Pigmentation PAX-ways: The role of Pax3 in melanogenesis, melanocyte stem cell maintenance, and disease

Abstract

Transcription factors initiate a program of gene expression and are catalysts in downstream molecular cascades that modulate a variety of cellular processes. Pax3 is a transcription factor that is important in the melanocyte and influences melanocytic proliferation, resistance to apoptosis, migration, lineage specificity and differentiation. In this review, we focus on Pax3 and the molecular pathways that Pax3 is a part of during melanogenesis and in the melanocyte stem cell. These roles of Pax3 are emphasized during the development of diseases and syndromes resulting from either too much or too little Pax3 function. Due to its key task in melanocyte stem cells and tumors, the Pax3 pathway may provide an ideal target for either stem cell or cancer therapies.

INTRODUCTION

The Paired box (PAX in human, Pax in mouse) proteins represent a highly conserved family of transcription factors essential to the development of many tissue types throughout embryogenesis and vital to the maintenance of several stem cell niches in the adult organism. There are nine different Pax proteins in man and mice, which contribute to regulation of proliferation, migration, resistance to apoptosis, and prevention of terminal differentiation within various tissues. The expression patterns of Pax proteins reflect the functions that these transcription factors play during development and in lineage specificity. Several members of the Pax family regulate development of the central nervous system including Pax2, Pax3, Pax5, Pax6, Pax7, and Pax8. The lineage specificity of these family members evolves as embryogenesis progresses. Pax1 and Pax9 contribute to skeletal development while Pax3 and Pax7 play a role in myogenesis. Other Pax proteins regulate development of internal organs and tissues such as Pax4 and Pax6 in the pancreas, Pax2 and Pax8 in kidney development, Pax8 in the thyroid, and Pax5 in B-cells (Lang et al., 2007). Pax3 is a particularly interesting member of the Pax family, which coordinates the development of the central nervous system, somites, skeletal muscle and an extensive list of neural crest-derived cell types including but not limited to cardiac tissue, gastrointestinal enteric ganglia, and melanocytes. The ability for Pax3 to control such a diverse set of developmental events is due to alternate splicing and the characteristics of the domains contained in the Pax3 protein. Alternative splicing is a crucial process by which a limited number of genes direct a diverse number of functions (Short and Holland, 2008). In the case of Pax3, alternative splicing most often affects the C-terminal end of the homeodomain, usually inducing a frame shift that in turn modifies transactivation domain functionality. Seven isoforms of Pax3 are currently described known as Pax3a through Pax3e, Pax3g, and Pax3h (Barber et al., 1999), all of which exhibit differential expression patterns suggesting unique functions (for review see(Wang et al., 2008)).

The pax3 transcripts encode proteins that are conserved throughout vertebrate species. The amino acid sequence between mouse and human display high conservation, and possess identity at 98% of the residues. The predicted amino acid sequence in quail shares 95% sequence identity with mouse and human, differing near the N-terminal end of the protein and intron 8 (Barber et al., 1999). The Pax3 protein is comprised of several domains, and these moieties mediate interactions with a host of different factors in alternate combinations, allowing for either activation or repression of the promoter region of downstream target genes. In this review, we will focus on the molecular pathways that Pax3 impacts. In addition, Pax3 interacts with several other proteins that act as transcriptional co-activators, co-repressors or operate as modifiers of Pax3 function. These molecular interactions greatly influence the melanocytic cell state during embryogenesis and in the adult. While Pax3 is essential in regulating apoptosis, proliferation and differentiation during these normal cell processes, the balance between terminal differentiation and proliferation becomes skewed when levels of Pax3 become dysregulated.

The Pax3 protein

Pax3 possesses four structural domains; the paired domain, octapeptide motif, homeodomain, and transactivation domain (Figure 1). The paired domain, named for the two helix-turn-helix (HTH) motif-containing sub-domains (PAI and RED) within it, comprises 128 amino acids in the amino-terminal region of Pax3 and binds DNA in addition to facilitating interactions with other proteins. Evidence suggests that the N-terminal HTH motif is necessary and sufficient for binding to the DNA (Jun and Desplan, 1996, Treisman et al., 1991, Xu et al., 1995) consensus sequence (G)T(T/C)(C/A)(C/T)(G/C)(G/C), which is used to recognize multiple target sites (Epstein et al., 1996, Chalepakis and Gruss, 1995, Chalepakis et al., 1994c, Chalepakis et al., 1994b, Jun and Desplan, 1996). While the PAI HTH motif is essential for DNA binding, the exact function of the RED HTH positioned near the C-terminal end of the paired domain within Pax3 remains to be fully elucidated. The RED HTH motif fails to interact directly with DNA (Treisman et al., 1991, Jun and Desplan, 1996) and does not influence the DNA binding ability of the N-terminal HTH motif (Jun and Desplan, 1996). The RED subdomain has, however, shown DNA recognition and binding capabilities when it is part of other proteins (Jun et al., 1998). In addition to mediating DNA binding, the paired domain also facilitates protein-protein interactions, which impacts the overall function of Pax3. The paired domain interacts with other homeodomain structural motifs (Jun and Desplan, 1996), with the High Motility Group (HMG) structural domains of Sox proteins (Lang and Epstein, 2003) and with the calcium binding protein, calmyrin (also known as CIB1) (Hollenbach et al., 2002) in order to control Pax3 activity. For example, the interaction with calmyrin represses transcriptional activity by interfering with the ability of Pax3 to bind DNA through an unknown mechanism (Hollenbach et al., 2002). In addition, Pax3 must bind to SOX10 in order to synergistically activate promoters such as Mitf (Watanabe et al., 1998, Bondurand et al., 2000) and c-Ret (Lang and Epstein, 2003, Lang et al., 2000).

Figure 1.

The Pax3 protein has four domains: the paired domain, octapeptide motif, homeodomain and transactivation domain. A. The paired domain (lavender) is made of PAI- and RED- subdomains. PAI is composed of two beta-sheets and a helix turn helix (HTH) motif, while RED contains a HTH motif. The homeodomain (pink) possesses three helicies designated I, II and III. The transactivating domain (TA) is located in the C terminal tail of the protein. The octapeptide motif is yellow. N-terminal (N) and C-terminal are indicated. Numbers correspond to amino acid sequence. B. The amino acid sequence of canonical Pax3. Odd numbered exons are denoted with black font, blue represents even numbered exons. Grey amino acids have a codon shared by the two surrounding exons. Specific regions of the protein are highlighted; the paired domain (lavender), the octapeptide (yellow), and the homeodomain (pink).

The homeodomain serves as another interface between Pax3 and DNA or other interacting proteins. The homeodomain is 60 amino acids in length and composed of three HTH motifs identified as helix I, II, and III. Helices I and II are positioned such that they are exposed to the surface of the protein allowing for interactions with other proteins, while helix III recognizes DNA containing a core TAAT sequence to mediate DNA binding (Wilson et al., 1995). Evidence suggests the Pax3 homeodomain also functions to regulate the DNA binding ability of the paired domain (Cao and Wang, 2000). Pax3 recognizes DNA containing consensus sequences for both the paired domain and the homeodomain, enabling synergistic activation between these domains for full transcriptional activation of downstream target genes (Cao and Wang, 2000, Chalepakis et al., 1994c, Treisman et al., 1991). The homeodomain is also capable of interacting with other proteins in order to modulate Pax3 activity. For example, the homeodomain binds to retinoblastoma (Rb) protein, which represses Pax3 transcriptional activation (Wiggan et al., 1998). Other co-repressors such as HIRA (Magnaghi et al., 1998) and Daxx (Hollenbach et al., 1999) also bind through the homeodomain. The inclusion of three different DNA-binding motifs within Pax3 allows for several combinations of DNA recognition sites and provides the flexibility necessary to control a host of developmental processes.

