|Home | About | Journals | Submit | Contact Us | Français|
We describe a mechanistic model whereby Foxd3, a forkhead transcription factor, prevents neural crest-derived precursors from acquiring a melanophore fate. Foxd3 regulates this fate choice by repressing the mitfa promoter in a subset of neural crest cells. mitfa is only expressed in a Foxd3-negative subset of neural crest cells, and foxd3 mutants show an increase in the spatial domain of mitfa expression, thereby suggesting Foxd3 limits the mitfa domain. Furthermore, foxd3:gfp transgenic zebrafish reveal foxd3 expression in xanthophore precursors and iridophores, but not in terminally differentiated melanophores. Luciferase experiments and embryo mRNA injections indicate Foxd3 acts directly on the mitfa promoter to negatively regulate mitfa expression. Taken together, our data suggests the presence of Foxd3 in a subset of precursors leads to mitfa repression and suppression of melanophore fate. MITF, the human mitfa ortholog, has recently been described as an oncogene and implicated in various forms of melanoma. Understanding the mechanisms that regulate mitfa and melanophore development could prove informative in the treatment and prevention of these human diseases.
The elucidation of the genetic mechanisms that transition a multipotent precursor towards a committed fate remains an important focus of developmental biology. The neural crest-derived pigment cells of the zebrafish, Danio rerio, are a remarkably tractable system for studying cell fate in vertebrate taxa. Pigment cells of the skin are derived from the neural crest, a transient embryonic population of cells that migrate through the animal to generate a diverse set of cell types (reviewed by Cooper and Raible, 2009). Zebrafish produce three varieties of neural crest-derived pigment cells: the black melanophore, the silver iridophore and the yellow xanthophore. These pigment cells migrate across the zebrafish and create coloration patterns that affect the behavior and ecology of the organism. Melanophores (or melanocytes) are broadly represented among species from each vertebrate class, while xanthophores and iridophores are absent in the homoeothermic vertebrates, birds and mammals (Braasch et al., 2006; Fujii, 2000). One hundred years of mouse genetic research have elucidated many aspects of melanophore biology, including the molecular pathways regulating melanoblast development, the biochemistry of melanin production and the dynamics of melanosome dispersal (Hoekstra, 2006; Kelsh, 2004; Lin and Fisher, 2007; Parichy, 2006). However, less is known about the mechanisms that act to dictate melanophore fate specification. In this study we describe an early event in pigment cell development in zebrafish, Danio rerio, which prevents a pigment precursor from attaining a melanophore fate.
Positive regulators of melanophore development are well defined. In many animals the basic helix-loop-helix/leucine zipper transcription factor Mitf acts as a master regulatory gene for black pigment cell differentiation (Levy et al., 2006). Mitf is amongst the earliest genes expressed in melanoblast precursors, and is necessary and sufficient for their development. Zebrafish mutant for the Mitf ortholog mitfa exhibit a complete absence of body melanophores (Lister et al., 2001). Previous work has placed Mitf upstream of multiple genes necessary for melanin production and terminally differentiated melanophores including dopachrome tautomerase, tyrosinase, tyrosinase-related protein-1, c-kit and bcl2 (reviewed by Steingrimsson et al., 2004). Ectopic expression of Mitf is sufficient to confer melanophore characteristics (Lister et al., 1999; Planque et al., 2004; Tachibana et al., 1996), demonstrating that it is necessary and sufficient for pigment cell specification.
Several positive regulators of zebrafish mitfa have been identified. The transcription factor Sox10 directly drives melanophore cell fate via the mitfa promoter (Elworthy et al., 2003), a regulatory relationship conserved in other animals (Bondurand et al., 2000; Lee et al., 2000; Potterf et al., 2000; Verastegui et al., 2000). Transcription of mitfa is also positively regulated by Wnt signals (Dorsky et al., 2000; Larue and Delmas, 2006; Saito et al., 2003). However, both Sox10 and Wnt signals play positive roles in multiple neural crest lineages (reviewed by Raible, 2006) suggesting negative regulators of Mitf are important for cell fate specification.
