Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Jul 1.
Published in final edited form as: J Invest Dermatol. 2021 Dec 3;142(7):1893–1902.e7. doi: 10.1016/j.jid.2021.11.025

Genome-scale DNA methylation analysis identifies repeat element alterations that modulate the genomic stability of melanocytic nevi

Meghan E Muse 1, Drew T Bergman 1, Lucas A Salas 1, Lisa N Tom 2, Jean-Marie Tan 2, Antonia Laino 2, Duncan Lambie 3,4, Richard A Sturm 2, Helmut Schaider 2,5, H Peter Soyer 2,6, Brock C Christensen 1,7,8,†, Mitchell S Stark 2,†,*
PMCID: PMC9163203  NIHMSID: NIHMS1763681  PMID: 34871578

Abstract

Acquired melanocytic nevi grow and persist in a stable form into adulthood. Using genome-wide methylation profiling, we evaluated 32 histopathologically and dermoscopically characterized nevi, to identify key epigenetic regulatory mechanisms involved in nevogenesis. Benign (69% globular and 31% non-specific dermoscopic pattern) and dysplastic (95% reticular/nonspecific dermoscopic pattern) nevi were dissimilar with only two shared differentially methylated (DM) loci. Benign nevi demonstrated an increase in both genome-scale methylation and methylation of Alu/LINE-1 retrotransposable elements, a marker of genomic stability, as well as global methylation. In contrast, dysplastic nevi showed evidence for genomic instability via hypomethylation of Alu/LINE-1 (Alu; P=0.00019 and LINE-1; P=0.000035). Using dermoscopic classifications, reticular/non-specific patterned nevi had 59,572 CpG DM loci (Q < 0.05), whereas globular nevi had no significant DM loci. In reticular/non-specific patterned nevi, the tumor suppressor PTEN had the greatest proportion of hypermethylated CpG loci in its promoter region compared to all other assayed gene promoters. The relative activity of reticular/non-specific nevi was evidenced by 50,720 hypomethylated loci being enriched for accessible chromatin, and 8,852 hypermethylated loci strongly enriched, for example, marks of active gene promoters, which suggests that gain of DNA methylation observed in these nevus types plays a role in gene regulation.

Keywords: CpG methylation, nevi, epigenetic, ALU, LINE-1, PTEN

INTRODUCTION

Acquired melanocytic nevi (AMN) share many phenotypic characteristics of early cutaneous melanoma and may be excised as a precautionary measure to be diagnosed via histopathology. The histopathological classification of an AMN is dependent upon the location of the melanocytic proliferations or nests (junctional, intradermal, or compound). Broadly speaking, if histopathological features of a melanoma are not identified, AMN are diagnosed as benign, or if architectural disorder or atypia are observed, then a classification of ‘dysplastic’ may be provided (Gerami and Busam, 2018). There are varying degrees of dysplasia (e.g. mild/moderate/severe or low/high grade) that may be observed (Gerami and Busam, 2018); however, this nomenclature is difficult to have a consensus amongst pathologists (Elmore et al., 2017). The decision to excise a melanocytic lesion is based upon phenotypical characteristics, and since AMN are a heterogeneous collection of lesions, dermoscopy is used as the primary method to aid in clinical decision making. AMN with a reticular or non-specific pattern are often flat, corresponding to junctional or compound melanocytic nevi. AMN with a globular pattern are frequently raised, or papillomatous, and primarily have intradermal melanocytic nests. Globular nevi that are raised, tend to be stable throughout life, and clinically, they are very different from reticular/non-specific nevi, which are acquired during puberty and adulthood (reviewed in (Tan et al., 2019)). Non-specific nevi with a peripheral rim of globules have been determined to be growing symmetrically with apparent growth cessation occurring during a span of 4 to 5 years (Bajaj et al., 2015, Kittler et al., 2000). Melanoma has been found to arise within both benign and dysplastic nevi which may be flat or raised; however, there is a lack of evidence of which type of nevus is more prone for melanoma development. Along with common phenotypic characteristics, nevi share numerous molecular features with cutaneous melanoma. The BRAF oncogene is well known to be mutated in nevi (Pollock et al., 2003), with the mutation found in up to 100% of nevi exhibiting growth hallmarks (Tan et al., 2018), indicating that this mutation is involved in proliferation. Since most AMN harbor this oncogenic mutation, this event does not constitute progression toward malignancy. We and others have investigated nevi using whole-genome and exome sequencing (Colebatch et al., 2019, Stark et al., 2018) to reveal common ultraviolet radiation (UVR) mutation signatures between nevi and melanoma. Importantly, some melanomas have been observed to arise from a pre-existing nevus, with molecular studies (Shain et al., 2015) identifying some early genomic events (e.g., TERT mutation and copy number loss of tumor suppressor genes) that occur prior to malignant transformation.

Despite the similarities described above, most nevi remain stable, and very few will transform into melanoma, so it is important to understand and recognize early molecular changes in AMN. DNA methylation profiling using a panel of genes has established that differential methylation profiles of only 22 genes can accurately determine (AUC > 0.95) a melanoma from an AMN (Conway et al., 2011). Moreover, genome-wide methylation profiling of healthy skin revealed a clear methylation signature associated with a high nevus count (Roos et al., 2017). We have previously determined that raised, papillomatous nevi, with a globular dermoscopic pattern, are genomically silent, with no copy number aberrations (CNA), whereas reticular/non-specific nevi had significant (P < 0.0001) CNAs (Stark et al., 2018) often observed in melanoma. Herein, we performed genome-wide methylation profiling of AMN with globular and reticular/non-specific patterns to reveal distinct methylation profiles.