The octapeptide motif, comprised of the amino acid sequence HSIDGILS (Goulding et al., 1991), functions primarily as a protein interaction epitope. Nestled between the paired domain and the homeodomain, the octapeptide shares similarity with the eh1 motif of the engrailed protein (Allen et al., 1991). The octapeptide motif directly interacts with calmyrin, and the complex impedes the ability of Pax3 to bind to DNA (Hollenbach et al., 2002). The transactivation domain is a S/G/T-rich region located in the carboxy-terminal end of the Pax3 protein and functions to mediate the integrity of Pax3-DNA interactions. In the absence of either the paired domain- or the homeodomain-responsive elements, the transactivation domain can block the homeodomain from binding DNA to ensure sequence specificity (Cao and Wang, 2000). These studies suggest the transactivation domain may directly interact with both the paired and homeodomains.

Although little is known about the mechanisms regulating Pax3 transcriptional activity, several studies provide insight into the post-translational modifications of Pax3. Pax3 activity is modulated by PKC during embryonic myogenesis and there are 8 putative sites for S/T phosphorylation by PKC (Brunelli et al., 2007). This suggests a possible functional relationship between these proteins. The kinase inhibitor PKC412 also reduces rhabdomyosarcoma translocation product Pax3/FKHR transcriptional activity, suggesting that phosphorylation is necessary for the full activity Pax3/FKHR (Amstutz et al., 2008). A new report indicates that the phosphorylation of Pax3 occurs on serine 205 by an unknown kinase in myoblast precursor cells. The phosphorylated state, however, is rapidly lost concurrent with the onset of differentiation (Miller et al., 2008). Collectively, these reports indicate that phosphorylation of Pax3 may influence DNA binding affinity. The ubiquitination-proteasomal degradation pathway also modulates Pax3 activity during myogenic development. Pax3 mRNA levels must be down-regulated during the differentiation of myogenic progenitors for terminal differentiation to occur. In order to degrade the remaining Pax3 proteins within these cells, Pax3 receives a monoubiquitin tag during adult stem cell differentiation, causing it to be targeted for degradation by the proteasome (Boutet et al., 2007). Due to the functions and biological consequences dictated by Pax3, it is important to modify the ability of Pax3 to activate or repress the transcription of downstream genes.

Protein-Protein Interactions Involving Pax3

Cofactors that interact with PAX3 and act as transcriptional modulators

Pax3 can serve as a transcriptional activator or repressor (Chalepakis et al., 1994b) by acting as either: 1) an activator or a repressor for the specific enhancer, 2) both an activator or a repressor based on Pax3 protein concentrations, or 3) a molecular toggle to repress or activate genes based on the collaborating cofactors located at the enhancer. Here, we will focus on this third possibility: the relationship between Pax3 and other proteins (Table 1). Some of these interactions are known to influence gene expression and cell state of melanocytes, while others have yet unexplored roles in pigment cells.

Table 1.

PAX3 Binding Partners

Cofactor	Consequence	Pax3 Interaction Domain*	Reference
Cofactors as transcriptional modulators
SOX10	Synergistic activation of Pax3 target genes (c-Ret, Mitf)	PD	(Lang and Epstein, 2003, Smit et al., 2000)
TAZ	Synergistic activation of Pax3 target genes (Mitf)	Multiple	(Murakami et al., 2006)
GRG4	Co-repression of Pax3 target gene (Dct)	Undetermined	(Lang et al., 2005)
HP1γ	Attenuation of Pax3 repression on target gene	PD	(Hsieh et al., 2006)
KAP1	Co-repression of Pax3 target gene	PD	(Hsieh et al., 2006)
TBX18/TBX22	Skeletal development	PD	(Farin et al., 2008)
Pax3 functional modulators
CIB1	Blocks Pax3-DNA interactions	PD, O	(Hollenbach et al., 2002)
RB/p130/p107	Possible block of c-Met activity	HD	(Wiggan et al., 1998)
DAXX	Attenuation of Pax3 activation of target genes	HD, O	(Hollenbach et al., 1999)
HIRA	Attenuation of Pax3 activation of target genes	HD	(Magnaghi et al., 1998)
BRN2	Unknown mechanism	Undetermined	(Smit et al., 2000)
RAD23B	Pax3 degradation	Ubiquitin tag in TA	(Boutet et al., 2007)
MSX1	Blocks Pax3 transcriptional activity	PD	(Bendall et al., 1999)
MOX1/MOX2	Blocks Pax3-DNA interactions	HD	(Stamataki et al., 2001)
IPO13	Transports Pax3 into the nucleus	HD	(Ploski et al., 2004)

^*PD stands for Paired domain, HD is the homeodomain, O is for octapeptide, and TA represents the transactivation domain.

This influence on downstream genes is enhanced through the direct interaction of several co-activators and repressors. Proteins that enhance Pax3 transcriptional activity include both the Sox10 and TAZ proteins. Pax3 and Sox10 directly interact through their DNA binding domains (Lang and Epstein, 2003, Smit et al., 2000). Sox10 is a member of the Sox (Sry-box) family of transcription factors, all of which contain a High Mobility Group (HMG) DNA binding domain. At least two Sox proteins, Sox9 and Sox10, are expressed in melanoblasts and melanocytes. Sox9 is expressed early in the melanocyte neural crest precursors but is down-regulated before migration, while Sox10 is activated just prior to and during embryonic neural crest migration (Kuhlbrodt et al., 1998, Southard-Smith et al., 1998). Sox10 is expressed in melanoblasts, and is down-regulated as these cells differentiate. This is mirrored by a concurrent upregulation of Sox9 in differentiating cultured melanocytes (Cook et al., 2005). The importance of Sox10 to the developing melanoblasts is seen during mouse development, where loss of Sox10 expression leads to substantial pigmentation defects (Pingault et al., 1998, Southard-Smith et al., 1998). Conversely, Sox9 plays an active role in differentiating melanocytes through the activation of differentiation genes, and is up-regulated by melanocyte differentiation signals such as ultraviolet radiation and forskolin (Passeron et al., 2007). Both Sox9 and Sox10 can activate Mitf, a protein essential for melanogenesis. Pax3 directly interacts with Sox10 and can activate Mitf expression in a synergistic manner (Lang and Epstein, 2003, Bondurand et al., 2000, Potterf et al., 2000). Pax3 and Sox10 also directly bind to and synergistically activate the c-Ret gene, which encodes a tyrosine kinase receptor (Lang et al., 2000). This Pax3-Sox10 enhancer site is specifically targeted by histone acetylation that leads to modulated Ret gene expression (Puppo et al., 2002). While Pax3 and Sox10 synergistically activate both the Mitf and the c-ret promoter, the mechanisms by which these factors modulate the expression of these genes are not the same. While the c-ret enhancer requires only Pax3 to directly bind to DNA while recruiting Sox10 through protein-protein interaction, the Mitf promoter requires both proteins to bind to the genomic cis regulatory site (Lang and Epstein, 2003). Sox10 also exerts different effects depending on the enhancer and the binding partner. Sox10 can activate the melanocyte gene Dct together with Mitf through the same enhancer that Pax3 represses Dct (Lang et al., 2005, Ludwig et al., 2004, Potterf et al., 2001). While Sox9 can activate several melanocytic Sox10 targets, such as Mitf and Dct, and the region of Sox10 that binds to Pax3 is highly conserved between the two Sox proteins, a direct Sox9 interaction with Pax3 has not been characterized. Pax3 directly interacts with TAZ (also known as WW domain-containing transcriptional regulator 1 or WWTR1), a transcriptional co-activator with a PDZ-binding motif (Murakami et al., 2006). The WW domain of the TAZ protein interacts with Pax3 through multiple domains. TAZ acts as a potent transcriptional co-activator with Pax3 on promoters in a luciferase assay system, including the promoter for Mitf. However, it is not known if TAZ is expressed in a melanocytic lineage. While Pax3 contains its own transcriptional activation domain (Chalepakis et al., 1994b) interactions with other proteins provides an opportunity to synergistically activate downstream targets.