Foxd3, a winged helix transcription factor, is a good candidate for a negative regulator of melanophore development. Originally identified as a gene expressed in murine embryonic stem cells (Clevidence et al., 1993; Hanna et al., 2002; Sutton et al., 1996), Foxd3 was later found to exhibit highly conserved pre-migratory neural crest expression throughout vertebrates (Hromas et al., 1999; Labosky and Kaestner, 1998; Odenthal and Nusslein-Volhard, 1998; Pohl and Knochel, 2001; Sasai et al., 2001). More recently, Foxd3 has been shown to play a functional role in the specification of various downstream neural crest derivatives (Ignatius et al., 2008; Lister et al., 2006; Montero-Balaguer et al., 2006; Stewart et al., 2006; Teng et al., 2008) While Foxd3 has been described as a transcriptional activator in muscle (Lee et al., 2006), it likely acts as a transcriptional repressor in the neural crest (Pohl and Knochel, 2001; Sasai et al., 2001; Yaklichkin et al., 2007). Foxd3 represses the formation of melanocytes in chick embryos (Kos et al., 2001), suggesting it is a key regulator of melanogenesis. Recently, Ignatius et al. (2008) have shown that zebrafish foxd3 genetically interacts with histone deacetylase and that Foxd3 protein can bind the mitfa promoter in vitro, but no binding sites were identified or functional tests reported. Here we examine the functional consequence of this interaction.
We propose a model in which Foxd3 acts upon neural crest cells, via the mitfa promoter, to prevent melanophore fate. foxd3 mutants reveal an increase in the spatial domain of mitfa suggesting Foxd3 restricts mitfa expression. In a wild-type background, mitfa is only expressed in a sub-set of neural crest cells that are negative for Foxd3 protein, again suggesting Foxd3 limits the mitfa domain. Furthermore, our foxd3:gfp transgenic zebrafish reveals foxd3 expression in xanthophore precursors and iridophores but not terminally differentiated melanophores. Luciferase experiments quantify Foxd3's ability to repress the mitfa promoter in a winged helix domain dependent manner. Lastly, embryo mRNA injections recapitulate the luciferase results in the context of a live zebrafish. Taken together, these results support a mechanism of cell specification whereby the absence of Foxd3 protein in pigment precursor cells allows mitfa activation and, subsequently, melanophore cell fate. Reciprocally, the presence of Foxd3 protein in other neural crest cells maintains mitfa repression and, thereby, suppresses melanophore fate.
The Tg(mitfa:gfp)w47 transgenic line was generated by injecting one-cell embryos with an agarose gel-purified Sal I/Not I restriction fragment of the plasmid pNP-BEGFP, in which the CMV promoter from the vector pCS2+BE-GFPLT was replaced with a 931-bp fragment of the mitfa promoter and 5′ UTR via the SalI and HindIII sites. A single germline transgenic founder was identified from 28 adult fish. Adult fish of the *AB strain foxd3zdf1 (sym1; Stewart et al., 2006), the transgenic line Tg(mitfa:gfp)w47 and the transgenic line Tg(foxd3:gfp)zf15 (Gilmour et al., 2002) reporter strains were maintained on a 14h/10h light/dark cycle at 28.5°C. Embryos for all experiments were obtained through natural crosses and staged according to Kimmel (Kimmel et al., 1995). In some experiments, phenylthiocarbamide (PTU; Sigma) was added to embryo medium at a final concentration of 0.2 mM to inhibit melanin synthesis.
