Regulation of melanoblast and retinal pigment epithelium development by Xenopus laevis Mitf


  • Mayuko Kumasaka,

    1. Department of Developmental Biology and Neurosciences, Graduate School of Life Sciences, Tohoku University, Sendai, Miyagi, Japan
    Current affiliation:
    1. Developmental Genetics of Melanocytes, UMR146 CNRS, Institute Curie, Bat 110 Centre Universitaire, 91405 Orsay, Cedex, France
    Search for more papers by this author
  • Shigeru Sato,

    1. Division of Biology, Center for Molecular Medicine, Jichi Medical School, Minamikawachi, Tochigi, Japan
    Search for more papers by this author
  • Ichiro Yajima,

    1. Department of Developmental Biology and Neurosciences, Graduate School of Life Sciences, Tohoku University, Sendai, Miyagi, Japan
    Current affiliation:
    1. Developmental Genetics of Melanocytes, UMR146 CNRS, Institute Curie, Bat 110 Centre Universitaire, 91405 Orsay, Cedex, France
    Search for more papers by this author
  • Colin R. Goding,

    1. Signaling and Development Laboratory, Marie Curie Research Institute, The Chart, Oxted, Surrey, United Kingdom
    Search for more papers by this author
  • Hiroaki Yamamoto

    Corresponding author
    1. Department of Developmental Biology and Neurosciences, Graduate School of Life Sciences, Tohoku University, Sendai, Miyagi, Japan
    • Department of Developmental Biology and Neurosciences, Graduate School of Life Sciences, Tohoku University, 6-3 Aramaki-aza-Aoba, Aobaku, Sendai 980-8578, Miyagi, Japan
    Search for more papers by this author


Mitf is a central regulator of pigment cell development that is essential for the normal development of the melanocyte and retinal pigment epithelium (RPE) lineages. To understand better the role of Mitf, we have used the Xenopus laevis experimental system to allow a rapid examination of the role of Mitf in vivo. Here, we report the function of XlMitfα-M on melanophore development and melanization compared with that of Slug that is expressed in neural crest cells. Overexpression of XlMitfα-M led to an increase in melanophores that was partly contributed by an increase in Slug-positive cells, indicating that XlMitfα-M is a key regulator of melanocyte/melanophore development and melanization. Moreover, overexpression of a dominant-negative form of XlMitfα led to a decrease in the number of melanophores and induced abnormal melanoblast migration. We also observed an induction of ectopic RPE and extended RPE by overexpression of XlMitfα-M and possible interactions between XlMitfα and several eye-related genes essential for normal eye development. Developmental Dynamics 234:523–534, 2005. © 2005 Wiley-Liss, Inc.


In vertebrates, melanocytes (melanophores in poikilotherms) are derived from multipotent neural crest cells. In contrast, the retinal pigment epithelium (RPE) is derived from monolayer cells lying at the outer layer of the optic cup. In some vertebrates, the epiphysis derived from the brain vesicle also contains pigment cells, which in humans does not produce melanin but rather works as an endocrine gland secreting melatonin. Previous studies have identified well over 100 genes related to the development or function of pigment cells (reviewed by Bennett and Lamoreux, 2003). Of these, the basic/helix–loop–helix/leucine zipper (bHLH-LZ) transcription factor Mitf (Hodgekinson et al., 1993; reviewed by Goding, 2000) is essential for the normal development of all pigment cells.

In mice, at least 17 Mitf mutant alleles have been identified and genetically characterized (Green, 1989; Tachibana et al., 1992; Steingrimsson et al., 1994). These mice exhibit reduced numbers of or complete lack of melanocytes, leading to a white coat and sometimes to deafness, and most also develop microphthalmia (small eye) due to the abnormal unpigmented RPE (Hodgkinson et al., 1993; Opdecamp et al., 1997; Nakayama et al., 1998). In addition, some alleles in the mouse Mitf gene affect the development of other cell types, including osteoclasts resulting in osteoporosis, and mast cells. When Mitf is expressed ectopically in fish, ectopic pigment cells and expression of melanogenic genes can be induced (Lister et al., 1999), whereas ectopic expression of Mitf in fibroblasts has been reported to lead to the adoption of a melanocyte-like phenotype (Tachibana et al., 1996). Mitf can regulate the transcription of the melanogenesis genes, such as Tyr, Tyrp1, Dct, and QNR-71 (Bentley et al., 1994; Yasumoto et al., 1994; Yavuzer et al., 1995), Kit (Tsujimura et al., 1996), Slug (Sanchez-Martin et al., 2002), and Tbx2 (Carreira et al., 1998, 2000), but also appears to regulate proliferation through both p16INK4a (Loercher et al., 2005) and p21Cip1 (Carreira et al., 2005) expression. Thus, Mitf is a key regulator of proliferation, survival, and differentiation of two different pigment cell derivatives, melanocytes/melanophores and the RPE.

Mitf expression is controlled by a combination of the Pax3, Sox10, CREB, and Lef1 transcription factors (reviewed by Goding, 2000) that together appear to control Mitf expression in the neural crest cells with the activation of Mitf expression leading to the commitment of a subset of neural crest cells to the melanocyte lineages. Expression of Mitf subsequently leads to regulation of genes important for the survival, proliferation, migration, and differentiation of melanocytes.

Intriguingly Mitf is expressed as multiple isoforms each with a distinct distribution. In humans and mice, multiple Mitf isoforms produced from the different promoters have been reported (Hallsson et al., 2000; Takeda et al., 2002; Watanabe et al., 2002) that differ at their N-termini but have the same DNA-binding and dimerization domains. Of these, the melanocyte-specific isoform Mitf-M is the best characterized and is indispensable for the development of neural crest-derived melanocytes but not for diencephalon-derived RPE (its developmental origin from the telencephalon has been suggested recently; Fernandez-Garre et al., 2002). Therefore, it was strongly suggested that Mitf isoform(s) other than Mitf-M should have a key role for RPE development (Yajima et al., 1999). Although it was reported that Mitf-A is predominantly expressed in the RPE, this mRNA is ubiquitous like the heart-type, Mitf-H (Amae et al., 1998; Udono et al., 2000). In addition, in humans, MITF-B, MITF-C, and MITF-D have been isolated (Yasumoto et al., 1998; Fuse et al., 1999; Takeda et al., 2002); and now MITF-J/Mitf-J have been reported in human and mouse (Hershey and Fisher, 2005).