Pax3 can also become a potent repressor of gene expression through the recruitment of several cofactor repressor molecules, including Grg4, HP1, KAP1, and several Tbx family proteins. Pax3 directly interacts with Grg4 (Groucho-related protein 4 in mice, TLE4 in humans), a member of the groucho-homologous proteins that act as transcriptional repressors. Grg4 is expressed in melanocytes, and together with Pax3 represses the expression of Dct (Lang et al., 2005). Pax3 also interacts with both HP1 and KAP1 (Hsieh et al., 2006). HP1 (heterochromatin protein 1) is a heterochromatin-associated protein that silences genes by promoting heterochromatin. Several transcriptional repressors, including KAP1, directly bind to and recruit HP1 to regulatory enhancers. The paired domain of Pax3 binds to both factors and recruits them to target promoters, where KAP1 enhances the ability of Pax3 to act as a transcriptional repressor, while HP1 attenuates this transcriptional repression. HP1 and KAP1 compete for Pax3 binding, and may represent a toggle for Pax3 function. Pax3 also directly interacts with Tbx1 subfamily of T-box proteins including Tbx18, 15, and 22 (Farin et al., 2008). All T-box genes contain T-domains, which are involved in DNA binding and protein interaction. The interaction between Pax3 and the Tbx proteins is through the paired- and T-domains. Although it is not known to what effect, if any, this binding has in melanocytes, several of the T-box proteins are expressed in the neural crest and the resulting pigmented cells. For example, mutations in Tbx15 results in an altered dorso-ventral pigmentation pattern, and Tbx2, a related protein in another T-box subfamily, acts as a transcriptional repressor of the tyrosinase-like protein 1 gene (Candille et al., 2004, Carreira et al., 1998). Pax3 contains some ability to repress transcription alone, through a protein moiety comprising of the first 90 amino acids of the protein that includes part of the paired domain (Chalepakis et al., 1994b). This inhibition of transcription is further enhanced through the interaction with potent co-repressors.

Protein interactions that modulate PAX3 function

There are several examples of protein interactions that inhibit Pax3 function and lead to cellular terminal differentiation. Usually, there is either physical interference preventing Pax3 from binding to DNA regulatory elements or a promotion of Pax3 protein degradation. For example, calmyrin, a calcium binding protein containing four putative EF-hand domains, actively blocks Pax3 binding to DNA (Hollenbach et al., 2002). Pax3 and calmyrin are co-expressed in undifferentiated myoblasts, and as the cells differentiate calmyrin levels increase while Pax3 levels decrease. Calmyrin binds directly to the Pax3 paired domain and inhibits both the transcriptional and DNA-binding activity of Pax3. It is not known if this interaction also exists in melanocytes and melanoblasts. Furthermore, the first two helices of the paired-like homeodomain of Pax3 directly bind to the N-terminal and pocket domain of Rb (retinoblastoma protein) and Rb-related protein family members p107 and p130 (Wiggan et al., 1998). The Rb-interacting epitope of the homeodomain is highly homologous to a region in E2F proteins, and the E2F-Rb interaction through this moiety is well characterized. While this interaction results in a block of Pax3 mediated activation of the Met promoter, the mechanism of how Rb inhibits Pax3 is not known. The Rb-Pax3 relationship is further complicated with the findings that Pax3 related proteins (Pax2, 5, and 8) also bind to Rb, and this interaction can result in transcriptional blockage, repression, or co-activation (Eberhard and Busslinger, 1999, Miccadei et al., 2005, Yuan et al., 2002). The downstream transcriptional consequences of direct binding of Rb and Pax factors may be dependent on the specific transcriptional target, cell type, and/or the phosphorylation state of Rb.

Pax3 also interacts directly with Daxx (death domain-associated protein)(Hollenbach et al., 1999), a protein originally identified through its interaction with the transmembrane death receptor FAS/CD95 in the cytoplasm, which has both pro- and anti-apoptotic functions. In the nucleus, Daxx acts as a repressor of transcription factors. The function of Daxx can be inhibited through binding to the promyelocytic leukemia protein (PML) and sequestration into nuclear bodies. Daxx binds to Pax3 through epitopes in the homeodomain and the octapeptide region and attenuates the ability of Pax3 to activate promoters. This inhibition of Pax3 by Daxx is blocked by the recruitment of Daxx to the nuclear bodies by PML (Lehembre et al., 2001). Daxx also inhibits a Pax3 direct downstream target gene, Met, although it is unknown if this is through a Pax3-dependent mechanism (Morozov et al., 2008).

HIRA represents another protein that directly interacts with Pax3 (Magnaghi et al., 1998). HIRA is a WD repeat-containing chromatin regulator that drives the formation of senescence associated heterochromatin foci (SAHF)(Zhang et al., 2005). The formation of this complex, which contains the heterochromatin-associated protein HP1, promotes cellular senescence by concurrently blocking genes that stimulate proliferation and by promoting cell cycle exit and entry to G₀. Both HIRA and HP1 are transiently recruited to PML bodies during the senescence process. Pax3 and HIRA interact through domains at the C terminal portions of both proteins. The consequences of this interaction are not known, although some clues can be gathered through what is known about other Pax3 protein interactions. Pax3 directly interacts with Grg4, which has some protein similarity to HIRA, and the Pax3-GRG4 complex acts as a repressor complex. In addition, both Pax3 and HIRA interact with HP1, and while Pax3 can recruit HP1 to promoters, this leads to an inhibition of Pax3 transcriptional activity. HIRA is also involved in the PML, and perhaps the Pax3-HIRA interaction parallels the inhibition caused by Daxx-Pax3 binding. However, the real function of Pax3 and HIRA is presently unknown.

Pax3 can also bind to Brn-2, although the downstream consequences of this interaction in cells are not characterized (Smit et al., 2000). This will be quite interesting if this interaction occurs in melanocytes and melanoma cells, due to the emerging role of Brn-2 in these cellular subtypes. The POU domain transcription factor Brn-2 (also called N-Oct3 and POU3F2) plays a role in central nervous system and neural crest development, and neuronal differentiation(Fujii and Hamada, 1993). While not usually expressed in melanocytes, Brn-2 is expressed abundantly in melanoma (Eisen et al., 1995, Sturm et al., 1994, Thomson et al., 1995). Brn-2 expression in melanocytes is also up-regulated by an active mutant form of B-Raf, V600E, found in more than half of spontaneous occurring melanomas (Wellbrock et al., 2008, Goodall et al., 2004). Brn-2 activates the expression of Mitf, through an enhancer just 5' to the response element for both Pax3 and Sox10. One possibility is that Brn-2 and Pax3 work together to regulate Mitf and other genes in normal melanocytes and melanoma cells.

Pax3 is regulated and degraded through monoubiquitination. While proteasomal degradation usually requires polyubiquinization and recognition by an ubiquitin receptor protein, Pax3 directly interacts with Rad23B as a monoubiquitiniated substrate target for proteasomal breakdown (Boutet et al., 2007). Rad23B acts as a bridge for a complex formation including both Pax3 and the intrinsic ubiquitin receptor protein S5a. Inhibition of this pathway in myoblasts results in a delay of muscle differentiation, including cellular fusion and the expression of markers suggesting that Pax3 maintains an undifferentiated state and that active degradation of Pax3 allows progression of terminal differentiation.

There are also other factors that interact with Pax3 but are not usually found in pigment cells. These include Msx1 (muscle segment homeobox 1) that, at least in the case of the MyoD gene, actively blocks Pax3 transcriptional activation through direct protein binding (Bendall et al., 1999). The interaction is mediated through the paired domain of Pax3 and the homeodomain of Msx1. Another set of factors, Mox1 and 2 (also known as Meox1 and 2, Mesenchyme homeobox) are mainly expressed in mesodermal structures. Pax3 binds to both of these factors through homeodomain interaction (Stamataki et al., 2001). Not all proteins that modulate Pax3 function actively inhibit the function of the Pax3 protein. Pax3 also directly interacts with Importin 13 (IPO13, also known as karyopherin 13 or Kap13)(Ploski et al., 2004). Pax3 interacts with IPO13 through the homeodomain and with basic amino acids just C-terminal to this domain. This interaction facilitates the nuclear import of Pax3. Since Pax3 is a potent transcriptional regulator, these protein interactions are essential for the melanocytes to modulate Pax3 function and alter gene expression and cell morphological states.

