Melanocyte stem cells

2.1. Functions

Melanocytes are specialized melanin producing cells, and are responsible for skin, hair, and eye pigmentation in vertebrate organisms. Melanocytes synthesize pigment melanin within a special organelle termed a melanosome, where a number of specific enzymes and structure proteins are assembled to synthesize melanin from tyrosine or phenylalanine (Slominski et al., 2004). Once synthesized, melanin is deposited in melanosomes to form melanin granules. These melanin granules are transferred from melanocytes via their dendrite to adjacent keratinocytes, where melanins are accumulated to generate pigmented skin or hairs. While the physiological roles of pigmentation may vary among animal species, in humans the principal function of melanocytes is to protect skin form genotoxic stress of ultraviolet (UV) radiation. This protection is enabled by the ability of melanin granules to absorb UV radiation and quench the UV-induced intracellular free radicals. Indeed, it has been demonstrated that individuals carrying hypomorphic mutation in the melanocortin-1 receptor (MC1R), which is one of critical molecules for the induction of the melanin synthesis pathway, show not only abnormal pigmentation phenotypes, but also higher incidence of melanoma formation and other skin carcinogenesis (Box et al., 2001; Healy et al., 2001), indicating a critical link between melanogenesis and protection against skin carcinogenesis.

In addition to the UV-protection, melanocytes also play an essential role in hearing system (Tachibana, 1999). Patients with Waardenburg Syndrome exhibit varying combinations of abnormal pigmentation as well as auditory defect (Newton, 2002). These phenotypes are caused by a deficiency of melanocytes. It is known that melanocytes also present in the stria vascularis where they consist of an intermediate cell layer, and act to generate the electrophysiological potential in the endolymph fluid (Price and Fisher, 2001; Takeuchi et al., 2000). In this case, melanin pigment formation is not required for their auditory function, but a K⁺ channel family, Kir4.1, on melanocyte plasma membrane plays a critical role for the generation of K⁺ gradient between plasma membrane and the endolymph fluid, which is essential for the transduction of sound by hair cells (Ando and Takeuchi, 1999).

In contrast to the clear functions of human melanocytes, physiological functions of melanocytes in other organisms are less clear and appear to be variable depending on distinct organisms. In lower vertebrates such as reptiles, amphibians and fish, melanocytes can generate diverse pigment patterns in the skin. Such pigment patters could be important in mediating color adaptation, predator evasion, feeding priority, or sexual selection in individual animals. Hence, in these organisms, melanocytes appear to play more socio-environmental functions in addition to their direct functions in protecting the skin from UV radiation.

2.2. Melanocyte development

Melanocytes in vertebrates are derived from the neural crest, which arises during gastrulation of embryogenesis at the dorsal edge of the neural plate (Thomas and Erickson, 2008). Once emerged, these neural crest cells migrate intensively to the specific sites where they differentiate into a wide range of lineages including peripheral neurons, endocrine cells, bone, cartilage, connective tissue, and melanocytes (Anderson, 2000). While the mechanisms by which multi-potent neural crest cells are specified into the melanocyte lineage remain unclear, it has been suggested that immature precursors for melanocytes, referred as melanoblasts, are originated from bi-potent glial-melanoblast progenitors, whose existence has been demonstrated in avian and mouse at the neural crest premigratory and migratory stages (Dupin et al., 2000; Mollaaghababa and Pavan, 2003). Recent studies indicate that melanoblast fate decision in these bi-potent progenitors is regulated by, at least in part, Wnt signaling pathway (Dunn et al., 2000). Disruption of Wnt signaling in Wnt1^−/−;Wnt3a^−/− combinatory knockout mice result in a dramatic reduction of several neural crest derivatives, including melanoblasts (Ikeya et al., 1997). Similarly, the mice in which β-catenin, an essential component of the canonical Wnt signaling pathway, is conditionally ablated in neural crest exhibit loss of embryonic melanoblasts (Hari et al., 2002), indicating an indispensable role of the canonical Wnt signaling pathway in melanoblast development. By contrast, gain-of-function of Wnt signaling by forced expression of Wnt1 in mouse neural tube explants culture shows marked increase of melanoblasts at the expense of glial cells (Dunn et al., 2000). Consistently, in zebrafish, overexpression of a constitutive active form of β-catenin in migratory neural crest cells also induces pigment cell formation along with a dramatic reduction of neuronal and glia lineages (Dorsky et al., 1998). Hence, these data clearly indicate a crucial role of Wnt signaling in restricting the glial-melanoblast progenitors toward a melanoblast fate.

Once melanoblasts arise, they migrate along the dorsolateral pathway between the dermatome and the epidermis toward the ventral midline (Steel et al., 1992). In mammalians, most of the melanoblasts further invade the overlying epidermis, where they proliferate and migrate extensively to distribute the entire epidermis (Steel et al., 1992; Wilkie et al., 2002). This is clear contract to those in fish, in which all of these migrated melanoblasts stay in the dermis. In the hairy region of mammalian skin, melanoblasts further enter the newly developing hair follicles where they finally become localized (Figure 1). Once in the hair follicles, melanoblasts are segregated into two populations: one consists of hair matrix melanocytes, which is responsible for pigmentation of the initial hairs; the other population consists of melanocyte stem cells, which are localized at the lower permanent portion of the hair follicle (the bulge region) (Nishimura et al., 2002) and are responsible for the maintenance of the hair follicle pigmentary system in the subsequent hair cycles (Figure 1). In non-hairy regions of the human skin, melanoblasts stay immature and reside on the basement membrane of the epidermis where they undergo terminal differentiation into mature melanocytes upon stimulation from keratinocytes.