Although much is known about the role of Mitf in activating pigmentation genes in cell lines in culture or the general role of Mitf in melanocyte development based, for example, on the genetics of Mitf in mice and humans, the limitations of these experimental systems have left many questions concerning Mitf function unresolved. Here, we have taken the first step to examine the in vivo function of a Xenopus laevis Mitf homologue that we have recently isolated (Kumasaka et al., 2004). In Xenopus, XlMitfα is expressed in melanophore lineages, the RPE, and epiphysis, although we could not identify isoform-specific expression (Kumasaka et al., 2004). The results presented here obtained by injection of Xenopus laevis fertilized eggs indicate that XlMitfα-M (an orthologue of mammalian Mitf-M) can control both melanophore and RPE development. The results highlight the usefulness of Xenopus as a model system for the analysis of genes involved in melanocyte development and function.


Overexpression of XlMitfα-M and dnXlMitfα Influences the Number of Melanophores and Their Dendricity

To gain an insight into the function of XlMitfα-M during development, we performed overexpression studies by injecting XlMitfα-M or dominant-negative XlMitfα (dnXlMitfα) mRNAs synthesized in vitro into one blastomere of two-cell stage embryos. A dnXlMitfα was generated by changing the highly conserved amino acid glutamic acid within the basic domain necessary for DNA binding to alanine (Fig. 1A). A similar strategy has been used to generate a dominant-negative form of other bHLH-LZ family members (Fisher and Goding, 1992; Krylov et al., 1997).

Figure 1.

Effects of XlMitfα-M and dominant-negative XlMitfα-M mRNA microinjection on the development of melanophores. A: Schematic representation of the wild-type XlMitfα-M and the dominant-negative form of XlMitfα (dnXlMitfα). In the dnXlMitfα construct, the amino acid glutamic acid in the basic region is converted to alanine. B: Control embryo. C: In XlMitfα-M-injected embryos, the number of melanophores was increased and the melanin granules in these cells dispersed much more extensively. D: Overexpression of dnXlMitfα induced a decrease in the number of melanophores, and a repression of melanin dispersal resulted in punctate-shaped melanophores. B–D: Stage 44, these images are the magnifications of the areas indicated in each inset therein.

In XlMitfα-M-injected embryos, the number of melanophores seemed to be increased and most melanophores were more dendritic compared with those in control embryos (76% of 106 embryos were affected, compare Fig. 1B with C). In contrast, dnXlMitfα-injected embryos, the number of melanophores seemed to be decreased and the melanophores exhibited a punctate shape compared with the normal dendritic form (71% of 42 embryos were affected, compare Fig. 1B with D). These results suggest that XlMitfα controls not only melanophore number but also their dendricity.

Overexpression of XlMitfα-M and dnXlMitfα Influences the Expression of Melanogenic Genes and Neural Crest Markers

We next analyzed the expression of a melanogenetic gene Dct (a marker for melanin-producing cells) and two neural crest marker genes (Sox10 and Slug) in XlMitfα-M- or dnXlMitfα-injected embryos. At stage 31, normal Dct expression is detected in melanoblasts located on the dorsal part of the neural tube in the trunk region and RPE in the eye (Fig. 2A-1, B-1, C). XlMitfα-M overexpression caused ectopic Dct expression (Fig. 2A-2, arrowheads) in and around the region where ectopic melanin production was observed (Fig. 2A-2, arrows). In contrast, on dnXlMitfα-injected side, such ectopic melanin production was never observed, although ectopic Dct expression in the trunk region was still detected (Fig. 2B-2, arrowheads). Notably in the eye, its expression was lost (Fig. 2B-2, arrow). To understand at what developmental stage XlMitfα-M caused an increase in the number of melanophores, we examined expression of Sox10 and Slug that are usually expressed in neural crest cells (Fig. 2F,I) and have been used as neural crest markers (Aoki et al., 2003; Mayor et al., 1995). In XlMitfα-M-overexpressed embryos, the number of Sox10-positive cells was increased along the dorsal midline on the injected side (Fig. 2D) and similar results were obtained for Slug expression (Fig. 2G-2). Of interest, expression of Slug was not detected in the ectopic melanized cells in contrast to the Dct expression in this region (compare Fig. 2A-2 with Fig. 2G-1). These results suggest that XlMitfα-M has the ability not only to increase Slug-positive melanoblasts but also induce ectopic melanization in Slug-negative non-neural crest cells. In dnXlMitfα-injected embryos, the Sox10 and Slug signals were weakened in places on the injected side and the expression was shifted slightly laterally (Fig. 2E, H-1, and H-2, black arrowheads; note that the Slug-positive area is shifted laterally).

Figure 2.

AN: Effects of overexpression of XlMitfα-M or dnXlMitfα on Dct, Sox10 and Slug expression. A-1, B-1 and C: At stage 31, Dct was normally expressed in melanoblasts on the dorsal side of the neural tube and in the RPE. A-2: On the XlMitfα-M-injected side, an increase in the number of Dct-positive melanophores (A-2, arrowheads) was observed ectopically covering the brown-colored melanized cells (A-2, arrows). B-2: On the dnXlMitfα-injected side, Dct-positive melanoblasts/melanophores were detected on between the dorsal side of the neural tube and the surface of the yolk sac in the trunk (B-2, arrowheads) contrasting with the noninjected side (B-1). An arrow indicates the eye expressing no signal (B-2). F,I: At stage 23, expression of Sox10 (F) and Slug (I) was detected in neural crest cells at the dorsal midline and in migrating cranial neural crest cells. D: In XlMitfα-M-overexpressed embryos, the number of Sox10-expressing premigratory neural crest cells was increased on the injected side (D, arrowheads). G-1, G-2: Similarly, Slug-expressing premigratory neural crest cells was increased by XlMitfα-M overexpression (G-2, arrowheads), although Slug expression was not detected in ectopic melanized cells (G-1, arrow) contrasted with Dct expression in these cells (A-2,E,H-1). H-2: In dnXlMitfα-injected embryos, the expression patterns of Sox10 (E, black arrowheads) and Slug (H-1 and H-2, black arrowheads) were disturbed and the signals shifted laterally contrasted with the noninjected side (H-1 and H-2, blue arrowheads). White arrows (E and H-1) and a black arrow (H-2) show the midline of each embryo. Green arrowheads (H-2) show the regions between the strong Slug-expressing regions (H-2, black arrowheads). J,K,L: BrdU immunostaining detecting proliferating cells. M, N: single-strand DNA (ssDNA) immunostaining detecting apoptotic cells. These stainings were performed in transverse sections prepared from the same embryos (G-2 and H-2) after in situ hybridization. J: The number of BrdU-positive cells (shown by red cells) on the injected side was increased (black arrowheads, contrasting with those in the noninjected side pointed by blue arrowheads) in the region where Slug expression was promoted. In the Slug-expressing region shifted laterally on the dnXlMitfα-injected side (H-2, black arrowheads), both the number of BrdU-positive cells (K, black arrowheads) and ssDNA-positive cells (M, black arrowheads) were larger than those in Slug-expressing area on the noninjected side (K and M, blue arrowheads). In the region where Slug signal became very weak on the injected side (H-2, green arrowheads), neither a decrease in the number of BrdU-positive cells (L, green arrowheads) nor an increase in the number of ssDNA-positive cells (N, green arrowheads) were observed, contrasted with Slug-expressing regions on the noninjected side (L and N, blue arrowheads). C,F,I: Noninjected embryos. A-1,A-2,B-1,B-2,C: Stage 32. D,E,F,H-1,I: Stage 23. G-1: Stage 21. G-2,H-2: Stage 25. NO, notochord; NT, neural tube.