Downstream Targets of Pax3

Pax3 and the Neural Crest

A major function of Pax3 is to act as a transcription factor and regulate the expression of downstream genes (Table 2). This role of Pax3 is key during the development of the neural crest in the embryo. The neural crest consists of a population of cells that arise from a region dorsal to the neural tube and contributes to a diverse set of tissue types including but not limited to melanocytes, cardiac neural crest, dorsal root and sympathetic ganglia, and thymus (Auerbach, 1954) (Figure 2). The dependence of proper neural crest development on PAX3 function is clearly seen in humans and mice with deficient PAX3 gene expression levels. Heterozygous and homozygous mutations in the Pax3 gene result abnormalities in many neural crest-derived cell types (Epstein et al., 1991a, Epstein et al., 1991b, Franz et al., 1993, Tassabehji et al., 1993, Mansouri et al., 1996, Auerbach, 1954, Franz, 1990, Epstein et al., 1993, Goulding et al., 1993, Tajbakhsh et al., 1997, Goulding et al., 1994, Bober et al., 1994, Tremblay et al., 1995, Franz, 1989, Franz, 1993, Conway et al., 1997b, Conway et al., 1997a). Although the entire mechanism through which Pax3 exerts control over neural crest population development is not completely elucidated, at least one function is the direct control of key regulatory genes such as c-Ret, TGF-β2, and Wnt1.

Table 2.

Direct downstream targets of Pax3

Gene	Location of regulatory region*	Activity of Pax3	Functional Role	Binding sequence	Reference
Melanocytes
Mitf	-260 bp	synergistic activation with Sox10	melanocyte development	ATTAATACTACTGGAAC	(Watanabe et al., 1998, Bondurand et al., 2000)
Dct (Trp2)	−200 bp	repression	melanocyte development	TCATGTGCT	(Lang et al., 2005)
Trp1	−1 bp (MSEi) −237 bp (MSEu)	activation	melanocyte development	GTGTGA with 3’ AT-rich region	(Galibert et al., 1999)
Anti-apoptotic properties
Pten	undetermined	repression	resistance to apoptosis	ATTA	(Li et al., 2007)
Bcl-XL	−42 bp	activation	resistance to apoptosis	GGGGAAATTACACTAAA**	(Margue et al., 2000)
Neural crest
c-Ret	−3.3 kb	synergistic activation with Sox10	neural crest development	TGTCACACTGCC	(Lang et al., 2000, Lang and Epstein, 2003)
Wnt1	−130 bp	activation	neural crest development	CTCGC	(Fenby et al., 2008)
TGF-β2	−788 bp −247 bp −638 bp −1124 bp	activation	neural tube development	GTTAT CCTTG GTTCC ATTA	(Mayanil et al., 2006)
Other downstream targets
c-Met	−257 bp	activation	embryonic myogenesis	GGTCCCGCTT	(Epstein et al., 1996)
Fgfr4	19.152 kb	activation	embryonic myogenesis	GGGCCTGAAAT	(Lagha et al., 2008)
Myf-5	−57.5 kb	activation	embryonic myogenesis	AGGCATGACT	(Bajard et al., 2006)
MyoD	~4kb area approximately − 18 kb	activation	embryonic myogenesis	undetermined	(Woloshin et al., 1995, Bendall et al., 1999)
Msx2	−3.732 kb	repression	cardiac neural crest development	GTCACACA	(Kwang et al., 2002)
Hes1	55 bp	activation	embryonic neurogenesis	CCTTG	(Nakazaki et al., 2008)
Ngn2	−152 bp	activation	embryonic neurogenesis	CCTTG	(Nakazaki et al., 2008)
Sostdc1	−590 bp	activation	osteogenesis	AGAGACAAAA	(Wu et al., 2008)

^*site relative to transcriptional start.

^**homeodomain consensus sequence is underlined

Figure 2.

Pax3 is expressed during development in the migrating neural crest and in other structures. A. Schematic of an E12.5 embryo. Approximate anatomic location of the area where the section shown in (B) is located is boxed on the picture. B. Pax3 expression in a section of E12.5 embryo. Immunofluorescence reveals Pax3 expression in a restricted pattern limited to the medial-dorsal area of the neural tube, the dorsal root ganglia, myotome-derived migrating primitive myoblasts, and migrating neural crest cells (white arrowheads). Boxed area is shown in detail in (C). C. Neural crest cells are migrating from a region dorsal to the neural tube and under the ectoderm (white arrows). D. Schematic of photo shown in (B). Pax3-expression domains are shown in green. The neural crest (NC) cells migrate in two major populations; dorso-ventral traveling cells (NC (dv)) and cells that travel mediolaterally (NC (ml)) dorsal to the dorsal root ganglia (DRG) and the somites or the myotome cells (M). This later population is generally fated to become melanocytes.

The c-Ret tyrosine kinase receptor spans the cellular membrane of neural crest-derived cells and is essential for motility, cell survival, proliferation, and differentiation (de Groot et al., 2006). Early indicators of the existence of a link between Pax3 and c-ret derived from the study of c-ret expression in homozygous Splotch embryos. The Pax3-dependent neural crest gives rise to enteric ganglia, with loss of Pax3 leading to a deficiency of ganglia within the intestine. Reintroduction of Pax3, however, rescues enteric ganglia development and up-regulates c-Ret expression, indicating that c-Ret lies downstream of Pax3 in the genetic hierarchy. Pax3 directly binds to an enhancer element within the c-Ret promoter and synergistically activates this gene with Sox10 (Lang et al., 2000, Lang and Epstein, 2003) in a manner mediated by chromosome acetylation (Puppo et al., 2002). Pax3 stimulates the proper migration and development of enteric ganglia via regulation of c-Ret in this neural crest-derived population.

Transforming Growth Factor β2 (TGF-β2) is a member of the TGFβ superfamily of proteins, which is a key regulator of cell behavior influencing proliferation differentiation, motility and apoptosis (Soma T, 2002). The development of many neural crest derivatives also depend heavily upon the growth factor TFG-β2 since mice homozygous for a mutation in the TGF-β2 gene mimics some neurocristopathies with defects in the heart, craniofacial structures, skeleton, ear, and urogenital system (Sanford et al., 1997). Mouse embryos deficient in Pax3 revealed a marked decrease in TGF-β2 expression when compared to wild type embryos, suggestive of a functional relationship between these two factors. TGF-β2 proved to be a direct downstream target of Pax3 through the binding via both the Paired and Homeodomains to several cis regulatory elements within the TGF-β2 promoter (Mayanil et al., 2006). This regulatory relationship between Pax3 and TGF-β2 demonstrates that Pax3 expression in one cell can influence other cell populations through the activation of this potent secreted growth factor.

The Wnt proteins are involved in the induction of neural crest migration, proliferation and differentiation (Wu J, 2003). Wnt signaling takes part in neural crest development as evidenced by the diminished neural crest cell population in Wnt1 and Wnt3a double mutant mice (Ikeya et al., 1997, Dorsky et al., 1998). Wnt1 expression initially marks the dorsal neural tube and early migratory neural crest cells but then diminishes as these cells move away from the neural crest (Jiang et al., 2000). Mouse embryos lacking Pax3 display a reduction in Wnt1 expression in the dorsal neural tube, suggesting Wnt1 is downstream of Pax3 (Conway et al., 2000). The 5’ Wnt1 promoter possesses a Pax3 binding motif conserved between mice and humans. Pax3 uses these enhancer elements to activate Wnt1, implicating Pax3 as a direct regulator of Wnt1 in the development of neural crest populations (Fenby et al., 2008). In the neural crest, Pax3 regulates the expression of at least three key signaling molecules that facilitate the crest to migrate, survive, and proliferate.