RESULTS

Study population and mutation status

Thirty-two paired nevus and adjacent normal samples were collected from 24 study participants, with 8 study participants contributing two nevus-adjacent normal pairs (Table 1). All excised nevi were dermoscopically classified (Supplementary Materials and Methods) prior to histopathological diagnosis. It was confirmed that all specimens were melanocytic nevi and could be classified as benign (n=13) or dysplastic (n=19) with junctional, compound, or intradermal histopathological architecture (Table 1 and Supplementary Table S1). No statistically significant differences were observed for the age, sex, and UV exposure status of benign and dysplastic nevi (Supplementary Table S2). Nevi histopathologically diagnosed as benign either had a globular dermoscopic pattern (69% or 9 of 13) or a non-specific pattern (31% or 4 of 13) (Supplementary Table S1–2). Dysplastic nevi from this collection revealed primarily reticular/non-specific pattern or were nevi with a peripheral rim of globules (95% or 18 of 19) (Supplementary Table S1–2); with only one globular nevus exhibiting dysplastic histopathologic features. The BRAF and NRAS mutation status was confirmed in 30 of 32 nevi using droplet digital PCR (Tan et al., 2018) or whole-exome sequencing (Table 1 and Supplementary Table S1). Insufficient DNA was available for 2 of 32 nevi to perform BRAF/NRAS mutation detection. BRAF was confirmed to be frequently mutated at codon 600 in 29 of 30 nevi (96.66%), with BRAF V600E being the most frequent (83.33%), followed by V600K (10%) and V600R (3.33%). In BRAF wild-type nevi, the NRAS G12S mutation (3.33%) was present. Taken together, this nevus cohort was 100% mutated for either BRAF or NRAS, which is consistent with our prior observations (Tan et al., 2018).

Table 1.

Subject and Sample Characteristics

Subjects Samples
n = 24 n = 32
Sex n (%)
 Female 7 (29.2)
 Male 17 (70.8)
Age mean (SD) 51.88 (14.36)
Status n (%)
 Benign 13 (40.6)
 Dysplastic 19 (59.4)
Histopath n (%)
 Compound 13 (40.6)
 Intradermal 12 (37.5)
 Junctional 7 (21.9)
Dermoscopic Pattern n (%)
 Globular 10 (31.2)
 Reticular 4 (12.5)
 Non-specific (PRG) 10 (31.2)
 Non-specific (other) 8 (25.0)
Sun Exposure n (%)
 Exposed 25 (78.1)
 Shielded 7 (21.9)
BRAF n (%)
 BRAF V600E 23 (71.9)
 BRAF V600E c. 1799_1800delTGinsAA 2 (6.2)
 BRAF V600K 3 (9.4)
 BRAF V600R 1(3) 1 (3.1)
 NRAS G12S 1(3) 1 (3.1)
 unknown 2 (6.3)
Open in a new tab

Inferred proportions of melanocytes differed in histopathologically diagnosed benign and dysplastic nevi

We created a reference library for DNA methylation-based cell type deconvolution of skin biopsies using publicly available cell line data for epithelial cells, fibroblasts, and melanocytes (GSE74877) and purified B cells, CD4+ T cells, CD8+ T cells, NK cells, monocytes, and neutrophils (GSE110554). The optimized reference library for deconvoluting epidermal tissue cell types included 609 CpG probes for estimating proportions of melanocytes, epithelial cells, fibroblasts, B cells, CD4+ T cells, CD8+ T cells, monocytes neutrophils, and NK cells (Supplementary Table S3 and Supplementary Figure S1A). Using the generated library, estimated relative proportions of immune cells from DNA methylation measured on in silico mixtures of whole blood-derived immune cells demonstrated a strong correlation (R2 > 0.99) with the true relative proportions of included cell types (Supplementary Figure S2). Additionally, estimated relative proportions of all cell types included in the library from computationally generated in silico mixtures demonstrated a strong correlation (R2 > 0.99) to the true relative proportions (Supplementary Figure S3).

After projecting the library on the nevi and adjacent skin samples to estimate cellular composition, median melanocyte proportions were found to be higher in benign nevi (46.9%) compared with dysplastic nevi (14.0%) (Supplementary Figure S1B). The benign nevi in this collection were primarily nevi that had a raised clinical presentation, with a globular dermoscopic pattern (69%). Since these globular nevi all have an intradermal histopathologic architecture, corresponding to high nevus cell cellularity, the DNA methylation-based reference library was congruent with nevus biology. Furthermore, the dysplastic nevi, which primarily had a flat clinical presentation with a reticular/non-specific dermoscopic pattern (95%), are known to have low nevus cell cellularity. In fact, the proportion of melanocytes in the nevus samples estimated from deconvolution were strongly correlated (R2 = 0.82) with a dermatopathologist-provided (HPS) semi-quantitative assessment of the relative proportion of melanocytes in each nevus sample based on H&E staining (Supplementary Figure S1C).

