3.1.1. Ascl1 in the Sympathoadrenal Lineage

Ascl1 also marks the developing sympathoadrenal NC lineage, including thyroid C cells (NC-derived neuroendocrine cells with neuronal and serotonergic characteristics) and adrenal medulla chromaffin cells. Thyroid C cells and serotonergic enteric neurons may arise from a common Ascl1-expressing sympathoadrenal progenitor, and Ascl1, in addition to being required for differentiation of serotonergic enteric neurons, is needed to establish the C-cell phenotype; Ascl1-null mice have a greatly reduced number of C cells (Lanigan et al., 1998). Cultured rat C cells initially express Ascl1, but it quickly becomes undetectable after the C cells are placed in culture, consistent with observed neuronal differentiation (Clark et al., 1995). Murine Ascl1 is required for general chromaffin cell differentiation and catecholaminergic differentiation (Huber et al., 2002a; Huber et al., 2002b). During early development, the number of adrenal medullary Phox2b-positive progenitor cells is initially unaffected in Ascl1-null mice, but by birth, mutants only have one third the amount present in wild type, and tyrosine hydroxylase (Th)-expressing (catecholaminergic) cells are greatly reduced. Most cells in the adrenal medulla of Ascl1-null mice do not contain chromaffin granules and have an immature, neuroblast-like phenotype, with very few ever expressing the distinguishing chromaffin cell gene, Pnmt (Huber et al., 2002a; Huber et al., 2002b; Huber et al., 2005). Both Ascl1 and Phox2b are expressed by sympathoadrenal progenitors, but these two proteins are required independently and have distinct requirements for development of chromaffin cells and sympathetic neurons (Huber et al., 2005; Unsicker et al., 2005).

3.1.2. Regulation of Ascl1 by BMP, cAMP, and Notch Signaling

Ascl1, with Hand2, Phox2a, Phox2b, and Gata2/Gata3, is induced in NC-derived neuronal progenitors by BMPs (secreted from tissues to which autonomic neuron progenitors migrate) during development of sympathetic neurons (Anderson et al., 1997; Lo et al., 1998; Lucas et al., 2006; Pisano et al., 2000; Schneider et al., 1999). In cultured NCSCs, Bmp2 and Bmp4 induce Ascl1 and promote autonomic neuronal differentiation (Greenwood et al., 1999; Lo et al., 1997; Lo et al., 1998; McPherson et al., 2000). Expression of Ascl1 in postmigratory NCCs allows nonneuronal cells to remain competent for neuronal differentiation in the presence of Bmp2, and Bmp maintains Ascl1 expression (Lo et al., 1997). Constitutive Ascl1 expression maintains competence of NC progenitors to respond to later Bmp2 signals and undergo neuronal differentiation (Lo et al., 1997). Bmp2 promotes development of cultured primary NCCs toward the sympathoadrenal lineage, including induction of Ascl1, Phox2a, Phox2b, and the catecholaminergic markers Th and catecholamines (Bilodeau et al., 2001). Bmp2, Bmp4, Bmp7, and elevated cAMP synergistically stimulate development of the sympathoadrenal lineage in NCCs. Together with Bmp2, moderate activation of cAMP signaling promotes sympathoadrenal cell development and expression of Phox2a, a sympathoadrenal lineage-determining gene. In contrast, strong activation of cAMP signaling represses sympathoadrenal cell development and Ascl1, Phox2a, and Phox2b (Bilodeau et al., 2000). In murine NCSCs, Ascl1 is more sensitive to Notch-mediated inhibition of neurogenesis and cell cycle arrest than the sensory lineage neuronal genes Neurogenin1 (Neurog1) and Neurog2, suggesting a differential role for these genes regulated by Notch signaling (Lo et al., 2002). Ascl1 is suppressed by Notch signaling, a repression necessary to maintain ENS progenitors by restricting NC neuronal differentiation. The number of Ascl1-expressing enteric NCCs is increased in Protein O-fucosyl transferase (Pofut)-null embryos lacking this enzyme involved in Notch cleavage (Okamura and Saga, 2008).

3.1.3. Ascl1 Interaction with Other Transcription Factors and Targets

There is evidence of considerable interplay between factors of the regulatory network defining/determining autonomic neurogenesis that includes Ascl1, Phox2a, Phox2b, Hand2, and Gata2/3. In the ENS and sympathetic ganglia, Phox2b is needed to maintain Ascl1 expression (Pattyn et al., 1999), and Phox2b upregulates expression of Ascl1 in combination with Nkx2.2 (Dubreuil et al., 2002). Although Phox2a and Ret are induced by Bmp2, constitutive Ascl1 in NCSCs can induce these factors without Bmp2 induction. Phox2a in autonomic ganglia is strongly reduced in Ascl1-null embryos (Lo et al., 1998). The Phox2a promoter has an E-box element that can be bound by Ascl1 (which, as described above, is induced by BMPs) near elements activated by components of the cAMP signaling pathway, suggesting a possible mechanism for synergy between BMP and cAMP signaling (Benjanirut et al., 2006). Sox10 is required in vivo for induction of Ascl1 and Phox2b (Kim et al., 2003), but Ascl1 is a strong repressor of Sox10, indicating a negative feedback loop that allows Ascl1-expressing cells to continue neuronal differentiation (Okamura and Saga, 2008). In vitro, Ascl1 activates the Ret promoter, important for intestinal innervation (Bachetti et al., 2005). The noradrenergic marker genes Th and Dbh are also downstream of Ascl1 (Schneider et al., 1999).

