Defining the Conditions for the Generation of Melanocytes from Human Embryonic Stem Cells



Because of their undifferentiated nature, human embryonic stem cells (hESCs) are an ideal model system for studying both normal human development and the processes that underlie disease. In the current study, we describe an efficient method for differentiating hESCs into a melanocyte population within 4–6 weeks using three growth factors: Wnt3a, endothelin-3, and stem cell factor. The hESC-derived melanocytes expressed melanocyte markers (such as microphthalmia-associated transcription factor and tyrosinase), developed melanosomes, and produced melanin. They retained the melanocyte phenotype during long-term cell culture (>90 days) and, when incorporated into human reconstructed skin, homed to the appropriate location along the basement membrane in the same manner as epidermis-derived melanocytes. They maintained a stable phenotype even after grafting of the reconstructs to immunodeficient mice. Over time in culture, the hESC-derived melanocytes lost expression of telomerase and underwent senescence. In summary, we have shown for the first time the differentiation of hESCs into melanocytes. This method provides a novel in vitro system for studying the development biology of human melanocytes.


Human embryonic stem cells (hESCs) are primitive, undifferentiated cells with the capacity for unlimited self-renewal and the ability to differentiate into multiple cell lineages (pluripotency). Their potential for regenerating diseased and damaged tissues is well known and has generated much excitement within both the lay and scientific communities. Embryonic stem cells are powerful tools with which to dissect processes underlying embryonic development. For many years, the use of these cells in medical and developmental studies has been hampered by the tendency of hESCs to spontaneously differentiate. Although there is evidence that some of these technical limitations are being overcome [1], there is still a need to develop methods and define conditions necessary to generate reproducible and homogenous populations of distinct cell lineages from hESCs. Although many cell types have been derived from hESCs—including neural cells, hematopoietic cells, cardiomyocytes, trophoblasts, and β-cells [2, [3], [4], [5], [6], [7], [8], [9], [10]–11]—their populations are often heterogeneous, and the propagation methods are not suitable for high-throughput generation. As an added complication, differentiation of some cell types were achieved under coculture conditions with supporting cells, making it very difficult to determine the key signaling pathways in the differentiation process.

Epidermal melanocytes play a critical role in protecting human skin from harmful ultraviolet (UV) rays. Their primary function is to produce the pigment melanin, which is packaged into vesicles known as melanosomes. Once generated, the melanosomes are rapidly transported to the surrounding epidermal keratinocytes, giving the skin its characteristic pigmentation. Defects in melanocytes can lead to pigmentary disorders such as albinism, piebaldism, and vitiligo, which are characterized by depigmented areas of skin and hair.

In vertebrate development, melanocytes originate from the neural crest and undergo a complex process of fate-specification, proliferation, migration, survival, and differentiation before finally residing in the epidermis [12]. Experimental evidence suggests that the WNT family of proteins [13, 14], stem cell factor (SCF) [15], and endothelin-3 (EDN3) [16] are all involved in the differentiation from neural crest to pigmented cells. The Wnt family of proteins induce neural crest and pigment cell formation in Xenopus, zebrafish [17, [18]–19], birds [14], and mice [13]. The essential role of Wnt family proteins in mouse melanocyte development are demonstrated by studies on mice null for Wnt-1 and Wnt3a [20]. Mutations in the SCF receptor c-Kit cause pigmentation deficiencies in mice and humans (piebaldism) [21, [22]–23]. Disruptions in the EDN3 system are associated with pigment loss and aganglionic megacolon in mice [16, 24] and Waardenburg-Shah syndrome and Hirschsprung's disease type 2 in humans [25, 26].

Attempts to study human melanocyte development have long been hampered by the differences in skin architecture between human and mouse. In particular, human melanocytes rest on the basement membrane among keratinocytes, whereas mouse melanocytes are localized near the hair follicles, deep in the dermis. This difference in environmental niche makes it difficult to extrapolate mouse development studies to humans. Although we know little about the existence of melanocyte stem cells in human skin, their presence is suggested by recent reports identifying a melanocyte stem cell population in the bulge region of the mouse hair follicle [27, 28]. Possible evidence for the existence of a melanocyte precursor population in human skin also comes from an earlier study that identified an intraepithelial KIT-positive cell population with melanocyte-like characteristics [29].

It is hoped that through the study of differentiation from hESCs to melanocytes, we may learn more about the phenotype of melanocyte stem cells in human skin and whether these cells can transform more readily than mature epidermal melanocytes. In the current study, we have defined conditions for the efficient derivation of a human melanocyte population from hESCs. The resulting hESC-derived melanocytes were subjected to detailed characterization and were found to produce melanin, synthesize melanosomes, and express all major melanocyte markers. When put into a reconstructed human skin model, the hESC-derived melanocytes behaved as fetal melanocytes, homed to the correct niche on the basement membrane, and produced melanosomes.