Global methylation and inferred methylation of repeat element regions of the genome in nevi

We compared the difference in DNA methylation between nevi and paired adjacent normal skin at LINE-1 and Alu repeat elements as well as the mean methylation of all measured probes on the array. Compared with adjacent normal skin, genome-scale methylation and repeat element measures were higher in our collection of benign nevi and lower in dysplastic nevi. The mean methylation change differences between nevi types were statistically significant for all three measures (see Figure 1A). Additionally, assessing the inferred repeat element methylation in repeat elements tracking to the chr9p21 region, a region known to be rich with repeat elements as well having well-known association with melanoma, inferred methylation status of LINE-1 repeat elements tracking to tumor suppressor gene CDKN2A was increased in benign nevi relative to paired adjacent normal skin while decreased in dysplastic nevi relative to paired adjacent normal skin (Supplementary Figure S4).

Figure 1: DNA methylation differences in benign and dysplastic nevi relative to adjacent normal skin.

Figure 1:

Open in a new tab

(a) Distribution of the change in mean methylation between nevi and adjacent normal skin for LINE-1 and Alu repeat elements and overall methylation status. Epigenome-wide association analyses identifying CpG sites that are significantly differentially methylated in (b) dysplastic nevi and (c) benign nevi relative to adjacent normal skin, adjusted for estimated proportions of epithelial cells, fibroblasts, and melanocytes. Red dashed lines indicated an FDR significance threshold of Q < 0.05. (d) Overlap of identified hypermethylated and hypomethylated (Q < 0.05) in benign and dysplastic nevi relative to adjacent normal skin. (e) Enrichment analysis of all differentially methylated CpG loci in dysplastic nevi relative to adjacent normal skin.

Hypermethylated loci in dysplastic nevi demonstrate strong enrichment for CpG dense CpG island regions

When comparing nevi classified as dysplastic (n = 19) to paired adjacent normal skin, 15,617 loci were identified as differentially methylated between the two tissue types (FDR < 0.05, Figure 1B, Supplementary Table S4), and only seven loci were differentially methylated in benign nevi (n = 13) relative to paired adjacent normal skin (FDR < 0.05, Figure 1C, Supplementary Table S4). Of the significantly differentially methylated loci, only one hypermethylated (cg11264521) and one hypomethylated loci (cg02233071) were shared between benign and dysplastic nevi relative to adjacent normal skin (Figure 1D). Among the 15,617 differentially methylated loci in dysplastic nevi, 1,915 loci were hypermethylated relative to adjacent normal skin and these 1,915 hypermethylated loci demonstrated 5.6-fold enrichment (95% CI: 5.1, 6.1, p = 2.4 × 10−290) for CpG island regions (Figure 1E). Among the five most significantly hypermethylated loci were CpG loci mapping to the promoter regions of TTC21A, ASAP1, HES5, PREPL, and the documented melanoma tumor suppressor RASSF5 (Aoyama et al., 2004) (Supplementary Table S5A). Conversely, the most hypomethylated loci predominantly mapped to gene body regions (Supplementary Table S5B), including GALNT15, SSH1, COBL, and CYGB.

Differentially methylated loci in dysplastic nevi are enriched for gene regulatory regions and active gene promoters in skin

The 1,915 hypermethylated loci were strongly enriched for marks of active promoters in all skin cell types from the Roadmap Epigenomics Mapping Consortium (Bernstein et al., 2010), including chromatin accessibility assessed by DNase I hypersensitive sites sequencing (DNase-Seq, log-odds ratio up to 3.71 in foreskin melanocytes) and chromatin immunoprecipitation sequencing (ChIP-Seq) for histone marks including H3K4me3, H3K27ac, H3K9ac, as well as the active enhancer mark H3K4me1 (Figure 2A). Of these skin cell type histone mark ChIP-seq datasets, the hypermethylated loci were most enriched consistently in foreskin melanocytes. Using the Encyclopedia of DNA Elements (ENCODE) resource, the hypermethylated loci were also enriched for transcription start site genomic segments based on predictions of regulatory activity in 6 cell types (Figure 2B) and strongly enriched for the binding of RNA polymerase II (POL2RA) across 52 cell types and contexts (Consortium, 2012) (Figure 2C).

Figure 2: Enrichment of differentially methylated loci in dysplastic nevi in functional genomic regions.

Figure 2:

Open in a new tab

(a) log-odds ratio enrichment of hyper- and hypomethylated CpG loci in fibroblast, keratinocyte, and melanocyte Roadmap epigenomic datasets (* = Q < 0.05, Fisher’s exact test). (b) log-odds ratio enrichment of differentially methylated loci in ENCODE genome segments in 6 cell types (* = Q < 0.05 across all cell types, Fisher’s exact test). (c) log-odds ratio enrichment of differentially methylated loci in POL2RA ChIP-Seq peaks in 52 cell types from ENCODE.

