Tuominen R, Kaller M, et al. MC1R variation and melanoma risk in the Swedish population in relation to clinical and pathological parameters. Pigment Cell Melanoma Res 2009; 22:196e204. Kennedy C, ter Huurne J, Berkhout M, et al. Melanocortin 1 receptor (MC1R) gene variants are associated with an increased risk for cutaneous melanoma which is largely independent of skin type and hair color. J Invest Dermatol 2001;117:294e300. Bishop DT, Demenais F, Iles MM, et al. Genome-wide association study identifies three loci associated with melanoma risk. Nat Genet 2009;41:920e925. Duffy DL, Iles MM, Glass D, et al. IRF4 variants have agespecific effects on nevus count and predispose to melanoma. Am J Hum Genet 2010;87:6e16. Gudbjartsson DF, Sulem P, Stacey SN, et al. ASIP and TYR pigmentation variants associate with cutaneous melanoma and basal cell carcinoma. Nat Genet 2008;40:886e891.
5 10. Hoiom V, Tuominen R, Hansson J. Genome-wide linkage analysis of Swedish families to identify putative susceptibility loci for cutaneous malignant melanoma. Genes Chromosomes Cancer 2011;50:1076e1084. 11. Abecasis GR, Cherny SS, Cookson WO, et al. MERLINdrapid analysis of dense genetic maps using sparse gene flow trees. Nat Genet 2002;30:97e101. 12. Boyles AL, Scott WK, Martin ER, et al. Linkage disequilibrium inflates type I error rates in multipoint linkage analysis when parental genotypes are missing. Hum Hered 2005;59: 220e227. 13. Wu C, Miao X, Huang L, et al. Genome-wide association study identifies five loci associated with susceptibility to pancreatic cancer in Chinese populations. Nat Genet 2010; 44:62e66. 14. Yoon KA, Park JH, Han J, et al. A genome-wide association study reveals susceptibility variants for non-small cell lung cancer in the Korean population. Hum Mol Genet 2010;19: 4948e4954. 15. Howlin J, Rosenkvist J, Andersson T. TNK2 preserves epidermal growth factor receptor expression on the cell surface and enhances migration and invasion of human breast cancer cells. Breast Cancer Res 2008;10: R36. 16. Moriyama M, Osawa M, Mak SS, et al. Notch signaling via Hes1 transcription factor maintains survival of melanoblasts and melanocyte stem cells. J Cell Biol 2006;173: 333e339.
-
(2014)
-
Investigation of a putative melanoma susceptibility locus at chromosome 3q29 €ran Jo €nsson b, Charlotta Enerba €ck c, Frida Appelqvist d, Rainer Tuominen a, Go b e a €iom a,* , Christian Ingvar , Johan Hansson , Veronica H o H akan Olsson a b
Department of Oncology and Pathology, Karolinska Institutet, Stockholm, Sweden; Department of Oncology, Clinical Sciences, Lund University, Lund, Sweden; c Division of Cell Biology and Dermatology, Department of Clinical and Experimental Medicine, Linko€ ping University, Linko€ ping, Sweden; d Department of Clinical Genetics, Sahlgrenska University Hospital, Go€ teborg, Sweden; e Department of Surgery, Lund University, Lund, Sweden Malignant melanoma, the most fatal form of skin cancer, is currently increasing in incidence in many populations. Approximately 10% of all cases occur in families with an inherited predisposition for melanoma. In Sweden, only a minor portion of such melanoma families carry a mutation in the known melanoma gene CDKN2A, and there is a need to identify additional melanoma susceptibility genes. In a recently performed genome-wide linkage screen, novel loci with suggestive evidence of linkage to melanoma were detected. In this study, we have further analyzed one region on chromosome 3q29. In all, 89 affected and 15 nonaffected family members from 42 melanoma-prone families were genotyped for 34 genetic markers. In a pooled linkage analysis of all 42 families, we detected significant evidence of linkage, with a maximum heterogeneity logarithm of odds (HLOD) score of 3.1 with 83% of the families contributing to the linkage score. The minimum critical region of linkage (defined by a 1LOD score support interval) maps to chromosome 3q29, spans 3.5 Mb of genomic sequence, and harbors 44 identified genes. Sequence variants within this region have previously been associated with cancer susceptibility. This study reports the presence of a putative novel melanoma susceptibility locus in the Swedish population, a finding that needs to be replicated in an independent study on other individuals with familial melanoma. Sequencing of genes in the region may identify novel melanoma-associated mutations. Keywords Candidate genes, familial cancer, linkage analysis, melanoma, susceptibility locus ª 2014 Elsevier Inc. All rights reserved.