3.1.4. Ascl1 in Human Disease

Congenital central hypoventilation syndrome (CCHS) is often associated with other ANS dysfunction, suggesting involvement of genes expressed broadly in the ANS, such as those in the Ascl1-Phox-Ret pathway. A rare heterozygous nucleotide substitution in Ascl1 was found in three patients with CCHS. Reproduction of these mutant Ascl1 alleles in an in vitro model of NCC noradrenergic differentiation disrupted noradrenergic neurogenesis (de Pontual et al., 2003). Human neuroblastomas, sympathetic NC-derived tumors, often express Ascl1 (about two thirds of all neuroblastomas) (Arvidsson et al., 2005; Gestblom et al., 1999; Ichimiya et al., 2001). In differentiating neuroblastoma cells, Ascl1 is downregulated with a corresponding upregulation of the transcriptional repressor Hes1, known to bind the Ascl1 promoter (Jogi et al., 2002). Ascl1 transcripts are also downregulated in several neuroblastoma lines before RA-induced differentiation, but neuroblastoma cell lines without endogenous Ascl1 do not respond to RA, suggesting the downregulation of Ascl1 may be an important step in promoting neuronal differentiation of neuroblastomas (Ichimiya et al., 2001). Ascl1 is also expressed in pheochromocytomas, NC-derived adrenal medullary thyroid tumors, and small cell lung cancers, all of which have characteristics of both neuronal and endocrine cells (Ball et al., 1993; Cairns et al., 1997).

3.3.1. Hand2 in the Pharyngeal Arches and Derivatives

Hand2, which is expressed in mesoderm- and NC-derived mesenchyme, has been identified as a key regulator of pathways necessary for pharyngeal arch morphogenesis (Clouthier et al., 2000; Ruest et al., 2003). NCCs entering the pharyngeal arches first migrate from the NT to the arches, proliferate, and finally undergo differentiation into terminal structures such as the mandibular skeleton and surrounding connective tissue (Clouthier et al., 2000). A Hand2 pharyngeal arch enhancer driving expression of Cre recombinase activates a ROSA26R^lacZ reporter recapitulating pharyngeal arch expression. Labeled cells are first detected in postmigratory cranial NCCs within the pharyngeal arches and later in their derivatives (Ruest et al., 2003). In Hand2-null mouse embryos, the first and second arches are hypoplastic because of increased apoptosis and the third and fourth arches fail to form, but migration of NCCs appears normal. Msx1, one important downstream effector of Hand2, was undetectable in the mesenchyme of Hand2-null pharyngeal arches.

Interactions between the NC-derived mesenchyme and surrounding cells of the pharyngeal arches are crucial for development. An important early pathway controlling Hand2 expression in the distal first pharyngeal arch mesenchyme is the endothelin-1 (Edn1) signaling pathway. Mouse embryos mutant for Edn1, normally expressed in pharyngeal arch epithelia and the mesoderm-derived pharyngeal arch core (Fukuhara et al., 2004; Thomas et al., 1998), have a phenotype similar to Hand2-null embryos, and both Hand2 and Hand1 are downregulated in the pharyngeal and aortic arches of Edn1-null embryos. The Hand2 locus contains an Edn1-dependent pharyngeal arch enhancer with four essential homeodomain binding sites. The homeodomain transcription factor Dlx6 binds these sites in an Edn1-dependent manner, and is downregulated in pharyngeal arches from EdnrA-null embryos (Charite et al., 2001; Fukuhara et al., 2004), demonstrating that Dlx6 acts as an intermediary between Edn1 signaling and Hand2 transcription (Charite et al., 2001). The G-protein-coupled receptor for Edn1, Ednra, is expressed in NC-derived mesenchyme. Ednra-null mice and Dlx5- or Dlx6-null mice are born with severe craniofacial defects resulting in neonatal lethality, including a homeotic transformation of mandibular-to-maxillary-like structures due to expansion of proximal first arch genes into the distal zones (Clouthier et al., 2000; Ruest et al., 2004). Constitutive activation of Ednra or misexpression of Hand2 in the Ednra domain induced the reverse: a transformation of maxillary-to-mandibular-like structures (Sato et al., 2008). Although NCC migration in Ednra-null embryos appears normal, there is complete loss or reduced expression of Hand2, Hand1, Goosecoid, Dlx2, Dlx3, Dlx6, and Barx1 in the postmigratory NC-derived arch mesenchyme, resulting in hypoplastic arches, differentiation defects, and apoptosis of some mesenchymal cells, much like the phenotype observed in Hand2-null embryos (Clouthier et al., 2000). A small distal Hand2 expression domain remains in Ednra-null mutants, and these cells likely contribute to formation of mostly normal lower incisors (Ruest et al., 2004). Mice deficient for the intracellular mediators of Edn1 signaling Ga(q) and Ga(11) also have reduced pharyngeal arch expression of Dlx3, Dlx6, Hand2, and Hand1, but not Msx1 (Ivey et al., 2003). The defects observed when any part of this Edn1-Ednra-Dlx6-Hand2 pathway is disrupted are reminiscent of those seen in human 22q11.2 deletion / velocardiofacial syndrome (Thomas et al., 1998).