Candidate oncogenes and tumor suppressor genes in melanoma demonstrate altered DNA methylation profiles in benign and dysplastic nevi relative to adjacent normal skin

Mean methylation was assessed in candidate oncogenes and tumor suppressor genes. Please refer to Supplementary Methods for details of these analyses. After adjusting for cellular composition, whilst the effect size was considered small, MAP2K1 and CDKN2A demonstrated statistically significant (FDR < 0.05) hypomethylation of the gene body in dysplastic nevi relative to adjacent normal skin (Supplementary Table S5A and Supplementary Table S7). Next, PIK3CA demonstrated statistically significant promoter hypomethylation while MDM2 demonstrated statistically significant promoter hypermethylation (Supplementary Table S6A and Supplementary Table S7) in dysplastic nevi relative to adjacent normal skin. Additionally, MYC showed statistically significant hypomethylation of the gene body in benign nevi relative to adjacent normal skin (Supplementary Table S6B).

Nevi with a reticular/non-specific pattern have distinct DNA methylation profiles relative to adjacent normal skin

Independent of histopathological diagnosis, we next sought to determine if there were any global methylation differences based upon dermoscopic classifications. Unsupervised clustering of the top 5,000 most variable loci across all samples demonstrated no distinct clustering patterns of reticular and non-specific nevi based on dermoscopic pattern (Supplementary Figure S5). Therefore, for subsequent analyses, dermoscopic patterns were grouped as ‘globular’ vs ‘non-globular’. Interestingly, globular nevi (n = 10) had no identified differentially methylated CpG loci relative to paired adjacent normal skin at an FDR threshold of Q < 0.05 (Figure 3A). However, when assessing nevi classified as non-globular (reticular/non-specific; n = 22), there were 59,572 loci identified as differentially methylated relative to paired adjacent normal skin at an FDR threshold of Q < 0.05 (Figure 3B). Of these loci, 8,852 were hypermethylated in nevi relative to adjacent normal skin which demonstrated a 1.9-fold enrichment (95% CI: 1.8, 2.0; p = 3.2 × 10−146) for CpG dense CpG island regions (Figure 3C). 1,458 of these hypermethylated loci were also identified as hypermethylated in dysplastic nevi relative to adjacent normal skin (Supplementary Figure S6).

Figure 3: Epigenome-wide association analyses identifying CpG sites that are significantly differentially methylated in globular nevi and non-globular nevi relative to adjacent normal skin.

Figure 3:

Open in a new tab

The data represented in panel (a) and (b) illustrates CpG sites in globular (a) and non-globular (b) nevi that have been adjusted for estimated proportions of epithelial cells, fibroblasts, and melanocytes. Red dashed lines indicated an FDR significance threshold of Q < 0.05. (c) Enrichment analysis of all differentially methylated CpG loci in non-globular nevi relative to adjacent normal skin.

Differentially methylated loci in nevi with a reticular/non-specific pattern are enriched for gene regulatory regions and active gene promoters in the skin

To further investigate the location and function of the broad methylation changes observed in reticular/non-specific patterned nevi compared to adjacent normal skin, we tested the differentially methylated loci for enrichment in functional genomic regions using publicly-available datasets (Locus Overlap Analysis or LOLA) (Sheffield and Bock, 2016) as described above. Notably, the 8,852 hypermethylated loci (compared to 1,915 loci when grouped as ‘dysplastic’) were strongly enriched for marks of active promoters in all skin cell types from the Roadmap Epigenomics Mapping Consortium (Bernstein et al., 2010), including chromatin accessibility assessed by DNase I hypersensitive sites sequencing (DNase-Seq; log-odds ratio up to 2.76 in adult dermal fibroblasts) and chromatin immunoprecipitation sequencing (ChIP-Seq) for histone marks including H3K4me3, H3K27ac, H3K9ac, as well as the active enhancer mark H3K4me1 (Supplementary Figure S7A). Of these skin cell type histone mark ChIP-seq datasets, the hypermethylated loci were consistently most enriched in epidermal keratinocytes. Using the ENCODE resource, the 8,852 hypermethylated loci were also enriched for transcription start sites, promoter-flanking, and enhancer genomic segments based on predictions of regulatory activity in 6 cell types (Supplementary Figure S7B) and enriched for the binding of RNA polymerase II (POL2RA) across 52 cell types and contexts from the ENCODE consortium (Consortium, 2012) (Supplementary Figure S7C).

Investigation of the 50,720 hypomethylated loci revealed that these were not enriched for promoter-specific features. Instead, they were enriched for accessible chromatin in melanocytes (log-odds ratio 1.22) and fibroblasts derived from adult foreskins (log-odds ratio up to 1.09), as well as H3K4me1 in all Roadmap skin cell types (log-odds ratio up to 1.89 in foreskin melanocytes; Supplementary Figure S7A). Additionally, H3K27ac histone marks were significantly enriched in adult foreskin melanocytes (log-odds ratio 1.06) and adult dermal fibroblasts (log-odds ratio 1.03), suggesting this set of hypomethylated loci may contain enhancer elements (Supplementary Figure S7A). Further analysis indicated that the hypomethylated loci were enriched for ENCODE enhancer segments and strongly enriched for weak enhancer segments (Supplementary Figure S7B).

PTEN demonstrates promoter hypermethylation in nevi with a reticular/non-specific pattern relative to adjacent normal skin

Of the hypermethylated loci in non-globular nevi relative to adjacent normal skin that occur in gene promoter regions, PTEN had the highest proportion of hypermethylated promoter CpG loci among all genes (Table 2). Further investigation of the pattern of promoter hypermethylation demonstrated that the 9 significantly hypermethylated loci (Q < 0.05) in non-globular nevi relative to adjacent normal skin are grouped at the start of the promoter region (Figure 4).