Malignant melanoma is a malignancy showing a rapid increase in incidence among Caucasian populations. The annual increase is now over 5% and mortality is also rising (1). Approximately 10% of all melanoma cases belong to kindreds with an inherited susceptibility to melanoma. In the melanoma families from the StockholmeGotland region, less than 10% carry a mutation in the known high risk melanoma gene CDKN2A. There have also been a few reported mutations in CDK4 in other populations and some evidence of linkage to a susceptibility locus at 1p22, mainly in Australian families characterized by an early age of onset (2). Recently, a rare gene variant in the MITF gene was found to be
Received September 26, 2013; received in revised form January 14, 2014; accepted February 18, 2014. * Corresponding author. E-mail address: [email protected] 2210-7762/$ - see front matter ª 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.cancergen.2014.02.007
associated with increased risk of melanoma both in the familial setting and in the general population (3,4). This gene variant fits well with the criteria of a rare disease-related allele with intermediate penetrance. However, in the families originating from Stockholm, only 1% were found to carry this susceptibility variant (unpublished data). Thus, in the majority of families the genetic cause is still unknown. In addition to the high risk predisposing genes, there are low penetrance alleles of importance for melanoma risk. The most studied low penetrance gene is the pigmentation gene MC1R, variants of which have been associated with melanoma susceptibility and pigmentation traits in numerous populations (5,6). Moreover, sequence variants in genes such as ASIP, TYR, and IRF4 have been related to melanoma susceptibility in different populations (7e9). In order to identify novel melanoma genes of importance for melanoma susceptibility in the Swedish population, we conducted a genome-wide linkage screen on members of 32
2 melanoma-prone families using high density single nucleotide polymorphism (SNP) arrays. By using parametric linkage analysis, two chromosomal loci were detected with suggestive evidence of linkage (10). Here, we report linkage data from a fine-mapping study on the chromosome 3q28-q29 locus in Swedish melanoma-prone families.
R. Tuominen et al. between the closest SNP marker upstream and downstream with LOD scores more than one level below the peak LOD score. MERLIN and ProgenyLAB software were used to construct the most likely haplotypes shared by affected family members, using the informative microsatellite markers.
Results Materials and methods Families The study cohort consisted of 42 families from Stockholm (n €teborg (n Z 4), and Lund (n Z 10). The general Z 28), Go features regarding the families are described in Table 1. Among the Stockholm families, 24 families were included in the previously performed genome-wide linkage screen (10). Six of the families consisted of two second-degree relatives with melanoma, whereas the remaining families had two firstdegree relatives or three or more relatives with melanoma. Eight families from the genome-scan were excluded because they were considered low risk families, with only two thirddegree relatives affected with melanoma (n Z 5), multiple genotyping errors due to poor quality of DNA (n Z 2), or nonSwedish ethnicity (n Z 1). Of the genotyped family members, 89 had a melanoma diagnosis and 15 were classified as unaffected.
