Advertisement

Whole-Exome Sequencing of Acquired Nevi Identifies Mechanisms for Development and Maintenance of Benign Neoplasms

  • Mitchell S. Stark
    Correspondence
    Correspondence: Mitchell S. Stark, Dermatology Research Centre, The University of Queensland, UQ Diamantina Institute, Level 5, Translational Research Institute, 37 Kent Street, Woolloongabba, Brisbane, Queensland 4102, Australia.
    Affiliations
    Dermatology Research Centre, The University of Queensland, The University of Queensland Diamantina Institute, Translational Research Institute, Brisbane, Queensland, Australia
    Search for articles by this author
  • Jean-Marie Tan
    Affiliations
    Dermatology Research Centre, The University of Queensland, The University of Queensland Diamantina Institute, Translational Research Institute, Brisbane, Queensland, Australia
    Search for articles by this author
  • Lisa Tom
    Affiliations
    Dermatology Research Centre, The University of Queensland, The University of Queensland Diamantina Institute, Translational Research Institute, Brisbane, Queensland, Australia
    Search for articles by this author
  • Kasturee Jagirdar
    Affiliations
    Dermatology Research Centre, The University of Queensland, The University of Queensland Diamantina Institute, Translational Research Institute, Brisbane, Queensland, Australia
    Search for articles by this author
  • Duncan Lambie
    Affiliations
    IQ Pathology, Brisbane, Queensland, Australia
    Search for articles by this author
  • Helmut Schaider
    Affiliations
    Dermatology Research Centre, The University of Queensland, The University of Queensland Diamantina Institute, Translational Research Institute, Brisbane, Queensland, Australia

    Department of Dermatology, Princess Alexandra Hospital, Brisbane, Australia
    Search for articles by this author
  • H. Peter Soyer
    Affiliations
    Dermatology Research Centre, The University of Queensland, The University of Queensland Diamantina Institute, Translational Research Institute, Brisbane, Queensland, Australia

    Department of Dermatology, Princess Alexandra Hospital, Brisbane, Australia
    Search for articles by this author
  • Richard A. Sturm
    Affiliations
    Dermatology Research Centre, The University of Queensland, The University of Queensland Diamantina Institute, Translational Research Institute, Brisbane, Queensland, Australia
    Search for articles by this author
Open ArchivePublished:February 21, 2018DOI:https://doi.org/10.1016/j.jid.2018.02.012
      The melanoma transformation rate of an individual nevus is very low despite the detection of oncogenic BRAF or NRAS mutations in 100% of nevi. Acquired melanocytic nevi do, however, mimic melanoma, and approximately 30% of all melanomas arise within pre-existing nevi. Using whole-exome sequencing of 30 matched nevi, adjacent normal skin, and saliva we sought to identify the underlying genetic mechanisms for nevus development. All nevi were clinically, dermoscopically, and histopathologically documented. In addition to identifying somatic mutations, we found mutational signatures relating to UVR mirroring those found in cutaneous melanoma. In nevi we frequently observed the presence of the UVR mutation signature compared with adjacent normal skin (97% vs. 10%, respectively). Copy number aberration analysis showed that for nevi with copy number loss of tumor suppressor genes, this loss was balanced by loss of potent oncogenes. Moreover, reticular and nonspecific patterned nevi showed an increased (P < 0.0001) number of copy number aberrations compared with globular nevi. The mutation signature data generated in this study confirms that UVR strongly contributes to nevogenesis. Copy number changes reflect at a genomic level the dermoscopic differences of acquired melanocytic nevi. Finally, we propose that the balanced loss of tumor suppressor genes and oncogenes is a protective mechanism of acquired melanocytic nevi.

      Abbreviations:

      CNA (copy number aberration), LOH (loss of heterozygosity), SNV (single-nucleotide variants), TCGA (The Cancer Genome Atlas)

      Introduction

      Melanocytic nevi are acquired benign neoplasms of the skin derived from melanocytes. In both children and adults, new nevi can form, and existing ones change regularly (
      • Abbott N.C.
      • Pandeya N.
      • Ong N.
      • McClenahan P.
      • Smithers B.M.
      • Green A.
      • et al.
      Changeable naevi in people at high risk for melanoma.
      ,
      • Duffy D.L.
      • Box N.F.
      • Chen W.
      • Palmer J.S.
      • Montgomery G.W.
      • James M.R.
      • et al.
      Interactive effects of MC1R and OCA2 on melanoma risk phenotypes.
      ,
      • Menzies S.W.
      • Stevenson M.L.
      • Altamura D.
      • Byth K.
      Variables predicting change in benign melanocytic nevi undergoing short-term dermoscopic imaging.
      ) and can be closely monitored for diagnostic purposes in melanoma screening. Many benign melanocytic lesions are excised unnecessarily because most nevi, even with a change in morphology, will never develop into a melanoma. Despite the rarity of transformation of an individual nevus (
      • Tsao H.
      • Bevona C.
      • Goggins W.
      • Quinn T.
      The transformation rate of moles (melanocytic nevi) into cutaneous melanoma: a population-based estimate.
      ), it has been reported that approximately 30% (range = 4–72%) of melanomas have arisen from a pre-existing nevus (
      • Pampena R.
      • Kyrgidis A.
      • Lallas A.
      • Moscarella E.
      • Argenziano G.
      • Longo C.
      A meta-analysis of nevus-associated melanoma: prevalence and practical implications.
      ), hence the reasoning for close monitoring.
      • Shain A.H.
      • Yeh I.
      • Kovalyshyn I.
      • Sriharan A.
      • Talevich E.
      • Gagnon A.
      • et al.
      The genetic evolution of melanoma from precursor lesions.
      showed that in archival clinical specimens, which had a melanoma adjacent to a nevus, there were many shared mutations and copy number aberrations consistent with a linear pathway from melanocyte to transformed precursor to melanoma. However, the conundrum still exists in that approximately 70% of melanomas arise de novo (
      • Pampena R.
      • Kyrgidis A.
      • Lallas A.
      • Moscarella E.
      • Argenziano G.
      • Longo C.
      A meta-analysis of nevus-associated melanoma: prevalence and practical implications.
      ) without any histopathological evidence for a precursor lesion. Improving the understanding of nevus development and transformation is therefore the key to understanding the etiology of melanoma and more efficient prevention and early detection methods for this increasingly common malignancy.

