Links between Schwann cells and melanocytes in development and disease

Authors


CORRESPONDENCE Catherine D. Van Raamsdonk, e-mail: cvr@mail.ubc.ca

Summary

Melanocytes are pigment-producing cells that reside in the skin, eyes, ears, heart, and central nervous system meninges of mammals. Schwann cells are glial cells, which closely associate with peripheral nerves, myelinating, and sheathing them. Melanocytes and Schwann cells both arise from the neural crest during development, and some melanocytes arise directly from Schwann cell precursors lining developing spinal nerves. In this review, we explore the connections between melanocytes and Schwann cells in development and transformation.

Introduction

Melanocytes are pigment-producing cells that reside in the epidermis, dermis, eyes, ears, heart, and meninges of the central nervous system of mammals. Melanocytes transfer pigment-containing organelles called melanosomes to keratinocytes in the epidermis to pigment the hair and skin of humans and are required for normal vision and hearing. Schwann cells are a type of glia that associate with peripheral nerve axons. Schwann cells produce a multilayered myelin sheath that insulates nerves, allowing for impulse conduction. Schwann cells also ensheath multiple small caliber axons without producing myelin in Remak bundles.

Melanocytes and Schwann cells are closely related during development, as they both arise from the neural crest. Interestingly, Schwann cells and melanocytes exhibit some plasticity in adults, particularly during transformation. Melanotic schwannomas, for example, exhibit cells with features of both melanocytes and Schwann cells. How might this occur? One fascinating aspect of Schwann cells is their ability to dedifferentiate and re-enter the cell cycle in response to nerve damage. Thus, it is possible that Schwann cells could dedifferentiate to a precursor state and then acquire melanocytic characteristics or transdifferentiate into melanocytes. Alternatively, the regulatory networks shared by melanocytes and Schwann cells could fail to completely suppress the alternative fate. In this review, we will explore the links between melanocytes and Schwann cells during development and in human disease.

New connections between melanocytes and Schwann cells during mouse development

Schwann cells and melanocytes both arise from the neural crest (Rawles, 1947; Woodhoo and Sommer, 2008). The multipotent neural crest cells (NCCs) that delaminate from the developing neural epithelium of the dorsal neural tube migrate along several routes (Le Douarin and Dupin, 2003). One group of NCCs migrates along the dorsolateral pathway between the dermomyotome of the somites and the overlying ectodermal epithelium, while others take the ventromedial pathway between the neural tube and the sclerotome of the somites. These migrating cells are likely to be a mixture of multipotent and specified, perhaps even fate-restricted, cell types.

Beginning around embryonic day (E) 9.5, just prior to migration, the trunk NCCs that follow the dorsolateral pathway upregulate the transcription factor, microphthalmia (Mitf), a crucial regulator of melanocyte differentiation. These cells will also upregulate dopachrome tautomerase, Dct, a melanogenic enzyme that marks melanocytes at all stages, including the melanocyte stem cells in adults (Adameyko et al., 2009; Goding, 2000; Mackenzie et al., 1997; Nishimura et al., 2002). These cells are now defined as melanoblasts, immature melanocytes. They migrate from the dorsum toward the ventrum in the dermis, eventually invading the epidermis. Kit tyrosine kinase receptor and its ligand, steel/stem cell factor/kit ligand, and the endothelin receptor B and its ligand, endothelin 3, are hypothesized to provide migratory cues and survival support for melanoblasts on the dorsolateral pathway (Bonaventure et al., 2013).

On the second pathway, the ventromedial pathway, cells migrate to the coalescing dorsal root ganglia (DRG). Some of these cells are already committed to a neuronal fate and will become neurons, while others remain uncommitted neural crest progenitors (Ernfors, 2010). By E11.5, spinal nerve extensions from the DRG appear along with undifferentiated cells that are migrating along them from the DRG to the periphery. These cells are called Schwann cell precursors (SCPs) (Jessen and Mirsky, 2005; Woodhoo and Sommer, 2008). SCPs are present around peripheral nerves from E11.5 until E15.5 (Jessen and Mirsky, 2005). SCPs upregulate expression of the desert hedgehog signaling molecule, Dhh (Bitgood and McMahon, 1995; Jaegle et al., 2003; Parmantier et al., 1999). In addition, SCPs closely associate with nerve projections and are able to migrate long distances following them.