Figure 1.

Schematic drawing of melanocyte development in the mouse.

(A) During embryogenesis, melanoblasts emerge in the neural crest at embryonic day (E) 9.5–10.5. These earliest stages of melanoblasts are characterized by the expression of Dct, Kit, Mitf, Pax3, Sox10, and Si. Once arise from neural crest, melanoblasts migrate in the developing dermis toward the ventral midline. Subsequently, at E12.5–13.5, they enter the epidermis where they actively migrate and proliferate to distribute the entire body. In addition to the pre-migratory melanoblast markers, these migratory melanoblasts express several melanosomal proteins including Tyr and Tyrp1. Upon initiation of hair follicle morphogenesis, the melanoblasts enter the newly developing and are segregated into two populations: one locates in the hair matrix where melanoblasts are differentiated into mature melanocytes; and the other colonizes at the putative bulge region, where melanoblasts are restricted in a resting status to become melanocyte stem cells. Melanocytes in the hair matrix express Oa1 and Mcr1 in addition to the migratory melanoblast markers, whereas in the bulge melanoblasts, expression of several melanoblast markers including Kit, Mitf, Pax3, Sox10, Tyr, and Tyrp1 are gradually downregulated. (B) Melanoblast localization during hair follicle morphogenesis. Skin fragments from Dct-LacZ transgenic mice were whole-mount stained with LacZ to visualize melanoblasts in the developing hair follicle. Melanoblasts colonize at the putative bulge region at stage 5–6 of hair follicle morphogenesis, suggesting stem cell specification would take place from stage 5–6 of hair follicle morphogenesis onward.

Dysregulation in melanoblast development typically results in hypo-pigmentation phenotypes, which is readily characterized in animals as piebaldism, congenital unpigmented spotting in the skin and hairs (Bennett and Lamoreux, 2003). Interestingly, spotting patterns of these mutant animals do not appear to be random: these spotting mutants tend to exhibit characteristic spotting patterns including belly spots, head spots, and piebald spotting (Figure 2) (Baxter et al., 2004). Although the reasons how these characteristic patterns are developed during embryogenesis are not completely understood, recent observations of the process of melanoblast development have provided some mechanistic insights into the formation of these spotting phenotypes. It has been demonstrated that melanoblast distribution in the embryonic epidermis is not uniform: melanoblasts exist at higher density in the head, cervical, and tail regions; whereas their density is much lower in the trunk region (Wilkie et al., 2002). Intriguingly, this distribution pattern is well correlated with pigmented and unpigmented pattern of the mutant animals: pigmented areas coincide with regions containing larger numbers of melanoblasts, while white spotting patterns seem to be paralleled with the regions of lower melanoblast density (Yoshida et al., 1996). Hence, based these observations, it is proposed that the piebald-type white spotting phenotypes are caused by dramatic reduction of total melanoblast number, rather than selectively affecting melanoblast precursor cells responsible for populating a given area (Baxter et al., 2004). By contrast, the belly and head spotting phenotypes would be explained by the failure of melanoblast migration because melanoblasts should travel relatively longer distance to reach the belly and head areas from the areas where melanoblast density is high (Baxter et al., 2004).

Figure 2.

Defects in melanocyte development cause white spotting, while the stem cell defect results hair graying.

(A) A Ednrb^s-l/Ednrb^s-l mouse demonstrating extensive piebald spotting. (B) A Kit^W-2J/+ mouse demonstrating a white head blaze, and small dorsal spot on the back. (C) A Sox10^Lacz/+ mouse exhibiting the characteristic white belly spot. (D) A Mitf^vit/vit mouse (upper) exhibits gradual hair graying. A lower mouse is age-matched control. Adapted from the WEB site of the European Society for Pigment Cell Research (ESPCR), “Color genes”: [[UNSUPPORTED:p/uri]].

Through the extensive studies of the white spotting mutant animals, a number of key genes for melanoblast development have already been identified and characterized. Examples of these genes include Pax3 (Paired-box 3), Sox10 (Sex-determining region Y-box 10), Mitf (Microphthalmia-associated transcription factor), Edn3 (endothelin 3), Ednrb (endothelin receptor B), Kit (c-Kit tyrosine kinase receptor), Kitl (Kit ligand, also called as SCF or steel factor), and Snai2 (also called as Slug) (Bennett and Lamoreux, 2003) (Figure 3).

Figure 3.

Simplified schematic showing of key molecules and signaling pathways implicated in melanocyte-keratinocyte interactions.