Table 2:

Genes with the top 10 most differentially methylated promoter regions (sorted first by total number of hypermethylated promoter loci (Q < 0.05) and them by the proportion of CpG loci in the promoter that were hypermethylated

Gene Hypermethylated Loci Total Loci Proportion Hypermethylated
PTEN 9 42 0.21
PRR15 8 9 0.89
SPTBN1 6 17 0.35
PAMR1 5 9 0.56
NFIX 5 10 0.5
MIR375 5 12 0.42
LOC100132215 5 15 0.33
ADM 5 17 0.29
HES5 4 9 0.44
ATOH8 4 10 0.4
Open in a new tab

Figure 4: Distribution of methylation across the PTEN promoter in non-globular and adjacent normal skin.

Figure 4:

Open in a new tab

The zoomed in section of the PTEN promotor region highlights the first 15 loci of each sample which contains the 9 hypermethylated (Q < 0.05) loci in non-globular nevi relative to adjacent normal skin.

DISCUSSION

Using genome-scale methylation profiling, we have presented evidence that 13 benign nevi assessed in this study were significantly different from 19 dysplastic nevi, sharing only two differentially methylated loci relative to adjacent normal skin. As previously stated, the 13 benign nevi in our study primarily were raised or papillomatous and exhibited a globular (69%) dermoscopic pattern; whereas the dysplastic nevi were flat and had a reticular/non-specific (95%) pattern. We have previously illustrated that globular nevi are genomically stable, with no copy number aberrations (CNA) present, which contrasts with reticular/non-specific nevi, which have numerous CNAs (Stark et al., 2018).

To provide further support of their stability, we have now confirmed that our collection of benign nevi has significantly higher mean methylation at LINE-1 and Alu repeat elements (which are commonly methylated across the genome in ‘normal’ cells) compared to adjacent normal skin, which contrasts with dysplastic nevi that had lower mean methylation. LINE-1 and Alu repeat elements are commonly found in the intergenic regions of our genome. These retrotransposable elements are related to genomic stability via methylation regulation. Furthermore, hypomethylation of these elements has been associated with genomic instability (Ayarpadikannan and Kim, 2014, Su et al., 2012). These data confirm our prior CNA analysis of globular nevi having no CNAs and provide reasoning for the increased CNA observed in reticular/non-specific nevi (Stark et al., 2018). In addition, we focused specifically on the chr9p21 region (containing the CDKN2A gene locus) to identify any differential methylation patterns in intergenic regions, as this region has been previously associated with “flat” nevi using quantitative-trait locus (QTL) analysis (Zhu et al., 1999). In this study by Zhu et al, flat nevi were classed as junctional or compound (i.e., primarily reticular/non-specific) (Zhu et al., 1999). The authors concluded that since germline mutations were rare in CDKN2A, non-coding regions of this locus were likely responsible for high (flat) nevus count (Zhu et al., 1999). We provide evidence that hypomethylation of LINE-1 repeat elements was significantly associated with our collection of dysplastic nevi, which were 95% flat or reticular/non-specific. Interestingly, LINE-1 elements are located within ~0.5 MB from D9S942, the specific polymorphic marker described in the Zhu et al. study (Zhu et al., 1999).

A strength of our approach is the development of a library for the estimation of cellular composition of samples. While in our study, only modest differences are observed between dysplastic nevi and adjacent normal skin, after adjusting for cellular composition and accounting for multiple comparisons, loci that were identified as being hypermethylated in dysplastic nevi demonstrated strong enrichment for CpG dense CpG island regions. Importantly, while our cell type library performed well in validation analyses, there remains the potential for error in the estimation of cellular composition, thus possibly introducing bias into our findings as these results were heavily altered after adjustment for cellular composition. CpG islands are often hypermethylated in tumor tissue relative to adjacent normal skin and commonly overlap with gene promoter regions (Esteller, 2002). Therefore, the hypermethylation events observed in our set of dysplastic nevi may play a role in early carcinogenic processes in melanoma.

Additionally, when assessing oncogenes and tumor suppressor genes previously implicated in melanoma, we observed promoter hypomethylation, associated with increased transcription (Jones and Baylin, 2002), of documented oncogene PIK3CA and promoter hypermethylation, associated with transcriptional repression (Suzuki and Bird, 2008), in MDM2. Since MDM2 generally acts in an oncogenic manner, precisely why MDM2 would have increased promoter methylation remains to be determined. We also observed gene body hypomethylation, associated with decreased gene transcription (Suzuki and Bird, 2008), in the documented oncogene MAP2K1 (MEK) as well as the tumor suppressor gene CDKN2A, in dysplastic nevi relative to adjacent normal skin.

While significant differences were observed between dysplastic nevi and adjacent normal skin, vastly more differentially methylated loci were observed between reticular/non-specific nevi and adjacent normal skin. Similar to the analysis of the dysplastic nevus group, hypermethylated loci in reticular/non-specific nevi relative to adjacent normal skin were enriched for CpG island regions which commonly overlap with gene promoter regions (Esteller, 2002). Importantly, these hypermethylated loci were also enriched for gene promoter regions, enhancer regions, and regions enriched for RNA polymerase II binding, suggesting that the gain of DNA methylation observed in these nevi plays a role in the regulation of gene expression.