      Results and Discussion

      Study sample and somatic mutation burden in nevi

      Using whole-exome sequencing, we assessed the somatic mutational landscape, mutation signatures, and copy number aberrations (CNAs) in 30 acquired pigmented melanocytic nevi and matching adjacent normal skin (perilesional). The prospectively collected nevi were classified clinically as having a globular (n = 12) or reticular/nonspecific (n = 18) pattern and were histopathologically diagnosed (by DL) as benign/dysplastic lesions with common architectural features (see Supplementary Table S1 online). Somatic single-nucleotide variants (SNVs) were identified in all nevi lesional tissue from the exome pull-down region (coding and untranslated regions) a range from 30 to 668 mutations (median = 256), or 0–9 mutations per megabase. The total number of deleterious mutations (nonsense and nonsynonymous) ranged from 13 to 340 SNVs (median = 89), which is higher than those previously described in a study of 19 dysplastic nevi (range = 0–46, median = 19) (
      • Melamed R.D.
      • Aydin I.T.
      • Rajan G.S.
      • Phelps R.
      • Silvers D.N.
      • Emmett K.J.
      • et al.
      Genomic characterization of dysplastic nevi unveils implications for diagnosis of melanoma.
      ). The observed differences are likely due to the sample populations, with all of our nevi derived from a population that has a history of frequent sun exposure combined with a geographical region that has one of the highest level of UVR in the world (Brisbane, Australia).

      Somatic SNV subclasses in nevi and matching perilesional skin

      Not surprisingly, C>T transitions were confirmed to be the most prevalent SNV class (Figure 1a), and subsequent investigation into the trinucleotide context (e.g., N[C>T]N) showed that overall, the nevi analyzed here share many similarities with the mutation frequency distribution in cutaneous melanoma (see Materials and Methods) (Figure 1b). The point of difference between nevi and melanomas can be seen with the ratio of T[C>T]G to T[C>T]C being higher in melanoma (Figure 1b). This increase in T[C>T]G transitions corresponds to the presence of deleterious mutations in tumor suppressor genes such as ARID2, ATM, and NF1, all of which are absent in our nevi and in nevi analyzed by
      • Melamed R.D.
      • Aydin I.T.
      • Rajan G.S.
      • Phelps R.
      • Silvers D.N.
      • Emmett K.J.
      • et al.
      Genomic characterization of dysplastic nevi unveils implications for diagnosis of melanoma.
      . This supports the notion that loss of tumor suppressor gene function contributes to melanomagenesis.
      Figure 1
      Figure 1Single-nucleotide variants (SNVs) present in nevi and matching adjacent skin compared with cutaneous melanoma. (a) Box and whisker plots (5th–95th percentiles) of the percentage of the six SNV classes (C>A, C>G, C>T, T>A, T>C, and T>G) somatically observed in melanocytic nevi (n = 30). (b) Bar graph of the median C>T trinucleotide context proportion observed in melanocytic nevi (n = 30) highly correlates with those observed in cutaneous melanoma (n = 396). T[C>T]G is an example of a C>T trinucleotide context that can alternatively be written as TCG>TTG. (c) Box and whisker plots (5th–95th percentiles) of the percentage of the six SNV classes somatically observed in perilesional skin samples (adjacent to melanocytic nevi) (n = 30). (d) Bar graph of the median C>T trinucleotide context proportion observed in perilesional skin samples (n = 30) highlighting the low correlation compared with nevi and melanoma in b. (e) Bar graph of the median 96 trinucleotide mutation profile observed in nevi (n = 30) and matching perilesional skin (n = 30).
      In matching perilesional skin, C>T transitions were again the most prevalent (Figure 1c), yet this did not match the mutation profile observed in nevi (Figure 1d and e). Given that the proportion of C>T SNVs accounts for most of the variants observed, we next chose to focus on these variant types. When dermoscopic patterns were compared, globular nevi had a significantly (P = 0.013) higher proportion of T[C>T]A transitions versus reticular/non-specific nevi (Figure 2a). When histopathological characteristics were compared, it was shown that intradermal nevi had a significantly increased amount of G[C>T]C and T[C>T]A variants (P = 0.03 and P = 0.04 respectively) versus junctional/compound nevi (Figure 2b, and see Supplementary Table S1 for corresponding details). These data indicate that the differences observed in the dermoscopic patterns can be related to their histopathological classification, with the globular nevi in our dataset having primarily a dermal component. There were no significant differences in C>T transitions in dysplastic versus benign nevi, which suggests that the broad and nonspecific classification of dysplastic nevi is inaccurate. Overall, these data suggest that specific mutation types may contribute to the development of distinct dermoscopic and histopathological nevus phenotypes.
      Figure 2
      Figure 2The trinucleotide context of C>T single-nucleotide variants in different dermoscopic and histopathological subtypes of nevi. Box and whisker plots (minimum to maximum) of the percentage of C>T variant types that reached statistical significance (P < 0.05) when dermoscopic and histotype subtypes of nevi were compared (Mann-Whitney U test). (a) Nevi classified as globular have a significantly (P = 0.0132) higher proportion of T[C>T]A transitions versus reticular/nonspecific nevi and (b) intradermal nevi have a significantly increased amount of G[C>T]C and T[C>T]A variants (P = 0.03 and P = 0.04 respectively) versus junctional/compound nevi. P-values are indicated on the graphs.