In the melanocyte lineage, a series of transcription factors, including Pax3, Sox10, Creb, Lef1, and Mitf play crucial roles in the regulation of melanocyte proliferation, differentiation, and survival. Expression the melanocyte master regulatory gene Mitf is regulated by synergistic action of Pax3, Sox10, Lef1, and Creb on its promoter/enhancer. Mitf activates its own promoter by positive feedback loop. Once translated, Mitf protein is phosphorylated by the Erk kinase downstream activation of c-Kit signaling pathway. Phosphorylation of Mitf results in stabilization of Mitf-p300 transactivation complex, and thereby upregulating its transcriptional activity to stimulate expression of the target genes including Dct, Typ, and Tyrp1. c-Kit signaling also stimulates expression of Bcl2 to mediate melanocyte survival. Activation of melanocortin signaling pathway increases cytoplasmic cAMP concentration, which results in activation of CREB. This activated CREB directly binds the cAMP-responsive elements present on the promoter regions of Mitf and various melanosomal genes, and stimulates their gene expressions. Wnt signaling is required for melanocyte development. Activation of Wnt signaling results in the stabilization of β-catenin/Lef complex, which leads in transactivation of downstream target genes such as Mitf to promote melanocyte-fate specification and melanocyte differentiation. In contrast, keratinocyte expression of TGF-β plays a role in suppressing melanogenesis. Activation of TGF-β signaling results in the repression of Pax3 through phosphorylated Smads, which leads to Mitf repression to block melanocyte activation. Activation of Notch signaling is essential for the survival of melanoblasts, while underlying molecular mechanism is unclear. pSmad, phosphorylated Smads; Fzd, Frizzeled; cAMP, adenosine 3′:5′-cyclic monophosphate; CREB, cAMP responsive element binding protein, Pkc protein kinase C; Erk, extracellular signal-regulated kinase/mitogen-activated protein kinase, and pMitf, phosphorylated Mitf. Signaling molecules and transcription factors whose physiological roles in melanocytes are evidenced by genetic mutant animals are shown in red.

Pax3 is a member of paired-box homeodomain transcription factors which are highly conserved across species (Buckingham and Relaix, 2007). Mice harboring heterozygous loss of functional Pax3 mutation, which is encoded by the Splotch locus, exhibit a ventral white spotting phenotype, reflecting a role of Pax3 in melanoblast development (Moase and Trasler, 1992). In contrast, the homozygous mutant mice are embryonic lethal due to multiple defects during cardio-muscular and neural crest development (Moase and Trasler, 1992). Hence, Pax3 plays key roles in neural crest development as well as muscle and cardio-vasculature formation. Requirement of Pax3 in specification of neural crest is conserved throughout vertebrates (Buckingham and Relaix, 2007). Although the exact role of Pax3 in melanoblast development is less clear, it has been recently suggested that Pax3 may play a key role in maintaining the balance between undifferentiated and differentiated status of melanoblasts (Lang et al., 2005; Watanabe et al., 1998). For instance, several melanocyte-specific genes such as Mitf and Dct have a consensus-binding motif for Pax3 in their promoter/enhancer regions, and their expression appears to be directly regulated by Pax3 (Lang et al., 2005). Mitf, a master regulatory transcription factor for a melanocyte lineage, is transactivated by the synergistic action of Pax3 and Sox10 (Watanabe et al., 1998) (Figure 3), which is critical in mediating cell fate commitment into the melanocyte lineage. By contrast, Dct, which is one of the downstream targets of Mitf, is transactivated by the synergism between Mitf and Sox10 (Jiao et al., 2004), while Pax3, acting as a transcriptional repressor, antagonizes Mitf/Sox10-mediated transactivation of Dct (Lang et al., 2005) to prevent terminal differentiation of immature melanoblasts. Thus, Pax3 may play a role in mediating melanocyte-fate commitment, while simultaneously maintaining an undifferentiated state of immature melanoblasts (Lang et al., 2005).

Sox10 is a member of the high mobility group (HMG) family of transcription factors (Kelsh, 2006). Sox10 expression in pre-migratory neural crest cells is conserved throughout vertebrates (Kelsh, 2006), and it expression is gradually restricted in glial and melanocyte lineages (Hou and Pavan, 2008). Sox10 has broad functions during neural crest development including neural crest formation, fate specification, and stem cell maintenance (Hou and Pavan, 2008; Kelsh, 2006). These functions of Sox10 are also well conserved in vertebrates. For instance, in mice, heterogeneous Sox10 dominant negative mutant, which is encoded by dominant megacolon (Dom) locus, exhibits white spotting and megacolon (Potterf et al., 2001) (Figure 2C). In homozygous, the mutant mice are embryonic lethal due to multiple failures of neural crest derivatives including loss of melanoblasts, enteric ganglia and peripheral nerve cells (Potterf et al., 2001). Intriguingly, these similar defects are also found in homozygous Sox10 mutants in zebrafish (Dutton et al., 2001). Hence, in the melanocyte lineage, Sox10 plays indispensable roles in promoting survival, migration and differentiation (Wegner, 2005). Although its downstream targets are not completely clarified, it has been demonstrated that Sox10 preferentially binds enhancer regions of several melanogenic genes including Mitf, Dct, Tyr, and Tyrp1, and enhances their expression (Murisier and Beermann, 2006). Thus, Sox10 represents one of key players in orchestrating melanocyte differentiation and maturation by directly regulating expression of genes required for melanogenesis (Wegner, 2005).