In gene-level assessments of DNA methylation alterations, we identified PTEN as having the most significantly hypermethylated loci in its promoter region of any tested gene. This unbiased assessment confirms a recent study that found that the PTEN promoter is commonly methylated in nevi, albeit significantly lower than melanoma (Salgado et al., 2020). Given that promoter hypermethylation is associated with gene silencing (Suzuki and Bird, 2008) and that PTEN is a well-documented tumor suppressor gene across multiple tumor types, including melanoma (Wu et al., 2003), this suggests that alterations to the regulation of PTEN occur early on in the development in melanoma and can be detected in flat acquired nevi, potentially priming these nevi for subsequent carcinogenesis. The MAPK pathway in nevi is constitutively activated, commonly by mutations in BRAF, leading to increased proliferation. Some studies have shown that this overactivation leads to oncogene-induced senescence (OIS) (Michaloglou et al., 2005) and, in turn, the stability of nevi. This assumption cannot be applied to all benign vs dysplastic, or globular vs reticular/non-specific nevi, and only partially explains their relative stability. We have previously documented that BRAF was 100% mutated in nevi that are actively growing (the presence of globules around the periphery of the nevus) (Tan et al., 2018). Our observation of PTEN promoter hypermethylation in nevi with a reticular/nonspecific pattern (which includes nevi that are actively growing) provides further support for the activation of the PI3K pathway, being a crucial step to overcome BRAF V600E-induced senescence. This notion is supported by an elegant study using a BRAFV600E-driven mouse model, which induced nevus growth (Vredeveld et al., 2012). The authors showed that depletion of PTEN using shRNA led to tumor progression via the activation of PI3K. Melanoma is commonly mutated for BRAF and NRAS and rarely for PIK3CA or its subunits (Curtin et al., 2006) which is mirrored in AMN. The growth of a nevus is often a very slow process, with active growth spanning years before growth arrest occurs. PTEN promoter hypermethylation likely plays a role in this process, but other mechanisms warrant further investigation.

The limitations of this study are the number of nevi diagnosed as benign that have a reticular/non-specific dermoscopic pattern. All our nevi were clinically benign and selected based upon their dermoscopic pattern. Since 86% (19 of 22) of the nevi with a reticular/non-specific pattern were diagnosed as dysplastic following histopathology, it would be important to determine if this trend occurs in a larger sample collection. The benign nevi collection analyzed in this study has evidence of UVR damage following mutation signature analysis (Stark et al., 2020). Despite the continued exposure to UVR, known to contribute to hypomethylation in photo-aged skin (Vandiver et al., 2015), as observed in our collection of dysplastic nevi, the methylation status of our benign nevi did not fit this model, which is likely due to phenotypic differences as described. An additional limitation of this study is that by only measuring bisulfite converted DNA and not tandem bisulfite and oxy-bisulfite converted DNA, all measures of DNA methylation reflect the relative proportion of both methylated and hydroxymethylated DNA. Therefore, we are unable to draw specific conclusions as to whether the alterations observe reflect alterations in methylation or hydroxymethylation of DNA. Overall, the increased global methylation and methylation of repeat elements is likely one of the driving mechanisms that contribute to nevus stability and the rarity of malignant transformation.

MATERIALS AND METHODS

Study population

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. With written informed consent, 32 acquired nevi with the globular (n=10), reticular/non-specific (n=12), and non-specific nevi with a peripheral rim of globules (n=10) dermoscopic patterns identified from a database of prospectively imaged naevi (Daley et al., 2017) were shave excised from 24 participants (17 male, 7 female; mean age 51.88 years). Of note, 8 of the 24 participants provided consent for two nevi to be shaved excised (n=16). Please refer to Supplementary Text for more details.

Bisulfite modification for Illumina EPIC array and data processing

DNA extracted from each nevus and matching adjacent skin (n=64) was sent to the fee-for-service provider University of Southern California (USC) Norris Comprehensive Cancer Center Molecular Genomics Core. Please refer to Supplementary Text for more details.

Reference library development and cell-type estimation

We developed a reference library for DNA methylation-based cell type deconvolution of skin biopsies using publicly available cell line data for epithelial cells, fibroblasts, and melanocytes (GSE74877). The library also included primary purified B cells, CD4+ T cells, CD8+ T cells, NK cells, monocytes, and neutrophils from our prior work (GSE110554). Please refer to Supplementary Text for more details.

Repeat element estimation

The methylation status of Alu and long interspersed nucleotide element-1 (LINE-1) repetitive elements were predicted using the REMP package in R (Zheng et al., 2017). Median values of Alu and LINE-1 were calculated across all estimated regions for each sample. Please refer to Supplementary Text for more details.

Statistical analysis

Differential methylation was assessed using linear mixed-effects models, modeling the relationship between logit-transformed beta values (M-values) in nevi samples relative to paired adjacent normal samples, controlling for the estimated relative proportions of epithelial cells, fibroblasts, and melanocytes in each sample. Please refer to Supplementary Text for more details.