By E9.5, Hand2 and Dlx6 expression in the mandibular mesenchyme is no longer dependent on the above-described endothelin signaling pathway and is at least partly mediated by FGF signals (Fukuhara et al., 2004). In both mice and zebrafish, Hand2 expression in the pharyngeal arches also requires Mef2c (Miller et al., 2007; Verzi et al., 2007). Zebrafish Mef2c mutants have pharyngeal arch phenotypes resembling zebrafish Edn1 partial loss-of-function mutants, and Mef2c is needed for expression of other genes downstream of Edn1 such as Dlx5, Dlx6, Barx1, and Goosecoid. Mef2c cranial NC expression does not require Edn1 signaling and Mef2c interacts genetically with Edn1, suggesting that Mef2ca functions within the Edn1 pathway (Miller et al., 2007). Hand2 and Runx2, a master regulator of osteogenesis, are partially overlapping in their expression domains in the developing mandible. Hand2, which is downregulated before osteoblast differentiation, negatively regulates intramembranous ossification of the mandible by physically interacting with Runx2 to inhibit Runx2 DNA binding and transcriptional activity. Mice with a hypomorphic mutation in Hand2 have reduced mandibular mineralization and ectopic bone formation due to precocious osteoblast differentiation (Funato et al., 2009). A screen for genes dependent on Hand2 for expression identified Ufd1l, involved in degradation of ubiquitinylated proteins. Human Ufd1l maps to human 22q11 and was deleted in patients with 22q11 deletion syndrome, and mouse Ufd11 is specifically expressed in those tissues affected in patients with 22q11 deletion (Yamagishi et al., 1999). The pathway including Ufd1l appears to be distinct from the Edn1 pathway, suggesting that Hand2 is involved in at least three parallel pathways needed for pharyngeal arch morphogenesis. An additional downstream target of Hand2 in the pharyngeal arches is Nebulette (Nebl), an actin-binding protein in the fetal heart, the human ortholog of which is deleted in DiGeorge syndrome 2 patients with cardiac and craniofacial abnormalities (Villanueva et al., 2002).

3.3.2. Hand2 in the Heart

In the developing heart, Hand2 is first expressed in mesoderm-derived cardiac precursors in the cardiac crescent and linear heart tube, becomes restricted to the developing right ventricle as heart looping occurs, and is expressed in the pharyngeal arch NC contributing to craniofacial structures and aortic arch arteries (McFadden et al., 2000). Hand2-null mouse embryos die between E9.5 and E10.5 (Hendershot et al., 2008; Srivastava et al., 1997), but some assessment of the role of Hand2 in early heart development can be made based on the mutant phenotype. Hand2 is required for the formation of both the mesoderm-derived right ventricle and the NC-derived aortic arch arteries; both tissues are hypoplastic in mutants (Srivastava et al., 1997; Srivastava, 1999b). A Hand2-lacZ transgene containing lacZ under the control of a 1.5-kb cardiac and NC-specific enhancer recapitulates endogenous Hand2 expression in the heart (McFadden et al., 2000). When Hand2 is deleted specifically in the NC (in Hand2^flox/flox; Wnt1-Cre embryos), embryos die at E12.5 with severe cardiovascular and facial defects and all NC-derived Hand2-expressing tissues seem to be affected (Hendershot et al., 2008; Morikawa et al., 2007). The early lethality in these embryos is due to loss of norepinephrine synthesis and can be rescued by activating adrenergic receptors (Morikawa and Cserjesi, 2008). In rescued embryos, loss of Hand2 in the NC lineage leads to NC-related OFT and aortic arch artery defects including pulmonary stenosis, arch interruptions, retroesophageal right subclavian artery, and ventricular septal defects (Morikawa and Cserjesi, 2008). Hand2 functions in part by regulating signaling from the cardiac NC to other cardiac lineages but does not affect cardiac NC migration or survival. Loss of Hand2 in NC also indicates the role it plays within the cardiac NC to regulate differentiation and proliferation of the second heart field (SHF)-derived myocardium (Morikawa and Cserjesi, 2008). Hand2 also plays a role in preventing apoptosis of the SHF cells; an increase in ectopic SHF apoptosis in compound Msx1-null, Msx2-null embryos is associated with reduced expression of Hand1 and Hand2, although it is unclear whether Hand reduction occurs in the NC- or mesoderm-derived cells (Chen et al., 2007).

In humans, there may be a link between Hand2, which maps to human 4q33, and cardiovascular malformations commonly associated with terminal deletions of chromosome 4q, but strong evidence for this has not been demonstrated (Huang et al., 2002). In zebrafish, Hand2 expression is detected in the earliest precursors of all lateral mesoderm at early gastrula. Hand2 is later expressed in lateral precardiac mesoderm, pharyngeal arch NC derivatives, and posterior lateral mesoderm. At looping heart stages, cardiac Hand2 expression remained generalized with no apparent regionalization (Angelo et al., 2000). Knockdown of zebrafish dihydrofolate reductase (Dhfr), one of the enzymes needed for biological function of folic acid, results in cardiac defects similar to Hand2 knockdown. Expression of Hand2 is reduced in Dhfr knockdown embryos, and cardiac defects are rescued by Hand2 overexpression (Sun et al., 2007). Disruption of Hand and Mef2c genes and a switch in MHC gene expression are early events in diabetic cardiomyopathy, likely due to the effects of oxidative stress on related signaling pathways (Aragno et al., 2006).