Pax3 and Melanocytes

Melanocytes are the pigment-producing cells of the skin that are derived from the neural crest. Pax3 both promotes and inhibits melanogenesis within these cells through transcriptional regulation of Microphthalmia-associated transcription factor (Mitf), dopachrome tautomerase (Dct), and tyrosinase-related protein 1 (Tyrp1). Mitf represents a direct downstream target of Pax3. Mitf is often referred to as a ‘master regulator’ of melanogenesis, due to its ability to activate many melanocyte-specific genes and to regulate key cell fate decisions including differentiation or proliferation. Mitf activates the expression of the melanocyte differentiation genes Tyrosinase, Tyrp1 and Dct by binding through a M-box motif (Bertolotto et al., 1998, Aksan and Goding, 1998). Pax3 utilizes both the paired and homeodomains to bind to a cis regulatory enhancer located upstream of the Mitf transcriptional start site (Watanabe et al., 1998). Mitf gene activation occurs in a synergistic fashion when Sox10 and Pax3 bind to the consensus sites within its promoter (Bondurand et al., 2000, Potterf et al., 2000). Thus, Pax3 influences melanoblasts toward a melanocytic fate by initiating the expression of Mitf.

While Mitf activates Dct expression, Pax3 inhibits both Dct expression and the ability of Mitf to bind to the Dct promoter. Pax3 forms a repressor complex with Lef1 and Grg4 on the Dct enhancer sequence and actively blocks Mitf binding. When beta-catenin is present, Lef1 changes partners to form an activating complex composed of Lef1, Mitf, and beta-catenin, which displaces Pax3 from the Dct enhancer (Figure 3). Mitf can also synergistically activate Dct expression with Sox10 (Potterf et al., 2001, Jiao et al., 2004, Ludwig et al., 2004). Conversely, in a cell culture system, Sox10 did not synergistically activate Dct in combination with Pax3 (Lang et al., 2005).

Figure 3.

Canonical Wnt signaling affects downstream genes in melanogenesis. A. In the absence of Wnt signaling, Pax3, together with Sox10, can activate the expression of Mitf while simultaneously repressing Dct expression. Pax3 forms a transcriptional repressor complex with Lef1 and groucho-related proteins such as Grg4. There is also a Lef1 binding site in the Mitf promoter. B. Canonical Wnt signaling alters gene expression in melanocytes. Wnt signals activate beta-catenin (βcat) and, together with Lef1, upregulates expression of Mitf. Activated beta-catenin also forms an activator complex on the Dct promoter comprised of beta-catenin, Mitf, and Lef1 and displaces the repressor complex containing Pax3. Sox10 can also synergistically activate Dct with Mitf, and can bind to a site just 3' to the M site as well as several other sites further 5' of the enhancer region shown. Mitf can also activate other downstream targets either with Lef1 and beta-catenin (other loci A) or with beta-catenin alone (other loci B). For example, Mitf can activate both Tyrp1 and Tyrosinase with beta-catenin in the absence of Lef1. This figure is a summary of findings from several manuscripts (Lang et al., 2005, Schepsky et al., 2006, Takeda et al., 2000, Yasumoto et al., 2002, Jiao et al., 2004, Ludwig et al., 2004).

Pax3 further exerts its control over melanocyte development by influencing the activity of the tyrosinase-related protein-1 (Tyrp1) promoter. Investigation into the regulatory elements of the Tyrp1 promoter reveals several binding sites, including two melanocyte specific elements recognized by Pax3, which interacts with the consensus motif GTGTGA (Yavuzer and Goding, 1994). Pax3 drives Tyrp1 expression using both Pax3 recognition motifs to push the melanocytes toward melanin production (Galibert et al., 1999). Although the entire molecular pathway that modulates melanocyte development is not known, the regulation of these three genes represent part of the mechanism through which Pax3 influences melanocyte stem cells toward a melanocytic fate and ensures the proliferation of cells, thus sustaining this population.

Pax3: cell survival and growth

Pax3 also regulates down-stream genes to protect the cells from apoptosis. One demonstrated example is seen in rhabdomyosarcoma tumor cells, which express the Pax3/FKHR fusion protein and exhibit resistance to cell death. Reduction of Pax3/FKHR within these tumors results in significant cell death, while the introduction of ectopic Pax3 expression restores the resistance to apoptosis (Bernasconi et al., 1996). Cells lacking Pax3 are more prone to apoptosis when compared to cells with normal Pax3 gene function (Borycki et al., 1999). Mouse embryos without a functional Pax3 gene exhibit severe neural tube defects including exencephaly and spina bifida along with a high rate of apoptotic cell death in the unfused region of the neural tube, suggestive of a putative role for Pax3 in resistance against apoptosis (Phelan et al., 1997). This protection from apoptosis by elevated Pax3 expression in rhabdomyosarcoma cells correlates with upregulation of Bcl-X_L, a potent anti-apoptotic gene. Pax3 interacts with the enhancer element upstream of Bcl-X_L via the homeodomain to activate this gene (Margue et al., 2000). Consequently, Pax3 confers anti-apoptotic characteristics to cell in which it is expressed.

Pax3 additionally confers cellular resistance to apoptosis via the regulation of tumor suppressor protein PTEN. PTEN is involved in many pathways but is an integral negative modulator of the PI3K/AKT signal transduction pathway which regulates cell proliferation and resistance to apoptosis (Chow and Baker, 2006). Rhabdomyosarcoma cell lines expressing various levels of Pax3 display an inverse correlation with PTEN protein levels: an increase in Pax3 results in the down-regulation of PTEN while a decrease in Pax3 allows for PTEN levels to rise. An identical trend is evident with manipulation of the Pax3-FKHR fusion protein levels. Cells in which the Pax3-FKHR levels are knocked down exhibit an increase in apoptosis, suggesting that Pax3 may control cell survival by regulating PTEN activity. Pax3 interacts with a putative homeodomain binding motif within the PTEN promoter, providing a mechanism by which Pax3 modulates PTEN expression patterns and ultimately allows for resistance to apoptosis (Li et al., 2007).

Pax3 and Development

Pax3 plays a pivotal role during development and is critical for the proper formation of the human and mouse neural, cardiovascular, endocrine, and musculature systems. During development, Pax3 contributes to a wide array of different tissue types such as the central nervous system including the neural tube and the brain, skeletal muscle derived from the somites, and the neural crest (Goulding et al., 1991, Terzic and Saraga-Babic, 1999). The neural crest forms cells that will eventually comprise the peripheral nervous system including sensory and motor nerves as well as the enteric ganglia, adrenomedullary cells, cardiac smooth muscle and mesenchyme, and pigment cells of the skin, hair, and inner ear. In the mouse embryo, Pax3 expression begins at approximately embryonic day (E) 8.5, peaks between E9 and E12, declines starting at E13, and falls to low levels by E17. Expression of Pax3 during embryogenesis first appears at E8.5 in the dorsal neural groove and neural tube (Stoller et al., 2008, Goulding et al., 1991). On E9, Pax3 expression includes neuroepithelium near the closing region of the neural tube followed by expression in the nearby somites (Stoller et al., 2008, Terzic and Saraga-Babic, 1999). By E10, Pax3 expression extends throughout the entire length of the embryonic spinal cord but remains restricted to the dorsal half, and slowly diminishes through E13. Consistent with the important role Pax3 plays in neural crest development, Pax3 is expressed soon after the formation of the neural crest. The expression of Pax3 persists in the brain and spinal cord of the central nervous system (Terzic and Saraga-Babic, 1999) and in many neural crest-derived cell populations including melanocyte stem cells. This review focuses upon Pax3 in neural crest development and melanocytes, although a summary of several downstream targets involved in the development of other systems is provided in Table 1.