The microphthalmia (mi) mutant mice exhibit hypopigmentation, small eyes, and osteopetrosis phenotypes in homozygous (Levy et al., 2006). This mutant locus encodes Mitf, a member of the Myc supergene family of basic–helix–loop–helix–leucine–zipper (bHLH-Zip) transcription factors. Mitf is thought to be a master regulatory transcription factor for the melanocyte lineage (Levy et al., 2006). Mitf directly regulates various melanocyte-specific genes including Dct, Tyr, Tyrp1, Si, Aim1, Mc1r, Mlana, and Trpm1 (Levy et al., 2006) (Figure 3). In the melanocyte lineage, Mitf has broad implications for survival, cell cycle regulation, migration, and differentiation, and these functions of Mitf appear to be well conserved in vertebrates (Hou and Pavan, 2008). Fate mapping studies of the Mitf-deficient mice showed dramatic reduction of migratory melanoblasts soon after the emergence of melanoblasts (Nakayama et al., 1998), suggesting that Mitf is required for the survival of melanoblasts at the early stage of their development. Consistently, in the zebrafish mitfa/nacre mutants, melanocytes are specifically lost in both embryo and adult, indicating essential role of mitfa in the maintenance of the melanocyte lineage (Lister et al., 1999). Mitf expression is also seen in retinal pigment epithelial (RPE) cells, osteoclasts, and mast cells, and plays important roles in these cell lineages (Hershey and Fisher, 2004; Planque et al., 2004). Interestingly, Mitf-homologous genes are also found in Drosophila and C. elegans, and mutation analysis Drosophila Mitf (Dmel) showed its potential role in eye imaginal disc formation, suggesting that the role of Mitf in eye development may be partially conserved between flies and vertebrates (Hallsson et al., 2004).

Endothelin signaling plays an indispensable role at the early stage of melanocyte development. In fact, it is known that classical hair spotting loci, piebald and lethal spotting (ls) encode Ednrb and its ligand Edn3, respectively (Giller et al., 1997; Matsushima et al., 2002) (Figure 2A). Consistently, hypomorphic mutations in EDN3 or EDNRB genes result in Waardenburg syndrome in humans (Attie et al., 1995; Edery et al., 1996). Hence, it is clear from these phenotypes that endothelin signaling is implicated in the melanocyte development. More precise studies using the mice in which Ednrb gene was conditionally ablated in the melanocyte lineage have shown that endothelin singling is transiently required for melanoblast development between E10.5 and E12.5, when melanoblasts are dispersing from the neural tube (Shin et al., 1999). Thus, these data suggests an essential role of endothelin signaling in promoting proper migration of melanoblasts at the early stage of melanocyte development rather than mediating fate-specification of melanoblasts. Consistently, in avian embryo, Ednrb2 expression is upregulated in melanoblasts prior to migrating into the dorsolateral pathway, and loss of Ednrb2 results in dramatic reduction of migrating melanoblasts (Harris et al., 2008). This indicates conservatory function of endothelin signaling between mammalian and avian species in promoting melanoblast migration. Interestingly however, in zebrafish, loss of function mutant of ednrb1/rose exhibits abnormal pigmentation pattern only in an adult stage, while leaving early larval pigment patterns intact (Parichy et al., 2000). This suggests that there may be certain species-specific variations in the requirement of endothelin signaling during melanocyte development.

Both melanocytes and melanoblasts highly express c-Kit tyrosinase kinase receptor, whose activation is mediated by Kitl (also known as steel factor or SCF) expressed in the epidermis (Yoshida et al., 1996; Yoshida et al., 1996). c-Kit signaling is implicated in diverse aspects of melanocyte regulation including proliferation, survival, migration, and differentiation (Jordan and Jackson, 2000; Mackenzie et al., 1997; Yoshida et al., 1996). Administration of an antagonistic antibody against c-Kit into mice depletes actively proliferating melanoblasts/ melanocytes, indicating an essential role of c-Kit signaling in proliferation and survival of these cells (Nishikawa et al., 1991; Yoshida et al., 1996). c-Kit signaling activates the MAP kinase pathway, which eventually phosphorylates Mitf by Erk kinase (Hemesath et al., 1998; Wu et al., 2000). Phosphorylation of Mitf results in stabilization of Mitf-p300 transactivation complex, and thereby enhancing its transcriptional activity to stimulate expression of the target genes (Hemesath et al., 1998; Price et al., 1998; Wu et al., 2000). These target genes include Bcl2 and Snai2, whose expressions are critical for melanoblast survival and migration, respectively (McGill et al., 2002; Sanchez-Martin et al., 2002) (Figure 3). This activation of Mitf is also implicated in induction of melanocyte terminal differentiation. The pleiotropic role of c-Kit signaling in the melanocyte lineage seems to be highly conserved among vertebrate species. In the mouse, Kit and Kitl mutations correspond to dominant spotting W and Steel (Sl) loci, respectively (Chabot et al., 1988; Huang et al., 1990; Miller et al., 2007) (Figure 2B). In humans, KIT and KITLG mutations are identified in piebaldism, a dominant ventral depigmentary disorder (Giebel and Spritz, 1991; Miller et al., 2007). Analysis of zebrafish c-Kit mutant sparse has demonstrated crucial functions of zebrafish c-Kit in survival and migration of melanocyte precursors (Parichy et al., 1999). Thus, from these phenotypes of Kit signaling mutants it is evident that Kit signaling plays a crucial role in melanocyte development.

