Melanocytes, pigment cells of neural crest origin, are responsible for melanin production and epidermal UVR photoprotection [1–3]. While protective against UVR-induced DNA damage [4,5], normal melanocytes at sun-exposed sites exhibit high mutational burden [6]. Evidence suggests that melanocytes or melanocyte stem cells are the cell of origin for both benign melanocytic nevi and melanoma [7]. Cutaneous melanoma accounts for the majority of skin cancer fatalities and prevalence continues to rise [8,9].

At the molecular level, the most common melanocytic tumors, including acquired nevi, congenital nevi, and cutaneous melanoma, are most often initiated through activating mutations in the MAPK pathway, predominantly in BRAF or NRAS genes [10,11]. Nevi are senescent lesions, and malignant transformation from a benign nevus is a rare event that typically requires cooperating alterations such as CDKN2A inactivation, TERT promoter mutations, and/or loss of PTEN or TP53 [12]. Non-BRAF/NRAS driver mutations are commonly found in melanomas from non-cutaneous sites [13].

BRAF or NRAS activating mutations are present in both benign nevi and malignant melanoma. It is thought that BRAF or NRAS mutation in a melanocyte or melanocyte stem cell leads to a period of proliferation followed by senescence, where nevi maintain a stable size over time. There is strong evidence that nevus cells are largely senescent although evidence suggests that a population of non-senescent cells exist within most nevi [14]. The exact mechanisms of senescence in nevi are controversial, with proposed models of oncogene induced senescence, dysfunctional telomere related senescence, and growth arrest by interaction with neighboring cells.

The exact dynamics of growth arrest after BRAF^V600E activation are unknown. In a given individual, nevi can vary dramatically in size, suggesting that even in genetically identical backgrounds that cells respond differentially to BRAF^V600E activation. In cell culture systems, overexpression of BRAF^V600E leads to immediate growth arrest without a proliferative period, while in vivo BRAF^V600E leads to nevus formation in zebrafish and mouse models BRAF^V600E [7,15,16]. It is likely that expression level plays an important role (i.e. overexpression of BRAF leads to immediate growth arrest while endogenous levels of expression may promote proliferation and nevus formation), as it has been shown in cultured primary human melanocytes that CRISPR/Cas9 introduction of mutant BRAF^V600E (at endogenous levels) leads to a gradual growth advantage [17]. It is also possible that melanocytes from a given individual may have differential competence to the same degree of BRAF^V600E signaling, and that some melanocytes might immediately undergo growth arrest while others may proliferate and form nevi. Notably, nodular melanoma is an aggressive melanoma variant that only rarely has a nevus associated precursor. It is thought to be caused by oncogene activation in a highly mutated melanocyte or melanocyte stem cell that has acquired the genomic/epigenomic alterations to bypass senescence and respond aggressively to oncogene activation [18].

While acquired melanocytic nevi are benign lesions, they have the capacity to progress to melanoma at a low frequency [19]. At the genetic/genomic level this progression has been well characterized. CDKN2A deletion and TERT promoter lesions are early events found in melanoma in situ, and absent from precursor nevi, while other alterations including mutations of p53 and PI3K (phosphatidylinositol 3-kinase) pathway components (including PTEN mutation and deletion) are predominantly found at the stage of invasive melanoma [12,20].

Less common variants of benign melanocytic neoplasms include blue nevi and Spitz nevi, and deep penetrating nevi (DPN) each exhibiting clinical, histological, and genetic/genomic features that are distinct from common acquired nevi. Blue nevi are characterized by high prevalence GNAQ mutations and less commonly activating PKC genomic fusions [21,22]. Spitz nevi are characterized by mutually exclusive fusions in ROS1, NTRK1, ALK, BRAF, and RET [23]. In addition to activating BRAF fusions, activating truncating mutations in MAP3K8 are found in both spitzoid and conventional melanomas, suggesting shared origins and/or biology of these tumor types [24].