The relatively recent twist in this story is that some SCP's lining spinal nerves do not become Schwann cells, but instead differentiate into melanoblasts. This forms a second wave of melanoblast differentiation around E12.5 of mouse development (Adameyko et al., 2009). At E13.5, Mitf and Dct are upregulated in the former SCPs that chose a melanoblast fate (Adameyko et al., 2009). These cells are now no longer in direct contact with nerves. The transcription factor, Oct6, marks committed Schwann cells, still closely associated with axons (Jaegle et al., 2003; Woodhoo and Sommer, 2008). These cells are immature (i.e. not yet myelinating). Around E16.5, the expression of the zinc finger transcription factor, Krox20, will signal that the myelination program is initiated in maturing Schwann cells (Topilko et al., 1994; Woodhoo and Sommer, 2008). In light of this newly discovered relationship, calling the bipotential precursor cells ‘Schwann cell precursors’ could lead to confusion; however, this convention has been kept.

Plp1-creER was an important tool in the discovery of SCP-derived melanocytes in mammals (see Table 1 for an explanation of Cre/LoxP technology). Proteolipid protein 1 (Plp1) is a component of myelin expressed in SCPs. The clustering of fate mapped, Mitf-positive/Cre-negative cells at E13.5 around spinal nerves following tamoxifen injection to activate CreER at E11.5 demonstrated that Plp1-creER induces recombination in SCPs that then become melanocytes (Adameyko et al., 2009). Importantly, melanoblasts on the dorsal–lateral pathway in the trunk at E10.5 are not fate mapped by tamoxifen injection at E9.5 (Adameyko et al., 2009). Tamoxifen injection at E11.5 fate maps cells that eventually reside in the pelage hair follicles and the dermis, but not in the interfollicular epidermis (Adameyko et al., 2009; Deo et al., 2013). This suggests that certain melanocytes may have specificity in their migratory targets.

Table 1. Potential genetic tools for targeting melanocytes of the Schwann cell lineage in mice
 Wnt1-creSox10-crePlp1-creERDhh-creMpz/P0-creKrox20-cre
  1. N.D., not determined; HF, hair follicle; IF, interfollicular.

  2. A number of transgenic mouse lines drive Cre expression beginning at different stages of NCC or Schwann cell development and can be used to induce recombination of engineered LoxP sites, producing tissue-specific gene knockouts and expression of reporter constructs (Cre/LoxP technology). In particular, fate mapping can be performed by breeding mice that carry a tissue-specific Cre transgene to mice that carry a fluorescent or LacZ reporter gene, whose constitutive expression is initiated upon the removal of a stop cassette flanked by LoxP sites in cells that express Cre. Reporter expression is then expected to be maintained for the life of the cell and in all of its descendents. A variant of this technology uses a Cre/estrogen receptor fusion protein (CreER) to switch on CreER activity temporarily following tamoxifen injection.

Gene promoterWnt1Sox10Plp1DhhMpz/P0Krox20
Inducible?NoNoYesNoNoNo
Initiation of expressionIn CNS and neural tube before emigrationIn neural crest cells, just after emigrationBy E9.5, in many cells of the CNS and PNSIn Schwann cell precursorsIn maturing Schwann cellsIn maturing Schwann cells
Fate maps melanocytes in
Pelage HFsYesN.D.YesNoN.D.N.D.
WhiskersNoYesYesNoN.D.N.D.
IF-EpidermisN.D.N.D.NoN.D.N.D.N.D.
IF-DermisN.DN.D.YesN.D.N.D.N.D.
References(Danielian et al., 1998; Hari et al., 2012; Wong et al., 2006; Yoshimura et al., 2013)(Hari et al., 2012; Wong et al., 2006)(Adameyko et al., 2009; Deo et al., 2013; Hari et al., 2012; Michalski et al., 2011)(Hari et al., 2012; Wong et al., 2006)(Feltri et al., 1999)(Gambardella et al., 2000; Topilko et al., 1994; Voiculescu et al., 2000)

Interestingly, several reports have suggested that the expression of Plp1 is unexpectedly wide given its protein function. Plp1-cre, using the same regulatory regions as Plp1-creER, fate maps cells in the telencephalon, optic stalk, mesencephalon, rhombencephalon, and spinal cord at E10.5 (Michalski et al., 2011). Tamoxifen injection at E9.5 induces Plp1-creER-mediated recombination in the majority of the cells in the DRG at E10.5, which would presumably fate map melanocytes in the ventral pathway (Adameyko et al., 2009; Hari et al., 2012). Tamoxifen injection at any time between E9.5 and E14.5 fate maps whisker follicle melanocytes, although the origin of these facial melanocytes is not yet known (Adameyko et al., 2012; Hari et al., 2012).