2.3. Genetics of melanocytes

Melanocytes afford an advantageous model in understanding molecular basis of diverse cellular processes, since alterations in genes involved in the melanocyte regulation are easily identifiable as coat color mutant in animals. Indeed, currently more than 120 different loci have been identified as coat color mutants in mice (Bennett and Lamoreux, 2003). While some exceptions, these classical pigmentation mutants fall into roughly two distinct functional subgroups (Bennett and Lamoreux, 2003). The first group is comprised of mutations of genes involved in melanocyte development. As described in detail in the previous section, these mutations typically exhibit diverse degree of congenital white spotting of hair and skin due to reduction of melanoblast number and/or a defect in melanoblast migration during development. The other group consists of mutations of genes implicated in melanocyte functions including melanin biosynthesis, melanosome generation/maturation, and melanosome transfer. Mutations in such genes result in alterations in coat color, or hypo- or hyper-pigmentation phenotype without developing any white spotting. The examples of genes in this group are: Atrn, Dct, Mcr1, Myo5a, Oa1, P, Rab27a, Si, Tyr, and Tyrp1 (Bennett and Lamoreux, 2003).

In addition to these two subgroups, a new class pigmentation mutants, which exhibit a premature hair graying phenotype from the second hair cycles, has recently been proposed particularly from the gene knockout approach in the mouse (Mak et al., 2006; Moriyama et al., 2006; Nishimura et al., 2005; Ruzankina et al., 2007; Schouwey et al., 2007). In these mutants, hair pigmentation appears to be normal in the initial hairs, while it is prematurely lost from the subsequent hair cycles. As hair pigmentation from the second hair cycle is exclusively derived from melanocyte stem cells, this hair graying phenotype is attributed to the improper maintenance or emergence of melanocyte stem cells in the initial hair follicles. A classical mutant, Mitf^vit/vit is categorized in this group (Lerner et al., 1986; Nishimura et al., 2005) (Figure 2D). Although the responsible genes are not clarified, Faded (fe) (Oh et al., 1986) and London Gray (lgr) loci (Nishina et al., 1994) have been reported to exhibit hair graying phenotype. In addition to these spontaneous mutants, recent gene ablation studies indicate that loss of Bcl2, Notch1 and Notch2, or RBP-J results in premature hair graying starting from the second hair cycle (Aubin-Houzelstein et al., 2008; Moriyama et al., 2006; Schouwey et al., 2007) (Figure 6). Hence, by combing genetic manipulations and phenotypical identification of hair graying mutants, hair pigmentary system would offer an advantageous opportunity for elucidating genetic basis of stem cell regulation.

Taking advantage of mouse pigmentary system where genetic alterations affecting melanocyte regulation are readily identifiable by abnormal coat and/or skin color phenotypes, a large-scale chemical mutagenesis screen has been conducted to dissect new molecular pathways implicated in homeostatic regulation of mouse pigmentation (Fitch et al., 2003). This approach has allowed researchers to identify several new dark skin and dark coat color loci that have been previously unrecognized as pigmentation mutants (Fitch et al., 2003; McGowan et al., 2008; Van Raamsdonk et al., 2004). Importantly, this random mutagenesis approach offers a significant opportunity not only for identifying melanocyte endogenous molecules, but also for elucidating external molecular mechanisms of the surrounding microenvironment by which melanocyte homeostasis is regulated.

Because of visibility of the clear pigmentation pattern and feasibility of gene manipulations, zebrafish is another excellent model organism that allows identifying pigment genes by phenotype-driven genetics approaches. Because many of genes and biological processes specific for vertebrate development such as those in neural crest development appear to be conserved in zebrafish, genetic programs identified in zebrafish would be directly applied for human genetics. Indeed, lots of pigmentation mutants including rose/ednrb1, sparse/kit, nacre/mitfa, colourless/sox10, and golden/slc24a5 are found to be conserved human pigmentary disorders (Pickart et al., 2004). Hence, with large-scale forward genetics screens and identification and characterization of pigmentation mutants, the zebrafish offers invaluable opportunities in understanding of not only pigment cell biology, but also human disease.

4.1. Molecular characterization of melanocyte stem cells

Understanding molecular mechanisms underlying stem cell regulation within the niche is critical for stem cell biology. However, this has been largely hampered by the lack of systematic approaches to dissect the complete molecular makeup of stem cells at the niche. In an effort to characterize the molecular basis of the regulation of melanocyte stem cells at the niche, one attempt has been made to characterize molecular expression profile in melanocyte stem cells (Osawa et al., 2005). The gene expression and immunohistological analyses indicate that several key melanogenic transcription factors including Sox10 and Mitf are dramatically downregulated in melanocyte stem cell population (Osawa et al., 2005) (Figure 1). Intriguingly, in addition to these functional genes for melanocytes, the expression of several house keeping genes is also shown to be downregulated, suggesting an idea that basal transcription is globally down-modulated in the melanocyte stem cells.

Given critical roles of Mitf and Sox10 in the promotion of melanogenesis, their downregulation is supposed to lead to suppression of the melanocyte proliferation and differentiation. Indeed, it has been demonstrated that down-modulation of Mitf in a melanoma cell line results in induction of a G1 cell-cycle arrest which is associated with a dramatic morphological change into formation of a small and rounded-cell morphology (Carreira et al., 2006). Hence, these data indicate a crucial role of the Mitf downregulation in induction of the cell cycle exit in melanocytes (Carreira et al., 2006). Intriguingly, the fact that the Mitf downregulation induces a rounded-cell morphological change suggests the idea that Mitf downregulation may also contribute to the induction of a stem-cell like phenotype. Likewise, Sox10 has been shown to orchestrate melanoblast development and melanogenesis by directly upregulating melanogenic genes required for melanocyte maturation (Wegner, 2005). Thus, its downregulation is supposed to result in attenuation of melanocyte differentiation as it is seen in the Sox10^dom mutant mouse (Potterf et al., 2001).