The neural crest is initiated between the neural plate and the non-neural ectoderm, but soon migrates from this location to other sites in the embryo (Le Douarin and Dupin, 2003) (Figure 2). The crest migrates as two major groups. The first travels dorsoventrally between the neural tube and the somites and forms neural structures. The neural crest migrating mediolaterally dorsal to the somites but under the superficial ectoderm are generally fated to be melanoblasts. Pax3 is expressed in both of these populations. The importance of Pax3 expression to the neural crest-derived melanocytes during development is apparent by the loss of pigmentation in humans and mice with mutations in the PAX3/pax3 gene. Initiation of the neural crest to a melanocyte lineage is Pax3 independent, but is necessary for the embryonic melanoblasts to expand their numbers and to restrict the precursor cells to a melanocyte fate (Hornyak et al., 2001). Mitf, a downstream Pax3 target, is also necessary for melanogenesis, by facilitating survival of the melanoblasts during migration. The melanoblast cells migrate to the skin into developing hair follicles. One of the hallmarks of the neural crest is its ability to migrate throughout the developing embryo and to give rise to a variety of cell types. The adult melanocyte retains some of the traits of their neural crest roots, since melanocyte stem cells can migrate to and from niches (Nishimura et al., 2002). It may also be a reason why melanoma converts easily to a metastatic state.

Pax3 and melanocyte stem cells

Melanocyte stem cells are a source of transient amplifying cells and differentiated melanocytes. Stem cells in the mature mouse skin are tightly regulated to maintain multipotency, resist apoptosis, and to be quiescent until daughter cells are needed. The melanocyte stem cells were first tagged in a mouse model using the Dct-LacZ transgene, which expresses beta-galactosidase in Dct-expressing cells in the skin (Nishimura et al., 2002). These transgene expressing cells have distinctive characteristics; they are phenotypically immature, slow growing, self-renewing, and able to give rise to differentiating and melanin-producing daughter cells.

In mice, these melanocyte stem cells reside in a specialized niche located in the lower permanent portion of the hair shaft called the hair follicle bulge region (Nishimura et al., 2002). The niche is home to several skin stem cell populations, including keratinocyte and follicular stem cells. When cells from the bulge region are isolated and grown in culture, they give rise to several cell lineages represented in adult skin (Sieber-Blum et al., 2004, Amoh et al., 2005, Oshima et al., 2001). The niche melanocyte stem cells still retain qualities of their neural crest roots, in terms of migratory ability and plasticity. The melanocyte stem cell can exit their original niche, migrate in the epidermis, and colonize a neighboring hair follicle bulge region (Nishimura et al., 2002). In this new home, the melanocyte stem cells can again provide pigment-producing progeny. Neural crest derived-cells isolated from both mouse and human skin have been termed skin-derived precursor cells (SKPs) and are multipotent with an ability to differentiate into neural and mesodermal cell lineages (Toma et al., 2001, Toma et al., 2005).

Both melanocyte stem cells and SKP defined cell populations express Pax3 (Lang et al., 2005, Osawa et al., 2005, Toma et al., 2005). Isolated melanocyte stem cells express both Dct and Pax3 and do not express several other melanocyte markers including Sox10, Mitf, Lef1, Kit, Tyrosinase, Tyrp1, Silver, and Mki67(Osawa et al., 2005). Of note, Pax3 and two other transcription factors, Sox10 and Mitf, are critical for melanocyte development and differentiation. These factors are part of the same pathway, where Pax3 and Sox10 activate the expression of Mitf (Bondurand et al., 2000, Potterf et al., 2000). This pathway may be part of a balance that toggles between a stem cell phenotype and an amplifying population of melanocytes. The Pax3(+), Sox10(−), Mitf (−) phenotype may promote the lineage-restricted stem cell phenotype, where the cells exist in an undifferentiated state, are generally quiescent, and resist apoptosis (Nishikawa and Osawa, 2007, Osawa et al., 2005).

The Wnt signaling pathway influences this molecular balance of Pax3, Sox10, and Mitf in melanocyte stem cells (Figure 3). There is evidence that Wnts, or its canonical downstream activator protein beta-catenin, can drive both the melanocyte stem cells toward differentiation or to escape from senescence. Some of the ability of beta-catenin to drive cells toward differentiation is through modulating the function of Pax3 and Mitf (Lang et al., 2005, Schepsky et al., 2006, Takeda et al., 2000). The prevention of differentiation by Pax3 is antagonized by the presence of beta-catenin. In the case of the Dct gene, Pax3 recruits the co-repressor Grg4 to an upstream enhancer and inhibits Dct expression (Lang et al., 2005). Paradoxically, both Pax3 and Dct are expressed in the melanocyte stem cell. However, if activated beta-catenin is present in the nucleus, a newly formed beta-catenin/Mitf/Lef1 activator complex displaces the Pax3/Grg4 repressor complex and allows expression of Dct. Inhibition of beta-catenin, either through gene deletion or through overexpression of the Wnt inhibitor DKK1, allowed the formation of immature melanoblasts but not melanocytes that expressed Dct (Lang et al., 2005). Hence, active beta-catenin is required for Dct expression in Pax3-expressing melanocyte precursors. Wnt signaling also activates Mitf expression directly, via a downstream signal of Wnt3a through a Lef1 binding site located 3' to the Pax3 response element in the Mitf promoter (Takeda et al., 2000). Mitf can also form a complex with Lef1, beta-catenin, or both and activate downstream genes such as Dct, Tyrp1, and tyrosinase (Schepsky et al., 2006, Yasumoto et al., 2002). The importance of repressing Wnt signaling for the melanocyte stem cell is the presence of several Wnt inhibitors present in the niche, including DKK3, Sfrp1, and Dab2 (Morris et al., 2004, Tumbar et al., 2004). The melanocyte stem cells also express several Wnt inhibitors themselves, such as DKK4, Sfrp1, Dab2 and Wif1 (Nishikawa and Osawa, 2007, Osawa et al., 2005). These experiments support a model where inhibition of Wnt signaling promotes the melanocyte stem cell phenotype and activation leads to melanocyte differentiation.

There is some evidence that Wnt signaling aids melanoblasts and maintains stem cells. For example, two Wnt factors, Wnt1 and 3a, are necessary for melanoblast development in the embryo (Ikeya et al., 1997, Dorsky et al., 1998). Pax3 can directly activate expression of the Wnt1 gene through a genomic enhancer, though in what cell type or context this occurs is unknown (Fenby et al., 2008). In addition, overexpression of an activated form of beta-catenin in Tyrosinase-expressing melanocytes promotes immortalization through repression of the p16^Ink4a, although this is coupled with reduced number of melanoblasts (Delmas et al., 2007). Although these experiments are only indirectly related to the melanocyte stem cell, they suggest that there is a differential response to Wnt signaling by the melanocyte, and this can be modulated through overall levels of Wnts and Wnt inhibitors. Coupled with the findings that median levels of Mitf leads to melanocyte proliferation while higher levels lead to differentiation (Gray-Schopfer et al., 2007), a possible model, albeit a simplistic one, for melanocyte stem cells can be created. This Wnt-Pax3-Mitf model is that low levels of Wnts support quiescent melanocyte stem cells that are Pax3(+), Mitf(−), median levels of Wnts lead to the expansion of daughter cells that are Pax3(+), Mitf(+), and high levels of Wnts promotes the differentiation of melanocytes that are Pax3(−), Mitf(+). This model still needs experimental support to further support this hypothesis, or lead to an alternate model. What is clear, however, is that Wnt signaling plays an important role in regulating melanocyte stem cells.

Pax3 and disease

Mutations or dysregulation of the Pax3 gene provide, on the one hand, valuable insights into how essential Pax3 is during many processes of embryonic development, but also cause deleterious syndromes and links with diseases. The importance of Pax3 is demonstrated by either the loss of Pax3, as exhibited in murine Splotch mice and Waardenburg syndromes I and III, while hyperactivity of Pax3 is exhibited in melanomas and translocation events as found in rhabdomyosarcoma.

