Germline Variants in the POT1-Gene in High-Risk Melanoma Patients in Austria

Risk of melanoma is in part determined by genetic factors. Currently the only established high penetrance familial melanoma genes are CDKN2A and CDK4. Recent studies reported germline variants in POT1 in melanoma families. In the present study, we sequenced the entire POT1 gene in 694 patients from the M3-study. Patients with multiple primary melanomas (n = 163) or with a positive family history (n = 133) were classified as high-risk melanoma patients. Additionally, 200 single primary melanoma patients and 198 non-melanoma controls were sequenced. For prediction analysis 10 different tools were used. In total 53 different variants were found, of which 8 were detected in high-risk melanoma patients, only. Two out of these 8 variants were located in exons and were non-synonymous: g.124510982 G>A (p.R80C) and g.124491977 T>G (p.N300H). While g.124491977 T>G was predicted to be neutral, 80% of the prediction tools classified g.124510982 G>A as deleterious. The variant, g.124467236 T>C, which possibly causes a change in the splice site was identified in a case with a positive family history in the present study. Another variant in the 5-UTR, g.124537261 A>G, was found in 2 high-risk patients. So, in conclusion, melanoma associated POT1 germline variants seem to be rare. Further studies are required to evaluate the role of POT1 for genetic counseling.

Approximately 10-15% of all melanoma patients report a positive family history, multiple primary melanomas or early onset of melanoma diagnosis (Müller et al. 2016). The most important high penetrance gene is the cyclin-dependent kinase Inhibitor 2A (CDKN2A), responsible for about 30% of all familial melanoma cases. Melanoma associated mutations in cyclin-dependent kinase 4 (CDK4), which were also classified as high penetrance mutations, seem to be very rare as only a few families were reported since the initial report in 1996 (ZUO et al. 1996). Only recently, a mutation in the telomerase reverse transcriptase gene (TERT) was described in melanoma patients, adding further data to the already existing evidence that stability of telomeres is important in melanoma biology.
Shelterin, a protein complex composed of six subunits, is involved in the protection of the chromosome ends and in the regulation of the telomerase activity (AOUDE et al. 2015). Recently this complex gained particular interest in melanoma genetics as germline variants were found in 3 shelterin genes in melanoma prone families (ROBLES-ESPINOZA et al. 2014;SHI et al. 2014;AOUDE et al. 2015): POT1, ACD and TERF2IP. The human POT1 gene is located at 7q31.33 and has 19 transcripts. The isoform 1 of the protein, where the variants were originally found, consists of 19 exons and of 634 amino acids. Since the initial description of POT1 as a predisposition gene for hereditary melanoma (ROBLES-ESPINOZA et al. 2014;SHI et al. 2014), no further variants associated with melanoma has been described except for one in a single melanoma prone family in the U.S.A. (WILSON et al. 2017). Therefore, the frequency of these variants in other populations remains unclear. This information is crucial to decide whether high-risk patients should be tested for POT1 in a routine genetic counseling of melanoma families (GOLDSTEIN et al. 2007). Here we present for the first time data of POT1 variants in high-risk melanoma patients in Austria.

Study participants
In total, DNA of 694 participants was analyzed. All participants were Caucasians with European ancestry and were recruited in Austria as described elsewhere (BURGSTALLER-MUEHLBACHER et al. 2015). High-risk melanoma patients (n = 296), included patients with multiple primary melanomas (n = 163) and patients with a positive family history (n = 133) and were compared to a reference group of single melanoma (n = 200) and non melanoma patients (n = 198). Descriptive data were shown for gender, age at diagnosis, Breslow index, tumor localization and histological subtype in Table 1. In multiple primary melanoma patients, data (date of surgery, localization, histological description such as histological subtype and Breslow index) refers to the first primary melanoma. Informed consent was obtained from all individual participants included in the study. The study was approved by the ethics committee of the Medical University of Vienna.

