Cooperative and indispensable roles of endothelin 3 and KIT signalings in melanocyte development



The development of melanocytes from neural crest-derived precursor cells depends on signaling by the receptor tyrosine kinase KIT and the G protein-coupled endothelin receptor B (EDNRB) pathways. Loss-of-function mutations in either of these two signaling receptor molecules cause a loss or a marked reduction in the number of melanocyte precursors in the embryo and finally lead to loss of the coat color. Using cultures of embryonic stem (ES) cells to induce melanocyte differentiation in vitro, we investigated the requirement for EDNRB signaling during the entire developmental process of the melanocyte, in association with that for KIT signaling. During the 21-day period necessary for the induction of mature melanocytes from undifferentiated ES cells, endothelin 3 (EDN3), a ligand for EDNRB, increased the number of melanocytes in proportion to the period during which it was present. We tested the compensatory effect of EDNRB signaling on KIT signaling in vivo by using KitW-LacZ/KitW-LacZ ES cells and confirmed that the ectopic expression of EDN3 in the skin reduced the white spotting of KitW57/KitW57mice. KIT ligand (KITL) and EDN3 worked synergistically to induce melanocyte differentiation in vitro; however, the complete lack of EDNRB signaling attained by the use of EDN3−/− ES cells and an EDNRB antagonist, BQ788, revealed that the resulting failure of melanocyte development was not compensated by the further activation of KIT signaling by adding KITL. Simultaneous blockade of EDNRB and KIT signalings eliminated melanocyte precursors completely, suggesting that the maintenance or survival of early melanocyte precursors at least required the existence of either EDNRB or KIT signalings. Developmental Dynamics 233:407–417, 2005. © 2005 Wiley-Liss, Inc.


Melanocytes are good cells for basic studies on the sophisticated interactions between environmental cues and intracellular genetic programs that enable proper control of biological processes required for the growth and differentiation of cells of distinct lineages from stem cells or precursors (Hall, 1999; Le Douarin and Kalcheim, 1999). KIT ligand (KITL, also called mast cell growth factor, steel factor, or stem cell factor) and endothelin 3 (EDN3) are two major signaling molecules thought to be indispensable for the early phase of melanocyte development both in vivo and in vitro. Null mutant mice for these two factors, such as KitlSl and Edn3ls, show overlapping white spotting or white coat color phenotypes that result from the loss of skin melanocytes (Geissler et al., 1988; Baynash et al., 1994). Mice with mutations in the receptor for KITL (KIT) or in that for EDN3 (EDNRB) showed the same coat color defects (Copeland et al., 1990; Hosoda et al., 1994). Not only EDN3 but also EDN1 and EDN2 were also suggested to act through EDNRB (Pla and Larue, 2003). Because these two receptors transduce different types of intracellular signals—KIT is a receptor tyrosine kinase and EDNRB is a G protein-coupled receptor—differences expected in their specific functions, if any, are issues yet to be reconciled.

Melanocytes are the immediate descendant of neural crest cells, which give rise to several vertebrate cell types (Hall, 1999; Le Douarin and Kalcheim, 1999). With respect to these cell types, KIT expression is restricted to melanocytes of the hair, skin, and choroids (Manova and Bachvarova, 1991; Bernex et al., 1996), whereas EDNRB expression is detected in enteric ganglia as well as in melanocytes (Opdecamp et al., 1998; Lee et al., 2003). Efforts to accurately track the lineage determination of pluripotent neural crest to specific lineage cells have revealed the more or less stochastic or asynchronous nature of this process, although melanocyte lineages are supposedly specified as a distinctive sublineage (Henion and Weston, 1997). Wnt signaling was suggested to be relevant to this determination process (Dorsky et al., 1998; Jin et al., 2001). Whereas KIT and EDNRB signalings are also suggested to influence this determination process (Kunisada et al., 1998, 2002), these two factors are thought to mainly act as growth and/or migration factors for melanocyte lineage cells (Kunisada et al., 1998; Jordan and Jackson, 2000; Pla et al., 2004). As for EDN3, its indispensable roles in melanocyte development during a narrow developmental stage from embryonic day (E) 10.5 to E12.5 was established in vivo (Shin et al., 1999). However, later distinctive roles beyond these stages were also clearly indicated (Lahav et al., 1996; Hou et al., 2004).