3.3.3. Hand2 in the Sympathetic Nervous System

Hand2 is expressed in several NC-derived tissues including components of the PNS. Hand2 is expressed in both the sympathetic and the parasympathetic divisions of the ANS (Dai et al., 2004). During human embryogenesis, Hand2 is expressed by sympathetic neuronal cells, extraadrenal chromaffin cells, and immature chromaffin cells in the adrenal gland, and is downregulated during chromaffin cell differentiation (Gestblom et al., 1999). In the chicken embryo, neuronal expression of Hand1 and Hand2 is restricted to sympathetic and enteric NC-derived ganglia (Howard et al., 1999; Howard et al., 2000). Sympathetic neurons are specified from NC precursors by a network of transcription factors including Hand2, Ascl1, Phox2a, and Phox2a. In the chicken embryo, overexpression of Hand2 also induces Gata2 expression (Tsarovina et al., 2004). Knockdown of Hand2 in NC-derived cells causes a significant reduction in neurogenesis and differentiation of catecholaminergic neurons but does not affect NT-derived neurons. In vitro, constitutive Hand2 expression is sufficient to induce catecholaminergic differentiation (Howard et al., 1999; Howard et al., 2000), and in vivo, Hand2 is sufficient to induce ectopic sympathetic neurons (Howard et al., 2000). Hand2 expression in chicken embryo sympathetic neurons is controlled by BMPs in vitro and in vivo and is induced downstream of Phox2b (Howard et al., 2000). In the mouse embryo, loss of Hand2 causes embryonic lethality by E9.5 and progressive loss of neurons and Th expression. Hand2 affects neural progenitor proliferation and expression of Phox2a and Gata3 (Hendershot et al., 2008). NC-specific deletion (Hand2^loxP/loxP; Wnt1-Cre) in the mouse results in lethality at E12.5 with severe cardiovascular and craniofacial defects, but NC-derived cells still populate regions of sympathetic nervous system (SNS) development and proliferate normally. Sympathetic precursors differentiate into neurons, demonstrating that Hand2 is not essential for SNS neuronal differentiation. However, the norepinephrine biosynthetic enzymes, Th, and dopamine b-hydroxylase (Dbh) were dramatically reduced in mutant embryos suggesting that the primary role of Hand2 in the SNS is determination of neuronal phenotype. Loss of Hand2 in the NC did not affect the expression of other genes regulating SNS development, including Phox2a, Phox2b, Gata2, Gata3, and Ascl1, but Hand2 was necessary for Hand1 expression (Morikawa et al., 2007). The major role of Hand2 during SNS development is most likely to permit sympathetic neurons to acquire a catecholaminergic phenotype (Morikawa et al., 2007). Hand2 acts together with Phox2a to regulate neurogenesis and noradrenergic marker gene expression. In these NC-derived cells, Hand2 and Phox2a are regulated by canonical and noncanonical BMP signaling. Protein kinase A activation by the noncanonical BMP pathway is needed to support neurogenesis and phosphorylation-dependent regulation of the Dbh promoter by Hand2. Activity of MapK is required for Hand2 transcription. Signaling affected by cAMP is necessary for regulation of Dbh promoter transactivation by Phox2a and Hand2 (Liu et al., 2005). MAP kinase activation and Bmp4-Smad1 signaling also mediate the differentiation of catecholaminergic neurons at least partly through induction of Hand2 (Liu et al., 2005; Wu and Howard, 2001). In zebrafish embryos with a Hand2 deletion (the Hand2^{hands off} mutation), sympathetic precursor cells aggregate to form normal sympathetic ganglion primordia as marked by the expression of Phox2b, Phox2a, and Ascl1, but these cells have reduced expression of the noradrenergic marker genes Th and Dbh and the transcription factors Gata2 and AP-2a. By contrast, generic neuronal differentiation seems to be unaffected, showing an essential and specific function of Hand2 for noradrenergic differentiation of sympathetic neurons, and implicating AP-2a and Gata2 as downstream effectors (Lucas et al., 2006). Further illustrating the role of Hand2 in catecholaminergic differentiation, mesencephalic NCCs form cholinergic parasympathetic neurons in the ciliary ganglion, whereas trunk NCCs form both catecholaminergic and cholinergic neurons in sympathetic ganglia. Mesencephalic NCCs do not express the catecholaminergic transcription factor Hand2 in response to cranial BMP expression. Quail–chicken transplant experiments show that mesencephalic NCCs have a reduced capacity to undergo sympathetic differentiation; a subset of these cells never express Hand2 (Lee et al., 2005b).

3.3.4. Hand2 in the Enteric Nervous System

Hand2 is expressed in NC-derived precursors of enteric neurons and affects neurogenesis and specification of noradrenergic sympathetic ganglion neurons. Hand2 gain-of-function in NC progenitors results in increased neurogenesis and more neurons expressing vasoactive intestinal polypeptide (VIP). NC-specific deletion of Hand2 results in loss of all VIP-expressing gut neurons, reduction in Th-expressing neurons, abnormal Th-expressing neuron morphology, and abnormal patterning of the myenteric plexus (D’Autreaux et al., 2007; Hendershot et al., 2007). NC-derived cells are present, but neurons do not develop in explants from Hand2-null embryonic gastrointestinal (GI) tracts, and instead only glia are generated. Terminally differentiated enteric neurons do not develop after knockdown or conditional inactivation of Hand2 in migrating NC-derived cells (D’Autreaux et al., 2007). In the chicken embryo, Hand2 is expressed in neurons of the myenteric and submucosal ganglia. Hand2 is increased in NC-derived cells cocultured with the GI tract, suggesting a GI tract-derived factor regulates expression of Hand genes. Exposure of GI tract-derived NC-derived cells to Bmp4 significantly increases expression of Hand2 in all gut segments (Wu and Howard, 2002). Gdnf could not induce expression of Hand2 in NCCs but caused a modest increase in Hand2 in gut-derived NCCs from the esophagus and colon (Wu and Howard, 2002). In zebrafish, Hand2 is also required for normal development of the ENS and intestinal smooth muscle (Reichenbach et al., 2008).