Unfortunately, another cre line that might have fate mapped SCP-derived melanocytes, Dhh-cre, does not appear to be efficient (Hari et al., 2012; Wong et al., 2006). It is possible that Dhh might not be upregulated quickly enough to efficiently drive recombination in presumptive SCP-derived melanoblasts, compared with Plp1-creER, which is expressed prior to the SCP stage, but whose protein is not activated until tamoxifen is administered (See Table 1 for a summary of Cre lines expressed at various stages of development along the ventromedial pathway).

Mechanisms specifying cell fate

What are the mechanisms that specify a Schwann cell fate versus a melanocyte fate within the neural crest? Although this area of research is still evolving, signaling pathways and transcriptional networks have been implicated (Figure 1).

Figure 1.

A model of melanocyte and Schwann cell differentiation from the trunk neural crest, showing proteins known to be expressed at each stage. Proteins whose expression initiates at the same stage are grouped by color.

Neuregulin signaling

Neuregulin-1 (Nrg1) is an EGF-like ligand for the ErbB family of tyrosine kinase receptors that is pivotal for Schwann cell development. There are at least 31 known isoforms of neuregulin-1, generated by multiple promoter usage and alternative splicing (Birchmeier and Nave, 2008; Fleck et al., 2012). The neuregulin-1 protein is processed by disintegrin and metalloprotease (ADAM) family members, including Bace1 (beta-site amyloid precursor protein cleaving enzyme 1) (Luo et al., 2011) and Tace1 (tumor necrosis factor-α-converting enzyme) (La Marca et al., 2011). Neuregulin-1 protein can be membrane bound, secreted (Fricker and Bennett, 2011), or directed to the nucleus (Adilakshmi et al., 2011) and participates in paracrine signaling (type i, ii, and iv isoforms), juxtacrine signaling (type iii isoforms) and autocrine signaling (type i isoforms). ErbB2, ErbB3 and ErbB4 receptors bind neuregulin-1 as heterodimers or homodimers (Ferguson et al., 2000). Downstream effects of ErbB activation include the activation of (i) phosphatidylinositol-3-OH kinase, (ii) PLCγ, and (iii) MAP kinase (Kao et al., 2009; Maurel and Salzer, 2000; Meintanis et al., 2001).

Neuregulin signaling is absolutely required for Schwann cell development (Newbern and Birchmeier, 2010). Early in the process, axons produce membrane-bound, type iii neuregulin-1 and undifferentiated cells adjacent to the axons express ErbB receptors. Through juxtacrine signaling, the NCCs close enough to touch the axon surface receive neuregulin-1 (Fricker and Bennett, 2011). Sustained neuregulin signaling is required throughout Schwann cell differentiation to stimulate cell proliferation (Levi et al., 1995), to direct migration along the axon (Heermann and Schwab, 2013), to upregulate Krox20 expression, which is critical for myelin expression (He et al., 2010), and to produce the proper myelin thickness (Michailov et al., 2004). Interestingly, neuregulin is not required to maintain myelin in adults (Fricker et al., 2011). In zebra fish models, the role of neuregulin signaling in Schwann cell proliferation and migration can be uncoupled, misexpression of Nrg1 type iii causes ectopic Schwann cell migration, and neuregulin signaling directs radial sorting, which produces a 1:1 ratio of Schwann cells to axons (Lyons et al., 2005; Perlin et al., 2011; Raphael et al., 2011).

ErbB3 is expressed in NCCs as they migrate from the dorsal neural tube and is maintained in glial cells and melanocytes, but not other NCC lineages (Meyer and Birchmeier, 1995). ErbB3 mutant embryos fail to produce Schwann cells. At E12.5, ErbB3 null embryos exhibit a greatly reduced number of SCPs (Adameyko et al., 2009; Riethmacher et al., 1997). By E16.5, there are no detectable Schwann cells lining nerves (Riethmacher et al., 1997). In contrast, ErbB3 null embryos exhibit a 178% increase in the number of melanoblasts around the distal ends of the dorsal spinal nerves at E12.5, with no difference in the number of melanoblasts migrating on the dorsal–lateral pathway (Adameyko et al., 2009; Buac et al., 2009). This suggests that in the absence of neuregulin signaling, SCPs around spinal nerves become melanocytes. Tissue-specific knockouts of Erk1/2 suggest that neuregulin signaling in SCPs is transduced through the MAP kinase pathway (Newbern et al., 2011). In zebra fish, Erbb3b is required to generate melanocyte stem cells (Hultman et al., 2009).