Neural crest cells that have emigrated from dissected developing neural tubes provide an excellent manipulatable culture system to investigate the process of cell fate determination; however, considerably hard-to-prepare homozygous embryos with defects in melanocyte development and limited cell numbers obtained from individual tubes may hamper such experimental research. In addition, neural crest cells might have been determined before delineation from the neural tube, which, therefore, would make it impossible to investigate the signals required for this process (Henion and Weston, 1997). Because of their potential to give rise to virtually all cell lineages when introduced into a blastocyst, embryonic stem (ES) cells differentiate into a variety of cell types in vitro when appropriate environmental cues are reproduced in the culture system (O'Shera, 1999; Prelle et al., 2002; Turksen, 2002; Wichterle et al., 2002; Kim et al., 2002). In these efforts, culture systems have been refined to selectively induce specific cell types and not to generate other cell types by using a combination of growth factors and forced expression of lineage-specific transcription factors. Among these efforts, we previously established a culture system in which cells of the melanocyte lineages can be induced from ES cells. When undifferentiated ES cells were cocultured with a bone marrow-derived stromal cell line ST2, cutaneous melanocyte lineage cells were efficiently and reproducibly generated in a strictly KIT signaling-dependent manner (Yamane et al., 1999).

In this study, we aimed at revealing the extrinsic control of melanocyte development by using cultures of ES cells in which the developmental process, including the determination of melanoblasts from their immediate precursors, is on-going in a fully manipulatable manner. Genetically modified ES cell lines defective in KIT and EDNRB signalings and also antagonists for these signaling molecules enabled us to create a completely null status of these signaling pathways. We show that EDNRB signaling is indispensable for melanocyte development irrespective of the KIT signaling and that these two pathways work cooperatively to maintain melanocyte precursor cells.


Roles of EDN3 in Derivation of Melanocytes From ES Cell Cultures

We previously established the culture conditions conducive to the induction of normal cutaneous melanocytes from ES cells (Yamane et al., 1999). In this culture system, ES cells are inoculated onto ST2 stromal cell monolayers; and 2 to 3 weeks later, pigmented melanocytes are induced. Induction of melanocyte precursors or melanoblasts was suggested to start on day 8 of the culture. To test the inductive role of EDN3, we added recombinant EDN3 to this culture system in combination with other factors to test their influence on melanocyte induction. As shown in Figure 1A, dexamethasone (Dex) alone acted as a melanocyte inducer; but other factors, including fibroblast growth factor 2 (FGF2), cholera toxin (CT), and EDN3, did not show inductive activities when added alone or in various combinations. This requirement for Dex in melanocyte induction is detectable only from day 12 of the culture and, thus, is not relevant to the early events of melanocyte differentiation, as previously reported (Yamane et al., 1999). When Dex was combined with FGF2 and CT, twice the number of melanocytes was induced; whereas EDN3 did not increase the number of melanocytes in combination with Dex, Dex + FGF2, or Dex + CT, as shown in Figure 1A. We, thus, speculate that EDN3 was already supplied from this culture system, possibly from other cell lineages induced from ES cells.

Figure 1.

Effects of endothelin 3 (EDN3) and various other factors on melanocyte differentiation from embryonic stem (ES) cells. A: One-thousand D3 ES cells were inoculated onto ST2 monolayers and incubated in the presence of the indicated factors. On day 21 of culture, the cells were trypsinized and the number of melanocytes containing melanin granules was then counted. Each bar represents the average value of triplicate cultures in a representative experiment, and horizontal lines indicate standard deviations. Dex, dexamethasone; FGF2, fibroblast growth factor 2; CT, cholera toxin. B: D3 ES cells were cultured under the standard condition on ST2 monolayers in medium supplemented with Dex, CT, and FGF2 and on the days indicated, reverse transcriptase-polymerase chain reaction was carried out to detect EDN1, EDN3, EDN receptor A (EDNRA), EDN receptor B (EDNRB), and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) messages. C:EDN3−/− ES cells lines were induced to differentiate into melanocytes. As in A, 1,000 EDN3−/− ES cells were inoculated onto ST2 monolayers in medium supplemented with the factors indicated; and on day 21 of culture, the number of melanocytes was counted.