Genotyping
The DNA was purified from whole blood as described previously (BURGSTALLER-MUEHLBACHER et al. 2015). Next generation-sequencing of POT1 was performed at the Genome Centre, Queen Mary, University of London (http://www.smd.qmul.ac.uk/gc/). For the preparation of DNA libraries 0.5 mg of genomic DNA was used. Amplicon libraries were created with the Fluidigm Access Array according to the manufacturer's protocol. The 150-bp paired-end sequencing was done on the Illumina MiSeq v2 platform.
The datasets generated during the current study are available in the NCBI Sequence Read Archive (https://www.ncbi.nlm.nih.gov/sra).

Prediction analysis of non-synonymous POT1 variants
Two non-synonymous POT1 variants, found in high risk melanoma patients only, were analyzed using 10 prediction tools as described previously ( For the latter 2, the tables of the UCSC genome browser tb_allHg19RS_BW and phyloP46wayPlacental were used. Most of those tools provide information about the effect of an amino acid exchange on the protein function. GERP++ and phyloP give a score depending on the conservation by comparing different species. The cut-off score for PROVEAN was -2.5, values below indicate the prediction as deleterious. In SIFT, values have a range from 0 to 1, whereas a score below 0.05 means that the variant is predicted to n be deleterious. In CADD values above 15 were classified as deleterious. The range Polyphen2 scores is from values of 0 to 1; higher scores are more likely to be found in deleterious variants with a cut-off score of 0.5. SNAP 2 has output scores between -100 (strong neutral prediction) to 100 (strong effect prediction). PANTHER calculates the preservation time to give a prediction. Longer times indicate a more likely functional impact.
As protein sequence for the data input, the POT1 isoform 1 (ENST00000357628) was used.

Data availability
All raw sequencing data are deposited in the NCBI Sequence Read Archiv (SRA) under the BioProject ID PRJNA400454.
-  Table 2); 48 were detected in melanoma patients exclusively and 5 additional variants in the control group only. Out of 53 variants, 27 were located in introns, 7 in the 59 untranslated region (UTR), 10 in the 39 UTR and 9 in exons (see Table 2). Of the latter, 8 resulted in an amino acid exchange and 1 was synonymous. Three non-synonymous variants were located at exon 9, 2 at exon 11 and 1 at exon 7, 14 and 17, respectively. The most common variants in the exons were p.G404V (21 participants), followed by p. D185E (4 participants) and p.V183G (3 participants). All 3 variants were found in cases as well as controls and were listed in the dbSNP (SHERRY et al. 2001).

POT1 variants in high-risk patients
Eight variants were exclusively found in high-risk melanoma patients (see Table 3). Four of these 8 variants were not listed in the dbSNP. Of all variants detected in high-risk melanoma patients exclusively (n = 8), 2 were located in the 59UTR, 3 in the 39 UTR, 1 in an intron and 2 in exons; 1 in exon 7 and another in exon 11. The latter 2 (g.124510982 G.A and g.124491977 T.G) were both found in one multiple primary melanoma patient each. The carrier of g.124510982 G.A, was a male patient, diagnosed with an amelanotic melanoma at the age of 33 with a second melanoma excised 35 years later and was tested wild type for CDKN2A. The carrier of the other non-synonymous variant, g.124491977 T.G, was 57 years old when his first primary melanoma was excised. Nine years later, an in-situ melanoma was found on his back.
Of the variants listed in the public SNP databases, g.124467236 T.C, which was described in a patient with multiple primary melanomas before (SHI et al. 2014), was found in our study in a female patient with a positive family history. She was diagnosed at the age of 22 while her mother had her diagnosis at the age of 40 (which could be confirmed by medical records), conforming with the criteria for inherited risk of melanoma.
The variant g.124537261 A.G located in the 59UTR, was the only one found in 2 high-risk patients. Both were diagnosed for melanoma before the age of 50 and tested wild type for CDKN2A mutations. One had a positive family history for melanoma and the other patient was diagnosed with 4 primary melanomas.
Two variants located in the 39UTR, g.124463018 C.T and g.124463400 T.C, were found in early onset patients with multiple primary melanomas each, both tested wild type for CDKN2A mutations.