When peripheral nerves are injured in adult mice, Schwann cells respond by dedifferentiating, proliferating, and covering the new axon that regrows (Figure 2) (Fricker and Bennett, 2011). Recently, it was shown that when the sciatic nerve is crushed and the axonal membrane degenerates (i.e. Wallerian degeneration), Schwann cells themselves upregulate neuregulin-1 expression, specifically isoform type i (Stassart et al., 2013). Expression of neuregulin-1 within Schwann cells is required for efficient myelin repair following nerve injury, even though it is not required during embryogenesis for normal development (Fricker et al., 2011; Stassart et al., 2013). Neuregulin-1 type i expression in Schwann cells is downregulated when Schwann cells are exposed to membrane-bound neuregulin-1 type iii on regenerated axons (Fricker et al., 2011; Stassart et al., 2013). Interestingly, ERK/MAP kinase signaling appears to be the primary trigger for Schwann cell dedifferentiation. Expression of activated RAF in adult Schwann cells produces dedifferentiation, even in the absence of a damaged nerve (Napoli et al., 2012).

Figure 2.

A model for the pathways activated in Schwann cells following nerve injury.

Melanocytes are produced from Schwann cells at the site of cut and deflected sciatic nerves in adult mice, leaving pigmented ‘streaks’ in the dermis (Adameyko et al., 2009; Rizvi et al., 2002). This is due to the transdifferentiation of Schwann cells at the cut nerve. Sciatic nerve crush is less likely to produce pigmentation streaks (Rizvi et al., 2002). Haploinsufficiency of neurofibromin, a negative regulator of RAS in the MAP kinase pathway, increases the amount of pigmentation occurring around cut and deflected sciatic nerves (Rizvi et al., 2002). Perhaps, Nf1 haploinsufficiency makes Schwann cells more likely to activate the MAP kinase pathway and dedifferentiate.

A few studies have suggested a growth-promoting role for neuregulin-1 on human pigment cells. Neuregulin-1 is expressed by cultured human fibroblasts. In one study, a highly expressed 60-kDa neuregulin-1 band was observed in Western blots of phototype IV (dark skin) fibroblasts, but not in fibroblasts isolated from lighter colored skin (phototypes I and III), and neuregulin-1 treatment of 3D reconstructed skins led to increased pigmentation (Choi et al., 2010). In addition, ERBB receptors are expressed in cultured human melanocytes (Gordon-Thomson et al., 2005). In human biopsies, phosphorylated ERBB3 was found in the cytoplasm of 18% of primary melanomas, but not in benign nevi or melanoma metastases (Buac et al., 2009). In a similar fashion, enhanced NRG1/ERBB signaling has been reported in human schwannomas and constitutive expression of neuregulin-1 in Schwann cells causes malignant peripheral nerve sheath tumors in mice (Kazmi et al., 2013; Stonecypher et al., 2006). Possibly, a quantitative effect of neuregulin/MAP kinase activation dictates the exact response that melanocytes and Schwann cells make during development versus transformation (Newbern and Snider, 2012).

Transcriptional networks

A number of transcription factors have been implicated in the cell fate commitment and differentiation of both Schwann cells and melanocytes. These include (i) Pax3, a paired box homeodomain transcription factor, able to bind other transcription factors through its paired domain to either activate or repress transcription, (ii) FoxD3, a forkhead box transcriptional repressor, and (iii) Sox proteins, SRY high-mobility group (HMG) box family members. These proteins play multiple roles in Schwann cell and melanocyte development. It should be noted that the exact role of these networks with respect to the two lineages of melanoblasts remains to be clarified.

The transcriptional regulators, Pax3, Sox10, and FoxD3, are all expressed in the dorsal neural tube prior to NCC migration. Pax3 and Sox10 positively regulate the expression of Mitf, which is required for melanoblast specification (Bondurand et al., 2000; Goding, 2000; Potterf et al., 2000). In avians, it is hypothesized that FoxD3 interacts with Pax3 to prevent it from binding to the Mitf promoter in uncommitted NCCs (Kos et al., 2001; Thomas and Erickson, 2009). In zebra fish, Foxd3 also suppresses Mitf expression (Curran et al., 2010). Zebra fish with homozygous loss of Foxd3 exhibit an initial decrease in melanophore numbers, followed by a significant increase (Hochgreb-Hagele and Bronner, 2013). In mice, FoxD3 is expressed in migratory neural crest cells and in Schwann cell precursors and so could continue to suppress melanocyte cell fate in the ventromedial pathway (Dottori et al., 2001; Labosky and Kaestner, 1998).

Interestingly, studies of melanocyte stem cells in adult hair follicles have shown that Pax3 also binds to the DCT promoter, in this case preventing Sox10 and Mitf from activating transcription (Lang et al., 2005). Pax3-mediated Dct repression is released when beta-catenin is activated. If this pathway functions during development as well, Pax3 could suppress differentiation of melanoblasts, possibly to allow more time for proliferation. Mice lacking Pax3 exhibit greatly reduced numbers of melanoblasts at E12.5, in both the trunk region and around the otic vesicle (Hornyak et al., 2001), and this is presumably due to an inability of potential melanoblasts to upregulate Mitf.