Genetic analysis Genomic DNA was extracted from the peripheral blood leukocytes by standard methods. In all, 29 single nucleotide variants (SNVs) and five microsatellite markers covering a region of 6.9 Mb of the chromosome 3q28-q29 region were analyzed. The region of interest was determined by including markers with LOD scores two levels below the peak LOD score in the original study. More information about the genetic markers can be found in Supplementary Table 1. The genotyping was performed at deCODE Genetics (Reykjavik, Iceland) using their in-house genotyping technique developed by Nanogen. Linkage format files were generated with the ProgenyLAB software (Progeny, South Bend, IN). Inconsistent genotypes were removed by MERLIN 1.1.2 (11), which also was used for calculating multipoint parametric linkage estimates. The linkage analyses were performed assuming an autosomal-dominant inheritance pattern using two different genetic models. One model assumed a rare disease allele with a population frequency of 0.0001, with complete penetrance for disease-gene carriers and 0.1% for noncarriers. The second model postulated a more common disease gene with an allele frequency of 0.003, with 90% penetrance for disease gene carriers and 2% for noncarriers. SNVs showing evidence of linkage disequilibrium (LD) (r2 > 0.16) were also removed, since high LD markers may inflate linkage data (12). Linkage data from the families were combined and analyzed together as well as in subgroups (divided into the original genome-scan families and the fine-mapping families, respectively), to calculate multipoint LOD scores allowing for heterogeneity (HLOD). In order to estimate the chromosomal region with the greatest probability of linkage, we defined the minimum critical region (MCR)
When analyzing the complete set of families, we detected significant evidence of linkage to 3q29 with a maximum multipoint HLOD score of 3.10. More than 80% of the families showed linkage to the locus, peaking at SNP rs1055161 (Figure 1). The size of the region of linkage was determined to be 3.5 Mb using a 1LOD support interval approach, bounded by SNPs rs3109183 and rs12487782. According to the GRCh37.3 human genome build, this region harbors 44 protein coding genes and 245 transcripts. All genes mapped to this region are listed in Supplementary Table 2. When the families included in the original genome-scan were analyzed separately, a maximum HLOD score of 2.53 at 3q29 was detected with all families linked to the region. The MCR covered 3.9 Mb, harboring some additional genes, in total 56 protein coding genes and 324 transcripts. In both analyses, the highest LOD score was obtained when we assumed a rare disease model (see Figure 1AeB). More details about the linked chromosomal regions are shown in Table 2. This table also shows that, whereas all original genome-scan families showed linkage to the identified region, only 54e62% of the additional fine-mapping families were linked, depending on the analysis model. It was possible to identify a shared haplotype among affected family members in 30 out of 42 families (Figure 1C). Parts of the haplotypes were also shared between families. Constructing three marker haplotypes (using D3S3663, D3S2748, and D3S1305), we found two common haplotypes (218/117/197 and 218/117/203) to be shared by 23% and 13% of the families, respectively.
Discussion We report here a novel locus at chromosome 3q29 with significant evidence of linkage to melanoma susceptibility in 42 melanoma-prone families originating from Sweden. In approximately 10e40% of melanoma families worldwide, germline mutations are present in the tumor suppressor gene CDKN2A. In Sweden, a founder mutation (dupArg112) accounts for most of the known melanoma-associated mutations. This mutation has been found in various parts of Sweden, but so far not elsewhere, suggesting that the Swedish population is relatively genetically homogenous. Segregating CDKN2A mutations are far more common in families with many affected family members. However, the majority of the Stockholm families consist of three or fewer affected family members, among which only 5% (R. Tuominen, personal communication) carry a mutation in CDKN2A (including low risk families with two affected relatives irrespective of their relationship). Thus, in a majority of the Stockholm families, the genetic background is still unknown. Following more stringent criteria established for familial melanoma (at least two first- or second-degree relatives with melanoma or three or more cases of melanoma), we
3q29 melanoma susceptibility locus Table 1
3
Description of the Swedish melanoma families included in the fine-mapping study
All families Gene scan families Fine map families Lund €teborg Go Stockholm
No. of genotyped unaffected family members
No. of families with DN
Mean age at diagnosis, y
No. of families
No. of affected individuals (mean/family)
No. of genotyped affected (mean/family)
42 24
103 (2.45) 55 (2.3)
89 (2.1) 50 (2.1)
15 13
NA 22
NA 46.5
10 4 4
27 (2.7) 10 (2.5) 11 (2.75)
22 (2.2) 9 (2.3) 8 (2.0)
1 0 1
NA 3 4
NA 48.9 67.1
Abbreviations: NA, not available; DN, dysplastic nevi.
included 75% of the genome-scan families in the finemapping analysis. One limitation with the study is that the families consist mainly of two case families, which are less informative for linkage analyses. However, simulation
analyses indicated that detection of a putative susceptibility gene was possible, if enough families share a common susceptibility gene (data not shown). Indeed, parametric multipoint linkage analysis revealed significant evidence of
Figure 1 Linkage and haplotype analyses of the 3q28-q29 chromosomal region. (A) Multipoint linkage analyses assuming a rare disease gene frequency of 0.0001 with full penetrance for carriers and 0.1% for noncarriers for all families combined (n Z 42), and for the genome-scan families (n Z 24) and the fine-mapping families (n Z 18) analyzed separately. (B) The same analyses as in part A but using the more common disease gene frequency of 0.003 with 90% penetrance for disease gene carriers and 2% for noncarriers. (C) The most likely haplotypes, estimated by MERLIN and ProgenyLAB, that were shared between affected individuals in each family. The same color represents the same allele.Abbreviation: NA, not available.