3.3.5. Hand2 in Neuroblastoma

Neuroblastomas are tumors derived from the SNS that exhibit NC properties, perhaps as a result of impaired differentiation (Gestblom et al., 1999; Pietras et al., 2008). All examined neuroblastoma specimens and cell lines have detectable Hand2 mRNA levels (Gestblom et al., 1999). In vitro studies from neuroblastoma samples show that Hand2 can homodimerize and forms heterodimers with another family of bHLH factors, the E proteins, and these heterodimers bind E-box DNA sequences (Dai and Cserjesi, 2002). High levels of another bHLH protein, Hif2a (a.k.a. Epas1), correspond to less-differentiated neuroblastoma cells and poorer prognosis. These high Hif2a-expressing cells tend to have higher expression levels of NC and early sympathetic progenitor marker genes such as Hand2 (Pietras et al., 2008).

3.3.6. Hand2 Partners and Function

Transcriptional activity of Hand2 can be modulated depending on its phosphorylation state and its choice of heterodimer partner. Hand2 forms heterodimers with at least three E proteins: Tcf3, Tcf4, and Tcf12, and these bind to E-box elements with different affinities (Murakami et al., 2004b). Another partner of Hand2 is Jun activation domain-binding protein (JAB1/Cops5), which augments Hand2 transcriptional activity by enhancing binding through the HLH domain (Dai et al., 2004). Akt, a serine/threonine protein kinase involved in cell survival, growth, and differentiation, phosphorylates Hand2 and inhibits Hand2-mediated transcription (Murakami et al., 2004a).

3.4.1. Targets of Mitf

The main targets of Mitf during melanogenesis are Dct, Tyr, and Trp1, members of the tyrosinase gene family important for melanin synthesis. In embryos with severe Mitf mutations, NC-derived cells that would normally express Mitf lack Dct expression and soon disappear, showing that melanocyte development and survival require Mitf function. In contrast, neuroepithelial-derived Mitf-expressing cells of the retinal pigment layer are retained but are unpigmented. These cells express Dct, but not Tyr and Trp1 (Jiao et al., 2004; Nakayama et al., 1998). The murine Dct gene is expressed early in melanocyte development during embryogenesis, before Tyr and Trp1 (Jiao et al., 2004). The Tyr, Trp1 and Dct promoters all contain a conserved E-box motif normally bound by Mitf. Mitf transactivates promoters of the tyrosinase gene family in both pigment cell lineages. These promoters also have specific DNA motifs corresponding to binding sites for RPE-specific factors such as Otx2 or melanocyte-specific factors Sox10 or Pax3 (Murisier and Beermann, 2006). A region of the Dct promoter critical for high Dct expression in melanocytes contains candidate binding sites for Sox10 and Mitf. Transfections into 293T and NIH3T3 cells show that Sox10 and Mitf can independently activate Dct expression, and when cotransfected, synergistically activate the Dct promoter (Jiao et al., 2004). Combinations of Pax3, Sox10, and Mitf regulate Dct transcription directly and synergistically (Jiao et al., 2006). Homozygous null Mitf mutants do not express Dct, and the number of cells expressing Kit is significantly reduced (Opdecamp et al., 1997). Wild-type NCC cultures rapidly give rise to cells expressing Mitf, Kit, and Dct, and with time, Kit expression increases concomitant with the appearance of pigmented cells. In contrast, cells cultured from Mitf mutant embryos did not increase Kit expression with time, suggesting Kit is downstream of Mitf. Mutant cells did not express Dct, never produced, and expression from Mitf itself was rapidly lost, suggesting a feedback mechanism. Mitf may initially play a role in the transition of precursor cells to melanoblasts and later regulating melanoblast survival by controlling Kit expression (Opdecamp et al., 1997). Mitf is required for maintenance of Kit expression in melanoblasts and Kit signaling in turn modulates Mitf activity and stability in melanocyte cell lines (Hou et al., 2000). In primary NCC cultures, initiation of Mitf expression in melanoblasts does not require Kit. A small number of Kit-null Mitf-positive cells can be maintained for at least 2 weeks in culture; these cells express several pigment cell-specific genes activated by Mitf (Dct, Silver(Si), and Trp1), but lack expression of Tyr, which encodes the rate-limiting enzyme in melanin synthesis. This demonstrates that the presence of Mitf alone is not sufficient for Tyr expression in melanoblasts. However, elevation of cAMP levels in these cultures increases the number of Mitf-positive cells, induces Tyr expression, and results in differentiation of melanoblasts into mature, pigmented melanocytes (Hou et al., 2000).

Mitf mutant embryos also have reduced expression of the membrane-associated transporter protein (Matp/Slc45a2) (Baxter and Pavan, 2002). Matp expression was similar to the melanogenic enzymes Dct and Trp1, suggesting similar regulation, and Matp mutations result in pigment alterations in mice (underwhite/uw mutants), fish (medaka b-locus), and humans (Oculocutaneous Albinism Type 4). Further evidence suggesting Matp acts in a pigment cell autonomous manner is the expression of mouse Matp in presumptive RPE starting at E9.5, and in NC-derived melanoblasts starting at E10.5 (Baxter and Pavan, 2002). Additionally, Mitf mutant embryos lose expression of the melanosomal gene Si, also normally expressed in the developing RPE starting at E9.5, and in NC-derived melanoblasts starting at E10.5. Si expression in dorsal regions precedes Dct expression, suggesting Si is an earlier melanoblast marker (Baxter and Pavan, 2003).