The heterozygous Splotch mouse possesses a partial loss of functional Pax3 and therefore exhibits a white belly spot on the abdomen, tail and feet (Mansouri et al., 1994, Mansouri et al., 1996, Auerbach, 1954, Russell and Roscoe, 1947). This phenotype results from a semidominant mutation within the Pax3 gene (Epstein et al., 1991b, Chalepakis et al., 1994a, Epstein et al., 1993). The Splotch locus and the Pax3 locus (Goulding et al., 1991), originally mapped closely together on chromosome 1 in mouse (Snell et al., 1954), were later identified as the same gene (Epstein et al., 1991b, Epstein et al., 1993). Six Splotch mutants exist resulting from different mutations or deletions of Pax3 that present a range of phenotypic severity. The original Splotch (Sp) (Russell and Roscoe, 1947) and the Splotch delayed (Sp^d) resulted from random mutations within Pax3, whereas the Sp retarded (Sp^r), Sp^1H, Sp^2H, and Sp^4H mutants developed from X-irradiation (Dickie, 1964, Vogan et al., 1993, Epstein et al., 1991b, Epstein et al., 1993, Goulding et al., 1993). Splotch mice possess a Pax3 mutation in the third intron producing four alternate transcripts, three of which result in truncated mRNAs with incomplete or missing domains due to early termination. The final transcript lacks the end of the paired domain and the octapeptide as exon 4 gets spliced out (Epstein et al., 1993). The homozygous Sp^2H mutation results from the deletion of 32bp within exon3 and results in death by day E16 (Epstein et al., 1991b). Mouse homozygotes die midgestation due to cardiac neural crest defect but also display anomalies in the neural tube and neural crest derivatives (Mansouri et al., 1994, Mansouri et al., 1996). The neural crest derivatives that suffer developmental difficulties include the central nervous system, melanocytes, dorsal root and sympathetic ganglia, Schwann cells (Franz, 1989, Franz, 1993), thymus, thyroid (Conway et al., 1997b, Conway et al., 1997a), and cardiac tissue such as the aortic arches (Franz, 1989, Franz, 1993, Conway et al., 1997a), as well as myotome-derived limb muscles (Epstein et al., 1991a, Epstein et al., 1991b, Franz et al., 1993, Tassabehji et al., 1993, Mansouri et al., 1996, Auerbach, 1954, Franz, 1990, Epstein et al., 1993, Goulding et al., 1993, Tajbakhsh et al., 1997, Goulding et al., 1994, Bober et al., 1994, Tremblay et al., 1995). Pigmentation abnormalities arise due to inefficient proliferation and migration of melanocyte precursors, the melanoblasts (Henderson et al 1997). Splotch (Sp), Sp^1H, and Sp^2H mutants share similar phenotypes, which suggests that they result from Pax3 loss of function (Epstein et al., 1993, Vogan et al., 1993). The Splotch-delayed (Sp^d) homozygous mouse develops as a result of a spontaneous glycine to arginine transversion within the paired domain that still allows for a full length protein (Vogan et al., 1993). Homozygous Sp^d leads to spina bifida and allows survival until birth (Dickie, 1964). Examination of the homozygous Sp^d embryo reveals that neural tube defects are restricted to the posterior end versus the entire neural tube as in Sp, Sp^1H, and Sp^2H mutants (Vogan et al., 1993, Dickie, 1964). Sp^d mutants display a reduction in size and number of spinal ganglia unlike Sp mice, which possess no spinal ganglia (Vogan et al., 1993, Moase and Trasler, 1989, Moase and Trasler, 1990). Sp^d mice also exhibit a decrease in limb bud muscle primordia but still develop the septum of the truncus arteriosis in the heart (Franz, 1993, Franz, 1989). The reduced severity displayed by Sp^d mutants when compared to Sp mutants suggests that Sp^d is the result in a partial loss of function. The homozygous Splotch-retarded mutant (Sp^r) presents the most severe phenotype and causes the embryo to die prior to implantation. Sp^r results from a large chromosomal deletion of between 14 and 16 centimorgans including all of Pax3 (Epstein et al., 1991b). Heterozygous Sp^r mice display pigmentation problems and growth delays (Vogan et al., 1993). Similar to Sp^r, Sp^4H arises from a complete deletion of Pax3 (Epstein et al., 1991b, Epstein et al., 1991a, Goulding et al., 1993, Vogan et al., 1993) and results in embryonic death immediately following implantation (Goulding et al., 1993). The various types of Splotch mice demonstrate the essential roles Pax3 plays within embryonic development with loss of Pax3 affecting many neural crest derivatives.

Waardenburg syndrome (WS) type I and type III affects humans possessing a mutation in Pax3 resulting in several phenotypes including abnormalities of the central nervous system, eye and nose as well as pigmentation abnormalities affecting skin, hair and otic pigment cells important for normal hearing (Tassabehji et al., 1992). The similarities between the WS phenotypes in humans and those in the Splotch mouse caused early speculation that these two syndromes involved either the same gene orthologues or similar molecular pathways (Asher and Friedman, 1990, Foy et al., 1990). The identification of other WS patients and linkage studies lead to verification that WS patients possess mutations within Pax3 and that this gene is located on chromosome 2 (Ishikiriyama et al., 1989, Foy et al., 1990, Farrer et al., 1992, Hoth et al., 1993, Tassabehji et al., 1992). There are four recognized types of WS but only WS type I (WS I) and WS type III (WS III) result from mutations within Pax3. The symptoms of WS I include sensorineural hearing loss, lateral displacement of the inner corner of the eye, a broad nasal bridge, and pigmentation defects resulting in early hair graying, heterochromia of the irises, and patchy hypopigmentation of the skin (Waardenburg, 1951, Hoth et al., 1993, Gad et al., 2008, Hageman and Delleman, 1977). Several different mutations within Pax3 result in WS. However, most of these mutations occur as missense or frameshift mutations within the highly conserved region of exon 2 that gives rise to part of the paired domain (Baldwin et al., 1992, Hoth, 1993 #166, Tassabehji et al., 1992, Morell et al., 1992). The amino acid change that occurs in these mutants alters the DNA binding affinity of the paired domain that most likely causes the loss of function. WS III, also known as Klein-Waardenburg syndrome, presents all of the phenotypes exhibited in WS I but also gives rise to musculoskeletal abnormalities. WS III also results from mutations within the paired domain of Pax3 including deletions and missense mutations within the paired domain (Hoth et al., 1993, Tassabehji et al., 1995, Zlotogora et al., 1995, Read and Newton, 1997, Tekin et al., 2001, Wollnik et al., 2003). The reduction of functional Pax3 within humans, as demonstrated by Waardenburg syndrome types I and III, results in anomalies in several neural crest-derived cell populations.