As mentioned above, Pax3 is expressed in the dorsal neural tube prior to NCC migration, and it continues to be expressed in immature Schwann cells through E14.5 (Goulding et al., 1991). Pax3 is downregulated when Schwann cells initiate myelin expression (Kioussi et al., 1995). Non-myelinating types of Schwann cells do not downregulate Pax3. Pax3 has been shown to repress the expression of the myelin basic protein (Kioussi et al., 1995) and to stimulate proliferation (Doddrell et al., 2012). Pax3Splotch/Pax3Splotch mutant embryos die by E13.5, at which point no immature Schwann cells can be detected (Franz, 1990). Thus, it seems that in both melanocytes and Schwann cells, Pax3 is required at an early stage, but is downregulated to permit terminal differentiation.

The binding of both Sox10 and Pax3 to the Mitf promoter synergistically activates the expression of Mitf in melanoblasts (Bondurand et al., 2000; Potterf et al., 2000). Sox10 expression is maintained in melanocytes in mice, but not in zebra fish (Britsch et al., 2001; Greenhill et al., 2011). In mice, the loss of Sox10 appears to be more severe than the loss of Pax3 in terms of melanocyte development. No Mitf-positive or Dct-positive cells can be found migrating on the dorsal–lateral pathway in Sox10LacZ/Sox10LacZ embryos at E12.5, perhaps because Sox10 directly regulates the expression of both Mitf and Dct (Britsch et al., 2001). A full gene dosage of Sox10 is critical for the survival of tumor cells in a hyperactive N-ras melanoma mouse model and is expressed in the majority of human congenital nevi and melanoma (Shakhova et al., 2012).

Sox10 expression is required at multiple stages for Schwann cells. In Sox10LacZ/Sox10LacZ embryos, no brain fatty acid binding protein (Bfabp/Fabp7/Blbp)–positive SCPs are present along spinal nerves (Britsch et al., 2001). Later knockouts of Sox10 at either the immature Schwann cell stage (Finzsch et al., 2010) or in mature Schwann cells (Bremer et al., 2011) have demonstrated that Sox10 is required for Schwann cell function throughout life. Sox10 activates myelin protein 0 and peripheral myelin protein 22 expression (Jones et al., 2011). Sox10 also appears to be harnessed by the leprosy bacterial pathogen to cause Schwann cells to dedifferentiate, so that pathogen-laden cells can be disseminated. An early event following infection is the exportation of Sox10 from the nucleus to the cytoplasm and the suppression of Sox10 mRNA expression by DNA methylation (Masaki et al., 2013). Similarly, human schwannoma cells exhibit reduced expression of Sox10 (Doddrell et al., 2013). This is in contrast to melanoma, above, which seems to require Sox10 for oncogenesis.

In addition, NCCs express another Sox transcription factor, Sox2, which is a stemness factor (Chambers and Tomlinson, 2009). Sox2 promotes proliferation and suppresses differentiation. Sox2 directly represses Mitf expression and must be downregulated before melanoblasts can be specified from SCPs (Adameyko et al., 2012). Interestingly, Sox2 expression decreases from proximal to distal along the spinal nerves, suggesting that melanoblasts might be more likely to arise from distal SCPs. Sox2 is downregulated in myelinating Schwann cells, but is re-expressed following nerve injury in adults. Following nerve injury, Sox2 activation causes the relocalization of N-cadherin, promoting the formation of Schwann cell cords in the nerve wound (Le et al., 2005; Parrinello et al., 2010).

In summary, Pax3, Sox10, and Sox2 are all important for both Schwann cell and melanocyte development and play major roles during cell fate commitment and dedifferentiation; however, there are many more questions to be answered. For example, what is the primary trigger specifying a melanocyte cell fate versus a glial cell fate within SCPs? Does FoxD3 play a critical role in mammals? A tissue-specific pan-NCC knockout of FoxD3 resulted in a catastrophic failure of the cranial neural crest, peripheral nervous system, and enteric neural crest (Teng et al., 2008). Plp1-creER could be used to knockout FoxD3 in SCPs to address whether more melanoblasts are made.