To detect the messages of endothelins in the present culture system, we performed reverse transcriptase-polymerase chain reaction (RT-PCR) at various times during the culture period (Fig. 1B). The EDN3 message was detected as early as day 8 in the present ES cell cultures, whereas no message was detected in the control ST2 stromal cell cultures, suggesting that EDN3 in these cultures was mainly produced from cells derived from ES cells. EDNRB was clearly detected starting from day 8, although it was detected in a small amount before then, including in undifferentiated ES cells (day 0). Therefore, it is likely that melanocyte lineage cells arising from ES cells in this culture system expressed EDNRB and that the ligand EDN3 is mainly supplied from cells derived from ES cells. EDN1, which is a potent growth and differentiation factor for melanocyte lineages, was also detected in the ES cell cultures. However, this message was also detected in the control ST2 cell cultures lacking ES cells. No EDN2 message was detected in the present culture system (data not shown).

Effect of EDN3 on EDN3−/− ES Cells: Melanocyte Development Is Keenly Dependent on EDN3 Signaling

In the previous section, we showed that the possible EDN3 source in this culture system is ES cell derivatives and not ST2 cells. Therefore, to avoid the effect of endogenous EDN3, we used ES cells in which both alleles of the EDN3 gene had been inactivated (Baynash et al., 1994). As shown in Figure 1C, EDN3−/− ES cells did not differentiate efficiently into melanocytes in the presence of Dex, FGF2, and CT or in most of their combinations, in contrast to the wild-type ES cells shown in Figure 1A. When EDN3 was added to the cultures, nearly 50,000 melanocytes per well were generated but only in the combinations that included Dex. The addition of FGF2 or CT did not increase the number of melanocytes. Thus, we confirmed that endogenous EDN3 is required for melanocyte differentiation from the wild-type ES cells. Of interest, without EDN3, the presence of Dex, FGF2, and CT produced a small but considerable number of melanocytes, in this case 5,000 per well, indicating that FGF2 and CT in combination with other endogenous factors could replace part of the EDN3 signaling. It also should be noted that EDN3 dependence of this culture system in turn demonstrated the minimal roles of EDN1 and the receptor EDNRA, whose expression was detected as was shown in Figure 1B, for melanocyte development in the cultures, as is clearly the case for in vivo development of melanocytes. According to these results, we regard the EDN3−/− ES cell as a necessary tool to evaluate the effect of EDN3 in the present culture system. Most of the experiments using EDN3−/− ES cells hereafter were carried out in the presence of Dex, FGF2, and CT, since the combination of these factors resulted in a stable and maximum response to the additional factors, including EDN3.

We then tested the requirement of EDN3 signaling throughout the life of the cultures by having EDN3 present for various times, as shown in Figure 2A. Cytological observation of the mature melanocytes observed at the end of the culture period revealed fully pigmented cells with a dendritic shape, as shown in Figure 2B. In all cases, pigmented melanocytes were first detected on day 15 or 17 of the culture and then increased in number until day 21. Based on Figure 2A, the total number of melanocytes induced strictly depended on the length of time that EDN3 was present, irrespective of the timing of EDN3 addition. In other words, as to the final number of melanocytes induced, EDN3 was equally effective during any phase of the culture. Although we did not measure the number of melanocyte precursor cells or melanoblasts during the culture, Figure 2A clearly indicates that melanocytes, including melanoblasts could respond to EDN3 signaling during their entire life cycle. Unexpectedly, the presence of EDN3 only from day 0 to day 3 of the culture period increased the melanocyte number fivefold over the control number, although no melanocyte precursors were detected during this early stage in the culture (Yamane et al., 1999). This finding might be partly attributed to the fact that EDNRB was detected in undifferentiated ES cells, as was shown in Figure 1B; although we did not investigate this EDN3 effect on the very early phase of the cultures.

Figure 2.

Effects of the time of addition or removal of endothelin 3 (EDN3) on the generation of melanocytes from embryonic stem (ES) cells. A: Two-thousand EDN3−/− ES cells were inoculated onto ST2 monolayers, and EDN3 was added or removed according to the schedule indicated at the left of each bar. For removal of EDN3 from the cultures, EDN3-containing medium was aspirated, and the cells were then washed two times with phosphate buffered saline and then refed with the medium lacking the factor. B: Representative higher magnification of melanocytes in one of the cultures was made.