Dysregulation of the Pax3 gene can also have deleterious effects in childhood development. Alveolar rhabdomyosarcoma (aRMS) is a common childhood soft-tissue tumor derived from skeletal muscle precursor mesenchymal cells but sometimes appears in locations atypical of skeletal muscle development (McLean and Castellino, 2008, Gallego Melcon and Sanchez de Toledo Codina, 2007). Most of these tumors express a chimeric gene resulting from a translocation event between chromosome 2 and chromosome 13. Examination of this reciprocal translocation results in a gene t(2;13)(q35,q14) that creates a protein with the N-terminal end of Pax3 including both the paired domain and the homeobox domain (Barr et al., 1993) fused to the C-terminal portion of the fork head DNA binding domain and the transactivation domain of a fork head family protein FOXO1 (formerly FKHR)(Davis et al., 1995, Barr, 1997, Barr, 2001, Galili et al., 1993, Shapiro et al., 1993, Barr et al., 1998). A high percentage of alveolar rhabdomyosarcomas express this 97 kDa protein with DNA-binding abilities similar to Pax3 but acts as a stronger transcription factor than either Pax3 or FOXO1(Bennicelli et al., 1995, Ludolph and Konieczny, 1995, Fredericks et al., 1995, Galili et al., 1993). Pax3-FOXO1 confers oncogenic potential to cells in which it is expressed by dysregulating genes involved in differentiation, proliferation, apoptosis, and metastasis. Expression of Pax3-FOXO1 within myoblasts cultured in differentiation medium effectively prevents differentiation into myotubes. The down-regulation of p57KIP2 may provide a mechanism through which Pax3-FOXO1 prevents differentiation and promotes proliferation. In normal myogenic precursor cells, p57KIP2 prevents cell cycle progression by inhibiting cyclin E-CDK2 and promotes cell myogenic differentiation by stabilizing MyoD1, a protein necessary for myogenesis (Roeb et al., 2007). The Pax3-FOXO1-dependent downregulation of p57KIP2 would thus provide favorable conditions for uncontrolled growth of undifferentiated cells. Pax3-FOXO1 may provide the tumorigenic characteristics to alveolar rhabdomyosarcomas since ectopic expression increases cell proliferation rates (Anderson et al., 2001), while down-regulation of this chimeric gene slows proliferation rates(Kikuchi et al., 2008). Pax3-FOXO1 represents a viable candidate for stimulating proliferation and metastasis in tumors since c-Met lies downstream of Pax3-FOXO1(Ginsberg et al., 1998). The Pax3-FOXO1 chimera may also aid cancerous cells in avoiding apoptosis in order to propagate. Reduction of Pax3 and Pax3-FOXO1 expression reveals an increase in cell death, suggesting that the expression of this chimeric gene may provide resistance to apoptosis (Bernasconi et al., 1996). Expression of Pax3 or Pax3-FKHR corresponds to an increase of the anti-apoptotic genes BCL-X_L (Margue et al., 2000) and TFAP2B (Ebauer et al., 2007), further supporting the anti-apoptotic role of Pax3-FOXO1. Knockdown of Pax3-FOXO1 transcripts in aRMS cells corresponds to reduced cell motility, allows differentiation to proceed and decreases proliferation rates (Kikuchi et al., 2008). Despite the array of genes manipulated by Pax3-FOXO1 that implicate this fusion gene in providing oncogenic characters to cells, introduction of Pax3-FOXO1 into an animal model is not sufficient to induce tumor formation (Scheidler et al., 1996, Anderson et al., 2001, Lagutina et al., 2002). This suggests that Pax3-FOXO1 must work in concert with other mutations to direct tumor growth(Keller et al., 2004).

Corruption of the mechanism through which Pax3 maintains control over melanocyte differentiation plays a vital role in the development of melanoma. Melanoblasts express Pax3 until the onset of differentiation into melanocytes when Pax3 expression diminishes, suggesting that Pax3 moves melanocyte stem cells toward a melanocytic fate while preventing terminal differentiation (Scholl et al., 2001, Lang et al., 2005). However, most primary melanoma tumors and melanoma cell lines overexpress Pax3 (Plummer et al., 2008, Muratovska et al., 2003, Parker et al., 2004, Barr et al., 1999, Scholl et al., 2001, He et al., 2005) with high Pax3 levels indicative of more aggressive melanomas (Ryu et al., 2007). Reduction of Pax3 transcripts within melanoma cell lines reveals the importance of Pax3 to melanoma cell survival since this reduction results in an increase in cell death from apoptosis (Scholl et al., 2001, He et al., 2005, Muratovska et al., 2003, Bernasconi et al., 1996). Several possible mechanisms exist through which Pax3 overexpression promotes a melanoma phenotype. The down-regulation of Pax3 corresponds with an increase in caspase 3 (a protein involved in programmed cell death) and p53 tumor suppressor protein (He et al., 2005). The increase in apoptosis exhibited by mouse embryos without functional Pax3 is partially rescued when heterozygous Splotch mice are crossed with p53 mutant mice (Pani et al., 2002). This suggests that Pax3 overexpression may allow melanoma to persist and avoid cell death by suppressing caspase 3 and p53 expression. Pax3 may also prevent apoptosis by regulating Bcl-X_L expression. Cell lines treated with reduced Pax3 expression undergo apoptosis at a higher rate but are rescued with overexpression of Pax3 or Bcl-X_L(Margue et al., 2000). These characteristics of Pax3 expression in melanoma suggest that hyperactivity of Pax3 may recapitulate a stem cell state to these cells, allowing for tumor formation and immortalization.

Summary

Pax3 function is essential for normal embryonic melanogenesis and for the melanocyte stem cell. Regulatory proteins dictate the development and maintenance of the melanocyte state, and several of these are direct down-stream targets of Pax3 (Figure 4). These factors regulate qualities essential for melanocyte development and preservation of the stem cell state, such as resistance to apoptosis, the ability to migrate, cellular lineage specificity, and a block of terminal differentiation. In order for the cells to expand and survive, the melanocyte stem cells require resistance to apoptosis, and activation of anti-apoptotic genes such as Bcl-X_L and a repression of PTEN by Pax3 provide resistance to cell death. Pax3 also aids in the ability for melanoblasts to travel to distant locations in the body and for melanocyte stem cells to migrate to and from niches by activating receptor and ligand genes such as Ret, Wnt1, TGF-β2, and Met. In addition, Pax3 contributes in specifying a specific tissue lineage by activating cell-type specific genes such as Mitf (in melanocytes) as well as MyoD and myf5 (in myoblasts) while concurrently blocking terminal differentiation. Pax3 can block differentiation directly by inhibiting the expression of differentiation factors such as Dct in melanocytes. Pax3 can further obstruct differentiation by activating inhibitory proteins, such as Sostdc1 (sclerostin domain-containing protein 1, also known as ectodin) (Wu et al., 2008). Sostdc1 acts a soluble inhibitor of bone morphogenic protein (BMP), and actively blocks BMP-induced differentiation. Although portions of the mechanism through which Pax3 regulates such diverse functions are beginning to emerge, much work remains. For example, the list of downstream targets provided in this review only includes genes with demonstrated direct Pax3 binding to enhancer regions. There are, however, a large number of putative downstream genes that appear to be modulated by Pax3 but have not been confirmed as direct Pax3 targets. Elucidation of the complex molecular pathways that allow Pax3 to regulate such a diverse set of functions will provide insight into the intricacies of development, stem cell regulation, and disease.

Figure 4.

Summary of Pax3 function during melanocyte development and in the melanocyte stem cell. Cellular responses are listed, and examples of downstream targets regulating these traits are in parentheses. Protein binding partners, such as transcriptional coactivators and repressors as well as modifying proteins can inhibit or enhance the function of Pax3.

Table 3.

Splotch Variants and Phenotypes

Splotch Variant	Notation	Mutation	Phenotype	Morbidity
Splotch (original)	Sp	-Random -Loss of function	no spinal ganglia, cardiac neural crest defect, limb muscle defect, neural tube and neural crest anomalies (CNS, melanocytes, dorsal root and sympathetic ganglia, Schwann cells, thymus, thyroid)	Lethal midgestation
Splotch delayed	Spd	-Random -Partial loss of function	spina bifida, reduced size of spinal ganglia, decreased limb muscle primordia, growth delays	Survival to birth
Splotch retarded	Spr	-X-irradiation -Loss of function	death	Death pre- implantation
Splotch1H	Sp1H	-X-irradiation -Loss of function	cardiac neural crest defect, limb muscle defect, neural tube and neural crest anomalies (CNS, melanocytes, dorsal root and sympathetic ganglia, Schwann cells, thymus, thyroid)	Lethal midgestation
Splotch2H	Sp2H	-X-irradiation -Loss of function	cardiac neural crest defect, limb muscle defect, neural tube and neural crest anomalies (CNS, melanocytes, dorsal root and sympathetic ganglia, Schwann cells, thymus, thyroid)	Lethal midgestation
Splotch4H	Sp4H	-X-irradiation -Loss of function	death	Death at implantation

References: (Moase and Trasler, 1989, Moase and Trasler, 1990, Russell and Roscoe, 1947, Dickie, 1964, Vogan et al., 1993, Epstein et al., 1991b, Epstein et al., 1993, Goulding et al., 1993, Mansouri et al., 1994, Mansouri et al., 1996, Franz, 1989, Franz, 1993, Franz et al., 1993, Conway et al., 1997a, Conway et al., 1997b, Epstein et al., 1991a, Tassabehji et al., 1993, Auerbach, 1954, Franz, 1990, Tajbakhsh et al., 1997, Goulding et al., 1994, Bober et al., 1994, Tremblay et al., 1995)