To examine the effects of overexpression of these two mRNAs on the proliferation of melanoblasts, we performed a bromodeoxyuridine (BrdU) incorporation assay. In this assay, we used the same embryos in Figure 2G-2 and 2H-2, with sections being prepared from the region indicated by the white lines. In the region where Slug was activated by overexpression of XlMitfα-M, the number of BrdU-positive cells was also increased (Fig. 2G-2 and J, black arrowheads), although it was not possible to determine whether the BrdU-positive cells expressed Slug. As for dnXlMitfα-injected embryos, we prepared transverse sections from two different regions (Fig. 2H-2, two white lines). In transverse sections where strong Slug expression was observed (Fig. 2H-2 black arrowheads), the number of BrdU-positive proliferating cells was increased (Fig. 2K, black arrowheads) compared with the Slug-expressing regions on the noninjected side (Fig. 2H-2 and K, blue arrowheads). In contrast, in the regions where the Slug signal was substantially weakened (Fig. 2H-2, green arrowheads), the number of BrdU-positive cells (Fig. 2L, green arrowheads) did not seem to be influenced compared with Slug-expressing region on the noninjected side (Fig. 2H-2 and L, blue arrowheads). In addition, to determine whether this reduction in Slug expression was related to the activation of apoptosis or to any abnormal migration ability of Slug-positive cells, we detected apoptotic cells by anti-single strand DNA antibody. Based on this assay, we could not detect any activation of apoptosis in the areas where Slug expression was reduced (Fig. 2N, compare the region pointed by blue arrowheads with that indicated by green arrowheads). However, in the regions where Slug expression was obvious (Fig. 2H-2, black arrowheads), many apoptotic cells were detected (Fig. 2M, black arrowheads). This result raised the possibility that the appearance of the regions expressing reduced levels of Slug was not related to the activation of apoptosis but to the abnormal migration of neural crest cells, including melanoblasts resulting in decreased numbers of melanophores present at later developmental stages.

Overexpression of Slug Increases the Number of XlMitfα-Positive Cells Resulting in Increase in the Number of Melanophores

Forced expression of XlMitfα-M induced an increase in the number of Slug-positive cells resulting in an increase in melanophores. It has been reported that Slug has the ability to enhance its own expression (LaBonne and Bronner-Fraser, 1998). Overexpression of Slug induces an increase in the number of melanophores as shown in Fig. 3B (compare Fig. 3A with B). However, it is unclear whether the increase in melanophores induced by Slug overexpression is caused by a direct induction of ectopic XlMitfα expression or by an indirect increase in XlMitfα-positive cells after induction of Slug-positive neural crest cells by Slug itself. We therefore investigated the influence of overexpression of Slug on XlMitfα expression. We found that, in Slug-injected embryos, the number of XlMitfα-expressing neural crest cells increased dramatically on the injected side (Fig. 3C,D, a yellow arrow indicates the midline); on the other hand, neither ectopic XlMitfα-positive cells nor ectopic melanization that was observed in XlMitfα-M–injected embryos was detected. Similarly, Sox10 was also up-regulated by Slug overexpression (Fig. 3E). These results suggest that overexpression of Slug does not induce ectopic XlMitfα- and/or Sox10-positive cells and that the increase in melanophores in Slug-overexpressed embryos was the result of the increase in neural crest-derived XlMitfα-positive cells.

Figure 3.

Effects of Slug overexpression on melanophore development. A,B: In Slug-injected embryos, the number of melanophores was increased (B) compared with noninjected embryos (A). C,D: Overexpression of Slug induced an increase in the number of XlMitfα-positive cells on the injected side (C,D, arrowheads). D: A transverse section of the trunk region of C, plane D. E: The number of Sox10-positive cells was also increased on the injected side (E, arrowheads). Yellow arrows in C and E show the midline. NT, neural tube.

XlMitfα-M Induces Slug, But It Does Not Result in the Induction of Neuronal Marker Genes

To infer the differences between the molecular mechanism(s) underlying an increase in melanophores caused by overexpression of XlMitfα-M and Slug, we performed reverse transcriptase-polymerase chain reaction (RT-PCR) analysis using animal cap explants. In XlMitfα-M–overexpressed animal cap explants, RT-PCR analysis using EF1α as a control, revealed that Slug was already slightly induced at embryonic stage 12 (Fig. 4A). At embryonic stage 17, Sox10 (a marker for some neural crest lineages including melanophores), Tyr and Tyrp1 (melanin-producing cell markers at later developmental stages compared with Dct), and Dct (a melanin-producing cell marker) were also induced, whereas NeuroD and N-tubulin (neuronal marker genes), both of which were inducible by Slug (Fig. 4C; also reported by LaBonne and Bronner-Fraser, 1998), were not (Fig. 4B). These results suggest that XlMitfα-M has an ability to induce Slug expression, and in the cell population of these Slug-positive cells, melanoblast precursors were included but neuronal precursor cells were not.

Figure 4.