Materials and Methods

Cell Culture, DNA Fingerprinting, and Transmission Electron Microscopy

Human embryonic stem cell (ESC) lines H1 and H9 were obtained from the WiCell Research Institute (Madison, WI, and were cultured on mitotically inactivated mouse embryonic fibroblast (MEF) feeder layers [4, 30]. Human ESCs were passaged every 7–10 days by scraping colonies off the dish. Human epidermal melanocytes, keratinocytes, and dermal fibroblasts were isolated from neonatal foreskins and cultured [31]. All human materials, including hESCs, were processed according to the guidelines of the internal Institutional Review Board.

DNA fingerprinting was performed by the Molecular Diagnostic Core Facility of University of Pennsylvania Cancer Center using the GenePrint Fluorescent STR system (Promega, Madison, WI, Polymerase chain reaction (PCR) products were electrophoresed using an ABI PRISM Genetic Analyzer 3100 (Applied BioSystems, Foster City, CA, and then analyzed with GENESCAN and Genotyper software (Applied Biosystems) for allele identification. Transmission electron microscopy (TEM) was performed by the Biomedical Imaging Core Laboratory of University of Pennsylvania Cancer Center using standard techniques.

Differentiation Induction

The L-Wnt3a cell line and its parental control L cells were obtained from American Type Culture Collection (Manassas, VA, Conditioned media from L-Wnt3a cells (Wnt3a-CM) and L cells (L-CM) were generated in Dulbecco's modified Eagle's medium (DMEM) supplemented with 1% fetal bovine serum. Compared with L-CM, Wnt3a-CM upregulated the expression of total β-catenin in L cells (data not shown). To induce melanocytic differentiation, embryoid bodies (EBs) were derived from hESCs by suspending cells in growth factor-depleted medium (80% knockout DMEM/Ham's F-12 medium [Invitrogen, Carlsbad, CA,], 20% knockout serum replacer [Invitrogen], 200 mM l-glutamine [Sigma-Aldrich, St. Louis,], 1% 10 mM minimal essential medium nonessential amino acids [Invitrogen]) [4]. After 4 days, EBs were collected and plated on 10 ng/ml fibronectin (Becton, Dickinson and Company, Franklin Lakes, NJ, flasks in differentiation medium Mel-1, containing 0.05 μM dexamethasone (Sigma-Aldrich), 1× insulin-transferrin-selenium (Sigma-Aldrich), 1 mg/ml linoleic acid-bovine serum albumin (Sigma-Aldrich), 30% low-glucose DMEM (Invitrogen), 20% MCDB 201 (Sigma-Aldrich), 10−4 M l-ascorbic acid (Sigma-Aldrich), 50% Wnt3a-CM, 50 ng/ml SCF (R&D Systems Inc., Minneapolis,, 100 nM EDN3 (American Peptide Company, Sunnyvale, CA), 20 pM cholera toxin (Sigma-Aldrich), 50 nM 12-O-tetradecanoyl-phorbel 13-acetate (TPA) (Sigma-Aldrich), and 4 ng/ml basic fibroblast growth factor (bFGF) (Invitrogen). Medium was changed every other day. Differentiating cultures were passaged when adherent hESCs reached 60% confluence. In Wnt blocking experiments, soluble recombinant human Dickkopf-1 (DKK-1) (R&D Systems) at 20 μg/ml or recombinant mouse secreted frizzled-related protein-2 (sFRP-2) (R&D Systems) at 25 μg/ml was added to culture medium. In differentiation experiments that tested the effects of individual growth factors, EBs were induced in basal growth medium (the same as that used for EB generation) supplemented with each growth factor at the concentration employed in the complete differentiation medium.

Human Skin Reconstructs and Mouse Transplantation

Human skin reconstructs were generated as described [31]. Briefly, dermal reconstructs consisted of collagen type 1 (Organogenesis, Canton, MA) embedded with fibroblasts. Epidermal melanocytes or hESC-derived melanocytes were seeded into tissue reconstruct trays (Organogenesis) together with keratinocytes onto dermal reconstructs at a ratio of 1:5 melanocytes to keratinocytes. Reconstructs containing epidermal melanocytes or depleted of melanocytes were used as positive and negative controls, respectively. Two weeks later, reconstructs were harvested or grafted onto severe combined immunodeficient (SCID) mice, as reported [32]. Harvested reconstructs were fixed in 10% neutral buffered formalin, processed by routine histological methods, and embedded in paraffin. Grafted reconstructs were harvested 4 weeks after transplantation, embedded in OCT compound embedding medium (Sakura Fineteck, Torrance, CA), and frozen at −80°C. Frozen sections were cut and fixed in cold acetone for staining.