Supplementary Material

Figure S3
Figure S2
Figure S1
Figure S4
Figure S6
Figure S5
Figure S7
Supplementary Materials and Methods
Supplementary Tables S1, S2, S3, S5, S6
Supplementary table S4
Supplementary Table S7

ACKNOWLEDGMENTS

The authors would like to thank the study participants and Dr Harald Oey for his technical assistance. This work was funded by a grant from 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. This work was additionally supported by grants R01CA216265 and R01CA253976 to BCC. LAS was supported by CDMRP/Department of Defense (W81XWH-20-1-0778) and NIGMS (P20GM104416-09 8299). MSS holds a fellowship (APP1106491) from the NHMRC. HPS holds an NHMRC MRFF Next Generation Clinical Researchers Program Practitioner Fellowship (APP1137127).

Abbreviations:

AMN

acquired melanocytic nevi

AUC

area under the curve

CAN

copy number aberrations

CpG

5’ C phosphate G 3’

DNA

deoxyribonucleic acid

DM

differentially methylated

ENCODE

Encyclopedia of DNA Elements

FDR

false discovery rate

LINE-1

class I transposable elements

LOLA

Locus Overlap Analysis

OIS

oncogene-induced senescence

PCR

polymerase chain reaction

QTL

quantitative-trait locus

RNA

ribonucleic acid

shRNA

short-hairpin RNA

UVR

ultraviolet radiation

Footnotes

CONFLICT OF INTEREST

HPS is a shareholder of MoleMap NZ Limited and e-derm consult GmbH and undertakes regular teledermatological reporting for both companies. HPS is a Medical Consultant for Canfield Scientific Inc and MoleMap Australia Pty Ltd, and a Medical Advisor for First Derm

DATA AVAILABILITY

The datasets generated and analyzed during the current study are available in the Gene Expression Omnibus under the accession number GSE188593 (https://www.ncbi.nlm.nih.gov/geo/).