To further investigate the presence of a “very sensitive” or “ more strictly dependent” culture period for the EDN3 effect during the induction of melanocytes in this culture system, we kept the cells in the presence of EDN3 for various 3-day periods during the course of the culture period. As shown in Figure 3A, day 6 to 9, day 12 to 15, and day 18 to 21 seemed to be the significantly more-effective periods, for numbers of melanocytes up to one third of the control (in which EDN3 was present throughout the culture period) were generated. In a reverse experiment, we then deleted EDN3 for various 3-day periods during the culture period. As shown in Figure 3B, deletion from day 6 to 9 or during later periods significantly reduced the final number of melanocytes, which reduction became progressively greater as the deletion period occurred later during the culture period. Still, none of the 3-day deletions fully reduced the induction of the melanocytes to the level found in the control (no EDN3 added). Finally, we kept the cells in the presence of EDN3 for two periods of time: first, for 2-day periods from day 0 to day 14 and, second, from day 18 to 21. In this experiment, again the number of melanocytes was not restored and, at most, 2 times more melanocytes were induced compared with the control cultures in which EDN3 was present only from day 18 to 21 (Fig. 3C). Thus, although there might be relatively more EDN3-sensitive or -dependent periods during the entire culture period, those periods must not be sensitive enough to prevent melanocyte development in the absence of EDN3.

Figure 3.

Effects of the timing and periods of the endothelin 3 (EDN3) addition on generation of melanocytes from embryonic stem (ES) cells. A–C: A total of 1,000 (A) or 2,000 (B,C) EDN3−/− ES cells were inoculated onto ST2 monolayers in medium supplemented with dexamethasone, cholera toxin, and fibroblast growth factor 2. EDN3 was present during the time periods shown at the left of each bar.

Lack of KIT Signaling Was Compensated by EDN3 During the Development of Melanocytes From ES Cells

Strict dependency of melanocyte lineage cells in vivo on KIT signaling was clearly indicated in the case of Kit mutant mice (Copeland et al., 1990; Nishikawa et al., 1991); and this dependency was also confirmed in our ES cell culture system (Yamane et al., 1999) by using ACK2 antibody to block KIT signaling. In fact, ES cells established from embryos homozygous for a Kit null allele produced by an insertion of the LacZ gene (KitW-LacZ/KitW-LacZ ES cells, Bernex et al., 1996) did not differentiate into melanocytes (Fig. 4A). In this case, although ST2 stromal cells are known to produce an adequate amount of KITL, a ligand for KIT, to support melanocyte differentiation from wild-type ES cells, endogenously produced EDN3 expected to be produced from ES cell-derived cells could not complement the lack of KIT signaling in melanocyte lineage cells derived from KitW-LacZ/KitW-LacZ ES cells, as is the case for in vivo melanocyte differentiation. The small amount of EDN3 message observed at day 8 of culture (PCR band in Fig. 1B) suggests that a sufficient amount of EDN3 to compensate the lack of KIT signaling may not be supplied from the cells derived from wild-type ES cells in this critical culture period for the development of melanocyte precursors. However, increasing the amount of exogenous EDN3 supported the differentiation of the melanocytes from KitW-LacZ/KitW-LacZ ES cells, as shown in Figure 4A. Whereas 50 ng/ml EDN3 was enough to fully induce melanocytes from wild-type ES cells (200,000 melanocytes per well, in this experiment), 200 ng/ml EDN3 could induce nearly 100,000 melanocytes from KitW-LacZ/KitW-LacZ ES cells. Representative cultures of melanocytes from KitW-LacZ/KitW-LacZ ES cells and those from wild-type controls are shown in Figure 4B. Thus, signaling exerted by the exogenous EDN3 compensated the defective induction of melanocytes from KitW-LacZ/KitW-LacZ ES cells.

Figure 4.

Effects of endothelin 3 (EDN3) on the generation of melanocytes from embryonic stem (ES) cells with KIT mutation. A: Two-thousand KIT-null (KitW-LacZ/KitW-LacZ) ES cells and control ES cells were cultured in the presence of various concentrations of EDN3, and the number of mature melanocytes on day 21 of culture was counted. B: Photographs of the representative dishes for each EDN3 concentration are shown. C: A white spot on the belly of a C57BL/6-KitW-57J/KitW-57J mouse and a reduced white spot in a hk14-EDN3/?; KitW-57J/KitW-57J double mutant mouse is also shown. All three of the double-mutant mice established showed a similar phenotype.