Why are these transcription factors downregulated in melanocyte stem cell population? One possibility is the implication of Wnt signaling. Recent molecular expression studies have revealed the preferential expression of Wnt inhibitors in the bulge epidermal cells, suggesting that Wnt signaling might be constitutively downregulated within the bulge region (Morris et al., 2004; Tumbar et al., 2004) (Figure 5B). Given Wnt/Lef1 pathway-dependent expression of Sox10 and Mitf genes in melanocytes (Aoki et al., 2003; Saito et al., 2003), the inhibition of Wnt signaling may represent one cause of the downregulation of Sox10 and Mitf in the bulge melanocayte stem cells. In addition to the Wnt signaling, the c-Kit signaling pathway also appears to be downregulated at the same region (Mak et al., 2006) (Figure 5C, D), which may also lead in down-modulation of Mitf transcriptional activity (Hou et al., 2000; Wu et al., 2000). Although the physiological role of the downregulation of Sox10 and Mitf is unclear, their downregulation might be important for the maintenance of immature and resting states of the stem cells to ensure their long-term survival (Osawa et al., 2005).

Figure 5.

BMP, Wnt, and c-Kit signaling are activated in melanocytes in the hair bulb, but not in melanocyte stem cell population at the bulge region.

Skin sections from Dct-LacZ transgenic mice were stained with antibody against either phosphorylated Smad1/5/8, Lef1, c-Kit, or phosphorylated MEK1/2 (shown in green). Melanoblasts/melanocytes in the hair bulb or the bulge region are visualized by anti-LacZ antibody (shown in red). Melanocytes in the hair bulb are positively stained with these antibodies, indicating that BMP, c-Kit, and Wnt signaling pathways are activated in these cells. In contrast, none of these markers are detectable in the bulge melanoblasts, suggesting that these signaling pathways may not be implicated in melanocyte stem cell regulation.

Melanocyte stem cells appear to be maintained in a quiescent status at the niche (Nishimura et al., 2002). One of the central questions of stem cell biology is why and how stem cell quiescence is maintained at the niche. Recent loss of function studies have clarified key roles of negative cell-cycle regulators, such as p21, p27, and Pten, in the maintenance quiescent stem cells (Cheng et al., 2000; Groszer et al., 2001; Walkley et al., 2005; Yilmaz et al., 2006). In the melanocyte lineage, loss of Pten expression is commonly (∼20%) observed in human melanoma (Wu et al., 2003), suggesting a role of Pten in the regulation of melanocyte growth. Melanocyte-specific ablation of Pten gene ameliorates the hair depilation-induced premature hair graying (Inoue-Narita et al., 2008). This is due to abnormal expansion of the stem cell number in the Pten-deficient mice. In vitro studies, Pten-deficient melanocytes showed a dramatic reduction of the cell cycle inhibitor p27^kip1, suggesting a role of Pten through p27^kip1 in the induction of cell cycle arrest (Inoue-Narita et al., 2008). Similarly, loss of Pten results in expansion of stem-cell pool in hematopoietic stem cells and neural stem cells, but stem cells are eventually depleted due to exhaustion (Groszer et al., 2001; Yilmaz et al., 2006). Therefore, these data indicate that Pten negatively regulates stem cell proliferation by restricting cell cycle entry. Importantly, disruption of such mechanism would ultimately lead to stem cell exhaustion, suggesting a critical role of relative quiescence in the maintenance stem cell population.

Despite intensive studies, molecular mechanisms underlying regulation of stem cells remain largely unclear. Melanocytes afford an advantageous model by which to define molecules required for the stem cell maintenance, because genetic alterations controlling the stem cell viability are easily identifiable by the hair graying phenotype of the animal. Hence, by integrating melanocyte-specific gene manipulations and phenotype-based screening systems, melanocyte stem cell system can provide an invaluable tool to dissect crucial molecules involved in stem cell regulation.

4.2. Cell-to-cell interaction

The fact that melanocyte stem cells are randomly scattered within the bulge region without forming into cell clusters reminds us a possibility that the bulge region itself serves a niche for the melanocyte stem cells rather than depending on specific niche cells that might be present at the bulge. Histological analysis of melanocyte stem cells indicates their direct cell-cell contact with the surrounding keratinocytes (Nishimura et al., 2002), suggesting the involvement of the cell-to-cell interaction in the regulation of the stem cells.

It has been widely accepted that the homeostatic regulation of melanocytes may occur through cell-cell interactions between keratinocytes and melanocytes within the epidermis (Hirobe, 2005). Keratinocytes provide a complex of paracrine factors and adhesion molecules for melanocytes and plays a dominant role in regulating melanocyte survival, proliferation, and differentiation (Figure 3). Although molecular details of the keratinocyte-melanocyte interactions remain largely elusive, a series of recent studies have provided some mechanistic insight into molecular basis of melanocyte regulation by the epidermal keratinocytes (Cui et al., 2007; Yang et al., 2008).