Reverse transcriptase-polymerase chain reaction (RT-PCR) analyses using animal cap explants after injection of XlMitfα-M or Slug mRNAs. A,B: XlMitfα-M–injected animal cap explants (AC). A: Stage 12. B: Stage 17. XlMitfα-M injection induced neural crest markers (Slug and Sox10) and melanogenic genes (Tyr, Tyrp1, and Dct) but did not induce neuronal markers (NeuroD and N-tubulin) and Slug was induced earlier than the others. C,D: Slug-injected animal cap explants. C: Stage 17. D: Stage 22/23. C: In Slug-injected animal cap explants, Sox10, N-tubulin, and NeuroD were induced at the time corresponding to stage 17 of embryos. D: In contrast, XlMitfα expression was not, but by the time corresponding to the stage 22/23, it became detectable. Dct expression was not detected in Slug-injected animal caps, although that was induced in XlMitfα-M–injected animal cap explants (compare B and D). EF1α was amplified to show the cDNA used were not degraded.

In Slug-injected animal cap explants, Sox10 and N-tubulin and NeuroD were induced at the time corresponding to embryonic stage 17, when these genes were normally detected, but in contrast, XlMitfα was not induced at this stage (Fig. 4C). However, at the time corresponding to stage 22/23 when XlMitfα expression was detectable in vivo, it became detectable in Slug-injected animal caps (Fig. 4D). These results suggest that forced expression of Slug induces XlMitfα and/or Sox10-positive cells but does not have the ability to hasten the onset of expression of these genes in contrast to induction of Slug expression by the overexpression of XlMitfα-M.

Overexpression of XlMitfα-M and dnXlMitfα Influences Eye Development and Expressions of Eye Marker Genes

At later developmental stages, we found that some embryos (16% of 106 embryos were affected) injected with XlMitfα-M also displayed several eye-related defects (Fig. 5B–E). In the majority of embryos having such defects (11 of the 17 affected embryos), the optic fissure failed to close properly, leaving a gap in structures of the eye such as the RPE (Fig. 5B) compared with the control side (Fig. 5A). The phenotype observed most frequently was that ectopic RPE-like structures were formed in the brain and along the optic stalk in 15 of the 17 affected embryos (Fig. 5C). Most embryos showed combinations of these two defects. A transverse section of the head of this phenotype revealed that the abnormal RPE extends along the optic stalk toward the ventral part of the brain (Fig. 5E), and this phenomenon was observed in almost all the embryos with an open optic fissure. In some XlMitfα-M–injected embryos, the eye size was reduced (in 7 of the 17 affected embryos; Fig. 5D).

Figure 5.

Effects of XlMitfα-M and dnXlMitfα mRNA microinjection on eye development. A: Control. B–E: XlMitfα-M-injected embryos. B: The optic fissure did not close (arrow). C: Ectopic retinal pigment epithelium (RPE, arrows) and RPE extension (arrowheads) were observed on the XlMitfα-M-injected side. D: The size of the eye on the injected side was small (arrow). E: A section of XlMitfα-M–injected embryos with an open optic fissure shows RPE extension toward the ventral part of the brain vesicle (arrow). FI: dnXlMitfα-injected embryos. In dnXlMitfα-injected embryos, a lack of ventral RPE (F, black arrow), a reduction of eye size and a defect of optic stalk (G,H, black arrows) were observed. In severe cases, the eye became very small and fused to the brain (H, black arrow). In addition, in some embryos, the phenotype with a lack of almost all the RPE was observed on the injected side (I, black arrow). A–C: Stage 45. D,E: Stage 40. F–I: Stage 44. BR, brain vesicle.

Also in dnXlMitfα-injected embryos, the phenotypes of embryos with eye malformations were observed (74% of 42 embryos were affected). Most of them (28 of the 31 affected embryos) showed a combination of lack of the ventral part of the RPE (Fig. 5F) and small eye (Fig. 5G), and an abnormal distance between the brain and the eye on the injected side (Fig. 5G). In severe cases, the size of the eye became very small (microphthalmia) and the eye fused to the brain (Fig. 5H). In some embryos, the RPE was completely absent (4 of the 31 affected embryos, Fig. 5I). Overexpression of dnXlMitfα induced similar phenotype with that induced by XlMitfα-M such as partial lack of the RPE and small eye but the percentage of these phenotypes were higher than those induced by overexpression of XlMitfα-M. Dominant-negative XlMitfα never induced ectopic RPE or RPE extension, and the reduction of the distance between the brain and eye caused by the lack of the optic stalk was specific in the dnXlMitfα-injected embryos.

To understand better the cause of these eye-related abnormalities, we studied the expression of several eye-related genes. Both Pax6 and Pax2 encode paired-box class transcription factors and are essential for normal eye development. Pax6 in particular is a key regulator of eye development and is expressed in the entire optic vesicle. In contrast, Pax2 is expressed in the ventral part of the optic cup and in the optic stalk (Heller and Brandli, 1997), although in mice, it is expressed in the entire optic vesicle like Pax6 (Nornes et al., 1990; Baumer et al., 2003). In XlMitfα-M–injected embryos, the eye field normally defined by Pax6 expression was reduced on the injected side of several embryos (Fig. 6A), which might account for the reduced eye size. In addition, expression of other eye-related genes, Rx1 (a paired-like homeobox gene, Fig. 6B), homeodomain-containing genes such as Six6 (Fig. 6C), and Otx5b (a bicoid-class homeobox gene, Fig. 6D) were reduced similarly. In contrast, Pax2 expression in the optic vesicle on the XlMitfα-M-injected side was up-regulated, although its expression in the otic vesicle and midbrain–hindbrain boundary was reduced (Fig. 6E). Otx5b is expressed not only in the optic vesicle but also the epiphysis in normal embryos, and expression in both regions was repressed after XlMitfα-M injection (Fig. 6D).

Figure 6.