3.4.2. Regulation of Mitf by Sox10 and Pax3

Sox10 can transactivate the Mitf promoter up to 100-fold by binding to a region conserved between mouse and human Mitf promoters. A dominant-negative Sox10 mutant reduces wild-type Sox10 induction of Mitf (Potterf et al., 2000). Both Sox10 and Pax3 directly regulate Mitf, and the strong transactivation of Mitf by Sox10 is further stimulated by synergy with Pax3 as each protein binds independently (Lang and Epstein, 2003; Potterf et al., 2000). At E11.5, mouse embryos homozygous for the Sox10^Dom mutation entirely lack NC-derived cells expressing the lineage markers Kit, Mitf, or Dct. In Sox10^Dom heterozygous embryos, melanoblasts expressing Kit and Mitf do occur, albeit in reduced numbers, and pigmented cells eventually develop in almost normal numbers in culture and in vivo. Dct is not expressed, suggesting that Sox10 acts as a critical transactivator of Dct (Potterf et al., 2001). Melanophore defects in zebrafish Sox10^colourless mutants can be explained by disruption of Mitf expression. Sox10^colourless NCCs adopt mesenchymal fates, and NCCs that would normally adopt nonmesenchymal fates generally fail to migrate, do not differentiate, and instead undergo apoptosis. All defects of affected NC derivatives are consistent with a primary role for Sox10 in specification of nonmesenchymal NC derivatives (Dutton et al., 2001). Mitfa expression is undetectable in Sox10-null zebrafish. The zebrafish Mitfa promoter contains Sox10 binding sites necessary for activity in vivo and in vitro, consistent with studies in mammalian cell cultures demonstrating Sox10 directly regulates Mitf. Reintroduction of Mitfa expression in NCCs can rescue melanophore development in Sox10-null embryos in a manner quantitatively indistinguishable from rescue in Mitfa-null embryos, suggesting the essential function of Sox10 in melanophore development is transcriptional regulation of Mitfa (Elworthy et al., 2003). A melanocyte-specific Mitf enhancer contains binding sites for Sox10. Sox10 and Mitf-M are coexpressed in melanoblasts migrating toward the otic vesicle of mouse embryos but are separately expressed in different cell types of the newborn cochlea suggesting that Sox10 regulates transcription from the Mitf promoter in a developmental stage-specific manner (Watanabe et al., 2002a). Mitf regulation by Sox10 is disrupted by sumoylation of Sox10, which represses Sox10 transcriptional activity on Mitf (Girard and Goossens, 2006). In the melanocyte lineage, Pax3 regulates Mitf and Trp1 (Corry and Underhill, 2005).

Mitf is also activated by a-melanocyte-stimulating hormone (Msh/Pomc) through a conserved cAMP response element (CRE). cAMP-mediated CRE-binding protein activation of the Mitf promoter requires a second nearby DNA element that is bound and activated by Sox10. In NC-derived melanoma and neuroblastoma cells lacking Mitf-M and Sox10 expression, Mitf promoter responsiveness to cAMP depends on Sox10, and Sox10 transactivation is likewise dependent on the CRE. Ectopic Sox10 expression, in cooperation with cAMP signaling, activates the Mitf promoter and expression of endogenous Mitf transcripts in neuroblastoma cells. This activation does not occur with the Sox10^Dom allele (Huber et al., 2003). High-level activation of cAMP signaling attenuates Bmp2-induced sympathoadrenal cell development and induces melanogenesis by inducing transcription of Mitf. Dominant-negative Creb1 inhibits Mitf expression and melanogenesis, suggesting Creb1 activation is necessary for melanogenesis. However, constitutive activation of Creb1 without PKA activation is insufficient for Mitf expression and melanogenesis, indicating PKA regulates additional aspects of Mitf transcription (Ji and Andrisani, 2005). In vitro studies demonstrate that the phorbol ester 12-tetradecanoylphorbol 13-acetate (TPA) is another factor that induces NCC differentiation into melanocytes and stimulates proliferation and differentiation of melanocytes. In primary NCC cultures, TPA is necessary for Mitf upregulation and melanin synthesis. In an immortalized melanocyte cell line that proliferates in the absence of TPA, TPA significantly increases the mRNA levels of the tyrosinase gene family (Tyr, Tyrp1, and Dct) and the expression of Mitf (Prince et al., 2003). When overexpressed in melanocytes, Emx1 and Emx2 downregulate Mitf, Tyrp1, Dct, and Tyr, suggesting that these genes may restrict expression of Mitf by inhibiting activation in neuroepithelial derivatives other than melanocytes (Bordogna et al., 2005).

3.4.3. Regulation of Zebrafish Mitf by Foxd3

The zebrafish Hdac1^colgate mutant has increased and prolonged expression of Foxd3, resulting in reduced numbers, delayed differentiation, and decreased migration of NC-derived melanophores and their precursors due to a severe reduction in the number of Mitfa-positive melanoblasts. Hdac1 is required to suppress Foxd3 expression in the NC, thus de-repressing Mitfa resulting in melanogenesis in a subset of NC-derived cells (Ignatius et al., 2008). Foxd3 acts directly on the Mitfa promoter to negatively regulate Mitfa expression (Curran et al., 2009). Mitfa is only expressed in a Foxd3-negative subset of NCCs, and Foxd3 mutants have more cells expressing Mitfa. Foxd3 prevents a subset of NC-derived precursors from acquiring a melanophore fate by repressing the Mitfa promoter and is not expressed in terminally differentiated melanophores. Foxd3 is, however, expressed in xanthophore precursors and iridophores (Curran et al., 2009).