B16 and NIH-3T3 cells were cultured in Dulbecco's Modified Eagle's Medium (DMEM) containing 5% fetal bovine serum (Gibco) supplemented with 1× penicillin/streptomycin and incubated at 37°C in an atmosphere of 5% CO2. Fresh culture medium was provided every 2 days and the cells were sub-cultured before reaching 70% confluency. The melb-a mouse melanoblast cells (Sviderskaya et al., 1995) were cultured in RPMI 1640 medium (Gibco) containing 10% Fetal bovine serum, Stem cell factor (10 ng/ml) and FGF2 (40 pM) supplemented with 1× penicillin/streptomycin and incubated at 37° in an atmosphere of 10% CO2. The pCS2-myc-tag Foxd3 plasmid used for the melb-a cell overexpression assay was described previously (ZDB-GENE-980526-143; Lister et al., 2006). The DNA-binding mutant form of pCS2-myc-tag Foxd3 (Foxd3 BM) was created using the Quikchange mutagenesis protocol (Stratagene) with the following oligonucleotides: forward, 5′-cca ttc gcc ata act ttt cgc tca acg act gct-3′; reverse, 5′-agc agt cgt tga gcg aaa agt tat ggc gaa tgg-3′. The mutation generated a single amino acid substitution in the winged-helix domain: leucine to phenylalanine (L147F). The pCS2-snail1b plasmid was made by cloning a full length open reading frame (ZDB-GENE-980526-514; Thisse et al., 1995) into pCS2 via HindIII and BamHI sites. For the NIH-3T3 luciferase assay, the coding sequence of zebrafish sox10 (ZDB-GENE-011207-1; Dutton et al., 2001) was amplified and subcloned into the vector pCS2-FLAG with three amino terminal FLAG epitope tags. The mitfa proximal promoter (GenBank accession # AF211890; Dorsky et al., 2000) was mutated using the Quikchange mutagenesis protocol (Stratagene). Changed base pairs are in lower case. Site 1 (nucleotides 579-588):5′ ATGCTGAtggCAggggATGTTT 3′. Site 2 (nucleotides 750-761): 5′- CGTTTGGGTAgggAAggagATATGA 3′. All constructs were confirmed by sequencing. About 1 × 105 cells were seeded into each well of a 24-well plate (Falcon 3047) 24 hours prior to transfection. Cells were transfected by the lipofectamine method (Invitrogen) according to the manufacturer's instructions. Transfection mixtures contained 1 μl of lipofectamine and 0.8 – 1 μg of plasmid constructs per well in 24-well plates. Reagents for the Galacto-Light chemiluminescent reporter assay were from Tropix (Bedford, MA, USA) and D-Luciferin was from Analytical Luminescence Laboratories (San Diego, CA, USA). Luciferase output was standardized with beta-galactosidase activity. All transfections and luciferase assays were performed independently and in quadruplicate.
Digoxigenin-labeled riboprobe for mitfa (ZDB-GENE-990910-11; Lister et al., 2006) and dct (ZDB-GENE-000508-1; Kelsh, 2000) has been characterized previously. In situ hybridization was performed as described previously (Lister et al., 1999), using NBT/BCIP as a chromagenic substrate. Fluorescent in situ hybridization was performed as described previously (Julich et al., 2005) using anti-Dig POD for both dct and mitfa, Alexa-Fluor tyramide substrate 568 (Invitrogen) and Roche blocking reagent and buffer. The following antibodies were used for immunohistochemistry at the indicated dilutions: rabbit polyclonal anti-Foxd3 (Lister et al., 2006), 1:500; mouse monoclonal anti-myc (Abcam), 1:1000; mouse monoclonal anti-Flag (Sigma), 1:1000; mouse monoclonal anti-Pax3/7 (DP312) (Minchin and Hughes, 2008), 1:500; rabbit polyclonal anti-MITF (Novus Biological), rabbit polyclonal anti-Sox10 (Park et al., 2005), 1:1000; 1:500; Anti-mouse (Alexa 488) and anti-rabbit (Alexa 568) secondary antibodies (Invitrogen), 1:750; mouse monoclonal anti-Green Fluorescent Protein (Invitrogen) and rabbit polyclonal anti-Green Fluorescent Protein (Invitrogen), 1:500 in conjunction with anti-mouse (Alexa 488), 1:1000 and anti-rabbit (Alexa 568), 1:1000 to boost signal. Brightfield images were obtained on a Nikon dissecting microscope with a Spot RT Slider digital camera (Diagnostic Instruments). Fluorescent confocal images were obtained on a LSM 5 Pascal confocal microscope (Zeiss). Whole images were processed for color balancing and brightness/contrast using Photoshop CS4 (Adobe) and formatted with Illustrator CS4 (Adobe).
Full length and capped foxd3 (100pg/embryo), foxd3 DNA-binding mutant (100pg/embryo) and sox10 mRNA (18 pg/embryo) were generated using mMessage mMachine (Ambion). Embryos for injection were obtained from natural matings, injected at the one-cell stage and imaged at shield stage (6-8 hours post fertilization; hpf) using a fluorescent Nikon dissecting microscope with Spot camera.