Connections between melanocytes and Schwann cells in human disease

In human pathological conditions, Schwann cell–based tumors can exhibit moderate to heavy amounts of pigmentation, and at times, melanomas exhibit histopathological features suggestive of neural/glial differentiation. A mixed morphology of cells suggests that dedifferentiation or transdifferentiation might play a role in their development. Furthermore, melanin and melanosomes have been described in Schwann cells of normal human skin (Carbonell et al., 1992; Garcia and Szabo, 1979) and some melanocytic nevi exhibit Schwannian differentiation (Fullen et al., 2001; Reed et al., 1999). Melanocytes in these ‘C-type’ nevi penetrate the dermis, express early NCC markers, associate with nerves, and produce corpuscles similar to those in Schwann cells. All mixed morphology neoplasms can be challenging to diagnose accurately. In this section, we will review some links between melanocytes and Schwann cells in human disease (Table 2).

Table 2. Features of melanotic schwannoma, neurofibromatosis type 1, and desmoplastic melanoma
 Melanotic Schwannoma–Carney complexPigmented plexiform neurofibroma – NF1Cafe-au-lait macule– NF1Desmoplastic melanoma
Cell type of originSchwann cellSchwann cellMelanocyteMelanocyte
What is it?Pigmented Schwann cell tumors, with myxomas, endocrine overactivity, and male subfertilityHeterogeneous tumor with associated pigmentationLight brown, uniformly colored, clearly defined, epidermal maculeAmelanotic melanoma with spindle-shaped tumor cells invading the dermis, usually in association with SSM or LMM
Gene(s) mutated PRKAR1A NF1 +/− with second hit in Schwann cellsNF1 +/− with second hit in melanocytesBRAF and Kit (rare); NF1?
Histological featuresSpindle and epithelioid cells, conspicuous macronuclei, nesting fascicles, psammomatous bodiesPigment cells rounded and dispersed singly and in small nestsIncreased melanin content, macromelanosomesSpindle-shaped cells
ImmunoreactivityPositive for HMB-45, S-100, vimentin; negative for GFAP; positive for PASFocally positive for tyrosinase and HMB-45, along with other markers for neurofibromasPositive for melanocyte markersNegative for HMB-45, positive for NCC markers, Sox10, and S100

Melanotic schwannoma

Melanotic schwannoma is a rare type of Schwann cell tumor that appears black, brown, blue or gray. Less than 1% of schwannomas are melanotic (Murali et al., 2010). Ten per cent of melanotic schwannomas are malignant with metastasis, while the remainder are benign (Shields et al., 2011). Melanotic schwannomas usually appear near the paraspinal ganglia and are associated with spinal or autonomic nerves, but can be located in other places (Chetty et al., 2007; Saavedra et al., 2006). Melanotic schwannomas arise sporadically or in association with dominantly inherited Carney complex, with roughly half of cases occurring in patients with Carney complex (Carney, 1990). Prominent features observed during histological analysis include rare mitotic figures, spindle and epithelioid shaped cells, the presence of conspicuous macronucleoli, cells arranged as nesting fascicles, sheets, or lobules, and if associated with Carney complex, the presence of psammomatous bodies.

Melanotic schwannomas are positive for melanoma markers, such as tyrosinase and HMB-45, and contain melanosomes in various stages of maturation (Arvanitis, 2010; Boyle et al., 2002). They also possess Schwann cell features such as interdigitating cytoplasmic processes and basement lamina formation (Arvanitis, 2010). Thus, melanotic schwannoma appear composed of ‘hybrid’ cells (Font and Truong, 1984). Melanotic schwannoma is sometimes mistaken for melanoma (Shields et al., 2011). A deep or paraspinal location in the absence of a cutaneous melanoma and spindle cells with heavy pigmentation are clinical findings that should prompt a consideration of melanotic schwannoma (Murali et al., 2010). Melanotic schwannomas lack GNAQ mutations, which might help distinguish them from blue nevi and leptomeningeal melanoma (Kusters-Vandevelde et al., 2010; Van Raamsdonk et al., 2009).

Carney complex is a disorder in which myxomas, endocrine overactivity, psammomatous melanotic schwannomas, and male subfertility are co-expressed, with variable penetrance and expressivity (Carney, 1990; Veugelers et al., 2004). In addition, cutaneous and ocular hyperpigmentation are associated with Carney complex, including lentigines (freckling), particularly on the face, pigmented lesions of the conjunctiva, and epithelioid blue nevi/pigmented epithelioid melanocytomas (Carney, 1990; Cohen et al., 2000; Zembowicz et al., 2007). Melanotic schwannomas in Carney complex develop on average approximately 10 yr earlier than in sporadic cases (Shields et al., 2011).