Immunocytochemical and Immunohistochemical Staining

Monolayer cells were fixed with 4% paraformaldehyde and stained with primary antibodies specific for microphthalmia-associated transcription factor (MITF) (monoclonal; Lab Vision, Fremont, CA,, tyrosinase (TYR) (monoclonal; Novocastra Ltd., Newcastle upon Tyne, U.K.,, tyrosinase-related protein-1 (TYRP1) (monoclonal; Signet, Dedham, MA), dopachrome tautomerase (DCT) (polyclonal; a gift from Dr. V.J. Hearing, Bethesda, MD), KIT (monoclonal; BD Pharmingen, San Diego,, SCF (polyclonal; Santa Cruz Biotechnology Inc., Santa Cruz, CA,, silver protein (SILV/HMB45) (monoclonal; Dako, Carpinteria, CA), and S100 protein (polyclonal; Dako) proteins. Human ESC colonies on the feeder layer were directly stained with monoclonal antibodies for stage-specific embryonic antigen (SSEA)-3 and SSEA-4 (The Wistar Institute). Human ESC colonies were fixed with 100% ethanol and stained with monoclonal antibodies for TRA-1–60 and TRA-1–81 (The Wistar Institute). Isotype-matched mouse antibodies and normal rabbit IgG were used as controls. After washings, primary antibody binding was detected by the corresponding Alexa Fluor 488 secondary antibodies (Invitrogen). Staining was observed by a Nikon E600 fluorescence microscope. Immunohistochemical studies for TYRP1, HMB45, and S100 were performed on paraffin or frozen sections using standard immunoperoxidase techniques.

Telomerase Activity Assay

Telomerase activity was determined by the TRAPeze enzyme-linked immunosorbent assay (ELISA) kit (Serologicals, Norcross, GA) according to the manufacturer's instructions. Briefly, cell lysates were obtained using a provided lysis buffer, and their protein concentrations were determined using the Bio-Rad II protein assay reagent (Bio-Rad, Hercules, CA, Equal amounts of protein were used for PCRs. PCR products were subjected to ELISA, and the absorbance of each sample was measured at 450 nm and 690 nm. Final absorbance (units) = A450A690. Aliquots of each sample were heat-inactivated to serve as controls.

TUNEL and Alkaline Phosphatase Staining

Cell viability was examined using the terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) kit (Roche Diagnostics, Indianapolis, IN). TUNEL label-only and DNase I-treated samples were included as negative and positive controls, respectively. Apoptotic cells were evaluated by a Nikon E600 fluorescence microscope. Alkaline phosphatase was detected by using the Vector Blue alkaline phosphatase substrate kit (Vector Laboratories, Burlingame, CA,

Purification and Validation of Wnt3a Protein

Soluble Wnt3a protein was purified from Wnt3a-CM following the procedure described previously [33]. For details, see Dr. R. Nusse's website at∼rnusse/assays/W3aPurif.htm. The activity of purified Wnt3a protein was evaluated by dose-response upregulation of total β-catenin in L cells.

Western Blotting, RNA Isolation, Reverse Transcription-PCR, and Sequence Analysis

Western blot analyses were performed as described [34] using antibodies against MITF, TYR, DCT, TYRP1, SOX10 (Abcam, Cambridge, U.K.,, PAX3 (Active Motif, Carlsbad, CA), β-catenin (BD Biosciences, San Diego, A monoclonal antibody against β-actin (Sigma-Aldrich) was used as a control. Band intensities on Western blots were measured by densitometry and normalized to β-actin control. RNA isolation and reverse transcription-PCR amplification of the MITF gene were described previously [34]. Sequence analysis was performed on an ABI PRISM 377 DNA sequencer (Applied Biosystems).

Statistical Analysis

Statistical analysis was performed using SAS software version 9.1 (SAS Institute Inc., Cary, NC). Data from three independent experiments presented in Figure 1 were analyzed. General linear modeling with repeated measures was used to test significant differences between the nine different media. Least square mean was used to test whether there are significant differences between media. In this model, total EB numbers or the surface area of adherent EBs differentiated under each medium condition was used as dependent variables, and medium and day were used as other independent variables. Statistical significance was defined as p < .05.