REFERENCES

  1. Aoyama Y, Avruch J, Zhang XF. Nore1 inhibits tumor cell growth independent of Ras or the MST1/2 kinases. Oncogene 2004;23(19):3426–33. [DOI] [PubMed] [Google Scholar]
  2. Ayarpadikannan S, Kim HS. The impact of transposable elements in genome evolution and genetic instability and their implications in various diseases. Genomics Inform 2014;12(3):98–104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Bajaj S, Dusza SW, Marchetti MA, Wu X, Fonseca M, Kose K, et al. Growth-Curve Modeling of Nevi With a Peripheral Globular Pattern. JAMA Dermatol 2015;151(12):1338–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bernstein BE, Stamatoyannopoulos JA, Costello JF, Ren B, Milosavljevic A, Meissner A, et al. The NIH Roadmap Epigenomics Mapping Consortium. Nat Biotechnol 2010;28(10):1045–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Colebatch AJ, Ferguson P, Newell F, Kazakoff SH, Witkowski T, Dobrovic A, et al. Molecular Genomic Profiling of Melanocytic Nevi. J Invest Dermatol 2019;139(8):1762–8. [DOI] [PubMed] [Google Scholar]
  6. Consortium EP. An integrated encyclopedia of DNA elements in the human genome. Nature 2012;489(7414):57–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Conway K, Edmiston SN, Khondker ZS, Groben PA, Zhou X, Chu H, et al. DNA-methylation profiling distinguishes malignant melanomas from benign nevi. Pigment Cell Melanoma Res 2011;24(2):352–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Curtin JA, Stark MS, Pinkel D, Hayward NK, Bastian BC. PI3-kinase subunits are infrequent somatic targets in melanoma. J Invest Dermatol 2006;126(7):1660–3. [DOI] [PubMed] [Google Scholar]
  9. Daley GM, Duffy DL, Pflugfelder A, Jagirdar K, Lee KJ, Yong XLH, et al. GSTP1 does not modify MC1R effects on melanoma risk. Exp Dermatol 2017;26(8):730–3. [DOI] [PubMed] [Google Scholar]
  10. Elmore JG, Barnhill RL, Elder DE, Longton GM, Pepe MS, Reisch LM, et al. Pathologists’ diagnosis of invasive melanoma and melanocytic proliferations: observer accuracy and reproducibility study. BMJ 2017;357:j2813. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Esteller M CpG island hypermethylation and tumor suppressor genes: a booming present, a brighter future. Oncogene 2002;21(35):5427–40. [DOI] [PubMed] [Google Scholar]
  12. Gerami P, Busam KJ. Acquired Melanocytic Nevi. In: Busam KJ, Germani P, Scolyer RA, editors. Pathology of Melanocytic Tumors. Philadelphia, PA 19103–2899: Elsevier; 2018. p. 8–25. [Google Scholar]
  13. Jones PA, Baylin SB. The fundamental role of epigenetic events in cancer. Nat Rev Genet 2002;3(6):415–28. [DOI] [PubMed] [Google Scholar]
  14. Kittler H, Seltenheim M, Dawid M, Pehamberger H, Wolff K, Binder M. Frequency and characteristics of enlarging common melanocytic nevi. Arch Dermatol 2000;136(3):316–20. [DOI] [PubMed] [Google Scholar]
  15. Michaloglou C, Vredeveld LC, Soengas MS, Denoyelle C, Kuilman T, van der Horst CM, et al. BRAFE600-associated senescence-like cell cycle arrest of human naevi. Nature 2005;436(7051):720–4. [DOI] [PubMed] [Google Scholar]
  16. Pollock PM, Harper UL, Hansen KS, Yudt LM, Stark M, Robbins CM, et al. High frequency of BRAF mutations in nevi. Nat Genet 2003;33(1):19–20. [DOI] [PubMed] [Google Scholar]
  17. Roos L, Sandling JK, Bell CG, Glass D, Mangino M, Spector TD, et al. Higher Nevus Count Exhibits a Distinct DNA Methylation Signature in Healthy Human Skin: Implications for Melanoma. J Invest Dermatol 2017;137(4):910–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Salgado C, Oosting J, Janssen B, Kumar R, Gruis N, van Doorn R. Genome-wide characterization of 5-hydoxymethylcytosine in melanoma reveals major differences with nevus. Genes Chromosomes Cancer 2020;59(6):366–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Shain AH, 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(20):1926–36. [DOI] [PubMed] [Google Scholar]
  20. Sheffield NC, Bock C. LOLA: enrichment analysis for genomic region sets and regulatory elements in R and Bioconductor. Bioinformatics 2016;32(4):587–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Stark MS, Denisova E, Kays TA, Heidenreich B, Rachakonda S, Requena C, et al. Mutation Signatures in Melanocytic Nevi Reveal Characteristics of Defective DNA Repair. J Invest Dermatol 2020;140(10):2093–6 e2. [DOI] [PubMed] [Google Scholar]
  22. Stark MS, Tan JM, Tom L, Jagirdar K, Lambie D, Schaider H, et al. Whole-Exome Sequencing of Acquired Nevi Identifies Mechanisms for Development and Maintenance of Benign Neoplasms. J Invest Dermatol 2018;138(7):1636–44. [DOI] [PubMed] [Google Scholar]
  23. Su J, Shao X, Liu H, Liu S, Wu Q, Zhang Y. Genome-wide dynamic changes of DNA methylation of repetitive elements in human embryonic stem cells and fetal fibroblasts. Genomics 2012;99(1):10–7. [DOI] [PubMed] [Google Scholar]
  24. Suzuki MM, Bird A. DNA methylation landscapes: provocative insights from epigenomics. Nat Rev Genet 2008;9(6):465–76. [DOI] [PubMed] [Google Scholar]
  25. Tan JM, Tom LN, Jagirdar K, Lambie D, Schaider H, Sturm RA, 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(1):191–7. [DOI] [PubMed] [Google Scholar]
  26. Tan JM, Tom LN, Soyer HP, Stark MS. Defining the Molecular Genetics of Dermoscopic Naevus Patterns. Dermatology 2019;235(1):19–34. [DOI] [PubMed] [Google Scholar]
  27. Vandiver AR, Irizarry RA, Hansen KD, Garza LA, Runarsson A, Li X, et al. Age and sun exposure-related widespread genomic blocks of hypomethylation in nonmalignant skin. Genome Biol 2015;16:80. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Vredeveld LC, Possik PA, Smit MA, Meissl K, Michaloglou C, Horlings HM, et al. Abrogation of BRAFV600E-induced senescence by PI3K pathway activation contributes to melanomagenesis. Genes Dev 2012;26(10):1055–69. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Wu H, Goel V, Haluska FG. PTEN signaling pathways in melanoma. Oncogene 2003;22(20):3113–22. [DOI] [PubMed] [Google Scholar]
  30. Zheng Y, Joyce BT, Liu L, Zhang Z, Kibbe WA, Zhang W, et al. Prediction of genome-wide DNA methylation in repetitive elements. Nucleic Acids Res 2017;45(15):8697–711. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Zhu G, Duffy DL, Eldridge A, Grace M, Mayne C, O’Gorman L, et al. A major quantitative-trait locus for mole density is linked to the familial melanoma gene CDKN2A: a maximum-likelihood combined linkage and association analysis in twins and their sibs. Am J Hum Genet 1999;65(2):483–92. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Figure S3
NIHMS1763681-supplement-Figure_S3.jpg (64.7KB, jpg)
Figure S2
NIHMS1763681-supplement-Figure_S2.jpg (49.7KB, jpg)
Figure S1
NIHMS1763681-supplement-Figure_S1.jpg (125.7KB, jpg)
Figure S4
NIHMS1763681-supplement-Figure_S4.jpg (58.8KB, jpg)
Figure S6
NIHMS1763681-supplement-Figure_S6.jpg (25.5KB, jpg)
Figure S5
NIHMS1763681-supplement-Figure_S5.jpg (76.4KB, jpg)
Figure S7
NIHMS1763681-supplement-Figure_S7.jpg (38.4KB, jpg)
Supplementary Materials and Methods
NIHMS1763681-supplement-Supplementary_Materials_and_Methods.docx (27.7KB, docx)
Supplementary Tables S1, S2, S3, S5, S6
NIHMS1763681-supplement-Supplementary_Tables_S1__S2__S3__S5__S6.xlsx (98.3KB, xlsx)
Supplementary table S4
NIHMS1763681-supplement-Supplementary_table_S4.xlsx (1.4MB, xlsx)
Supplementary Table S7
NIHMS1763681-supplement-Supplementary_Table_S7.xlsx (27.6KB, xlsx)

Data Availability Statement

The datasets generated and analyzed during the current study are available in the Gene Expression Omnibus under the accession number GSE188593 (https://www.ncbi.nlm.nih.gov/geo/).

RESOURCES