Our established transgenic mice expressing EDN3 in their skin also confirmed this compensation in vivo. Transgenic mice in which EDN3 cDNA was driven by the cytokeratin 14 promoter facilitated melanocyte generation in interfollicular skin, and the entire skin was occupied by a large number of melanocytes throughout the life of these animals (Kunisada et al., 2002). When these EDN3 transgenic mice were crossed with KitW57/KitW57 mice, in which only a minimal amount of KIT tyrosine kinase activity was retained and most of the coat color of the trunk was lost (Sawada et al., 1991), the pigmentation defect was mostly reversed, as shown in Figure 4C, indicating that the absence of KIT signaling was compensated by the exogenously supplied EDN3 in vivo.

Complete Lack of EDNRB Signaling Was Not Compensated by KIT Signaling

When EDN3−/− cells were supplemented with KITL, significant recovery of the melanocyte number was observed (Fig. 5A). This effect was also observed in the presence of hepatocyte growth factor (HGF), a ligand for c-met receptor tyrosine kinase, which also is known to affect melanocyte development (Kunisada et al., 2000), suggesting that EDN3 signaling was compensated by the signal transduction machinery activated by these receptor tyrosine kinases. Evidently, the combined treatment with KITL and EDN3 caused a marked increase in the number of melanocytes, to more than twofold the number observed with EDN3 alone or with IRL1620, a strong agonist of endothelins, alone (Fig. 5A).

Figure 5.

Effects of endothelin 3 (EDN3), KIT ligand (KITL), hepatocyte growth factor (HGF), and their antagonists on the generation of melanocytes from embryonic stem (ES) cells. A: KITL, EDN3, IRL1620 (EDNRB agonist), and HGF and certain combinations of them were present from the start of the cultures. B: BQ788, an antagonist of EDNRB; ACK2 antibody, an antagonist of KITL; and the other combinations indicated were added from the start of the cultures. Note that A and B are parts of a single experiment and that the bars representing no factor added (indicated as “no”) are common to both figures. C: After the cells had been in the presence of BQ788 for the periods indicated, the cultures were washed and refed with medium containing EDN3 until day 21 of the culture. Two-thousand EDN3−/− ES cells were inoculated into each culture dish. Asterisks indicate P < 0.05

Several hundreds to thousands of melanocytes were reproducibly induced from EDN3−/− ES cells under the standard conditions without EDN3, as shown in Figure 5A. This emergence of a small number of melanocytes likely resulted from the small amount of EDNRB signaling stimulated by EDN1 produced by ST2 stromal cells, as shown in Figure 1B, because the addition of BQ788, an antagonist for endothelins acting through EDNRB, almost completely abolished melanocyte development (Fig. 5B). The minimal amount of EDNRB signaling was not adequate to induce mature melanocytes, because the addition of ACK2 antibody alone fully suppressed their induction. The effect of EDN3 or IRL1620 shown in Figure 5A was not suppressed by the simultaneous treatment with ACK2 (Fig. 5B), indicating that the lack of KITL signaling was compensated by EDNRB signaling, as suggested by the experiments in the previous section. The stimulative effect of KITL shown in Figure 5A was completely blocked by the BQ788 treatment, as shown in Figure 5B. In this experiment, any small amount of EDNRB signaling through any possible endogenous endothelins was eliminated by adding BQ788 and the use of EDN3−/− ES cells. Thus, extremely stringent depletion of EDNRB signaling may not be compensated even by the additional activation of KIT signaling by KITL. It is likely that EDN3 signaling governs a crucial developmental step that could not be rescued by the KIT signaling at least in the present culture system.

Maintenance of the Early Progenitors of Melanocytes Depends on Either EDNRB or KIT Signaling

As shown previously (Yamane et al., 1999), the emergence of precursors of melanocytes characterized by their dependence on KITL and expression of dopachrome tautomerase (DCT) and KIT markers were suggested to occur as early as day 8 of the cultures. In the presence of BQ788 from the initiation of the cultures up to day 8, no significant effect was observed when EDN3 was added after the removal of the BQ788 (Fig. 5C). Although the presence of BQ788 for 10 days or longer in the cultures progressively suppressed the differentiation of melanocytes even followed by the addition of EDN3, significant number of melanocytes were detected even when BQ788 was removed and EDN3 was present from day 16 (Fig. 5C; 17,000 ± 3,700 mature melanocytes were induced on day 21). This finding suggests that melanocytes precursors that readily differentiate into mature melanocytes within 6 days were maintained without EDNRB signaling. Based on the results shown in Figure 4A, melanocyte precursors were also maintained without KITL signaling, since the addition of EDN3 successfully induced melanocytes from KitW-LacZ/KitW-LacZ ES cells.