Figure Figure 1..

Highly efficient differentiation of human embryonic stem cells (hESCs) into pigmented melanocytes in melanocyte differentiation medium. (A): hESC colonies grew on lethally irradiated mouse embryonic fibroblast feeder layers. (B): Embryoid bodies (EBs) formed in suspension under feeder-free conditions at day 4. (C): In differentiation medium, cells migrated out from an EB adhered to the substrate by day 6. (D): Differentiating cells appeared with morphology typical of melanocytic lineage migrated out of an adherent EB by day 16. (E): By day 30, differentiated populations homogeneously displayed melanocytic morphology after splitting. (F): Unpigmented EBs prior to differentiation (left) contrast with a highly pigmented cell pellet from hESC-derived melanocytes collected 9 weeks after differentiation (middle). Pigmentation was sustained in hESC-derived melanocytes in long-term cultures (6 months) maintained in differentiation medium without TPA (right). In each case, 5 × 105 cells were spun down in each tube and photographed. (G): A transmission electron microscopy (TEM) image shows an hESC-derived melanocyte with many melanosomes in the cytoplasm 8 weeks after differentiation (arrow). (H): Premelanosomes (white arrow) and mature melanosomes (black arrows) are seen near the Golgi apparatus under TEM. (I): Melanocytes derived from hESCs proliferated rapidly. A typical growth rate (cells per plate) during week 10 in the absence of TPA is shown (mean ± SD from triplicate samples). Bars = (A), 3 mm; (B–D), 200 μm; (E), 100 μm; (G), 10 μm; and (H), 500 nm.


Differentiation of hESCs into Homogeneous Melanocytic Populations

Human ESCs (H1 and H9 lines) were routinely grown as colonies on mitotically inactivated MEF feeder layers (Fig. 1A). The cells were prepared for differentiation by mechanical removal from the feeder layer, resuspension in hESC media (without bFGF), and transfer to untreated tissue culture flasks. Under these conditions, the hESCs grew in suspension as EBs (Fig. 1B). Any contaminating feeder layer cells rapidly adhered to the original flask and were discarded following serial passage of the cells into new flasks. After 4 days of culture, EBs were plated onto fibronectin-coated flasks containing complete differentiation media (which contained multiple growth factors including Wnt3a, SCF, and EDN3, as described in detail in Materials and Methods). After 24–48 hours, some of the EBs formed adhesive colonies, out of which bipolar cells migrated (Fig. 1C, 1D). The differentiating hESCs continued to proliferate and reached ∼60% confluence after 3 weeks of continuous culture. At this point, the cultures were dissociated into single-cell suspensions using trypsin and were replated onto fibronectin-coated flasks. Homogenous cultures of cells with a melanocyte-like dendritic morphology were established following an additional 4–6 weeks' maintenance (Fig. 1E). Whereas EBs were unpigmented, hESC-derived melanocytes were highly pigmented by 9 weeks (Fig. 1F) and retained their pigmented phenotype in long-term culture over a 6-month period (Fig. 1F). This demonstrates the stable and efficient induction of melanocytic differentiation from hESC cells. The fact that melanocytic differentiation was induced in both the H1 and H9 hESC cell lines suggests that this method is broadly applicable to embryonic stem cell populations.

Characterization of hESC-Derived Melanocytes

Critical tests for melanocyte differentiation include pigmentation and melanosome synthesis. Melanosomes were identified in the hESC-derived melanocyte population by TEM (Fig. 1G, 1H; data not shown). The proliferation characteristics of hESC-derived melanocytes were then compared with those of human epidermal melanocytes isolated from newborn foreskin. Like the epidermal melanocytes, hESC-derived melanocytes proliferated continuously over a 6-month period, eventually undergoing senescence at passage 40 (a representative proliferation rate over one passage cycle is shown in Fig. 1I).

To confirm the melanocytic phenotype of hESC-derived cells, we determined expression of melanocyte-specific markers using immunostaining and Western blotting. Immunofluorescence studies demonstrated the presence of all tested melanocyte markers, including MITF, c-KIT, DCT, TYR, TYRP1, SILV, and S100 (Fig. 2A). MITF expression was predominantly present in nuclei, whereas the other proteins were localized to the cytoplasm. Dermal fibroblasts did not express any of the markers tested, and all newborn-derived epidermal melanocytes used as positive controls were positive (data not shown). Western blot analyses further confirmed that the hESC-derived melanocytes express melanocyte markers MITF, sex-determining region Y (SRY)-related transcription factor SOX10, the paired homeodomain transcription factor PAX3, TYRP1, DCT, and TYR (Fig. 2B). No specific bands were detected that corresponded to the melanocyte markers in either EBs or human dermal fibroblasts (except for PAX3 expression in MEFs).