Linkage analysis of some families affected by Carney complex have identified loss-of-function mutations in PRKAR1A (Kirschner et al., 2000). PRKAR1A encodes a regulatory subunit of protein kinase A. The protein kinase A holotetramer consists of two regulatory subunits and two catalytic subunits (Kim et al., 2007). The regulatory subunits restrain the activity of the catalytic subunits until cAMP is present; however, the mechanism by which PRKAR1A mutations lead to tumorigenesis is not yet known (Wilkes et al., 2005). PRKAR1A might require a ‘second hit’ to promote oncogenesis, because somatic loss of heterozygosity has been reported in some human tumors; however, a second hit is not always found (Bossis et al., 2004; Casey et al., 2000; Kirschner et al., 2000). Sarcomas/myxomas, but not schwannomas, develop in Prkar1a +/− knockout mice aged 9–19 months, with no evidence of loss of the wild-type allele in the tumors (Veugelers et al., 2004). If the loss of the PRKAR1A regulatory subunit causes hyperactivity of protein kinase A, then the melanization of schwannomas and cutaneous hyperpigmentation could be due to aberrant expression of Mitf, which is activated by the PKA/CREB pathway (Roh et al., 2013).

Neurofibromatosis type 1

Neurofibromatosis type 1 is a dominant genetic disorder that features both Schwann cell and melanocyte-based phenotypes. Neurofibromatosis type 1 patients bear heterozygous mutations in the NF1 gene, which is a tumor suppressor that negatively regulates RAS in the MAP kinase pathway (Fountain et al., 1989; Wallace et al., 1990). NF1 is expressed in neural crest cells prior to emigration and continues to be expressed throughout Schwann cell and melanocyte development (Daston and Ratner, 1992; De Schepper et al., 2008; Maertens et al., 2006). One of the features of neurofibromatosis type 1 is the development of neurofibromas, which are benign Schwann cell-based tumors, driven by somatic loss of heterozygosity of NF1 in Schwann cells (Maertens et al., 2006). Neurofibromas consist of a heterogeneous population of Schwann cells, neurons, fibroblasts, mast cells, and endothelial cells (McClatchey, 2007). Neurofibromas can be dermal or plexiform. Dermal neurofibromas originate from peripheral nerves in the dermis, are observed in 95% of neurofibromatosis type 1 patients, and rarely undergo malignant transformation. Plexiform neurofibromas, on the other hand, can originate from any nerve of the body, are found in 50% of neurofibromatosis type 1 patients, and have the potential to transform into malignant peripheral nerve sheath tumors (MPNST's) (McClatchey, 2007).

Occasionally, hyperpigmentation resembling giant congenital melanocytic nevi overlies plexiform neurofibromas in neurofibromatosis type 1 (De Schepper et al., 2005). In addition, there are cases of melanotic neurofibromas in which the pigmentation is located within the neurofibroma (Fetsch et al., 2000). Pigmented neurofibromas are focally positive for the melanoma markers, tyrosinase and HMB-45 (Boyle et al., 2002; Fetsch et al., 2000). It has not been resolved whether the pigmented cells in melanotic neurofibromas are melanocytes or Schwann cells producing melanin. The dermis of human skin does not normally possess melanocytes, so this pigmentation is ectopic.

Cafe-au-lait macules (CALMs) are another important diagnostic feature of neurofibromatosis type 1 (De Schepper et al., 2005). CALMs are benign, light brown, uniformly colored and clearly defined epidermal macules. CALMs are produced through a somatic second hit of NF1 within melanocytes (De Schepper et al., 2008). Other pigmentary manifestations of neurofibromatosis type 1 include mild, generalized skin hyperpigmentation due to NF1 haploinsufficiency, which is visible in mosaic individuals, and skin-fold freckling (De Schepper et al., 2005; Maertens et al., 2007). Similar to humans, generalized epidermal skin hyperpigmentation is observed in Nf1 heterozygous mice and also in mice in which both copies of Nf1 have been knocked out specifically in melanocytes, as in CALMs (Deo et al., 2013).

The Nf1-haploinsufficient environment around neurofibromas plays a role in tumor formation (Yang et al., 2008). Krox20-Cre/+; Nf1flox/Nf1null mice develop neurofibromas, while Krox20-Cre/+; Nf1flox/Nf1flox mice do not (Zhu et al., 2002). Nf1 null Schwann cells secrete elevated levels of stem cell factor, which causes hyperproliferation of Nf1+/− mast cells (Yang et al., 2003). The Nf1+/− mast cells in turn secrete elevated levels of TGF-β and cause hyperproliferation of Nf1+/− fibroblasts (Yang et al., 2006). Thus, a Nf1-haploinsufficient microenvironment mediates a number of enhanced paracrine signaling events within the developing neurofibroma. In contrast, a haploinsufficient environment is not required for plexiform neurofibroma formation in Plp1-creER Nf1 knockout mice when tamoxifen is administered in adulthood (Mayes et al., 2011). Perhaps, this difference lies in the recombination efficiency of Krox20-cre versus Plp1-creER. A sufficiently large cluster of Nf1/ mutant Schwann cells could produce its own tumor accelerating microenvironment. The Krox20-Cre/+; Nf1flox/Nf1null mouse might be more relevant to neurofibromatosis type 1, because only isolated Schwann cells would be expected to receive a second hit in people.