To observe zebrafish melanoblasts in vivo we generated a GFP reporter transgenic line using 931 bp. of the mitfa promoter and 5′ UTR upstream of the mitfa start site. At 20 hpf mitfa:gfp is expressed in lateral stripes along the dorsal aspect of the hindbrain and trunk region (Fig. 1A). This pattern is indicative of migratory neural crest cells arising from the closing neural tube. By 26 hpf mitfa:gfp positive cells are present dorsally at the posterior end of the zebrafish and ventrally throughout the head and anterior trunk region (Fig. 1B). To confirm the fidelity of our mitfa:gfp transgenic, we co-labeled transgenic embryos with anti-GFP and a mitfa riboprobe and counted cells along the dorsal-lateral stripes. At 24 hpf, 98.4% of mitfa:gfp cells (n=311) are also positive for mitfa transcript (Fig. 1C,E). To confirm that cells positive for mitfa:gfp are melanoblasts and fated to become melanophores, we examined transgenic embryos for co-expression of GFP and dopochrome tautomerase (dct) mRNA (Fig. 1D,F). dct, also known as tyrosinase-related protein-2, is a melanogenic enzyme downstream of mitfa that resides in the melanosome membrane within melanoblasts (Kelsh et al., 2000; Lister et al., 1999).Therefore, at 24 hpf, one would expect a proportion of mitfa positive cells to have upregulated dct expression. We found 55% of mitfa:gfp positive cells (n= 193) are dct positive. In conclusion, our mitfa:gfp transgenic accurately marks mitfa positive melanoblasts at these stages.
If Foxd3 acts as a transcriptional repressor, one would expect to observe higher overall mitfa expression or ectopic mitfa expressing cells in foxd3 mutant zebrafish. To explore this regulatory relationship, we assessed possible spatial and temporal expansions of mitfa expression in foxd3 mutant embryos using in situ hybridization. The foxd3zdf1mutation is a nucleotide deletion that disrupts the forkhead DNA-binding domain of the foxd3 gene (Stewart et al., 2006). At 40 hpf, we observe a spatial expansion of mitfa in the head region and anterior trunk (Fig. 2 A,D). At later time points, 48 and 60 hpf, the increase in mitfa expression is noticeable, although less pronounced (Fig. 2 B,C,E,F). Similar results were recently reported by Ignatius et al. (Ignatius et al., 2008), who found mitfa signal in melanoblasts to be more robust in foxd3 morpholino-injected zebrafish than in uninjected controls.
Foxd3 expression is a robust marker of pre-migratory neural crest but is down-regulated in a subset of cells as development proceeds, possibly via an autoregulatory negative feedback mechanism (Lister et al., 2006; Stewart et al., 2006). If Foxd3 acts to repress mitfa expression, we would predict their distribution to be mutually exclusive. To assess the relationship between mitfa and Foxd3 expression, we used Tg(mitfa:gfp) zebrafish and anti-Foxd3 antibody on a field of neural crest cells along the dorsal aspect of the anterior trunk region of zebrafish embryos (Fig. 3A,B). At 24 hpf, we find that expression of Foxd3 protein does not substantially overlap with mitfa:gfp. Analysis of mitfa transcript by in situ hybridization confirms this relationship (Fig. 3C). To quantify the mutually exclusive relationship between mitfa and Foxd3 expression, we looked at the earliest stages of GFP transgene initiation to determine whether there was any overlap in expression, and found that approximately 90% of GFP positive cells were Foxd3 negative (Fig. 3D). In contrast, Sox10, a known mitfa activator (Dutton et al., 2001) shows extensive overlap (Fig. 3E). Cell counts reveal that 91.8% of mitfa:gfp positive cells (n=250) are Sox10 positive at this stage.