4
R. Tuominen et al.
Table 2 Summary of the linkage results with multipoint parametric LOD scores for all families and the genome-scan and finemapping families separately 1LOD support interval SNPs rsID, bp Study cohort All families n Z 42
Disease modela
Rare Common Genome-scan families Rare n Z 24 Common Fine-mapping families Rare n Z 18 Common
Maximum HLOD score a valueb
Peak SNP Position, bp Left-flanking SNP
Right-flanking SNP
rs1055161 rs1055161 rs3732474 rs724767 rs2550240 rs753815
rs12487782 (195600227) 3.10 rs1045075 (195968643) 2.0 rs1045075 (195968643) 2.53 3qter 1.19 3qter 0.85 3qter 0.78
194310731 194310731 194063153 193722330 195495916 193359205
rs3109183 rs1024946 rs3109183 rs2193880 D3S1314 D3S1314
(192102141) (190347523) (192102141) (190262799) (190092160) (190092160)
0.83 0.82 1.0 1.0 0.54 0.62
a
The rare disease model assumes a disease gene frequency of 0.0001, with complete penetrance for carriers and 0.1% noncarriers, whereas the common disease model uses a disease gene frequency of 0.003, with 90% penetrance for carriers and 2% for noncarriers. b The a value indicates the proportion of linked families.
linkage when all families were analyzed (83% of the families were contributing to the linkage signal). The linkage was mainly based on the families included in the original genomewide linkage scan, and therefore, this locus would need to be replicated in an independent cohort before being considered a confirmed melanoma susceptibility locus. The additional fine-mapping set of families contributed to the linkage signal but was considerably more heterogeneous than the original family set, since only 54% of the families were linked. One explanation for the increased heterogeneity among the finemapping families may be that they originate from various parts of Sweden (families were recruited from three different centers), whereas the genome-scan families were solely recruited from Stockholm. There was also a difference in age of onset between the Stockholm families, with a higher age in the fine-mapping set compared with those included in the genome scan. Whether this is just random or if this difference in age of onset could be consistent with a genetic background different from that of the original genome-scan families is difficult to elucidate given the low number of families in the fine-mapping cohort. We did not observe any difference in other clinical variables between these family groups. The 3q29 region has not previously been implicated as a susceptibility locus for hereditary melanoma. However, sequence variants near the TFRC and C3orf21 genes that lie within or in close proximity to this locus have previously been associated to pancreatic carcinoma and nonesmall-cell lung carcinoma, respectively (13,14). Furthermore, six of the 42 protein coding genes that mapped within the region of linkage showed somatic mutations in melanoma cells (three in tumor samples and three in cultured samples) according to the Catalogue of Somatic Mutations in Cancer (COSMIC) database (www.sanger.ac.uk/genetics/CGP/cosmic/). Interestingly, there is also some biological plausibility that this region may harbor a novel risk gene for hereditary melanoma, since there are several possible candidate genes in this 3.5-Mbp region. Candidate genes include the HRAS-like suppressor (HRASLS ), the hairy enhancer of split 1 (HES1), the ArfGAP with coiled-coil, ankyrin repeat and PH domains (ACAP2) and the tyrosine kinase, nonreceptor 2 (TNK2). The TNK2 gene, for example, exhibits three somatic mutations in melanoma, according to the COSMIC database. TNK2 regulates the activity of proteins involved in cell survival and proliferation by tyrosine phosphorylation, and has also been shown to enhance invasion in breast cancer cells (15). The other suggested candidates have been reported to be
mutated, not in melanoma but in other cancer types. HRASLS is somatically mutated in squamous cell carcinoma. ACAP2, which is involved in membrane trafficking by functioning as a GTPase-activating protein, shows somatic mutations in several tumor types, for example, ovarian carcinoma. HES1 carries mutations in lung and ovarian cancer tissue and encodes a protein functioning as a transcriptional repressor. Interestingly, increased NOTCH1 signaling through Hes1 has been found to play an important role in melanocyte survival by preventing initiation of apoptosis (16). The 3q29 locus and other candidate loci are currently being investigated further by massive parallel sequencing of candidate genes mapped within these chromosomal loci in order to detect novel disease-associated mutations in melanoma patients. In summary, we have fine-mapped a region on chromosome 3q29 that is likely to harbor a novel risk gene for melanoma development. Genomic variants within this region may explain some of the melanoma-susceptibility segregation in Swedish melanoma-prone families. However, an independent replication of this finding is crucial to confirm this region as a novel melanoma risk locus. Identifying the causative gene, or genes, would be invaluable for the identification of healthy family members at increased risk for disease.