If biallelic loss of Nf1 occurs in immature Schwann cells using either Dhh-cre or Plp1-creER with tamoxifen injection at E12.5, plexiform neurofibromas develop rapidly in 100% of the mice (Le et al., 2011; Mayes et al., 2011; Wu et al., 2008). Interestingly, 39% of the Dhh-cre knockout mice were reported to exhibit melanized neurofibromas and/or pigmentation overlying the spinal cord, even though Dhh-cre does not appear to induce recombination in melanocytes (Hari et al., 2012; Wong et al., 2006; Wu et al., 2008). This suggests that Nf1/ Schwann cells can either express melanocytic traits or recruit normal melanocytes to ectopic sites.

Desmoplastic melanoma

Desmoplastic melanoma is an uncommon amelanotic melanoma variant characterized by spindle-shaped tumor cells invading the dermis. Desmoplastic melanoma is typically observed in older individuals, where it arises on sun-exposed skin in the head/neck region. It can be confused with neurofibromas, dermatofibromas, and even scars. The tumor cells form fascicles that resemble Schwann cells in a myelin sheath (Huttenbach et al., 2002). In addition, a fibrotic stroma surrounds these tumors, with prominent collagen tissue formation.

Desmoplastic melanomas are considered to arise from melanocytes for several reasons. First, 83% possess an epidermal component, usually lentigo melanoma or superficial spreading melanoma (Jaimes et al., 2013). They also possess melanoma-specific features, such as atypical vascular structures, peppering, blue-white veil, atypical globules, crystalline structures and atypical pigment networks (Jaimes et al., 2013). Desmoplastic melanoma cells typically lack melanosomes, including premelanosomes, and are negative for HMB-45, an antibody specific to the gp100 protein present on melanosomes (Orchard, 2000). Desmoplastic melanomas express a pattern of early NCC markers including p75NGFR, N-CAM, and peripherin, which could indicate a NCC progenitor like state, and sometimes are focally positive for tyrosinase and Mitf (Huttenbach et al., 2002). They do not express myelin or other markers of Schwann cell differentiation (Penneys et al., 1984). They are usually positive for S100 and Sox10, which label both melanocytes and Schwann cells (Mohamed et al., 2013; Palla et al., 2013).

No specific oncogenic mutations are characteristic of this tumor type. BRAF and KIT mutations have been reported in single desmoplastic melanoma tumors (Chen et al., 2013; Miller et al., 2012). Because desmoplastic melanoma shows some similarity to nerve sheath tumors, NF1 has been analyzed. 67% of desmoplastic melanomas exhibited loss of heterozygosity for NF1 compared with 5% of common melanomas (Gutzmer et al., 2000). However, mutation of the remaining NF1 allele was not shown and desmoplastic melanoma is not more frequent in neurofibromatosis type 1 patients.

Future directions

Melanocytes can arise directly from the neural crest or from SCPs. The discovery of two different lineages of melanocytes poses new questions for the field. Are these two lineages predisposed to behave differently throughout life or are they indistinguishable from one another? For example, do they respond differently to certain oncogenic mutations, possibly giving rise to different subtypes of melanoma? Do they respond differently to different signaling factors? Do these two lineages of melanocytes differ in the final location that they prefer to home in the skin (i.e. the hair follicles, the interfollicular epidermis, or the dermis)? The mouse is well suited to addressing this latter question, because the tail skin possesses a uniform and differentiated population in both the adult epidermis and dermis (Fitch et al., 2003). Although the environmental cues produced by keratinocytes and other surrounding cell types must play a major role, some intrinsic difference in melanocytes themselves related to their origin could influence their behavior in specific ways.

We do not yet know the underlying trigger that differentiates melanocytes from Schwann cells during development, whether it is a stochastic event or an environmental cue. We do not completely understand the genotype–phenotype relationships in human disease, for example, how mutations in different genes result in different, specific pigmentary lesions. The range of human phenotypes involving melanocytes is fascinating, and we have only scratched the surface molecularly. A better understanding of the underlying connection between melanocytes and Schwann cells is necessary both to describe normal NCC development and to understand cases of transdifferentiation.

Ancillary