Acquired melanocytic nevi have long been classified histopathologically as junctional, compound, or intradermal. These histologic subtypes correspond to clinical and dermatoscopic features: junctional nevi are flat and exhibit reticular dermatoscopic patterns, while intradermal nevi are raised, often non-pigmented, and exhibit a globular dermatoscopic pattern [25–28]. Emerging evidence suggests molecular differences between these nevus types, including oncogenic mutation spectrum.

Nevi with a globular pattern or peripheral rim of globules (PG), and/or those having intradermal growth pattern almost universally harbor BRAF mutations (92–100%) [25,27,29,30]. Conversely, these same studies show that reticular-patterned nevi and/or nevi with junctional growth pattern more often contain NRAS mutations or lack BRAF^V600E mutations (33%), although BRAF mutations are also present in approximately 67% of reticular nevi. These features also correspond to anatomical distribution patterns. BRAF-mutant nevi are more commonly located on the face and trunk, especially in younger individuals, while NRAS-associated reticular nevi show no distinct site preference but predominate in older populations [25,27].

Histologically, BRAF-mutant nevi (especially globular-patterned) typically feature large, well-circumscribed melanocytic nests, periadnexal and perivascular proliferation, and congenital-type features [27,29]. NRAS-mutant nevi, conversely, exhibit lentiginous growth restricted to epidermal ridges, smaller and more irregular nests, and a markedly higher rate of architectural disorder and cytologic atypia [27,30]. Molecularly, reticular nevi display increased copy number alterations (CNAs), loss of heterozygosity (LOH), and greater mutational heterogeneity [31]. Importantly, dysplastic nevi are less likely to exhibit a globular dermatoscopic pattern, and dysplastic nevi demonstrate extensive differential methylation at transcriptionally active loci [32].

Congenital melanocytic nevi are present at birth, and are classified based on projected adult size [33]. Giant congenital melanocytic nevi (GCMN) are defined as adult size ≥20cm and these lesions are disfiguring and at greater risk for transformation to melanoma. While it was initially reported that all GCMN are solely due to NRAS mutation, subsequent studies have revealed that BRAF^V600E mutations and BRAF activating fusions cause GCMN at a lower frequency, and BRAF/NRAS status impacts melanocytic nevus growth and architecture [34]. The largest studies to date have shown that about 68% of GCMN are due to NRAS mutation, BRAF^V600E in 7%, and approximately 7% with activating BRAF genomic fusions [34,35].

While in acquired nevi BRAF mutations lead to only a brief period of proliferation, in GCMN their presence defines a more dynamic, hyperproliferative pathway of development. BRAF-mutant GCMN usually have a multinodular appearance: numerous small, closely packed nodules of the same size with a hard, mobile center within a softer mass [35]. Additionally, BRAF fusion rearrangements are associated with intense pruritus, hyperproliferation, and the development of satellite lesions to mucosal sites such as the oral cavity [36–38]. Interestingly, in GCMN patients having the largest lesions (predicted adult size (PAS) >60 cm), NRAS mutations were overrepresented (91% of cases). As larger lesions are thought to form earlier during development, this suggests the possibility that non-NRAS driver mutations are not tolerated during early embryogenesis [35]. Interestingly, two case reports have linked ALK and NRTK1 fusions (common drivers in Spitz nevi) as rare drivers of congenital nevi; these lesions were BRAF/NRAS negative and exhibited aggressive clinical features of proliferation and or melanoma formation [39,40].

Histologically, BRAF-mutant lesions show an abnormal growth pattern characterized by the subcutaneous infiltrations of nevus cells in adipose tissue in a string-like formation between adipocytes [41]. Immunohistochemical staining has intense cytoplasmic expression of BRAF^V600E, which demonstrates clonal proliferation [35]. Invasive growth into deeper tissues, large numbers of nests, and higher Ki-67 proliferation indices of BRAF-positive lesions are shown, compared to their BRAF-negative counterparts, reflecting augmented mitotic activity. Conversely, NRAS-mutant GCMNs, exhibit disordered pigmentation, varied hairiness, surface irregularities, and asymmetric margins, without consistent architectural characteristics that would be suggestive of mutation status [37,42].