Effects of XlMitfM and dnXlMitfα mRNA microinjection on expression of eye-related genes. A–E: XlMitfα-M–injected embryos. A: The facial region expressing Pax6 on the injected side (black arrow) was smaller than that on the noninjected side (blue arrow). Black dashes show the midline, and the eye region where Pax6 is normally expressed is shown by white dashes. B,C: Similarly, Rx1 (B) and Six6 (C) were repressed on the injected side (black arrow). D: Expression of Otx5b was repressed both in the optic vesicle (black arrow) and the pineal body (white dashes) on the injected side. Blue arrow shows normal Otx5b expression in the optic vesicle on the noninjected side, and black dashes show the midline. E: In contrast, Pax2 expression was activated in the optic vesicle (black arrow), although those in the otic vesicle (green dashes) and midbrain–hindbrain boundary (pink dashes) were repressed on the injected side compared with those on the noninjected side (blue, green, and pink arrows, respectively). FJ: dnXlMitfα-injected embryos; overexpression of dnXlMitfα induced the reduction of Pax6 (F), Rx1 (G), Six6 (H), Otx5b (I), and Pax2 (J) in the optic vesicle on the injected side (black arrows). Pax2 expression (J) in the otic vesicle (green dashes) and the midbrain–hindbrain boundary (pink dashes) and Otx5b expression in the pineal body (I, white dashes) were also repressed on the injected side. A–C,E–H,J: Stage 23. D,I: Stage 25. CG, cement gland; MH, midbrain–hindbrain boundary; OT, otic vesicle; PB, pineal body.

In dnXlMitfα-overexpressed embryos, expression of all genes examined (Pax6, Rx1, Six6, Otx5b, and Pax2) was also repressed in the optic vesicle (Fig. 6F–J) and Otx5b expression in the pineal organ was also down-regulated on the injected side (Fig. 6I). These results were similar to those obtained from XlMitfα-M–overexpressed embryos, but expression of Pax2 in the eye was regulated to the contrary (compare Fig. 6E with J).


In this study, we have explored the use of Xenopus as a system to analyze the role of factors required for melanocyte development. Overexpression of XlMitfα-M or dominant-negative XlMitfα led to an altered number and dendricity of melanophores, effects on eye development, and the expression patterns of several marker genes.

Function of XlMitfα on Melanophore Development and Melanogenesis

While overexpression of XlMitfα-M induced both ectopic melanization and an increase in the number of melanophores, in contrast, Sox10-injected Xenopus embryos exhibited an increase in the number of Dct-positive cells and melanophores but did not produce ectopic melanin (Aoki et al., 2003). In mice, Sox10 binds and activates Mitf promoter (Bondurand et al., 2000; Lee et al., 2000; Potterf et al., 2000; Jiao et al., 2004) and it is likely that a failure to correctly activate the Mitf promoter accounts for the pigmentation defects observed in Dominant megacolon (Dom) mice expressing a truncated version of Sox10. If a similar relationship occurs in Xenopus, it is likely that Sox10-mediated activation of Mitf may be responsible for the elevated numbers of melanophores observed in Sox10-injected embryos. The inability of Sox10, unlike Mitf, to target pigmentation genes such as Tyr and Tyrp1 would also account for the fact that Mitf, but not Sox10, can induce ectopic melanization. This interpretation is supported by our in vitro analyses using animal cap explants, where in XlMitfα-M–injected animal caps, melanogenic genes such as Tyr, Tyrp1, and Dct were induced before the initiation of normal Dct expression in embryos, whereas in Sox10-injected animal caps (Aoki et al., 2003), Dct expression was not induced until the corresponding stage (stage 25) when it initiates in embryos.

XlMitfα Mediates Increases in the Number of Melanophores

Overexpression of XlMitfα-M induced an increase in melanophores, whereas mutations in Mitf in other organisms led to a loss of melanocytes/melanophores (Steingrimsson et al., 1994; Lister et al., 1999; Yajima et al., 1999). Although it is clear that Mitf is essential for melanocyte/melanophore development, it is not yet evident precisely which downstream genes mediate its effect. One candidate Mitf target gene is Slug, which is a member of the Snail family of zinc-finger transcription factors that has an evolutionarily conserved function in mesoderm development and neural crest formation in vertebrates (LaBonne and Bronner-Fraser, 1998, 2000; Sefton et al., 1998; Del Barrio and Nieto, 2002). Slug is a useful marker for premigratory neural crest cells (Nieto et al., 1994; Mayor et al., 1995), and Slug overexpression causes up-regulation of other neural crest marker genes and Slug expression itself (LaBonne and Bronner-Fraser, 1998).

Significantly, overexpression of XlMitfα-M induced Slug in vivo and in vitro, strongly suggesting that Slug may play a role in the Mitf-mediated increase in melanophores. Recently, it was reported that Waardenburg syndrome type 2 (WS2), which shows pigmentary abnormalities and sensorineural deafness caused by abnormalities of MITF expression or function (Hughes et al., 1994; Tachibana et al., 1994; Tassabehji et al., 1994, 1995; Goding, 2000; and references therein), is also caused by deletions of human SLUG (Sanchez-Martin et al., 2002) and MITF that can bind the human SLUG promoter and up-regulate its expression in vitro (Sanchez-Martin et al., 2002). These reports complement our observations in the Xenopus system and suggest that Mitf-mediated activation of Slug may be an evolutionarily conserved event important for melanocyte/melanophore development. In agreement with this explanation, overexpression of Slug in Xenopus can induce an increase in the number of melanophores (LaBonne and Bronner-Fraser, 1998) and inhibition of Slug leads to reduced melanophore formation (Carl et al., 1999). Our results also indicate that forced Slug expression can induce an increase in XlMitfα-positive cells, raising the possibility that, in addition to being a downstream target of Mitf, Slug may also act upstream, either by directly activating the Mitf promoter or indirectly by increasing the neural crest population from which melanophores originate. Thus, among Slug-positive neural crest cells, some cells become XlMitfα-positive and are committed to being melanoblasts, and once committed, these XlMitfα-positive cells up-regulate Slug expression and as a result activate the proliferation of Slug/XlMitfα double-positive committed melanoblasts.

Decreased Melanophores Numbers

Depending on the severity of the allele, mice with Mitf mutations show a partial or complete lack of melanocytes, resulting in an entirely white coat or a white spotted coat, respectively (Steingrimsson et al., 2003). Moreover, in one of the zebrafish mutants nacre that has a mutation in Mitfa corresponding to Mitf-M in other vertebrates, only melanophores are absent, although xanthophores and iridophores, both of which are also neural crest-derived pigment cells in poikilotherms, exist (Lister et al., 1999). From these reports, Mitf appears to be crucial for normal melanocyte/melanophore development, although the mechanisms underlying the decrease or absence of melanocytes/melanophores remain unknown. Our results using dnXlMitfα may provide a hint as to the mechanisms involved.