In the human skin, UV irradiation causes skin pigmentation, which is attributed to acceleration of melanogenesis in the epidermis. By contrast, under a normal skin condition, majority melanocytes are maintained in an undifferentiated and resting status. Accumulating evidence indicates these melanocyte behavior is tightly regulated by the surrounding keratinocytes. For instance, under the normal condition, the growth and differentiation of the epidermal melanocytes is constitutively suppressed by the keratinocyte expression of TGFβ (Yang et al., 2008). Activation of TGFβ signaling in melanocytes leads to downregulation of Pax3, which is mediated direct repression of Pax3 promoter downstream TGFβ signaling pathway (Figure 3). As activation of the melanogenesis program requires the Pax3-Mitf transcription network, the repression of Pax3 by TGFβ signaling seizes the ability of melanocytes to respond to the external melanogenesis stimuli (Yang et al., 2008). Thus, the melanocytes are restricted in the undifferentiated and resting status on the normal keratinocytes. On contrary to the suppressive role of the normal keratinocytes, the keratinocytes following UV irradiation exhibit an intensive melanogenesis stimulatory activity. Several lines of studies have clarified a central role of the tumor suppressor protein p53 in the UV-induced melanogenic response (Cui et al., 2007). The activation of p53 leads to TGF β repression in the epidermal keratinocytes (Yang et al., 2008), allowing melanocyte to respond to the external melanogenic stimuli. In addition, the p53 also promotes marked induction of keratinocyte expression of αMSH and KITL, which is responsible for the acceleration of UV-induced melanogenesis in the epidermis (Cui et al., 2007; McGowan et al., 2008). Hence, by integrating TGF β repression and αMSH and KITL upregulation, the irradiated keratinocytes play a dominant role in stimulating proliferation and differentiation of the melanocytes (Figure 3).

In addition to these keratinocyte-derived paracrine factors, direct cell-cell adhesion between melanocytes and keratinocyte is also required for the proper regulation of melanocytes (Haass et al., 2005). Numerous in vitro culture studies demonstrate a dominant role of keratinocytes in promotion of the growth and survival of the primary melanocytes (Halaban et al., 1988; Sviderskaya et al., 1995; Yonetani et al., 2008). It is known that co-culture of melanocytes and keratinocytes can efficiently maintain the primary melanocytes in vitro, whereas in the absence of keratinocytes, melanocytes are hardly to grow in culture even in the presence of various keratinocyte-derived growth factors. This suggests the strict requirement of direct cell-cell contact to keratinocytes for the maintenance of the melanocytes (Yonetani et al., 2008). The spatial association between melanocytes and keratinocytes is partly mediated by homophilic interaction of E-cadherin molecules. Loss of E-cadherin expression, which is frequently seen in metastatic melanomas, results in aberrant growth and migration of melanoma cells, while conversely, forced expression of E-cadherin in melanoma cells renders them susceptible for keratinocyte-dependent growth controls (Hsu et al., 2000). Hence, these studies provide compelling evidence for a pivotal role of cell-cell interactions between melanocytes and keratinocyte in the proper regulation of melanocytes, while underlying molecular mechanisms remain largely elusive.

These data clearly indicate a predominant role of the keratinocytes in homeostatic regulation of melanocytes in the epidermis (Figure 3). Therefore, it would be speculated that keratinocytes might also be implicated in the regulation of melanocyte stem cells at the niche. In this regard, it is intriguing to note that TGFβ signaling appears to be preferentially activated in the bulge keratinocytes in the hair follicle (Tumbar et al., 2004). Given the important role of TGFβ signaling in restricting differentiation of the epidermal melanocytes (Yang et al., 2008), it would be possible that keratinocytes via TGFβ signaling may play a role in the maintenance of melanocyte stem cells at the niche. In addition to keratinocytes, dermal fibroblasts are also thought to be involved in melanocyte regulation. It is known that dermal fibroblasts secrete various stimulatory factors at the dermal papilla and stimulate proliferation and differentiation of melanocytes. The question of how melanocyte stem cells are regulated by the cells in other lineage(s) has been incompletely understood. This could be an important issue that should be addressed in future.

4.3. Notch signaling

Notch compromises a family of evolutionally conserved receptors, whose activation is mediated by specific cell-cell interaction with cells expressing the ligand, Jagged or Delta (Artavanis-Tsakonas et al., 1999). Once activated, the intracellular domain of the Notch receptors is cleaved and translocated into nucleus, where it forms transactivation complex with the RBP-J transcription factor to initiate transactivation of various target genes. Accumulating evidence indicates pleiotropic roles for Notch signaling in diverse cellular processes, including cell cycle arrest regulation, apoptosis/survival, differentiation, and stem cell maintenance, while the exact function of Notch signaling is highly cell context dependent (Hurlbut et al., 2007).