      Driver gene analysis in nevi

      Previous studies have used a targeted gene approach to identify likely driver genes among a sea of passenger genes (
      • Cheng D.T.
      • Mitchell T.N.
      • Zehir A.
      • Shah R.H.
      • Benayed R.
      • Syed A.
      • et al.
      Memorial Sloan Kettering-Integrated Mutation Profiling of Actionable Cancer Targets (MSK-IMPACT): A hybridization capture-based next-generation sequencing clinical assay for solid tumor molecular oncology.
      ,
      • Shain A.H.
      • Yeh I.
      • Kovalyshyn I.
      • Sriharan A.
      • Talevich E.
      • Gagnon A.
      • et al.
      The genetic evolution of melanoma from precursor lesions.
      ). Using a combined gene list from these studies, the number of driver genes that harbored a deleterious mutation present in our nevi ranged from 1 to 14 (see Supplementary Table S2 online). Mutually exclusive mutations in BRAF and NRAS were the most frequent (83% and 17%, respectively), which is in keeping with our prior study using droplet digital PCR (
      • Tan J.M.
      • Tom L.N.
      • Jagirdar K.
      • Lambie D.
      • Schaider H.
      • Sturm R.A.
      • et al.
      The BRAF and NRAS mutation prevalence in dermoscopic subtypes of acquired naevi reveals constitutive mitogen-activated protein kinase pathway activation.
      ). Other genes of note included MET with three mutations (10%; NM_000245: Glu266Lys, Pro712Thr, and Gly1151Arg), and GRIN2A with two mutations (7%; NM_000833: Arg899Trp and Pro1132Leu). Targeted gene panels allow for a focused approach but are limited to identifying known genes. In addition to these well-known genes, we identified genes such as HDAC9 (NM_058176: p.Ser612Phe), MYH11 (NM_002474:p.Gly743Glu), and DCC (NM_005215: p.Asp866Asn), which were flagged as predicted driver mutations using IntOGen (
      • Gonzalez-Perez A.
      • Perez-Llamas C.
      • Deu-Pons J.
      • Tamborero D.
      • Schroeder M.P.
      • Jene-Sanz A.
      • et al.
      IntOGen-mutations identifies cancer drivers across tumor types.
      ) (see Supplementary Table S3 online). These driver mutations are present at a similar variant allele frequency as BRAFV600E and BRAFV600K, respectively, which indicates that they most likely occurred at the same time during nevus formation. BRAF mutations have been observed to occur in all cells present in a nevus; as such, nevi are considered to be clonal, originating from a single initiated melanocyte (
      • Yeh I.
      • von Deimling A.
      • Bastian B.C.
      Clonal BRAF mutations in melanocytic nevi and initiating role of BRAF in melanocytic neoplasia.
      ). However, alternative to this is the theory that nevi are not always clonal but can be polyclonal (
      • Lin J.
      • Takata M.
      • Murata H.
      • Goto Y.
      • Kido K.
      • Ferrone S.
      • et al.
      Polyclonality of BRAF mutations in acquired melanocytic nevi.
      ). In our study, we found evidence for both notions (see Supplementary Table S2), although in most cases in which a BRAF mutation appears in a clonal nevus, other gene mutations co-occur at a similar mutation frequency, as noted in Supplementary Tables S2 and S3. In cases in which BRAF mutations occur at a subclonal frequency (< % nevus cell estimate), it is unclear if this is due to a late acquisition of the BRAF mutation or an observation of oncogene-induced senescence (
      • Michaloglou C.
      • Vredeveld L.C.
      • Soengas M.S.
      • Denoyelle C.
      • Kuilman T.
      • van der Horst C.M.
      • et al.
      BRAFE600-associated senescence-like cell cycle arrest of human naevi.
      ) in the initial nevus cell population followed by an outgrowth of cells that have acquired a selective advantage via a new mutation. Given that BRAF V600 mutations are the most prevalent, it is likely that they are the initiator and that the other genes contain private deleterious mutations that provide a permissive environment for nevus development.

      TERT promoter mutation analysis in nevi

      TERT promoter mutations are prevalent in melanoma (60–86%) (
      The Cancer Genome Atlas Network
      Genomic classification of cutaneous melanoma.
      ,
      • Hayward N.K.
      • Wilmott J.S.
      • Waddell N.
      • Johansson P.A.
      • Field M.A.
      • Nones K.
      • et al.
      Whole-genome landscapes of major melanoma subtypes.
      ,
      • Horn S.
      • Figl A.
      • Rachakonda P.S.
      • Fischer C.
      • Sucker A.
      • Gast A.
      • et al.
      TERT promoter mutations in familial and sporadic melanoma.
      ,
      • Huang F.W.
      • Hodis E.
      • Xu M.J.
      • Kryukov G.V.
      • Chin L.
      • Garraway L.A.
      Highly recurrent TERT promoter mutations in human melanoma.
      ) and have also been observed in intermediate melanocytic lesions (
      • Shain A.H.
      • Yeh I.
      • Kovalyshyn I.
      • Sriharan A.
      • Talevich E.
      • Gagnon A.
      • et al.
      The genetic evolution of melanoma from precursor lesions.
      ) that occur adjacent to a melanoma. We did not detect any of the common TERT promoter mutations (data not shown) in our benign or dysplastic lesions, which was not an unexpected finding considering there was no evidence for malignant transformation observed clinically or histopathologically.