In dnXlMitfα-injected embryos, a decrease in the number of melanophores was observed and the expression pattern of Dct was disturbed, whereas the numbers of Dct-positive cells appearing to be increased as in XlMitfα-M–injected embryos. One possible explanation is based on the observation that, in dnXlMitfα-injected embryos, Sox10 and Slug expression is shifted laterally preceding Dct expression and that, as a result, Dct-positive cells abnormally developed in this position instead of being restricted to their normal location.

Moreover, in dnXlMitfα-injected embryos, the pattern of Slug and Sox10 expression was disturbed with areas of weaker signals appearing between regions with intense expression. It was possible, therefore, that overexpression of dnXlMitfα caused apoptosis, repression of proliferation, and/or abnormal migration of neural crest cells. However, in the region where the Slug signal was weaker, we observed no significant down-regulation of proliferation or activation of apoptosis compared with the Slug-expressing region on the noninjected side. On the other hand, at the region where Slug expression was clearly seen, the number of BrdU-positive cells and apoptotic cells was larger than those on the noninjected side. These results may suggest that the weakening of the Slug signal arises from the abnormal migration along the anteroposterior axis rather than by the activation of apoptosis and the repression of proliferation of neural crest cells. In light of this possibility, the increase in BrdU-positive cells and apoptotic cells in Slug-expressing regions on the dnXlMitfα-injected side may originate from the aggregation of Slug-positive cells and neural crest cells with an abnormal migration ability may die later, leading to a decrease in the number of melanophores at later developmental stages. This possibility remains to be elucidated.

XlMitfα Controls Dendricity of Melanophores

In addition to the role of XlMitfα in regulating melanophore numbers, our results also suggest a possible role for XlMitfα in melanophore dendricity. Thus, overexpression of dnXlMitfα led the surviving melanophores to adopt a punctate phenotype that appeared to reflect as change in cell shape rather than simply a redistribution of melanophores, while expression of normal XlMitfα appeared to increase dendricity. This result is consistent with observations made using ectopic expression of mammalian Mitf-M in cultured cells, which appears to increase dendricity (Tachibana et al., 1996; Carreira et al., 2005). However, although we believe that Mitf is likely to regulate genes intimately connected to regulation of the cytoskeleton, we cannot rule out additional effects on regulation of melanin dispersal. Indeed the processes are closely connected because melanin dispersal cannot occur without a compatible cytoskeleton organization.

XlMitfα Is Essential for Normal Eye Development

In the assays used here, the XlMitfα used appears to be an orthologue of the mammalian Mitf-M that is a neural crest-specific Mitf isoform. Previously, we reported that XlMitfα is expressed in melanophore lineages, as well as the RPE and epiphysis (Kumasaka et al., 2004). However, we were unable to identify any isoform-specific expression, and it remains unclear whether any isoform-specific orthologue of Mitf-M is specifically expressed only in the neural crest-derived melanophore lineage as in mammals. In most of the XlMitfα-M–overexpressed embryos, ectopic RPE and RPE extensions were observed. Thus, two explanations are possible for the phenotype induced in the eye: (1) the XlMitfα-M used could function in RPE development as an alternative for other XlMitfα isoforms that are normally specifically involved in RPE development, or (2) XlMitfα-M is normally expressed in the RPE and forced overexpression of it induces multiple eye malformations.

Among XlMitfα-overexpressed embryos, a small number of embryos exhibited a small eye phenotype and a reduction of Pax6 expression. The small eye phenotype could arise either as a result of the capacity of XlMitfα to inhibit cellular proliferation, as suggested from results obtained using mammalian Mitf-M (Carreira et al., 2005; Loercher et al., 2005) or alternatively, but not mutually exclusively, by means of the suppression of Pax6 expression or activity by XlMitfα overexpression. Several lines of evidence tend to support the latter interpretation. First, Xenopus embryos injected with a dominant-negative form of Pax6 showed either complete loss of eye structure or a reduction in eye size (Chow et al., 1999); second, infection of chicken cultured RPE cells with a retrovirus-expressing Mitf, inhibited Pax6 expression (Mochii et al., 1998); and in quail, cultured RPE cells, the DNA-binding domains of Pax6 interact with the bHLH-LZ domain of Mitf and repress their respective target promoters (Planque et al., 2001).

A small eye phenotype was also observed by overexpression of dnXlMitfα, a result similar to that observed in mice bearing Mitf mutations (Green, 1989; Tachibana et al., 1992; Steingrimsson et al., 1994). Loss of XlMitfα function in the eye at early developmental stages could induce morphological eye malformation arising from repression of Pax6. In addition, the dnXlMitfα used in this study could potentially also interact with Pax6 protein and repress its activity in the same manner of XlMitfα-M overexpression. At present, we cannot discriminate between these possibilities.

The expression of other eye-related genes we examined (Rx1, Six6, and Otx5b) were similarly repressed by forced expression of both XlMitfα-M and dnXlMitfα. The mechanism underlying the repression of this set of genes remains unknown, although it is possible that they are indirectly regulated by Pax6 and consequently any inhibition of Pax6 function or expression by XlMitfα-M and dnXlMitfα would affect their expression.

In contrast to Pax6, Rx1, Six6, and Otx5b expression, overexpression of XlMitfα-M and dnXlMitfα had different effects on Pax2 expression. In XlMitfα-M-injected embryos, expression of Pax2 at the ventral part of the eye vesicle was up-regulated, whereas forced expression of dnXlMitfα down-regulated the expression of this gene. Thus, Pax2 may be a candidate target gene for XlMitfα in the optic vesicle. Consistent with this possibility, the ectopic RPE and RPE extensions induced by XlMitfα-M overexpression were observed along the optic stalk in which Pax2 was strongly expressed. When XlMitfα is overexpressed in this area, reciprocal activation of XlMitfα and Pax2 may occur. Leading to ectopic RPE and RPE extension. It is also possible that the interactions between Pax2 and Pax6 are one of the causes of abnormal Pax2 expression in XlMitfα up-regulated and down-regulated embryos. Schwarz et al. (2000) postulated a model for the reciprocal inhibition between Pax2 and Pax6 at the border of the optic stalk and optic cup for their spatial specification. In XlMitfα-M–overexpressed embryos, up-regulation of Pax2 expression appears to be caused by both down-regulation of Pax6 and up-regulation of XlMitfα. However, in contrast, in dnXlMitfα-overexpressed embryos, Pax6 is down-regulated but the up-regulation of Pax2 has not been observed. These results suggest that Pax2 expression could not be activated without XlMitfα even if Pax6 is down-regulated.