3.4.4. Mitf Downstream of Wnt Signaling

In cultured human melanocytes, exogenous Wnt3a upregulates expression of Mitf. Wnt3a signaling likely recruits b-catenin and Lef1 to the Lef1-binding site of the Mitf promoter (Takeda et al., 2000). In zebrafish, Wnt signals are necessary and sufficient for NCCs to adopt pigment cell fates, which also require Mitf. As in mammalian cells, a promoter region of zebrafish Mitfa containing Tcf/Lef binding sites mediates Wnt responsiveness. An Mitf reporter construct is strongly repressed by a dominant-negative Tcf in melanoma cells. Mutation of Tcf/Lef sites abolishes Lef1 binding and reporter function in vivo (Dorsky et al., 2000). Functional cooperation of Mitf with Lef1 results in synergistic transactivation of Dct, an early melanoblast marker. b-Catenin is required for efficient transactivation but is dispensable for the interaction between Mitf and Lef1. The interaction with Mitf is unique to Lef1 and not detectable with Tcf1. Lef1 also cooperates with Mitf-related proteins, such as the bHLH factor Tcfe3, to transactivate the Dct promoter (Yasumoto et al., 2002). Mitf functions downstream of the canonical Wnt pathway in mammalian melanocyte and zebrafish melanophore development and is regulated by b-catenin across species (Widlund et al., 2002). b-Catenin is expressed throughout development of the melanocyte lineage and contributes to regulation of Mitf expression (Larue et al., 2003). The Wnt/b-catenin canonical signaling pathway also plays a role in melanoma. Several components of the Wnt pathway are altered in melanoma tumors and cell lines. Activated b-catenin is found in approximately 30% of human melanoma nuclei and can induce genes regulating proliferation (Myc or CyclinD1) in addition to regulating cell lineage-restricted genes (Brn2) and melanocyte-specific genes (Mitf and Dct) (Larue and Delmas, 2006).

3.4.5. Posttranslational Regulation of Mitf

Mitf is posttranslationally regulated by phosphorylation and ubiquitination affecting protein activity and stability (Galy et al., 2002). Kit signaling is linked to phosphorylation of Mitf through activation of MAP kinase and Rps6ka1 (ribosomal protein S6 kinase), and IGF-1 and HGF/SF signaling may also be involved. Phosphorylation of Mitf is carried out by GSK3b (Tachibana, 2000). Mitf is also modified in melanoma cells by sumoylation: mutations affecting sumoylated residues, although displaying normal DNA binding, stability, and nuclear localization, result in a substantial increase in the transcriptional stimulation of promoters containing multiple (but not single) Mitf binding sites and enhanced cooperation with Sox10 on the Dct promoter. Sumoylation of Mitf regulates the protein’s transcriptional activity with respect to synergistic activation and sumoylation plays a significant role among the multiple mechanisms regulating Mitf (Murakami and Arnheiter, 2005). In melanocyte cultures, the MAPK pathway targets Mitf for ubiquitin-dependent proteolysis, resulting in a rapid degradation and downregulation (Galy et al., 2002).

3.4.6. Mitf in Human Disease

Mutations in human Mitf are found in patients with type 2 Waardenburg syndrome (WS2), a dominantly inherited syndrome associated with hearing loss and pigmentary disturbances, and sometimes occurring in conjunction with other NC syndromes (Boissy and Nordlund, 1997; Lalwani et al., 1998; Steingrimsson et al., 1994; Tachibana, 2000; Van Camp et al., 1995). WS2 is heterogeneous; 15–20% of cases are caused by mutations in Mitf, particularly affecting Mitf-M (Morell et al., 1997; Read and Newton, 1997; Tassabehji et al., 1995; Yajima et al., 1999). The dominant melanophore phenotype in type 4 Waardenburg syndrome individuals with Sox10 mutations likely results from failure to activate Mitf in a normal number of melanoblasts (Elworthy et al., 2003). In human melanoma, many pigmentation markers are lost, but Mitf expression is still active, even in unpigmented tumors, suggesting a role for Mitf beyond differentiation. Many primary human melanomas have aberrant nuclear accumulation of b-catenin, a potent growth mediator of melanoma cells dependent on Mitf, which is a downstream target of b-catenin. The suppression of melanoma clonogenic growth that occurs when b-catenin-Tcf/Lef regulation is disrupted can be rescued by constitutive expression of Mitf, suggesting a prosurvival mechanism for Mitf. b-Catenin regulation of Mitf represents a tissue-restricted pathway that significantly influences the growth and survival of melanoma cells (Widlund et al., 2002). Key proteins in melanogenesis such as Mitf-M, Sox10, Pax3, Trp1, and Tyr are absent or greatly reduced in the bulbs of graying hair compared to black hair. Hair graying is likely caused by defective migration of melanocyte stem cells into the bulb area of hair (Choi et al., 2008).