      Mutation signature analysis in nevi and matching perilesional skin

      The underlying causative mechanisms for most of the common malignancies investigated to date can be elucidated by analyzing somatic mutation signatures (
      • Alexandrov L.B.
      • Nik-Zainal S.
      • Wedge D.C.
      • Aparicio S.A.
      • Behjati S.
      • Biankin A.V.
      • et al.
      Signatures of mutational processes in human cancer.
      ). Each unique signature is numbered 1 through 30 and is defined by the pattern and proportion of specific trinucleotides present in each sample (
      Catalogue of Somatic Mutations in Cancer.
      ). In cutaneous melanoma, it has been established that UVR creates a well-known somatic mutation signature (signature 7). We confirm that the predominance of C>T transitions observed in our nevi resulted in a prevalent proportion of signature 7-related SNVs (29/30, 97%; range = 0–82%, median = 61%) (Figure 3a and see Supplementary Tables S1 and S4 online). Three of the 30 nevi (10%) had a low signature 7 proportion (0–13%) (see Supplementary Tables S1 and S4), which correlated with no or limited UVR exposure based on body location and clinical evidence of sun exposure. This suggests that acquired melanocytic nevi are not always initiated via UVR exposure. An age-related signature (signature 1) (
      • Alexandrov L.B.
      • Nik-Zainal S.
      • Wedge D.C.
      • Aparicio S.A.
      • Behjati S.
      • Biankin A.V.
      • et al.
      Signatures of mutational processes in human cancer.
      ) was the next most prevalent (28/30, 93%; range = 0–61%, median = 26%) (Figure 3a, and see Supplementary Tables S1 and S4). However, we could find no significant association between signature 1 and age at lesion excision (data not shown). Remarkably, two samples had no signature 1, along with a low or absent signature 7 (13% and 0%, respectively). Instead of these common signatures, evidence for defective DNA mismatch repair (
      • Alexandrov L.B.
      • Nik-Zainal S.
      • Wedge D.C.
      • Aparicio S.A.
      • Behjati S.
      • Biankin A.V.
      • et al.
      Signatures of mutational processes in human cancer.
      ) was identified (signature 20, 23% and signature 6, 13%). This highlights an alternative mechanism for nevi development in the absence of UVR. Defects in DNA repair is in fact prevalent in our nevi (10/30, 33%) with many mutually exclusive signatures being identified (signatures 3, 6, 15, 20, and 26) (
      • Alexandrov L.B.
      • Nik-Zainal S.
      • Wedge D.C.
      • Aparicio S.A.
      • Behjati S.
      • Biankin A.V.
      • et al.
      Signatures of mutational processes in human cancer.
      ) (Figure 3a, and see Supplementary Tables S1 and S4), which confirms the notion that defective DNA repair contributes to nevogenesis.
      Figure 3
      Figure 3Mutation signatures present nevi and matching adjacent skin. Box and whisker plots (5th–95th percentiles) of the percentage of the common COSMIC mutation signatures (n = 30) present in our (a) nevi (n = 30) and (b) matching perilesional skin (n = 30). For descriptions of COSMIC mutation signatures, please refer to http://cancer.sanger.ac.uk/cosmic/signatures. COSMIC, Catalogue of Somatic Mutations in Cancer.
      Recent studies have established that sun-exposed skin derived from the eyelid have an excess of C>T mutations in a targeted panel of genes (
      • Martincorena I.
      • Roshan A.
      • Gerstung M.
      • Ellis P.
      • Van Loo P.
      • McLaren S.
      • et al.
      Tumor evolution. High burden and pervasive positive selection of somatic mutations in normal human skin.
      ). Using whole-exome sequencing, we were able to determine the somatic mutations present in 30 matching normal skin samples (see Supplementary Table S5 online). We too observed a predominance of C>T mutations; however, this did not culminate in a UVR signature in all samples (3/30, 10%) (Figure 3b and Supplementary Table S4). Defective DNA repair signatures (signatures 3 and 26) (
      • Alexandrov L.B.
      • Nik-Zainal S.
      • Wedge D.C.
      • Aparicio S.A.
      • Behjati S.
      • Biankin A.V.
      • et al.
      Signatures of mutational processes in human cancer.
      ) were, however, frequently observed (83% and 50%, respectively), along with signature 1 (100%) (Figure 3b and Supplementary Table S4).
      These mutation signature data suggest a number of scenarios. First, the accumulation of age-related mutations and defective DNA repair machinery leads to the accumulation of UVR signature mutations in a given melanocyte, which contributes to nevogenesis. Second, the absence of the classical UVR signature mutations in the perilesional skin, which predominates in the more pigmented (and thus more melanin-containing) nevi, suggests that the melanin content may contribute to the excess of C>T transitions specific to the UVR signature. This notion is supported by the study by
      • Premi S.
      • Wallisch S.
      • Mano C.M.
      • Weiner A.B.
      • Bacchiocchi A.
      • Wakamatsu K.
      • et al.
      Photochemistry. Chemiexcitation of melanin derivatives induces DNA photoproducts long after UV exposure.
      , which described that the many of cyclobutane pyrimidine dimers arise via the breakdown of melanin pigment.

      Somatic mutation type and mutation signature analysis of nevi relative to body location/degree of sun exposure

      Next, we investigated nevi located on different body sites in combination with clinical evidence of sun exposure. Indeed, nevi located on shielded sites (see Supplementary Table S1) were significantly (P = 0.006) more likely to have a higher proportion of insertion/deletions (Figure 4a) compared with nevi located on sun-exposed sites, which were more likely (P = 0.03) to have a higher proportion of deleterious mutations. Not surprisingly, nevi from sun-exposed sites had a higher proportion of signature 7 mutations (P = 0.001) (Figure 4a), which is consistent with UVR contributing to high somatic mutation burden. Moreover, nevi that had a signature 7 profile that was less than the observed median (61%) were more likely (P = 0.014) to have a defective DNA repair signature (Figure 4b), which in turn was more likely (P = 0.045) to be present in dysplastic nevi (Figure 4c). There were no significant differences (data not shown) in the different dermoscopic patterned nevi based on degree of body site/sun exposure, the proportion of signature 7 or DNA repair defect signatures, or the proportion of deleterious mutations present. However, reticular/nonspecific nevi where more likely (P = 0.002) to have a higher proportion of insertion/deletions (Figure 4d) versus globular nevi, which may suggest an alternative mechanism for development.
      Figure 4
      Figure 4Somatic mutation type and mutation signature of nevi relative to body location/degree of sun exposure. (a) Insertion/deletion mutations are more commonly found in sun-shielded nevi (P = 0.0062), and sun exposed nevi are more likely to have deleterious mutations (P = 0.03) and a higher proportion of signature 7 mutations (P = 0.0013). (b) The total number of nevi that have a signature 7 mutation profile of less than the median (61%) are more likely to have defective DNA repair signatures present (P = 0.0142). (c) Dysplastic nevi are more likely to have a defective DNA repair signatures (P = 0.045) present, whereas there are no significant differences in signature 7 mutations relative to benign nevi. (d) Reticular/nonspecific patterned nevi are more likely to have an insertion/deletion mutation (P = 0.0017), but there is no significant difference in deleterious mutations between dermoscopic subtypes. P-values are indicated on the graphs. ns, not significant (P > 0.05).