GCMN are associated at a low frequency with serious extracutaneous abnormalities, including neurological involvement, typical facial features, and endocrine disturbance [35]. Interestingly, both Polubothu et al. and Zhao et al. did not identify dissimilarities in the incidence of congenital neurological anomalies or melanoma development by mutation type.

When compared to cutaneous sites, which are driven by BRAF and NRAS mutations, other anatomic locations, are associated with distinct mutational landscapes of both benign nevi and malignant melanoma [22,43–45].

Acral sites

Melanocytic lesions from acral sites (soles, palms, nail beds) have a characteristic molecular profile. Acral nevi, while still most frequently BRAF-mutant (66.7%), have a broader mutational spectrum, such as GNAQ, KIT, NRAS, and NF1 [44,46]. Cytomorphologic equivalents show that the mutational subtype influences cell morphology - e.g., BRAF mutations being associated with epithelioid cells, and GNAQ/NF1 with spindle cells [44,46]. Interestingly, acral melanomas show increased mutational heterogeneity, with lower frequencies of BRAF mutations (34.4%) and increased frequencies of structural alterations and gene amplifications (CRKL, GAB2, PAK1, CCND1, TERT, MDM2) [47,48]. KIT mutations are particularly associated with amelanotic acral melanoma subtypes [44,46]. There is conflicting evidence on whether the number of acral nevi is a risk factor for acral melanoma [49]. In addition, it has been shown that while melanoma occurring on the dorsal hands and feet resemble other cutaneous melanoma in having high incidence of BRAF^V600E mutations and approximately 25% incidence of precursor nevus, those occurring one the volar surface have low frequency of BRAF mutation and ~10% precursor nevus association [50]. This divergence might suggest a decreased risk for acral melanomas progression from acral nevi that occur on palms/soles.

Mucosal sites

The genetic drivers of mucosal nevi have only been explored in a limited fashion, but there is evidence of divergence between nevi and melanoma patterns. In oral mucosal nevi, the BRAF^V600E mutation frequency is 42%, while the frequency in oral mucosal melanoma is only 6.42% [51]. Similarly, in the female genital tract, nevi were shown to be predominantly BRAF^V600E-positive (100% in common nevi, 75% in atypical nevi), while melanomas had <10% BRAF^V600E frequency [45,52].

Molecular studies of primary and metastatic melanoma are more numerous, and demonstrate a mutation spectrum including KIT mutations (16–26%), with lower frequencies of NRAS (15%) and BRAF (5–8%) mutations, and hardly any UVR-associated mutagenesis [53,54]. In addition, mucosal melanomas are most frequently found to have somatic structural alterations and chromosomal copy number changes rather than single-point mutations. A subset of mucosal melanomas (~9.5%) with GNAQ and GNAI mutations have worsened prognosis, thus also exhibiting evidence of molecular heterogeneity [55]. Overall, these findings argue that mucosal melanomas are less likely to evolve from benign precursor nevi.

As discussed previously, oncogenic driver mutations in melanoma differ based on anatomic site. For example, while BRAF/NRAS mutations are predominant in cutaneous melanoma, they are less common in acral melanoma and mucosal melanomas. The mechanism behind these differential mutation patterns is unclear. It is possible that the processes driving mutations differ based on anatomic site. Chronic ultraviolet radiation is thought to drive cutaneous melanoma at cutaneous sites, as evidenced by high mutational burden and predominance of “UV signature” mutations in the lesions [56]. High mutational burden and UVB signature are found in conjunctival melanomas (UV-exposed mucosal melanoma variant), and subungual acral melanomas (that occur on dorsal, sun-exposed surfaces), consistent with the exposure of these sites to UV radiation [47,57]. Conversely, mutational burden is lower, and UVB signature largely absent, in mucosal and sun-protected acral melanoma.