Coexistence of KIT and EDNRB Signalings Is Required for Maintenance of Early Melanocyte Precursors

We finally asked about the induction of melanocyte precursors under the situation in which KITL and EDNRB signalings were both depleted. After 16 days of culturing ET3−/− ES cells with ACK2 and BQ788 followed by returning them to the normal culture condition including EDN3, no mature melanocyte was observed after 21 days. Even after 30 days in the presence of EDN3, we could not detect any substantial number of mature melanocytes (Fig. 6B); whereas significant numbers of melanocytes were induced in the control cultures containing EDN3 from the initiation of the cultures (Fig. 6A). A reduction in the formation of melanocyte lineage cells or their precursors was also indicated by the RT-PCR analysis of melanocyte-specific MITF-M transcription factor (Yajima et al., 1999). On day 10, MITF-M was detected only in the cultures supplemented with EDN3; then, on day 16, MITF-M became detectable in the standard cultures without EDN3 (Fig. 6C). On day 20, MITF-M was detected in the cultures treated with ACK2; whereas cultures treated with BQ788 or BQ788+ACK2 both still showed a minimal amount of MITF-M (Fig. 6C). Thus, the cultures treated with BQ788+ACK2 up through day 16 did not form MITF-M–positive melanocyte precursors and produced no melanocytes after prolonged culture in the presence of EDN3.

Figure 6.

Effects of simultaneous addition of antagonists for endothelin receptor B (EDNRB) and KIT on the generation of melanocytes from embryonic stem (ES) cells. Two-thousand EDN3−/− ES cells were first cultured in the presence of BQ788 and ACK2 antibody for 16 days, washed with phosphate buffered saline, and refed with the standard medium containing EDN3. A: After 5 days, pigmented mature melanocytes were generated in the control cultures. B: On the other hand, even after 15 days in the presence of EDN3, BQ788- and ACK2-treated cultures generated no mature melanocytes. C:EDN3−/− ES cells were cultured with combinations of BQ788, ACK2, or their combination, as indicated under each lane for the indicated periods. Then, total RNA was isolated from the cultures and reverse transcriptase-polymerase chain reaction was performed for the measurement of melanocyte lineage-specific MITF-M messages. Lane m and ES indicate RNAs were isolated from the melbA melanoblast cell line and undifferentiated ES cells, respectively. GAPDH, glyceraldehyde-3-phosphate dehydrogenase.


In EDN3ls/ls mice, melanoblasts first normally migrate away from neural tube but remain in the dorsal area without proliferating or migrating until E14.5. Then, they become reactivated to proliferate and migrate to cover a significant part of the body; however, the middle to dorsal part of the trunk and head regions do not become pigmented (Yoshida et al., 1996). A reduced number of mature melanocytes differentiated from EDN3−/− ES cells in our presently described cultures, reflecting the above-mentioned white-spot phenotype of the EDN3ls/ls mice. Stage-specific expression of EDNRB detected by using a tetracycline-dependent transactivation system clearly revealed that there is critical EDNRB-dependent period between E10 and 12.5 during which neural crest cells become induced to develop into melanocytes (Shin et al., 1999). However, the presence or absence of EDN3 for a restricted period in the present EDN3−/− ES cell cultures (Fig. 3) did not demonstrate such an exclusively EDN3-dependent time point necessary for melanocyte differentiation, thus indicating that in vitro differentiation from ES cells proceeded not uniformly. Deletion of EDN3 signaling or KIT signaling did not eliminate the melanocyte precursors but just arrested or slowed down their proliferation, and EDN3 addition may have significantly restored the final number of mature melanocytes under the highly supportive in vitro culture conditions. Nevertheless, EDN3 addition or deletion at any point of the culture period affected the final number of melanocytes arising from EDN3−/− ES cells, suggesting that EDN3 signaling is necessary for the entire developmental stages of melanocytes. Using chicken or mouse neural crest cells and mouse skin melanocyte precursors, other investigators showed EDN3 to affect melanocyte development in vitro from the early phase of the development to the final maturation process (Lahav et al., 1996, 1998; Stone et al., 1997; Hirobe, 2001).