      Copy number aberration analysis of nevi

      In addition to high somatic mutation burden, cutaneous melanomas have an increased amount of CNAs. Specifically, melanomas often have focal deletions in tumor suppressor genes (e.g., CDKN2A) and focal amplifications in oncogenes (e.g., MITF), along with extensive regions of copy number loss of heterozygosity (LOH), often encompassing whole chromosomal arms (
      • Stark M.
      • Hayward N.
      Genome-wide loss of heterozygosity and copy number analysis in melanoma using high-density single-nucleotide polymorphism arrays.
      ). Complex genomic rearrangements are also the hallmark of non–UVR-induced mucosal and acral melanomas (
      • Hayward N.K.
      • Wilmott J.S.
      • Waddell N.
      • Johansson P.A.
      • Field M.A.
      • Nones K.
      • et al.
      Whole-genome landscapes of major melanoma subtypes.
      ). Our dermoscopically globular nevi were largely genomically silent, with minor CNAs observed (Figure 5a); however, in stark contrast we detected a high frequency of genome-wide CNAs in our reticular/nonspecific nevi (P < 0.0001) (Figure 5b, and see Supplementary Tables S6 and S7 online), which correlates with the higher proportion of observed insertion/deletions. Considering the extensive nature of the CNAs present in the reticular/nonspecific nevi, one would assume that this may herald the early beginnings of a melanoma. However, upon closer inspection of the genes involved, the CNAs appear to be balanced events. Most of the observed CNAs were large regional CNAs (mainly LOH) rather than focal regions of loss, which is a hallmark of cutaneous melanoma. These large regional events suggest that no specific gene is being targeted; instead, this is a random mutagenic process. To find evidence for LOH events effecting gene expression, interrogation of the melanoma TCGA data (
      The Cancer Genome Atlas Network
      Genomic classification of cutaneous melanoma.
      ) showed that on average, LOH of key melanoma genes can significantly result in a loss of expression in tumor suppressors and in oncogenes (see Supplementary Figures S1 and S2 online). We therefore postulate that the net effect of the loss of both gene functions in the same nevus specimen combine to have an overall copy neutral effect, thus keeping the nevus in a state of equilibrium so as to be protected against malignant transformation.
      Figure 5
      Figure 5Copy number aberrations (CNAs) in different dermoscopic patterned nevi. Genome-wide copy number aberrations in dermoscopic patterned (a) globular and (b) reticular/nonspecific nevi. (a) Globular nevi are largely genomically silent, with minor CNAs observed, and (b) reticular/nonspecific nevi have significantly (P < 0.0001) more CNAs, with mostly large regional LOH events being present. Chromosomes are represented from left to right in chromosomal order. Dark blue indicates copy number equivalent to homozygous deletion (homozygous deletion or deep deletion), light blue indicates copy number equivalent to LOH (LOH or shallow deletion), light red indicates copy number equivalent to three copies (gain), and dark red indicates copy number equivalent to more than three copies (amplified). CNV, copy number variant; LOH, loss of heterozygosity.
      In summary, this study has highlighted that acquired nevi occur in most cases after cumulative exposure to UVR, coupled with defective DNA repair mechanisms in normal skin melanocytes. This environment permits the accumulation of UV-related somatic mutations, which is evident in the mutation signatures (Figure 6, and see Supplementary Tables S1 and S4) observed in our nevi and matching adjacent skin. Moreover, we predict that CNA events that are balanced (or absent) in clonally expanded melanocytes lead to the nevogenesis pathway. Finally, if these CNA events become imbalanced in a given clonal expansion, then this contributes to the formation of early melanomas.
      Figure 6
      Figure 6Model for the fate of clonally expanded skin melanocytes. Somatic mutation signature and copy number aberration (CNA) analysis allows for insights into nevogenesis. This model shows that normal skin melanocytes accumulate mutations if there are defects in normal DNA repair mechanisms. Over time, clonal expansion occurs and in most cases forms a benign nevus that has balanced CNA events and is thereby protected against malignant transformation. If imbalanced CNA events occur, then the clonally expanded melanocytes follow the melanomagenesis pathway. Melanoma can directly arise from a benign nevus or via the intermediate neoplasm state.

      Materials and Methods

      Patients and tissue sampling

      This study was approved by the Metro South Human Research Ethics Committee (Brisbane, Australia; HREC/09/QPAH/162) and was carried out in accordance with the Declaration of Helsinki. After we received written informed consent from patients, 30 acquired melanocytic nevi with globular (n = 12), reticular (n = 14) and nonspecific (n = 4) dermoscopic patterns identified by a board-certified dermatologist (HPS) from a database of prospectively imaged nevi (
      • Daley G.M.
      • Duffy D.L.
      • Pflugfelder A.
      • Jagirdar K.
      • Lee K.J.
      • Yong X.L.
      • et al.
      GSTP1 does not modify MC1R effects on melanoma risk.
      ) were excised by shave excision from 20 participants. Of these, 10 donated a single nevus, and the following 10 participants had two nevi excised: 07LW, 28RH, 111CM, 158RP, 61JB, 510AN, 673PS, 736TP, 822MT, and 1346KJ, (see Supplementary Table S1). Each nevus was bisected, and one half of the tissue was formalin fixed and histopathologically diagnosed by a board-certified pathologist (DL). The second half of the tissue was dissected to isolate nevus from adjacent perilesional skin according to methods previously described (
      • Tan J.M.
      • Tom L.N.
      • Jagirdar K.
      • Lambie D.
      • Schaider H.
      • Sturm R.A.
      • et al.
      The BRAF and NRAS mutation prevalence in dermoscopic subtypes of acquired naevi reveals constitutive mitogen-activated protein kinase pathway activation.
      ). DNA extraction was performed as described previously (
      • Tan J.M.
      • Tom L.N.
      • Jagirdar K.
      • Lambie D.
      • Schaider H.
      • Sturm R.A.
      • et al.
      The BRAF and NRAS mutation prevalence in dermoscopic subtypes of acquired naevi reveals constitutive mitogen-activated protein kinase pathway activation.
      ). Saliva was collected, and germline DNA was extracted from each study participant as previously described (
      • Daley G.M.
      • Duffy D.L.
      • Pflugfelder A.
      • Jagirdar K.
      • Lee K.J.
      • Yong X.L.
      • et al.
      GSTP1 does not modify MC1R effects on melanoma risk.
      ).