A series of recent studies have elucidated that Notch signaling represents a key component among keratinocyte-melanocyte interactions (Aubin-Houzelstein et al., 2008; Kumano et al., 2008; Moriyama et al., 2006; Schouwey et al., 2007) (Figure 6). Gene expression and immunohistochemical analyses showed that Notch signaling and its obligate target Hes1 are activated in the embryonic melanoblasts as well as the bulge melanoblast population (Moriyama et al., 2006), suggesting implication of Notch signaling in the regulation of the melanocyte lineage. As one of the Notch lignad, Jagged2, is abundantly expressed in the surrounding epidermal keratinocytes (Moriyama et al., 2006), Notch signaling in the melanoblasts is supposed to be activated as a consequence of melanocyte-keratinocyte interactions. To investigate the physiological role of Notch signaling in the melanocyte lineage, the RBP-J gene was conditionally ablated in a melanocyte-specific manner using a Tyr-Cre driver transgenic mouse line (Aubin-Houzelstein et al., 2008; Moriyama et al., 2006; Schouwey et al., 2007) (Figure 6). As the Tyr-Cre transgene has been demonstrated to be activated in the melanoblasts from E11.5 onward, the Cre-mediated ablation of the RBP-J gene could take place in any cells of these melanoblast descendants including melanocyte stem cells and differentiated melanocytes in the hair follicle (Delmas et al., 2003). Therefore, this system allows investigating the Notch functions in the melanocyte lineage. Conditional ablation of Notch signaling results in a dramatic reduction of embryonic melanoblasts (via apoptosis), which is evident from the dilution of initial hair pigmentation (Moriyama et al., 2006) (Figure 6). These animals also exhibit premature hair graying in hairs grown from subsequent hair cycles, suggesting a role of Notch signaling in the regulation of melanocyte stem cells (Moriyama et al., 2006; Schouwey et al., 2007) (Figure 6). In fact, histological analyses of postnatal follicles revealed a progressive loss of the bulge melanoblasts during the first hair cycle (Moriyama et al., 2006; Schouwey et al., 2007). Hence these data suggest the indispensable role of Notch signaling in the establishment and/or maintenance of melanocyte stem cells. Despite the obvious phenotype of the stem cell loss, the question about whether Notch signaling is required for stem cell specification or maintenance is not easy to be addressed. This is mainly because of lack of a definitive maker that allows detection of melanocyte stem cells. Nevertheless, a series of recent studies support the latter possibility that Notch signaling play a role in the maintenance of the stem cells (Kumano et al., 2008). Notch1^+/−;Notch2^+/− compound heterozygous mice display normal hair pigmentation at the birth but exhibit gradual hair graying from the second hair cycle (Kumano et al., 2008) (Figure 7). Since the stem cell establishment is completed during the initial hair follicle morphogenesis, this result suggests that Notch signaling is critical for the maintenance of melanocyte stem cells rather than their establishment. Indispensable role of Notch signaling in the maintenance of melanocyte stem cells has been further confirmed by pharmacological inhibition of Notch signaling in adult mice (Kumano et al., 2008). Administration of a γ-secretase inhibitor into adult mice causes gray spots on their hair, which is associated with a significant reduction of the bulge melanocyte stem cells. Hence, taken together, these findings suggest the multiple role of Notch signaling in: the promotion of melanoblasts survival; and the maintenance of melanocyte stem cells (Aubin-Houzelstein et al., 2008; Osawa and Fisher, 2008).

Figure 6.

Notch signaling plays a crucial role in melanocyte development.

Melanocyte-lineage specific ablation of RBP-J, an essential transcriptional mediator for Notch signaling, results in severe dilution of hair pigmentation (A). This is due to apoptotic elimination of melanoblasts during development. To test whether Notch signaling plays a role in melanocyte stem cells, the first hairs were shaved to induce the second hair cycling (B, circled area). The mice also exhibit premature hair graying from the second postnatal hair cycle (B), indicating an indispensable role of Notch signaling in the maintenance of melanocyte stem cells.

Figure 7.

Genetic links between Notch signaling and c-Kit signaling.

Both Notch1^−/+ (A) and Notch2^−/+ (B) mice show no obvious defect in hair pigmentation, while combinatory attenuation of Notch and c-Kit signaling enhances the spotting phenotype of W ^v/+ mice, suggesting a link between Notch signaling and c-Kit signaling. (C), Notch1^−/+; W^v/+; (D), Notch2^−/+; W^v/+; (E), Notch1^−/+; Notch2^−/+; W^v/+. Pictures are provided courtesy of Dr. Keiki Kumano and Dr. Shigeru Chiba.

These studies indicate that Notch signaling compromises an essential component of epidermal–melanocyte interactions; but what is the underlying mechanism? One possible clue to address this question is provided by Kuman et al, who suggested the potential synergism between Notch signaling and c-Kit signaling pathway (Kumano et al., 2008). It has been shown that, whereas both Notch1^+/−;Notch2^+/− mice and W^+/v mice show subtle coat phenotypes, their combinatory mutants in Notch1^+/−;Notch2^+/−;W^+/v compound mice result in dramatic enhancement of white spotting and premature hair graying phenotypes (Kumano et al., 2008) (Figure 7). Thus, these results clearly indicate a synergistic cooperation of these signaling pathways in the regulation of the melanocyte lineage (Kumano et al., 2008). Given the fact that both Notch and c-Kit signaling pathways are essential components of keratinocyte-melanocyte interactions, it would be tempting to speculate that the collaboration between these signaling pathways may function as a safety lock mechanism, by which melanocytes are allowed to survive only under the strict control of keratinocytes, while the exact molecular interactions between Notch and c-Kit signaling pathways in the melanocyte lineage remains open question.

It has been recognized that dysregulation of normal homeostatic regulation may cause neoplasmic conversion of the cells. Because melanomas emerge primarily within epidermal melanoblasts, it is reasonable to speculate that both normal and malignant melanoblasts may share key pathways that regulate biological homeostasis and maintenance. Indeed, it has been recently demonstrated that Notch signaling participates in melanoma progression (Balint et al., 2005; Liu et al., 2006). Thus, the elucidation of molecular mechanisms downstream of Notch is important not only for melanoblast development but also for understanding the molecular mechanisms of melanomagenesis.