Figure Figure 2..

Expression of markers characteristic of the melanocytic lineage in human embryonic stem cell (hESC)-derived melanocytes. (A): Eight weeks after differentiation, hESC-derived melanocytes were positive for a panel of melanocyte markers (top rows). Phase-contrast images of the same field are shown in the bottom row. A representative image obtained from isotype matched mouse antibodies is shown as control. Bar = 100 μm. (B): Western blots show that hESC-derived melanocytes were positive for melanocyte markers SOX10, PAX3, TYRP1, DCT, TYR, and MITF. Lane 1, dermal fibroblasts; lane 2, mouse embryonic fibroblasts; lane 3, embryoid bodies; lane 4, hESC-derived melanocytes at day 35; lane 5, primary epidermal melanocytes; lane 6, melanoma cell line 451Lu. The loading variation is shown by β-actin. Abbreviations: DCT, dopachrome tautomerase; MITF, microphthalmia-associated transcription factor; SCF, stem cell factor; SILV, silver protein; TYR, tyrosinase; TYRP1, tyrosinase-related protein-1.

To ascertain whether the hESC-derived melanocytes were of neural crest rather than retinal pigmented cell phenotype, total RNA was extracted from the hESC-derived melanocytes, and MITF cDNA was amplified by RT-PCR using primers specific for the neural crest-derived MITF isoform [34], we confirmed that hESC-derived melanocytes were of neural crest origin. Sequence analysis of MITF cDNA confirmed their neural crest origin based on the expression of epidermal melanocyte-specific isoform MITF-M (data not shown). DNA fingerprinting analysis showed identical alleles at 12 different STR loci and contained both X and Y alleles at the amelogenin locus between hESCs and hESC-derived melanocytes, confirming that the hESCs were the originating cells (data not shown). The mature phenotype of the hESC-derived melanocytes was demonstrated by the lack of telomerase expression (supplemental online Fig. 1). Further work demonstrated that hESC-derived melanocytes did not express markers characteristic of hESCs, such as alkaline phosphatase, SSEA-3, SSEA-4, TRA-1–60, and TRA-1–81 (data not shown).

Localization of hESC-Derived Melanocytes in Reconstructed Human Skin

Examining the behavior of hESC-derived cells in vivo or in a tissue-dependent context can provide definitive evidence for our hypothesis. Because, as noted in the Introduction, mouse skin is not a suitable model with which to study human melanocytes, we used a reconstructed human skin xenograft model in which keratinocytes are overlaid onto a dermis of human skin fibroblasts and collagen [31]. The hESC-derived melanocytes were introduced into the human skin reconstructs. When the skin reconstructs were harvested and sectioned, the hESC-derived melanocytes were found to localize to the basal keratinocyte layer, as do melanocytes in the skin of newborns (Fig. 3A). TEM studies revealed that the hESC-derived melanocytes grown in human skin reconstructs also expressed melanosomes (Fig. 3B, 3C). To answer whether the hESC-derived melanocytes were stable for prolonged periods of time in an in vivo situation, the cells were introduced into human skin reconstructs and then grafted onto SCID mice. This allowed the grafts to become vascularized and prolonged their survival (Fig. 3D). These studies demonstrate that the hESC-derived melanocytes homed to the basement membrane and remained stable for over 4 weeks, as demonstrated by the expression of the melanocyte marker TYRP1 (Fig. 3D). Control reconstruct grafts, which lacked hESC-derived melanocytes, expressed no TYRP1.

Figure Figure 3..

Homing of human embryonic stem cell (hESC)-derived melanocytes to the basal layer of the epidermis in three-dimensional organotypic cultures that mimic human skin architecture. (A): Human skin reconstructs (organotypic cultures) are generated without melanocytes (left; negative control), with human epidermal melanocytes (middle; positive control) or with hESC-derived melanocytes (right). The epidermal and dermal layers contain keratinocytes and fibroblasts, respectively, and are labeled in one of the representative H&E staining images. Both epidermal and hESC-derived melanocytes (arrows) are dispersed normally among the basal layer keratinocytes (shown at larger magnification in insets). (B): A transmission electron microscopy image shows an hESC-derived melanocyte among basal layer keratinocytes in skin reconstructs (arrow). (C): A close-up of the boxed area in (B) shows melanosomes at various stages in cytoplasm. (D): Identification of hESC-derived melanocytes in human skin reconstructs after being grafted onto severe combined immunodeficient mice. Grafted melanocyte-free reconstructs do not contain TYRP1-positve cells (left), whereas TYRP1 immunoreactive cells are found among basal layer keratinocytes in reconstructs containing either epidermal (middle) or hESC-derived (right) melanocytes (arrows). Bars = (A), 100 μm; insets, 40 μm; (B), 4 μm; (C), 1 μm; and (D), 50 μm. Abbreviations: F, fibroblasts; K, keratinocytes; SILV, silver protein; TYRP1, tyrosinase-related protein-1.