In summary, we have demonstrated that the Xenopus system provides a rapid and efficient system for the analysis of genes involved in melanocyte and RPE development. This study provides new information for several ambiguous points about Mitf function in pigment cell development in vivo and in vitro. From XlMitfα-M overexpression analyses, it is clear that XlMitfα-M contributes to an increase in the number of melanophores by increasing the number of Slug-positive and Sox10-positive cells in vivo and this finding is supported by the in vitro RT-PCR analysis using animal cap explants. Furthermore, XlMitfα-M has the ability to induce melanization in non-neural crest cells. Forced expression of dnXlMitfα suggests that XlMitfα is also necessary for the melanoblast migration. In addition, overexpression of Slug shows that Slug contributes to the specification and proliferation of neural crest cells. As for eye development, the results indicate that XlMitfα has the ability to induce RPE and that the interactions between XlMitfα and other eye-related genes are indispensable for the normal eye development. The results provide an insight into the Mitf-dependent mechanisms underlying melanocyte/melanophore development, neural crest differentiation, and eye development.


Embryos and Microinjection

Xenopus laevis adults purchased from COPACETIC Co., Ltd. (Aomori, Japan) and HSK Co., Ltd. (Shizuoka, Japan) were maintained in our laboratory at 23°C. Fertilized eggs were obtained from adults injected with 250 U of human chorionic gonadotropin to induce egg laying. Embryos were dejellied in 2% cysteine (pH 7.5) and were washed twice in 1× MMR (100 mM NaCl, 2 mM KCl, 1 mM MgCl2, 2 mM CaCl2, 0.1 mM ethylenediaminetetraacetic acid, 5 mM HEPES, pH 7.5). RNAs (XlMitfα-M, dnXlMitfα and Slug; 10 nl at 5 ng/μl) were injected into one blastomere at the two-cell stage in 1× MMR+3% Ficoll and were maintained for 5 hr at 16°C. Embryos were transferred to 0.1× MMR+1% Ficoll and were maintained overnight at 16°C. The following day, embryos were transferred to 0.1× MMR and maintained until fixation. Staging was carried out according to Nieuwkoop and Faber (1967). Under the conditions we used, injection of lower concentration of XlMitfα-M or dnXlMitf mRNA (10 nl at 1 ng/μl) had no effect.

Plasmid Construction and In Vitro RNA Synthesis

To generate pCS2 XlMitfα-M, positions 1 to 852 and 640 to 1176 of the 1176-bp XlMitfα-M cDNA were PCR amplified using primers XlMi-forward1 (5′-cgggatccctcgagATGCTGGAAATGCTTGACTATAACC-3′, flanked with BamHI and XhoI sites) and XlMi-reverse1 [5′-GTCGAGCAGAGGCCAGTAGAAGG-3′], XlMi-forward2 (5′-atatgcggccgcAAAAGCTTCGGTGGATTACATTCGC-3′, flanked with a NotI site), and XlMi-reverse2 [5′-ctacttgtcatcgtcgtccttgtagtcACAGTGATCATTTTCTTCCAT-3′, sequences for FLAG epitope tag underlined], respectively. The amplified fragments were subcloned using the TA Cloning Kit, excised by digesting with XhoI/SphI (position 758) and SphI (position 758)/XbaI, respectively, and then cloned into the XhoI/XbaI sites of the pCS2.

Dominant-negative construct of XlMitfα is converted LIER to LIAR in the basic domain (Fig. 1A). To prepare XlMitfα dominant-negative construct, we first used PCR with primers XlMi-forward3 (5′-CAATTGGTTCAACCTGTGAAAAAGAG-3′, corresponding to TIGSTCEKE) and dXlMi-reverse1 [5′-CTTCGCCTGCGCGCAATGAGG-3′, complementary to LIARRR and flanked with a BssHII site], the conditions for PCR were as follows: 94°C for 5 min, 30 cycles of 94°C for 1 min, 58°C for 1 min, and 72°C for 1 min, then 72°C for 10 min. dXlMi-forward1 (5′-CCTCATTGCGCGCAGGCGAAG-3′, corresponding to LIARRR and flanked with a BssHII site) and XlMitf-reverse3 [5′-GTCGAGCAGAGGCCGGTAGAAG-3′, complementary to STGLCS], the conditions for PCR were the same as those mentioned above. Both fragments were cloned using the TA Cloning System (Invitrogen) and sequenced. The amplified fragments were subcloned using the TA Cloning Kit (Invitrogen), excised by digesting with BglII/BssHII and BssHII/SphI, respectively, and then cloned into the BglII/SphI sites of the pCS2 XlMitfα-M. To generate pCS2 Slug, full-length Xenopus Slug cDNA was PCR amplified using primers Slug-forward1 (5′-ATGCCCCGGTCATTTCTGGTCAAG-3′, corresponding to MPRSFLVK) and Slug-reverse1 (5′-ctcgagctacttgtcatcgtcgtccttgtagtc ATGTGCTACACAGCAACCAGA-3′, complementary to SGCCVAH, flanked with a XhoI site and the sequences for FLAG epitope tag underlined). The amplified fragment was subcloned using pGEM-T Easy Vector System I (Promega) and sequenced, and excised with EcoRI/XhoI and then cloned into EcoRI/XhoI sites of the pCS2 vector. Capped XlMitfα-M, dominant-negative XlMitfα, and Slug mRNA were synthesized with SP6 RNA polymerase using templates linearized with NotI using mMESSAGE mMACHINE High Yield Capped RNA Transcription Kit (Ambion). β-galactosidase mRNA coinjected was synthesized from pCS2+nβ-gal with SP6 RNA polymerase using template linearized with NotI in the same way.

Red-Gal Staining and Whole-Mount In Situ Hybridization

Embryos were coinjected with β-galactosidase mRNA (β-gal, 200 pg/embryo) and were fixed in MEMFA. Embryos were rinsed several times in phosphate buffered saline (PBS) after fixation. β-Galactosidase staining was carried out with 0.1 mM potassium ferricyanide, 0.1 mM potassium ferrocyanide, and 0.4% red-gal in PBS at room temperature. After staining, embryos were rinsed in PBS and dehydrated with 25%, 50%, 75% ethanol/PBT (PBS, 0.1% Tween-20), and 100% ethanol at room temperature, and then stored in 100% ethanol until applying whole-mount in situ hybridization. Whole-mount in situ hybridization was carried out as described previously (Kumasaka et al., 2003).