      Next-generation sequencing and data analysis

      Please refer to Supplementary Material (online) for detailed methodology.

      Somatic variant calling and filtering

      Somatic variants present in the nevi and perilesional samples were determined first by filtering out all variants that were present in the matching saliva-derived germline DNA, followed by all variants that were present in the 1000 genomes (1000g2014oct_eur build) and ExAC Non-Finnish European databases respectively. Variants present in dbSNP build 138 were not used as a filter because of the presence of somatic mutations (e.g., BRAFV600E). Accordingly, a proportion of the variants presented in Supplementary Tables S3 and S5 may indeed be polymorphisms. Next, any variants present in a pool of 30 perilesional samples were removed from each of the nevi samples, which resulted in a list of somatic variants present only in the lesion and not from perilesional contamination. Next, all somatic variants present in the nevi and perilesion were filtered further to include only those with a total of 10 or more reads, an alternative read frequency of approximately 3% (based on known a mutation in NRASQ61K previously determined by Droplet Digital PCR [Bio-Rad Laboratories, Hercules, CA] [
      • Tan J.M.
      • Tom L.N.
      • Jagirdar K.
      • Lambie D.
      • Schaider H.
      • Sturm R.A.
      • et al.
      The BRAF and NRAS mutation prevalence in dermoscopic subtypes of acquired naevi reveals constitutive mitogen-activated protein kinase pathway activation.
      ]), a variant P of 0.05 or less, and an alternate Phred base quality of 30.

      Detection of known mutations

      To test the sensitivity of the exome sequencing depth for identifying somatic mutations in BRAF or NRAS present in our nevi samples (
      • Tan J.M.
      • Tom L.N.
      • Jagirdar K.
      • Lambie D.
      • Schaider H.
      • Sturm R.A.
      • et al.
      The BRAF and NRAS mutation prevalence in dermoscopic subtypes of acquired naevi reveals constitutive mitogen-activated protein kinase pathway activation.
      ), we first used The Integrative Genome Viewer (Broad Institute, Cambridge, MA) to visualize the BAM files at specific codons. BRAF V600E/K mutations were detectable in all samples (26/26) (see Supplementary Table S2), and NRAS G13C and Q61L/K/R were detectable in the remaining 4 of 4 samples (see Supplementary Table S2). The digital PCR methodology used in our prior study (
      • Tan J.M.
      • Tom L.N.
      • Jagirdar K.
      • Lambie D.
      • Schaider H.
      • Sturm R.A.
      • et al.
      The BRAF and NRAS mutation prevalence in dermoscopic subtypes of acquired naevi reveals constitutive mitogen-activated protein kinase pathway activation.
      ) detected additional NRAS G13C mutations in samples with codon 61 mutations that were not detectable in our exome dataset because they were less than the limit of detection (approximately 1% mutant frequency). Furthermore, the NRAS Q61L mutation detected in 903JA-33 via exome sequencing was not validated using Droplet Digital PCR and instead was found to have an NRAS G13C mutation. The NRAS Q61L is possibly a false positive result. Next, we confirmed the presence of the validated mutations in the stringently filtered dataset as described. BRAF/NRAS mutations were detectable in 17 of 30 samples, with 13 of 30 not passing filtering because of low mutant allele frequency (10/13), which were filtered out during variant calling, or because the mutant call had a P-value greater than 0.05 (3/13) as determined by VarScan2 (
      • Koboldt D.C.
      • Zhang Q.
      • Larson D.E.
      • Shen D.
      • McLellan M.D.
      • Lin L.
      • et al.
      VarScan 2: somatic mutation and copy number alteration discovery in cancer by exome sequencing.
      ).

      TERT promoter sequence analysis

      The TERT promoter region was amplified using forward (5′-ACGAACGTGGCCAGCGGCAG-3′) and reverse (5′-CTCCCAGTGGATTCGCGGGC-3′) primer sequences designed using Primer3web (version 4.1.1; http://primer3.ut.ee/) software. To increase the specificity of the primers, a touchdown PCR protocol was used with a cycling routine of two cycles at each annealing temperature commencing at 69°C, decreasing by 1°C-steps (69°C to 63°C), followed by 20 cycles at the optimal temperature (62°C). All other cycling conditions were according to the manufacturer’s protocol (Thermo Fisher Scientific, Waltham, MA). PCR reactions consisted of 25–50 ng of DNA, 0.5 μmol/L of each primer, 200 μmol/L dNTPs (Thermo Fisher Scientific), together with 0.4 units of Phusion Hot Start II High-Fidelity DNA polymerase (#F-549, Thermo Fisher Scientific), 5X Phusion GC buffer (supplemented with MgCl2; #F-519, Thermo Fisher Scientific) and 5% DMSO. The 369-base pair PCR products were electrophoresed on a 2% Tris-acetate-EDTA agarose gel, purified using the QIAquick Gel Extraction Kit (Qiagen, Hilden, Germany), then sequenced using the BigDye Terminator v3.1 Cycle Sequencing Kit (Thermo Fisher Scientific) as per the manufacturer’s instructions. Agencourt CleanSEQ (Beckman Coulter Life Sciences, Pasadena, CA) purified products were run on the 3730-series Genetic Analyzer (Thermo Fisher Scientific) by the Australian Equine Genetics Research Centre (Brisbane, Australia). All TERT promoter sequence chromatograms were analyzed using Sequencher 5.4 software (Gene Codes Corporation, Ann Arbor, MI) compared with the hg19 reference sequence (chromosome 5: 1,295,022–1,295,495).