Using the previously characterized Tg(foxd3:gfp) zebrafish (Gilmour et al., 2002), we looked for foxd3 expression in the three zebrafish pigment cell types: xanthophores, iridophores and melanophores. The thick yellow pigmentation in xanthophores obscures cellular boundaries and makes accurate cell counts difficult. To better identify xanthophores at 27 hpf, we used the Pax3/7 antibody, recently determined to be exclusively expressed in the xanthophore lineage at this stage (Minchin and Hughes, 2008). We found 94% of Pax3/7 xanthoblasts (n=165) are foxd3:gfp positive (Fig. 4A,B). Terminally differentiated iridophores illuminate under incident light, allowing accurate cell counts. At 72 hpf, we observed 88.3% of differentiated iridophores (n=342) to be foxd3:gfp positive (Fig. 4C,D). Terminally differentiated melanophores lie adjacent to iridophores along the dorsal stripe at 72 hpf (Fig 4D). These melanophores display no overlapping foxd3:gfp expression, however, it is possible the melanin accumulation within melanophores masks GFP expression. To address this concern, we treated fish with the melanin synthesis blocker PTU at 24 hpf, then analyzed foxd3:gfp expression in partially melanized zebrafish at 72 hpf (Fig. 4E,F). We observed foxd3:gfp expression in only 5% of melanophores (n=314). We conclude that foxd3 expression is maintained in xanthophores and iridophores, however, foxd3 expression is down-regulated in melanophores.
To test whether Foxd3 expression is sufficient to repress endogenous Mitf expression we transfected pCS2+ plasmid expressing zebrafish foxd3 under the CMV promoter into melb-a mouse melanoblast cells. Melb-a cells constitutively express Mitf (Lei et al., 2002). We then assayed expression of endogenous Mitf expression with an anti-MITF antibody. Expression of Foxd3 strongly repressed Mitf expression (Fig. 5A). In contrast, foxd3 with a leucine to phenylalanine substitution in the DNA binding, winged-helix domain (L167F), a mutation of a highly conserved amino acid shown to block DNA binding of Fox proteins in other systems (Hacker et al., 1995), was incapable of repressing endogenous Mitf, implying Foxd3 must bind directly to DNA to repress Mitf (Fig. 5B). To quantify the relationship between Foxd3 and endogenous Mitf expression, cell counts were performed (Fig. 5C). 96.5% of cells expressing Foxd3 repressed Mitf expression (n=58). 2.2% of cells expressing Foxd3 binding mutant repressed Mitf expression (n=47). We conclude that zebrafish foxd3 prevents endogenous Mitf in cell culture.
To determine whether Foxd3 could directly regulate the mitfa promoter, we used a pGL3-SV40 luciferase reporter plasmid comprising 931 bp of the zebrafish mitfa proximal promoter (Dorsky et al., 2000) to perform luciferase assays in NIH-3T3 mouse fibroblasts and B16 mouse melanoma cell lines. In B16 melanoma cells, the mitfa promoter has high basal activity. Co-transfection of the foxd3 plasmid caused a 19-fold decrease in the activity of mitfa:luciferase as compared to basal mitfa:luciferase alone (Fig. 6A). A similar co-transfection using the DNA binding mutant version of foxd3 resulted in a return to basal mitfa:luciferase expression levels (Fig. 6A). Acting as a negative control, snail1b, a transcriptional repressor with no known mitfa activity, was transfected into the assay. snail1b failed to repress mitfa:luciferase (Fig. 6A). Foxd3 transfection had no effect on luciferase expression under control of an SV40 promoter (data not shown), suggesting that mitfa control elements were necessary for repression. Co-transfection with the transcriptional repressor snail1b (LaBonne and Bronner-Fraser, 2000), failed to repress mitfa:luciferase, suggesting that repression by Foxd3 was specific (Fig. 6A).
Transfections with an ectopic activator in NIH-3T3 cells confirmed the preceding results. NIH-3T3 cells have low basal activation of mitfa:luciferase, as would be expected from a fibroblast cell. To compensate for low basal mitfa expression, sox10, a known mitfa activator (Elworthy et al., 2003) was co-transfected to enact a 5 fold increase over basal mitfa:luciferase expression levels (Fig. 6B). Co-transfection of foxd3 with sox10 blocked the mitfa promoter activation normally observed with sox10 alone (Fig. 6B). These data suggest Foxd3 can antagonize Sox10's activation of the mitfa promoter. When foxd3 DNA binding mutant was used, we observed no repression of sox10 activation (Fig. 6B). Taken together, these results suggest Foxd3 represses the mitfa promoter by binding to DNA. Ignatius et al. (Ignatius et al., 2008) recently reported binding of Foxd3 to mitfa promoter DNA in vitro, consistent with direct activation, and identified two putative Forkhead binding sites but did not test them for function. We tested the function of these presumptive Foxd3 binding sites in the mitfa promoter and observed a partial relief of repression after the two sites were mutated (Fig. 6C). foxd3 transfections with NIH-3T3 cells yielded a 13 fold repression using the full length mitfa promoter (Fig. 6C). In contrast, foxd3 transfections using our mutagenized S1,S2 mitfa promoter yielded only a 2.5 fold repression (Fig. 6C). This suggests the S1 and S2 sites are involved with Foxd3 repression of mitfa, however, additional Foxd3 binding sites may also be present or additional indirect mechanisms may be at play.