For RNA probes, Slug, Sox10, Pax6, Pax2, Rx1, and Six6 were isolated by RT-PCR. Primers designed for this work were: Slug-forward1 (5′-ATGCCCCGGTCATTTCTGGTCAAG-3′, corresponding to MPRSFLVK) and Slug-reverse3 [5′-GC(GA)CCCAAGCT (CG) AC (AG) TACTCC-3′, complementary to KEYVSLGA], Sox10-foward1 (5′-TCAGGTCAAAGTCATGGACCCCC-3′, corresponding to SGQSHGP) and Sox10-reverse1 [5′-GAGATGGAGGGAAATGCTGAACC-3′, complementary to GSAFLSIS], Pax6-forward (5′-AACCTGGCGAGCGAGAAGCAGCAGA-3′, corresponding to NLASEKQQ) and Pax6-reverse [5′-TTCTTTTTCTAGCGCCTCTATTTG-3′ complementary to QIEALEKE], Pax2-forward (5′-TCTGATGGTTCTGGTCCAAATGG-3′, corresponding to SDGSGPNG) and Pax2-reverse[5′-GTGGCGGTCATAGGCAGTG-3′, complementary to ATAYDRH], Rx1-forward (5′-AGAAGAAACACAGAAGGAACCGG-3′, corresponding to KKKHRRNR) and Rx1-reverse [5′-CCAAGGCTTGCCAATAAACTGGAT-3′, complementary to IQFIGKPW], Six6-forward (5′-ATGTTTCAGCTGCCTATTCTGAAC-3′, corresponding to MFQLPILN) and Six6-reverse [5′-TCAGATGTCACATTCACTGTCGC-3′, complementary to DSECDI]. The conditions for these RT-PCR were as follows: 94°C for 5 min; 40 cycles of 94°C for 1 min, 56°C for 1 min, 72°C for 3 min; then 72°C for 10 min. All these fragments were cloned using the TA Cloning System (Invitrogen; Slug, Sox10, and Pax6) or the pGEM-T Easy Vector System I (Promega; Pax2, Rx1, and Six6) and were verified by sequencing. Digoxigenin-labeled antisense RNA probes for Dct, XlMitfα and Otx5b were described previously (Kumasaka et al., 2003) and for Slug, Sox10, Pax6, Pax2, Rx1, and Six6 were synthesized with SP6 (Slug, Sox10, Pax6, Pax2, and Six6), or T7 (Rx1) RNA polymerases using templates linearized with NotI (Pax6 and Six6), SphI (Pax2), XhoI (Slug and Sox10), or PstI (Rx1) using digoxigenin RNA labeling mix (Roche Diagnostics).

Animal Cap Assay and RT-PCR Analysis

Dejellied embryos were injected with XlMitfα-M in the animal pole at the two-cell stage and incubated in 1×MMR+3% Ficoll until at stage 8.5. After the removal of vitelline membranes, animal caps were cut out and cultured in 1× LCMR (66 mM NaCl, 1.33 mM KCl, 0.33 mM CaCl2, 0.17 mM MgCl2, 5 mM HEPES, pH 7.2) supplemented with 0.1% BSA and were then harvested at the stages indicated in the figure legends. At each stage, 10 animal cap explants were harvested and RT-PCR performed. We used RT-PCR with primers EF1α (forward primer, 5′-CAGATTGGTGCTGGATATGC-3′; reverse primer, 5′-ACTGCCTTGATGACTCCTAG-3′; 28 cycles), Tyr (forward, 5′-GCATCCCGAGATGCCTTCATAGG-3′; reverse, 5′-TGAAGTTGGCCGACCGATCCATG-3′; 28 cycles), Tyrp1 (forward, 5′-TCCGTGAGCAAAACCTTTCTGGGC-3′; reverse, 5′-TGTGTCAAACATATTCACATCAAG-3′; 28 cycles), Dct (forward, 5′-CTGTCCGGGACACATTGC-TCG-3′; reverse, 5′-AGTGGAATTCCTGAAAAAAGGAGG-3′; 28 cycles), Slug (forward, 5′-CAAACTTTCCG-ACTCGCATGCG-3′; reverse, 5′-CTAATGTGCTACACAGCAACCAGA-3′; 28 cycles), Sox10 (forward, 5′-TTCCACCTAATGGTCATGCTGGG-3′; reverse, 5′-GAGATGGAGGGAAATGCTGAACC-3′; 28 cycles), N-tubulin (forward, 5′-ACACGGCATTGATCCTACAG-3′; reverse, 5′-AGCTCCTTCGGTGTAATGAC-3′; 28 cycles), XNeuroD (forward, 5′-CCATGGGACAGCAGCGCATGCTGAC-3′; reverse, 5′-GTCCGCATTACTGCTCTCCGGAACAACG-3′; 28 cycles), XlM-itfα (forward, 5′-GAAAGCCTCAGTGGATTACATTCGCA-3′; reverse, 5′-GTGTATCGTCCATGAATATGTCT-TC-3′; 32 cycles).

BrdU Incorporation and Immunohistochemistry

BrdU was dissolved in 0.1× MMR at a final concentration of 50 μg/ml. Embryos were incubated in this BrdU solution for 1 hr, transferred to 0.1×MMR, and maintained for 2 hr. After fixation, these embryos were used for whole-mount in situ hybridization analysis and cryosections (10 μm) were made from frozen blocks as described previously (Kumasaka et al., 2003) for immunostaining. Immunostaining was performed using primary antibody, anti-BrdU (Developmental Studies Hybridoma Bank), anti–single-strand DNA (DAKO) and followed by ENVISION Kit/HRP (DAKO). Before primary antibody reaction, sections were incubated in 2 N HCl for 1 hr at 37°C to expose the DNA.


We thank Drs. Hiroyuki Ide and Koji Tamura for helpful discussions. We also thank all the members of the Yamamoto and Ide laboratories for technical advice. H.Y. was funded by a Grant-in Aid from the Ministry of Education, Culture, Sports, Science, and Technology, Japan.