Prediction analysis of non-synonymous POT1 variants
Prediction analysis was performed for non-synonymous variants in coding sequences which were only found in high-risk melanoma patients: g.124491977 T.G and g.124510982 G.A, respectively. While g.124491977 T.G was predicted to be neutral by all of the used prediction tools, the variant g.124510982 G.A was predicted to be deleterious by 8 of 10 prediction tools (80%). Results of all prediction analyses are shown in Table 4.

Coincidence of CDKN2A mutations
To exclude coincidence with CDKN2A mutations, we then examined the CDKN2A sequence of our cases carrying potential risk variants of POT1. One of the variants found exclusively in high-risk patients, n  .1 PM: patients with multiple primary melanomas, wt: wild type.

DISCUSSION
Only recently, novel disease associated germline variants in POT1 were reported in melanoma pedigrees (ROBLES-ESPINOZA et al. 2014;SHI et al. 2014). This finding is of particular interest as the established disease causing mutations in familial melanoma, i.e., mutations in CDKN2A and CDK4 account only for 30-40% of the melanoma pedigrees. Despite this, just one family with a POT1 germline variant associated with melanoma was published so far (WILSON et al. 2017).
In the present study, in which the entire POT1 gene was sequenced in cases at high risk of melanoma and in control patients, a total of 53 variants were found. Despite this, previously published POT1 variants described in melanoma pedigrees (ROBLES-ESPINOZA et al. 2014;SHI et al. 2014) were not detected in our study. However, we found the intronic variant, g.124467236 T.C, in a patient with a positive family history of melanoma which was described in a patient with multiple primary melanomas carrying the variant previously (SHI et al. 2014). The region of the variant g.124467236 T.C is highly conserved and according to in silico analyses, this variant possibly causes a change in the splice site. Taken together, this finding supports the idea that this variant is associated with melanoma (SHI et al. 2014). Our case with the g.124467236 T.C germline variant in POT1 harbored a non-effective variant in CDKN2A at the position g. 21971211 G.C (c.151-4 G.GC) (BURGSTALLER-MUEHLBACHER et al. 2015). As described previously, no effect on splicing could be confirmed when the transcript was analyzed (BURGSTALLER-MUEHLBACHER et al. 2015).
Of the 53 genetic variants found, 8 were exclusive in high-risk melanoma patients. Two of them, g.124491977 T.G and g.124510982 G.A, both non-synonymous variants, were tested for their alleged functionality. While g.124491977 T.G was predicted to be neutral by all 10 tools, g.124510982 G.A was predicted to be damaging by 80% of the prediction tools and is therefore very likely to be biologically functional. Comparing the wild type amino acid arginine with the resulting cysteine, there are differences in some amino acid features. The mutant residue is smaller and charged neutral, compared to the negatively charged wild type amino acid. Consequently, the correct folding of the protein could be influenced due to the more hydrophobic nature of the resulting amino acid (VENSELAAR et al. 2010).
One potential limitation of this study is the fact that family history was largely reported and histopathologic reports confirming the diagnosis of relatives were not available for all cases. In the current study, the potential effect of the variants was assessed by computational analyses. Naturally, functional analyses are required to determine the exact role of these variants in melanoma development.
In conclusion, melanoma driving POT1 germline variants might be rare. However, further studies are required to assemble comprehensive information on the frequency and the role of POT1 in familial melanoma. It is also important to note that germline variants in POT1 were reported to be associated with other types of cancer such as colorectal cancer ), glioma (BAINBRIDGE et al. 2015 and chronic lymphatic lymphoma (CALVETE et al. 2015;KARAMI et al. 2016;SPEEDY et al. 2016). As none of the variants described were found in melanoma cases, further studies might reveal that POT1 variants are specific to specific cancer types.

ACKNOWLEDGMENTS
We thank all participants of the M3 study for their contribution. This project was funded by the Anniversary Fund of the Austrian National Bank (grant number 15079) and the Medical Scientific Fund of the Mayor of the City of Vienna (grant number 10077). Study design, data collection, data analysis, manuscript preparation and/or publication decisions were not influenced by the funding sources. Conflicts of interest: None.