The Critical Role of Wnt in hESC-to-Melanocyte Differentiation

To date, little is known about the role of Wnt3a in human melanocyte differentiation. Here we show that after 2–3 weeks' growth in the complete differentiation medium containing 50% Wnt3a-CM, hESCs began to exhibit melanocyte morphology and expressed MITF (supplemental online Fig. 2A). They developed into pigmented melanocytes after 6–8 weeks (supplemental online Fig. 2B). When the concentration of Wnt3a-CM in the differentiation medium was lowered from 50%–20%, onset of melanocytic differentiation was delayed. After 4–5 weeks of culture, the hESCs exhibited melanocyte morphology, but the majority of cells lacked strong expression of MITF (supplemental online Fig. 2A).

When the hESCs were grown in the medium in which Wnt3a-CM (50%) was replaced by L-CM control media (which lacked any Wnt3a), the cells exhibited a different, nonmelanocytic morphology. These cells were negative for MITF, and the cell pellets remained unpigmented for 9–10 weeks (supplemental online Fig. 2B) and never acquired pigmentation even during long-term (>6 months) culture. Further proof of the indispensable role of Wnt3a in human melanocyte development was demonstrated by the use of WNT antagonists DKK-1 and sFRP-2 [35, 36]. In complete media (with 50% Wnt3a-CM), the differentiating hESCs proliferated normally for 3 weeks, even in the presence of DKK-1 and sFRP-2. After this point, however, proliferation ceased, and the cells could not be maintained for more than 5 weeks (data not shown). The cells also demonstrated extensive TUNEL staining, indicating the onset of apoptosis in the antagonist-treated hESC population (data not shown). As a final exploration of the role of Wnt3a, we replaced the Wnt3a-CM with purified Wnt3a. The activity of the purified Wnt3a was confirmed by its concentration-dependent upregulation of β-catenin levels (data not shown). It was also demonstrated that purified Wnt3a could substitute for Wnt3a-CM in the melanocyte differentiation medium and drive melanocytic differentiation, albeit with lower efficiency (data not shown).

Effects of the Individual Growth Factors on Melanocytic Differentiation

We investigated whether all three growth factors (Wnt3a, SCF, and EDN3) were required for melanocyte differentiation. We tested the growth factors individually and in every possible combination. The hESCs did not repopulate to a large-cell mass when Wnt3a was excluded from the differentiation media, suggesting that proliferation or cell survival is not supported in the absence of Wnt3a (Fig. 4A, 4B). Indeed, in the presence of Wnt3a alone or in combination with one other growth factor, the EBs survived, and ∼20% adhered to the substrate and underwent differentiation (Fig. 4A). The numbers of total differentiating EBs cultured in the nine different media were statistically analyzed. Although experimental time itself did not have a significant effect on the differentiating EB numbers (F = 1.01, p = .37), both time and medium demonstrated a significant combined effect on the EB numbers (F = 2.25, p = .03). Significant differences were observed between medium SCF/EDN3, SCF, EDN3, and the others at day 21, suggesting that lower numbers of differentiating EBs under these conditions may be caused by removal of Wnt3a-CM or L-CM. In Figure 4B, either experimental time or medium showed a significant effect on the surface area of adherent EBs (F = 5.18, p = .02; F = 3.48, p = .01, respectively). No significant differences were found among the media Wnt3a-CM, Wnt3a-CM/SCF, Wnt3a-CM/SCF/EDN3, and L-CM and complete medium; however, significant differences were identified between these media and medium SCF/EDN3, SCF, EDN3. These data further indicate that an unknown factor present in L-CM may act as a differentiation inducer or essential survival factor for differentiating hESCs.

Figure Figure 4..