Interestingly, BRAF^V600E and the majority of NRAS driver mutations in melanoma are not UV-signature mutations, which opens the question as to how these might be related to UV exposure. While evidence suggests that UVB can at low frequency cause DNA mutations identical to those that cause BRAF^V600E mutation, much evidence argues against a role of UVB in BRAF^V600E mutation [58]. Cutaneous melanomas that lack BRAF or NRAS mutations are most commonly occur on chronic sun-damaged sites (that receive the highest cumulative UVB dosage) and exhibit a very high mutational burden and strong UV signature, while BRAF^V600E mutant melanomas are less common on chronically sun-exposed sites and are associated with a lower mutational burden [59].

An alternative model is that melanocytes may have differential sensitivities to respond to different activating oncogenes. The first study to support this model demonstrated that after expression of BRAF^V600E along with additional mutations, the ability to transform to melanoma depended on the intrinsic transcriptional program present in the cell of origin [60]. In a melanoma prone p53^-/- genetic background, induction of BRAF^V600E in melanoblast and neural crest lineage caused melanoma induction, while BRAF^V600E did not cause melanoma when expressed in differentiated melanocytes. Enhanced expression of the chromatin modifier ATAD2 in melanoblast/neural crest lineages was linked to this differential response, suggesting that transcriptional program of the cell of origin, including chromatin modifiers such as ATAD2, regulate responses to oncogenic mutations.

Recent experimental work also suggests that melanocytes at different anatomic sites, despite a shared developmental origin, display distinct cell states that correlate with distinct oncogene response. This is a paradigm shift with profound implications: the same oncogene can initiate melanoma in one anatomical site and be biologically quiescent in another. Weiss et al., demonstrated that the CRKL gene is frequently amplified in human acral melanoma and that CRKL overexpression preferentially led to tumor development in the fins of zebrafish, the evolutionary equivalents of tetrapod limbs [61]. Tumorigenesis required the existence of HOX13-controlled positional identity (HOXA13, HOXB13, HOXD13), which controlled IGF signaling in conjunction with CRKL. Disruption of CRKL or HOX13 activity inhibited transformation in acral sites, emphasizing that oncogenic mutation can be transforming only in cells with the appropriate spatial transcriptional program. The relevance to humans is underscored by the enrichment of HOX13 expression and CRKL/GAB2 amplifications in human acral melanoma, in support of a conserved, site-specific susceptibility model [61].

A mucosal melanoma model was developed by Babu et al., where CCND1 overexpression in combination with PTEN and TP53 loss led to transformation solely in internal (mucosal-like) melanocytes, while cutaneous melanocytes harboring the identical mutations remained refractory [62]. Epigenomic and transcriptomic profiling revealed mucosal melanocytes possess an alternate chromatin structure, reduced antigen presentation, and elevated neural crest migration signatures. These elements were associated with PAX3 motif enrichment in open chromatin, suggesting that PAX3 confers a permissive transcriptional state for transformation. Overexpression of PAX3 increased the pool of internal melanocytes, supporting its role in mucosal melanocyte identity and oncogene competence [62].

Overall, recent evidence suggests that the interplay of genetic drivers and developmental and anatomic context impacts nevogenesis and melanoma pathogenesis. Identical oncogenic genetic alterations (e.g. BRAF^V600E or NRAS codon 61 mutations, CRKL amplification, CCND1 mutation) are able to exert different biologic effects depending upon melanocyte positional dependent transcriptional programs, epigenetic terrain, and microenvironmental signals as key determinants of malignant competence.

Further delineating the transcriptional, epigenetic, and microenvironmental programs specifying melanocytic positional identity will be critical in understanding the events responsible for melanoma formation, and may also provide important clues for differential therapy responses of melanoma types. For example acral and mucosal melanoma demonstrate much lower response to immunotherapy [63]. While low therapy response is correlated to lower mutational burden in these tumors, it is possible that epigenomic or other aspects of positional cell state may impact therapy responses. Through the integration of mutational, transcriptional, and epigenomic landscape of melanocytes, improved understanding of melanocytes positional and developmental identity will improve our understanding of melanocyte and melanoma biology.

KK, SMO, and DEF contributed to literature review and writing/editing of the manuscript.