Acknowledgment We are grateful to the family members belonging to the melanoma families for their participation in this project. We n for help with the clinical data. This project thank Diana Linde was supported by grants from the Swedish Research Council, the Swedish Cancer Society, the Karolinska Institutets research funds, and the Magnus Bergvall foundation.
Supplementary data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.cancergen.2014.02.007.
References 1. Swedish Cancer Registry. Cancer incidence in Sweden 2011. Stockholm, Sweden: Board of Health and Welfare. Available at,
3q29 melanoma susceptibility locus
2.
3.
4.
5.
6.
7.
8.
9.
http://www.socialstyrelsen.se/lists/artikelkatalog/attachments/ 18919/2012-12-19.pdf; 2012. Accessed: December 7, 2012. Gillanders E, Juo SH, Holland EA, et al. Localization of a novel melanoma susceptibility locus to 1p22. Am J Hum Genet 2003; 73:301e313. Bertolotto C, Lesueur F, Giuliano S, et al. A SUMOylationdefective MITF germline mutation predisposes to melanoma and renal carcinoma. Nature 2011;480:94e98. Yokoyama S, Woods SL, Boyle GM, et al. A novel recurrent mutation in MITF predisposes to familial and sporadic melanoma. Nature 2011;480:99e103. Hoiom V, Tuominen R, Kaller M, et al. MC1R variation and melanoma risk in the Swedish population in relation to clinical and pathological parameters. Pigment Cell Melanoma Res 2009; 22:196e204. Kennedy C, ter Huurne J, Berkhout M, et al. Melanocortin 1 receptor (MC1R) gene variants are associated with an increased risk for cutaneous melanoma which is largely independent of skin type and hair color. J Invest Dermatol 2001;117:294e300. Bishop DT, Demenais F, Iles MM, et al. Genome-wide association study identifies three loci associated with melanoma risk. Nat Genet 2009;41:920e925. Duffy DL, Iles MM, Glass D, et al. IRF4 variants have agespecific effects on nevus count and predispose to melanoma. Am J Hum Genet 2010;87:6e16. Gudbjartsson DF, Sulem P, Stacey SN, et al. ASIP and TYR pigmentation variants associate with cutaneous melanoma and basal cell carcinoma. Nat Genet 2008;40:886e891.
5 10. Hoiom V, Tuominen R, Hansson J. Genome-wide linkage analysis of Swedish families to identify putative susceptibility loci for cutaneous malignant melanoma. Genes Chromosomes Cancer 2011;50:1076e1084. 11. Abecasis GR, Cherny SS, Cookson WO, et al. MERLINdrapid analysis of dense genetic maps using sparse gene flow trees. Nat Genet 2002;30:97e101. 12. Boyles AL, Scott WK, Martin ER, et al. Linkage disequilibrium inflates type I error rates in multipoint linkage analysis when parental genotypes are missing. Hum Hered 2005;59: 220e227. 13. Wu C, Miao X, Huang L, et al. Genome-wide association study identifies five loci associated with susceptibility to pancreatic cancer in Chinese populations. Nat Genet 2010; 44:62e66. 14. Yoon KA, Park JH, Han J, et al. A genome-wide association study reveals susceptibility variants for non-small cell lung cancer in the Korean population. Hum Mol Genet 2010;19: 4948e4954. 15. Howlin J, Rosenkvist J, Andersson T. TNK2 preserves epidermal growth factor receptor expression on the cell surface and enhances migration and invasion of human breast cancer cells. Breast Cancer Res 2008;10: R36. 16. Moriyama M, Osawa M, Mak SS, et al. Notch signaling via Hes1 transcription factor maintains survival of melanoblasts and melanocyte stem cells. J Cell Biol 2006;173: 333e339.