Lack of KIT signaling was compensated significantly by the addition of EDN3 (Fig. 4), confirming the findings of a previous investigation using neural crest cells isolated from the neural tube (Hou et al., 2000). Also, when EDN3 transgenic mice showing melanocytosis in their skin were crossed with KitW57 mutant mice, most of the white spots on the KitW57/KitW57 mice turned into pigmented skin, as shown in Figure 4C, indicating the compensation of KIT signaling by EDN3 signaling. However, as shown in Figure 5B, the addition of KITL did not rescue the generation of melanocytes from EDN3−/− ES cells in the presence of the EDNRB antagonist BQ788, indicating that EDNRB signaling was not compensated by KIT signaling. Still, it is also possible that BQ788 induced reduction of EDNRB signaling indirectly affected the KIT signaling not fully recovered by the addition of KITL. A synergistic effect of KIT and EDNRB signalings was demonstrated (Fig. 5B), in which 10 times more mature melanocytes were generated when both signalings were activated by the addition of KITL and EDN3 to the cultures.

In our experiments, we did not directly assess the number of melanocyte precursors; therefore, the final lack of mature melanocytes does not necessarily mean the elimination of the precursor cells but may suggest the suppression of the growth and differentiation of precursor cells. In fact, as shown in Figure 5C, even after a long 16-day absence of EDN3 signaling, we could detect quick emergence of mature melanocytes on day 21, indicating the survival of melanocyte precursors without EDN3 signaling. In accordance with this observation, Lee et al. (2003) reported the emergence and early dorsolateral migration of the melanoblasts in the dorsal trunk area in EDNRB-null (EdnrblacZ/EdnrblacZ) embryos and, thus, concluded that the initial specification, maintenance, and migration of neural crest precursors destined to become melanoblasts are EDNRB-independent (Lee et al., 2003). In KIT-null (KitW-LacZ/KitW-LacZ) embryos, the emergence and location of β-Gal–positive melanoblasts was indistinguishable as in KitW-LacZ/+ embryos; but by E11.5 to 12.5, the number of labeled cells close to the surface ectoderm was reduced compared with that for KitW-LacZ/+ embryos, indicating KIT signaling-independent specification, maintenance, and migration of melanoblasts at least in their initial developmental stage (Hou et al., 2000). Taking these observations into account, EDNRB and KIT environmental signalings individually act as a developmentally indispensable signaling only after the initial specification of the melanocyte cell lineage. This early specification process might proceed at least in the absence of either EDNRB or KIT signalings.

Finally, we found that the simultaneous suppression of EDNRB and KIT signalings had an unexpected effect on the development of melanocytes. After the elimination of both signalings from the cultures by adding KIT and EDNRB antagonists simultaneously to EDN3−/− ES cells, no mature melanocytes differentiated, even after a prolonged culture period of supplementation with KITL and EDN3. Neither EDN3 nor KIT signaling is thought to be involved in the specification and maintenance of neural crest stem cells, and it is likely that these two signalings are used for the early development of melanocyte lineage cells after the establishment of neural crest lineages. It could be that specification of melanocyte lineage cells from neural crest stem cells requires these two signals, but it is also possible that either EDNRB or KIT signaling is necessary for the maintenance or survival of early precursors of melanocytes and that simultaneous loss of both signalings may lead to the loss of these precursors. In other words, provided that one of these two signals is available, early melanocyte precursors are specified and survive for a certain period of time. The notion that two different signals work together to regulate the early stages of melanocyte development might be the important framework to further investigate the mechanisms governing the transition of neural crest stem cells to their immediate descendants.

Because KIT defects affect the sterility of the mice even in the heterozygous mutation, EDN3 and KIT double null mutant mice are quite difficult to establish; thus, the in vitro ES cell culture system used in our experiments worked practically in combination with various factors and their antagonists. To determine the precise action of EDNRB and KIT signalings during the early phase of melanocyte development, including specification from neural crest stem cells, we are aiming to isolate and purify neural crest cells supposedly induced in the present culture system and melanocyte precursors from these cultures. In addition, obtaining neural crest cells from EDN3 and KIT double null mutant mice will be required to confirm our conclusion in vivo.