To examine mitfa promoter activity in live zebrafish embryos, we used a transient mRNA co-injection assay in the mitfa:gfp transgenic zebrafish. GFP is first observed in melanophore precursors at 15-18 hpf, reflecting endogenous mitfa expression (Fig. 3D). Under normal conditions, a 6 hpf shield stage embryo displays no GFP (Fig. 7B). Injection of sox10 mRNA into one-celled stage mitfa:gfp embryos results in robust precocious GFP expression at 6 hpf (Fig. 7D), recapitulating sox10's effect on mitfa expression in zebrafish (Elworthy et al., 2003). Upon co-injection of both foxd3 mRNA and sox10 mRNA, precocious mitfa:gfp expression is strikingly absent (Fig. 7F), while co-injection of DNA binding mutant foxd3 mRNA and sox10 mRNA results in a return to the robust precocious GFP expression observed with injections of sox10 alone (Fig. 7H). We conclude that, in an over-expression assay in a live zebrafish embryo, Foxd3 can prevent mitfa:gfp expression in a manner dependent on its DNA binding domain.
Our results demonstrate that Foxd3 can repress the activity of the zebrafish mitfa promoter and thus would inhibit melanophore differentiation. These results are consistent with previous results in avian embryos, where electroporation of Foxd3 inhibited melanocyte differentiation (Kos et al., 2001). We also find that transfection of foxd3 into melb-a mouse melanoblast cells blocks Mitf expression. However, we have not been able to test whether Foxd3 overexpression in zebrafish embryos alters melanophore differentiation because of its earlier effects on axis specification (Lister et al., 2006; Steiner et al., 2006). We observed enhanced levels of mitfa signal in foxd3 mutants, consistent with the idea that Foxd3 represses mitfa. However, this mitfa de-repression was spatially limited to the head and anterior trunk region. Consistent with these observations, there is little change in melanophore cell number in foxd3 mutants (Stewart et al., 2006). These observations suggest that additional negative regulators of mitfa might exist in other regions of the zebrafish embryo, such as the posterior trunk and tail. Additionally, specific activators are needed in addition to loss of repression to increase mitfa expression.
Our data are consistent with Foxd3 acting directly as a transcriptional repressor on the mitfa promoter. Foxd3 has been demonstrated to act as a repressor in other systems (Pohl and Knochel, 2001; Yaklichkin et al., 2007). Foxd3 might regulate mitfa indirectly through protein-protein interactions rather than directly by promoter binding. However, several lines of evidence support a direct interaction. We found that transfection of foxd3 inhibits the effects of exogenously supplied sox10 activator in our assays with NIH 3T3 cells, suggesting a direct action. Additionally, introducing the L167F mutation into the winged helix domain of Foxd3, a mutation of a highly conserved amino acid shown to block DNA binding of Fox proteins in other systems (Hacker et al., 1995), eliminated the effects of Foxd3 on mitfa. However, it is possible that Foxd3 interfered with an endogenous co-activator that was not sufficient to activate mitfa on its own. There remains the possibility that a properly configured winged-helix domain might be necessary for interaction with other proteins if Foxd3 were interacting independently of direct DNA binding. There is some precedent for this idea: Guo et al. (2002) utilized immunoprecipitation assays to show Oct-4 physically interacts with the DNA-binding domain of FoxD3. In this manner, Oct-4 acts as a co-repressor modulating Foxd3 repressive activity in mammalian endoderm development. It is possible similar protein co-repressors are responsible for modulating Foxd3 activity in our assays. However, we observed Foxd3 repression of mitfa in mouse melanoblast (melb-a cells), mouse melanoma (B16 cells), mouse fibroblast (NIH-3T3 cells) and zebrafish embryos. Therefore, if protein-protein co-repressors are involved, this indirect interaction must be present in each of these four systems.