Fate of EBs, both adherent and floating, during differentiation of hESCs cultured with individual or combined growth factors. Data obtained are from triplicate experiments using approximately 30 to 50 EBs in each trial. (A): Percentage of viable adherent (filled bars) and floating (blank bars) EBs during the first 3 weeks of differentiation. Basal medium containing SCF, EDN3, or both, resulted in only a small fraction of EBs adhering to substrate during the first 2 weeks. However, these EBs failed to proliferate into a large population. Consistent loss of floating EBs was also observed under these conditions. Wnt3a-CM alone and combined with other growth factors led to an increasing number of differentiating and adherent EBs. Complete differentiation media Mel-1 and L-CM were included as controls. (B): Percentage of flask surface areas occupied by differentiating EBs. EBs failed to differentiate and establish adherent cultures in basal medium containing SCF, EDN3, or both. In contrast, in any medium containing Wnt3a-CM, EBs progressively differentiated and produced adherent cultures. Using L-CM also resulted in proliferating adherent cultures. Abbreviations: EB, embryoid body; EDN3, endothelin-3; L-CM, conditioned medium from L cells; SCF, stem cell factor.

Although stable cultures are established in the presence of basal media supplemented with Wnt3a-CM, Wnt3a-CM + SCF, Wnt3a + EDN3, Wnt3a + SCF + EDN3, or L-CM, melanocytes were only established in the presence of Wnt3a + EDN3 and all three of the growth factors (Fig. 5). Melanocytes derived in the presence of all three growth factors persisted in culture for >90 days. Interestingly, an MITF+ population of cells was detected between 14–28 days in the presence of Wnt3a + EDN3, but these later regressed, which implies that the combination of Wnt3a + EDN3 may be sufficient for melanocyte fate determination. No melanocytes were established following any other combination of growth factors.

Figure Figure 5..

Requirement of Wnt3a, SCF, and EDN3 for melanocytic differentiation of human embryonic stem cells. Challenging embryoid bodies with individual or combined growth factors resulted in proliferating cell populations under several conditions, as indicated. Differentiated cells were examined at days 7, 14, 21, 28, and 90 by immunocytochemistry with melanocyte markers TYR (data not shown), MITF, and TYRP1. Images shown are from 28 days, with the exception of the insets. Melanocytes were identified at the early stage of differentiation (between 14 and 21 days) in the medium containing both Wnt3a-CM and EDN3 (insets); however, they were not detected after 28 days despite establishment of long-term cultures. Only a combination of Wnt3a-CM, SCF, and EDN3 sustained a long-term population containing melanocytic cells, as indicated by their morphology and immunoreactivity for MITF and TYRP1 (arrowheads). Medium containing L-CM also supported proliferation of nonmelanocytic cells. Abbreviations: EDN3, endothelin-3; L-CM, conditioned medium from L cells; MITF, microphthalmia-associated transcription factor; SCF, stem cell factor; TYRP1, tyrosinase-related protein-1.


This is the first reported derivation of a melanocyte population from human embryonic stem cells (hESCs). Using a novel feeder cell-free approach and three defined melanocyte growth factors (Wnt3a, SCF, and EDN3), we reproducibly generated a homogenous population of bona fide human melanocytes from two hESC lines (H1 and H9). The hESC-derived melanocytes exhibit the correct melanocyte morphology, are pigmented, synthesize melanosomes, and express all of the melanocyte markers tested.

A defining characteristic of melanocytes is their expression of melanocyte-specific transcription factors such as MITF. Of all of these transcription factors, MITF is best known as a critical player in melanocyte development [37]. Although the hESC-derived melanocytes expressed lower levels of MITF protein than newborn foreskin-derived melanocytes, they still expressed functional melanosomes and produced melanin, suggesting that they reached a critical expression level of MITF required for differentiation. In E10.5 mouse embryos, the appearance of MITF-positive cells at the dorsal midline of the neural tube is rapidly followed by melanocyte proliferation [38]. Two other melanocyte transcription factors, SOX10 and PAX3, also upregulate MITF expression [39, [40], [41], [42]–43]. Once activated, the major target genes of MITF are those of the melanocyte-specific tyrosinase family that encode three major pigment enzymes: TYR, TYRP-1, and DCT [44, 45]. Using hESCs as a pluripotent model system, we have shown for the first time that a network of transcription factors essential for melanocyte development and pigmentation—including MITF, SOX10, and PAX3—is upregulated during differentiation of human melanocytes.

The critical challenge in differentiating hESCs into cell populations of distinct lineages is defining the requisite cell culture conditions. The melanocyte differentiation medium used in this study contains multiple factors known to maintain melanocyte growth in culture, such as dexamethasone, TPA, and cholera toxin, as well as growth factors Wnt3a, SCF, EDN3, and basic fibroblast growth factor (bFGF). Among these factors, dexamethasone was recently reported to induce the differentiation of melanocytic cells from mouse ESCs [46]. TPA and cholera toxin, which activate protein kinase C and increase intracellular cyclic AMP, respectively, have long been included as growth enhancers in melanocyte media [47].