      Mutation signature analysis

      Filtered somatic SNVs present in the nevi and perilesional skin were imported into the deconstructSigs (
      • Rosenthal R.
      • McGranahan N.
      • Herrero J.
      • Taylor B.S.
      • Swanton C.
      DeconstructSigs: delineating mutational processes in single tumors distinguishes DNA repair deficiencies and patterns of carcinoma evolution.
      ) package using R 3.4.0 for Windows (https://github.com/raerose01/deconstructSigs). Mutation signatures were determined by using the framework of 30 COSMIC signatures (
      Catalogue of Somatic Mutations in Cancer.
      ) collated in the deconstructSigs package. Somatic mutations present in melanoma datasets were downloaded from ftp://ftp.sanger.ac.uk/pub/cancer/AlexandrovEtAl/somatic_mutation_data/Melanoma/ and imported into the deconstructSigs package.

      Copy number analysis

      CNAs were determined via the CNVkit (
      • Rieber N.
      • Bohnert R.
      • Ziehm U.
      • Jansen G.
      Reliability of algorithmic somatic copy number alteration detection from targeted capture data.
      ) package (https://github.com/etal/cnvkit) and run using Python 2.7 (https://www.python.org/download/release/2.7). Briefly, matching nevi and perilesion BAM files, with duplicates marked and sorted (see Supplementary Material), were analyzed with CNVkit according to standard methods. The matching perilesion was used as the background normal. Nevus cell content was estimated based on BRAF mutation frequency as determined by methods previously reported (
      • Shain A.H.
      • Yeh I.
      • Kovalyshyn I.
      • Sriharan A.
      • Talevich E.
      • Gagnon A.
      • et al.
      The genetic evolution of melanoma from precursor lesions.
      ). In the absence of BRAF mutation, tumor cell fraction was estimated from hematoxylin and eosin sections by an experienced dermatopathologist (HPS). A segmentation file (see Supplementary Table S7) was compiled from all lesions to allow for genome-wide visualization (Figure 5) in the Integrative Genome Viewer. Genes involved in regions of gain (3 or more copies) and loss (1 or 0 copies) summarized in Supplementary Table S6, were those commonly mutated in melanoma (
      The Cancer Genome Atlas Network
      Genomic classification of cutaneous melanoma.
      ). The effect of copy number on RNA expression in exemplar genes derived from Supplementary Table S6 and shown in Supplementary Figures S1 and S2 was determined using the melanoma TCGA datasets accessed via the cBioPortal for Cancer Genomics (http://www.cbioportal.org/) (
      • Cerami E.
      • Gao J.
      • Dogrusoz U.
      • Gross B.E.
      • Sumer S.O.
      • Aksoy B.A.
      • et al.
      The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data.
      ,
      • Gao J.
      • Aksoy B.A.
      • Dogrusoz U.
      • Dresdner G.
      • Gross B.
      • Sumer S.O.
      • et al.
      Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal.
      ).

      Statistical analysis

      Box plots and statistical analysis were performed using two-tailed t tests (Mann-Whitney) or Fisher exact test with GraphPad Prism version 7.03 for Windows (La Jolla, CA) or SPSS Statistics, version 24 (IBM, Armonk, NY). A P-value less than 0.05 was considered statistically significant.

      Conflict of Interest

      The authors state no conflict of interest.

      Acknowledgments

      The authors would like to thank the study participants and are grateful for the support of their colleagues, particularly Katie Lee and Clare Primiero. This work was funded by project grants (APP1062935, APP1083612); the Centre of Research Excellence for the Study of Nevi (APP1099021) from the National Health and Medical Research Council (NHMRC), Australia; and the Merchant Charitable Foundation. JMT holds a scholarship from the Australian Government Department of Education and Training. RAS and MSS hold fellowships (APP1043187 and APP1106491, respectively) from the NHMRC.