We have not been able to demonstrate consistent interactions of Foxd3 protein with the mitfa promoter in vitro, suggesting direct interactions may be of low affinity. We note that Ignatius et al. (2008) utilized electromobility shift assays to determine Foxd3 binds two sites from the mitfa promoter. When we mutated these sites and tested them in luciferase assays, we found that inhibition was alleviated from a 13 fold to a 2.5 fold repression. This partial relief of repression suggests that additional Foxd3 binding sites may be present on the mitfa promoter or that indirect, protein-protein interactions are at play.
The repression of mitfa by foxd3 sheds light on the possible mechanism behind the genetic interaction between foxd3 and kit, a receptor tyrosine kinase necessary for melanophore survival (Cooper et al., 2009). Inactivation of foxd3 partially rescues melanophore cell death seen in kit mutants. Since Mitf is a positive activator of the pro-survival gene bcl2 (McGill et al., 2002), inhibition of Mitf by Foxd3 would be expected to promote cell death. Inactivation of Foxd3 in turn would release this inhibition and have pro-survival effects that might partially overcome the loss of Kit function.
Our foxd3:gfp data suggests that, within pigment cells, Foxd3 repression is unique to melanophores. Xanthoblasts and iridophores maintain foxd3:gfp expression, suggesting Foxd3 does not interfere with their development and differentiation. Foxd3 appears to play a positive role in iridophore specification as iridophores are strongly reduced in both foxd3zdf1 mutant zebrafish (Stewart et al., 2006) and in foxd3 morphant zebrafish (Lister et al., 2006). Therefore, the restricted spatial expansion of mitfa observed in the foxd3zdf1 mutant may represent ectopic mitfa expression in a pigment precursor population of neural crest cells. Interestingly, mitfa exerts a repressive role on iridophore development, as seen in increased cell counts of larval trunk and tail iridophores in mitfa mutants as compared to wild type zebrafish (Lister et al., 1999). Fully understanding the possible mechanisms by which Foxd3 regulates transcription will be necessary to determine how Foxd3 acts to promote the development of other neural crest lineages, such as iridophores, peripheral neurons and glial cells.
While this work was under review, Thomas and Erickson (2009) described a similar inhibitory regulation of the avian MITF gene by FOXD3. In this study, FOXD3 regulated MITF indirectly through protein-protein interactions with PAX3, a positive regulator of Mitf in avian and mammalian species (Watanabe et al., 1996). In zebrafish, Pax3 does not regulate melanophores but rather xanthophores (Minchin and Hughes, 2008), and we find no evidence of regulation of the zebrafish mitfa promoter by Pax3 (J.A.L., unpublished). Thus while the regulatory interactions between Foxd3 and Mitf may be widely conserved, specific mechanisms may vary. Thomas and Erickson (2009) provide evidence for the regulatory interaction between Foxd3 and Mitf playing a role in regulating glial cell development, suggesting that this interaction may have more widespread effects on neural crest development than the regulation of pigment cell precursors described here.
Our work regarding Foxd3 repression of mitfa corroborates with previous findings from mammalian melanocyte research (reviewed by Harris and Erickson, 2007). The conservation of genetic mechanisms regulating melanogenesis across vertebrates makes the tractable zebrafish system an attractive tool to study human diseases associated with melanocytes. A recent study has correlated the occurrence of vitiligo, a chronic skin condition causing hypo-pigmentation, in one family to a single base change (G>T) which upregulates Foxd3 promoter activity in cultured cells (Alkhateeb et al., 2005). Furthermore, MITF has recently proven to be an amplified oncogene in a significant fraction of human melanomas and to play an oncogenic role in human clear cell sarcoma (Levy et al., 2006). Understanding the molecular mechanisms that properly repress mitfa expression may offer insight into improved strategies for treating these life-threatening diseases.
The authors wish to thank David White for fish care, John Kanki for the foxd3zdf1 fish, Bruce Appel for the Sox10 antibody, the Biology Imaging Facility for confocal assistance and Dot Bennett for melb-a cells. Additional thanks to Cynthia Cooper for laboratory training. This work has been supported by grants from NIH to D.W.R. and J.A.L. K.C. would like to thank Washington Research Foundation, ARCS Foundation and the Developmental Biology Training Grant (NICHD T32 HD007183-2681) for their financial support.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.