D3 ES cell and endothelin 3−/− ES cell (Baynash et al., 1994) lines were maintained under feeder cell-free conditions. KitW-LacZ/KitW-LacZ, KitW-LacZ/+, and +/+ ES cells, established from the (KitW-LacZ/+ × KitW-LacZ/+)F1 embryos (Bernex et al., 1996), were maintained on mouse embryonic fibroblasts. All the ES cell cultures were supplemented with leukemia inhibitory factor, as previously described (Yamane et al., 1997). For the induction of melanocytes from undifferentiated ES cells, 1,000 to 2,000 undifferentiated ES cells were harvested by trypsinization and inoculated into six-well plates previously seeded with ST2 cells and allowed to differentiate in α-minimum essential medium (α-MEM, Invitrogen) supplemented with 10% calf serum in 5% CO2 at 37°C. The following factors were added; 10−7 mol/L Dex (Sigma), 20 pmol/L FGF2 (R&D Systems), 10 pmol/L CT (Sigma), and 100 ng/ml of EDN3 (Peptide Institute, Inc.) as reported (Yamane et al., 1999; Kunisada et al., 2003). The medium containing serum, Dex, FGF2, and CT was designated as “standard medium.” The medium was changed every 3 days. For some experiments, 100 ng/ml of IRL1620 (Peptide Institute, Inc.), 100 ng/ml of BQ788 (Peptide Institute, Inc.), 100 ng/ml of recombinant mouse KITL (SLF, PEPRO TEC), 10 μg/ml of anti-KIT monoclonal antibody (ACK2, Nishikawa et al., 1991) or 10 μg/ml of HGF (R&D Systems) was added, as indicated in the figures.

RT-PCR Analysis

Total RNA was purified from cultures at various times by using Isogen (Nippon Gene). After treatment with DNase I (Amersham Biosciences), first-strand cDNA synthesis was carried out by using Super Script reverse transcriptase (Invitrogen) primed with random hexamer in a 20-μl reaction mixture containing 1 μg of total RNA. A total of 1.5 μl of the first-strand cDNA mixture was used for PCR with Taq polymerase (Takara or Toyobo) performed in a 50-μl volume. Seven microliters of each PCR product was electrophoresed on an agarose gel and stained with ethidium bromide. Primers used for PCR were as follows: Endothelin1(EDN1), 5′TAG ATG TCA GTG CGC TCA CC/5′-ATG-CCT TGA TGC TAT TGC TG; Endothelin 3 (EDN3), 5′-AGA CTG TGC CCT ATG GAC TGT CCA A/5′-CGC ATC TCT TCT GCA GCT GGC CTT T; Endothelin receptor type A (ENDRA), 5′-TAC ACC CTC ATG ACC TGT GAG ATG C/5′-CAC TGG ATA CTC TGT TCC ATT CAC GG; Endothelin receptor type B (ENDRB), 5′-TAC AAG ACA GCC AAA GAT TGG TGG C/5′-AGC CAT GTT GAT ACC AAT GTA GTC C; MITF-M, 5′-GCT GGA AAT GCT AGA ATA CAG/5′-TTC CAG GCT GAT GAT GTC ATC. PCR conditions used were an initial denaturation step at 94°C for 2 min followed by 35 cycles of 94°C, 0.5 min; 55°C, 0.5 min; and 72°C, 1 min.

Mice for Genetic Study

Transgenic mice expressing mouse EDN3 in their skin keratinocytes from a construct driven by human cytokeratin 14 promoter (C57BL/6-Tg(hk14-EDN3); Kunisada et al., 2002) and C57BL/6-KitW-57J/KitW-57J were maintained and crossed with each other to obtain hk14-EDN3/+; KitW-57J/+ double mutant mice. These double mutant mice were then crossed to obtain hk14-EDN3/?; KitW-57J/KitW-57J mice.


We thank Dr. Shin-ichi Nishikawa for providing the hybridoma cell line for ACK2 and ST2 stromal cell line and Dr. Masashi Yanagisawa for EDN3−/− ES cells. We also thank Dr. Shin-ichi Hayashi for continuous help.