      Supplementary Material

      References

        • Abbott N.C.
        • Pandeya N.
        • Ong N.
        • McClenahan P.
        • Smithers B.M.
        • Green A.
        • et al.
        Changeable naevi in people at high risk for melanoma.
        Australas J Dermatol. 2014; 56: 14-18
        • Alexandrov L.B.
        • Nik-Zainal S.
        • Wedge D.C.
        • Aparicio S.A.
        • Behjati S.
        • Biankin A.V.
        • et al.
        Signatures of mutational processes in human cancer.
        Nature. 2013; 500: 415-421
        • The Cancer Genome Atlas Network
        Genomic classification of cutaneous melanoma.
        Cell. 2015; 161: 1681-1696
      1. Catalogue of Somatic Mutations in Cancer.
        (Signatures of mutational processes in human cancer, http://cancer.sanger.ac.uk/cosmic/signatures)
        Date: 2018
        (accessed 20 December 2017)
        • Cerami E.
        • Gao J.
        • Dogrusoz U.
        • Gross B.E.
        • Sumer S.O.
        • Aksoy B.A.
        • et al.
        The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data.
        Cancer Discov. 2012; 2: 401-404
        • Cheng D.T.
        • Mitchell T.N.
        • Zehir A.
        • Shah R.H.
        • Benayed R.
        • Syed A.
        • et al.
        Memorial Sloan Kettering-Integrated Mutation Profiling of Actionable Cancer Targets (MSK-IMPACT): A hybridization capture-based next-generation sequencing clinical assay for solid tumor molecular oncology.
        J Mol Diagn. 2015; 17: 251-264
        • Daley G.M.
        • Duffy D.L.
        • Pflugfelder A.
        • Jagirdar K.
        • Lee K.J.
        • Yong X.L.
        • et al.
        GSTP1 does not modify MC1R effects on melanoma risk.
        Exp Dermatol. 2016; 26: 730-733
        • Duffy D.L.
        • Box N.F.
        • Chen W.
        • Palmer J.S.
        • Montgomery G.W.
        • James M.R.
        • et al.
        Interactive effects of MC1R and OCA2 on melanoma risk phenotypes.
        Hum Mol Genet. 2004; 13: 447-461
        • Gao J.
        • Aksoy B.A.
        • Dogrusoz U.
        • Dresdner G.
        • Gross B.
        • Sumer S.O.
        • et al.
        Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal.
        Sci Signal. 2013; 6: pl1
        • Gonzalez-Perez A.
        • Perez-Llamas C.
        • Deu-Pons J.
        • Tamborero D.
        • Schroeder M.P.
        • Jene-Sanz A.
        • et al.
        IntOGen-mutations identifies cancer drivers across tumor types.
        Nat Methods. 2013; 10: 1081-1082
        • Hayward N.K.
        • Wilmott J.S.
        • Waddell N.
        • Johansson P.A.
        • Field M.A.
        • Nones K.
        • et al.
        Whole-genome landscapes of major melanoma subtypes.
        Nature. 2017; 545: 175-180
        • Horn S.
        • Figl A.
        • Rachakonda P.S.
        • Fischer C.
        • Sucker A.
        • Gast A.
        • et al.
        TERT promoter mutations in familial and sporadic melanoma.
        Science. 2013; 339: 959-961
        • Huang F.W.
        • Hodis E.
        • Xu M.J.
        • Kryukov G.V.
        • Chin L.
        • Garraway L.A.
        Highly recurrent TERT promoter mutations in human melanoma.
        Science. 2013; 339: 957-959
        • Koboldt D.C.
        • Zhang Q.
        • Larson D.E.
        • Shen D.
        • McLellan M.D.
        • Lin L.
        • et al.
        VarScan 2: somatic mutation and copy number alteration discovery in cancer by exome sequencing.
        Genome Res. 2012; 22: 568-576
        • Lin J.
        • Takata M.
        • Murata H.
        • Goto Y.
        • Kido K.
        • Ferrone S.
        • et al.
        Polyclonality of BRAF mutations in acquired melanocytic nevi.
        J Natl Cancer Inst. 2009; 101: 1423-1427
        • Martincorena I.
        • Roshan A.
        • Gerstung M.
        • Ellis P.
        • Van Loo P.
        • McLaren S.
        • et al.
        Tumor evolution. High burden and pervasive positive selection of somatic mutations in normal human skin.
        Science. 2015; 348: 880-886
        • Melamed R.D.
        • Aydin I.T.
        • Rajan G.S.
        • Phelps R.
        • Silvers D.N.
        • Emmett K.J.
        • et al.
        Genomic characterization of dysplastic nevi unveils implications for diagnosis of melanoma.
        J Invest Dermatol. 2017; 137: 905-909
        • Menzies S.W.
        • Stevenson M.L.
        • Altamura D.
        • Byth K.
        Variables predicting change in benign melanocytic nevi undergoing short-term dermoscopic imaging.
        Arch Dermatol. 2011; 147: 655-659
        • Michaloglou C.
        • Vredeveld L.C.
        • Soengas M.S.
        • Denoyelle C.
        • Kuilman T.
        • van der Horst C.M.
        • et al.
        BRAFE600-associated senescence-like cell cycle arrest of human naevi.
        Nature. 2005; 436: 720-724
        • Pampena R.
        • Kyrgidis A.
        • Lallas A.
        • Moscarella E.
        • Argenziano G.
        • Longo C.
        A meta-analysis of nevus-associated melanoma: prevalence and practical implications.
        J Am Acad Dermatol. 2017; 77: 938-945
        • Premi S.
        • Wallisch S.
        • Mano C.M.
        • Weiner A.B.
        • Bacchiocchi A.
        • Wakamatsu K.
        • et al.
        Photochemistry. Chemiexcitation of melanin derivatives induces DNA photoproducts long after UV exposure.
        Science. 2015; 347: 842-847
        • Rieber N.
        • Bohnert R.
        • Ziehm U.
        • Jansen G.
        Reliability of algorithmic somatic copy number alteration detection from targeted capture data.
        Bioinformatics. 2017; 33: 2791-2798
        • Rosenthal R.
        • McGranahan N.
        • Herrero J.
        • Taylor B.S.
        • Swanton C.
        DeconstructSigs: delineating mutational processes in single tumors distinguishes DNA repair deficiencies and patterns of carcinoma evolution.
        Genome Biol. 2016; 17: 31
        • Shain A.H.
        • Yeh I.
        • Kovalyshyn I.
        • Sriharan A.
        • Talevich E.
        • Gagnon A.
        • et al.
        The genetic evolution of melanoma from precursor lesions.
        N Engl J Med. 2015; 373: 1926-1936
        • Stark M.
        • Hayward N.
        Genome-wide loss of heterozygosity and copy number analysis in melanoma using high-density single-nucleotide polymorphism arrays.
        Cancer Res. 2007; 67: 2632-2642
        • Tan J.M.
        • Tom L.N.
        • Jagirdar K.
        • Lambie D.
        • Schaider H.
        • Sturm R.A.
        • et al.
        The BRAF and NRAS mutation prevalence in dermoscopic subtypes of acquired naevi reveals constitutive mitogen-activated protein kinase pathway activation.
        Br J Dermatol. 2018; 178: 191-197
        • Tsao H.
        • Bevona C.
        • Goggins W.
        • Quinn T.
        The transformation rate of moles (melanocytic nevi) into cutaneous melanoma: a population-based estimate.
        Arch Dermatol. 2003; 139: 282-288
        • Yeh I.
        • von Deimling A.
        • Bastian B.C.
        Clonal BRAF mutations in melanocytic nevi and initiating role of BRAF in melanocytic neoplasia.
        J Natl Cancer Inst. 2013; 105: 917-919

      Linked Article

      • Erratum
        Journal of Investigative DermatologyVol. 138Issue 9
        • Preview
          Whole-Exome Sequencing of Acquired Nevi Identifies Mechanisms for Development and Maintenance of Benign Neoplasms
        • Full-Text
        • PDF
        Open Archive