Interestingly, when cholera toxin, dexamethasone, and TPA were removed from differentiation medium, the three growth factors (Wnt3a, EDN3, and SCF) still induced partial melanocytic differentiation. This suggests that the three growth factors are sufficient to induce some degree of melanocytic differentiation on their own, and that cholera toxin, dexamethasone, and TPA act more as differentiation enhancers.

Not all of the factors in our differentiation system are as well-defined. Although a role for Wnt3a in the development of avian [14] and murine [13] melanocytes has been demonstrated, little is known about its role in the human melanocyte. Here we demonstrate that decreasing the concentration of Wnt3a-CM in the differentiation media from 50%–20% delayed the onset of melanocytic differentiation and use of soluble Wnt antagonists induced apoptosis in the hESC population. These results are consistent with previous studies showing that blocking the Wnts using sFRP leads to death in developing embryonic tissues [48]. Other studies have also shown that Wnts are survival factors for neural crest progenitors [20, 49] and hematopoietic stem cells [50].

There is increasing evidence that cell fate is the result of many signals acting in a synergistic manner [51]. In agreement with this idea, our depletion experiments demonstrate that the combination of Wnt3a, SCF, and EDN3 is critical for full melanocyte differentiation from hESCs. Treating the hESCs with any one of these growth factors alone failed to induce expression of the mature melanocyte markers TYR or TYRP1.

The absence of melanocytes in animals that are deficient in either EDN3 or its receptor suggests that this pathway is critical in the development of neural crest-derived melanocyte populations [16]. In vitro studies show that EDN3 promotes the proliferation, survival, and differentiation of melanocyte precursors [52, [53]–54]. Our results show that EDN3 is required for melanocyte development.

The role of the KIT gene (encoding the SCF receptor) in melanocyte development is more complex. Mutations in KIT do not affect specification of melanocyte lineage but instead hamper melanoblast survival at later developmental stages [21, 22]. In particular, SCF/KIT signaling is essential for migration, proliferation, survival, and differentiation of the precursor melanoblasts [38, 55, [56], [57]–58]. SCF/KIT signaling also upregulates MITF through the activation of mitogen-activated protein kinase [59].

A situation can be envisaged in which the three growth factors perform subtle but differing roles in hESC-to-melanocyte differentiation. In this model of activity, Wnt3a signaling determines melanocyte fate of neural crest cells, EDN3 contributes towards cell fate, and SCF promotes proliferation/survival of the committed progenitors. Further evidence for the close interaction of the three growth factors in melanocyte development comes from mouse embryonic development studies where Wnt3A is expressed at E7.5 [60], Edn3 is expressed at E9.5 [53], and Kit and Scf are expressed between E9.5 and E10.5 [61, 62].

Previous studies have described the generation of melanocytes from mouse ESCs [46]. In this instance, differentiation was independent of Wnt3a signaling and instead required SCF, EDN3, TPA, and dexamethasone. There are a number of possible explanations for the discrepancies between the study of Yamane et al. [46] and our own. First, that study used mouse ESCs [46]. It is likely that mouse melanocytes occupy an environmental niche different from that of human melanocytes and may require different signals. Second, that model used a coculture system in which differentiation of mouse ESCs was induced in the presence of stromal cells [46]. It could be assumed that Wnt3a was produced by the stromal cells.

In summary, we have defined conditions and developed a highly efficient method of differentiating a homogenous human melanocyte population from hESCs. We demonstrate that a complex synergistic interplay of three growth factors—Wnt3a, SCF, and EDN3—is required for the differentiation of human melanocytes from embryonic stem cells. We hope that a greater understanding of the underlying biology of melanocytes will yield important new insights into pigmentary diseases and the development of melanoma.


The authors indicate no potential conflicts of interest.


We thank the WiCell Research Institute for providing hESCs and its staff members L.J. Crandall, K. Van Den Heuvel, and D. Manning for technical support in culturing the cells. We are very grateful to Dr. James Thomson for constructive discussions, Dr. V.J. Hearing for the DCT antibody, James Hayden for photographing the cell pellets, Akihiro Yoneta and Richelle Takemoto for assistance in mouse experiments, the Molecular Diagnostic Core Facility for DNA fingerprinting, and the Biomedical Imaging Core Laboratory for TEM. This work was supported by grants from the National Institutes of Health (CA25874, CA80999, and CA76674) and the Commonwealth Universal Research Enhancement Program, Pennsylvania Department of Health.