3.8.1. Twist1 in the Cranial Mesenchyme and Pharyngeal Arches

In the mouse, Twist1 is first expressed in extraembryonic tissues, then within some ectodermal cells of the primitive streak, and subsequently in mesoderm outside the primitive streak (Fuchtbauer, 1995; Stoetzel et al., 1995). Beginning around E8–E8.25, Twist transcripts accumulate in the cranial mesenchyme in both cranial NCCs and NC in the pharyngeal arches (Fuchtbauer, 1995; Gitelman, 1997; Stoetzel et al., 1995). During embryogenesis, Twist1 persists in mesodermal and NCC derivatives, often at sites of epithelial–mesenchymal interactions, and most prominently in progenitor cells of the muscle and cartilage/skeletal lineages, suggesting a role for Twist in inhibiting differentiation in these cell types (Fuchtbauer, 1995; Stoetzel et al., 1995). Twist1-null mouse embryos die at E11.5 with defects in the cranial mesenchyme; the cranial neural folds do not fuse as a result of cell-autonomous Twist1 function in cranial mesenchyme (Chen and Behringer, 1995; Gitelman, 1997). Malformation of the pharyngeal arches and defects in the somites and limb buds are also observed (Chen and Behringer, 1995; Soo et al., 2002). In the cranial mesenchyme, cells lacking Twist1 migrate abnormally and sometimes lack mesenchymal characteristics, suggesting Twist1 regulates migration and specification of the cranial mesenchymal cells needed for fusion of the cranial neural folds (Chen and Behringer, 1995; Gitelman, 1997). In the first pharyngeal arch, Twist1 is required in both the mesoderm-derived cranial mesenchyme for directing early migration of NCCs and in the NCCs themselves; without Twist1, NCCs stray from their normal migratory path (Soo et al., 2002; Vincentz et al., 2008). Twist1 is also necessary for differentiation of the first pharyngeal arch tissues into bone, muscle, and teeth (O’Rourke and Tam, 2002; Ota et al., 2004; Soo et al., 2002). Consistent with this, mutations in human Twist contribute to syndromic craniofacial abnormalities and perhaps also to nonsyndromic cleft lip and palate, one of the most common human birth defects (Bacon et al., 2007).

In addition to its requirement for NC emigration from the NT and migration into the pharyngeal arches, Twist1 is also required for later migration along cardiac NCC pathways. In Twist1-null mice, cardiac NCCs are delayed in their colonization of the endocardial OFT cushions, which later contain abnormal, nodular mesenchyme derived exclusively from the cardiac NC. A subpopulation of Twist1-null NCCs that successfully migrate to the OFT has defects in maturation, migration, and adhesion. There are similar nodules in pharyngeal arches, and the dorsal NT has an expanded domain of NCCs (Wnt1-Cre-lineage marked cells) (Vincentz et al., 2008). Loss of Twist1 also affects patterning of the cranial ganglia but not patterning of the peripheral ganglia derived from trunk NC (Ota et al., 2004).

3.8.2. Twist1 in the Skull Vault

The frontal and parietal bones that form the vertebrate skull vault are composed of cells of NC and mesodermal origin, respectively. These two divergent cell lineages come together at positions between the ectoderm and cerebral hemispheres. The boundary between these two mesenchymal cell populations becomes the developing coronal suture, a major growth center for the skull, where both lineages proliferate and differentiate toward an osteogenic fate (Ishii et al., 2003; Merrill et al., 2006). Craniosynostosis (also termed coronal synostosis, and for a less common subset, Saethre-Chotzen syndrome) is a relatively common human birth defect resulting from premature fusion of the developing frontal and parietal bones (Merrill et al., 2006). One cause of human craniosynostosis, calvarial foramina (persistent unossified areas within the skull vault), is linked to mutations in Twist1, Msx2, and the gene encoding the ephrin A4 ligand, Efna4 (Firulli and Conway, 2008; Ishii et al., 2003; Merrill et al., 2006; Paznekas et al., 1999), three factors that cooperate in regulation of skeletogenic mesenchyme proliferation and differentiation in the developing frontal bone. Heterozygous loss of murine Twist1 results in a defective boundary between frontal NC and parietal mesoderm mesenchymal cells, allowing the NCCs to invade the undifferentiated mesoderm of the coronal suture (Ishii et al., 2003; Merrill et al., 2006). This is accompanied by an expansion in Msx2 expression and reduction of Efna2, Efna4, and Epha4; Efna4 expression in Twist1-heterozygous embryos can be rescued by loss of one allele of Msx2, which in turn rescues the craniosynostosis defect (Merrill et al., 2006). In Twist1-null, Msx2-null double mutants, the quantity and proliferation of frontal bone skeletogenic mesenchyme are reduced compared to individual mutants. Msx2 and Twist1 likely cooperate in regulation of skeletogenic mesenchyme differentiation and proliferation, with each independently controlling NCC and mesoderm contribution (Ishii et al., 2003). Mice deficient for Efna4 exhibit defects in the coronal suture NC–mesoderm boundary much like Twist1 heterozygous mice, and compound Twist1 heterozygous, Epha4 heterozygous mice have more severe defects than those of individual heterozygotes, indicating a genetic interaction. DiI labeling of migratory osteogenic progenitors contributing to frontal and parietal bones reveals that Twist1 and Epha4 are required for exclusion of these cells from the coronal suture (Ting et al., 2009).

3.8.3. Twist in Zebrafish and Xenopus

Zebrafish Twist gene expression is detected in cranial NC, sclerotome, lateral plate mesoderm, and in the dorsal aorta (Germanguz et al., 2007). During early development, Twist1 is expressed in cranial mesenchyme and later in NC of the pharyngeal arches, consistent with a role in craniofacial development (Yeo et al., 2007). In Xenopus, Twist1 is expressed at the time of early NC specification together with Snail and Slug in the presumptive neural folds and marks the presumptive NC (Hopwood et al., 1989; Linker et al., 2000). Twist1 expression follows expression of Snail and Slug and can be induced by Snail during NC specification (Aybar et al., 2003; Linker et al., 2000).

3.8.4. Twist as an NC Marker and in Cancer and NC-Derived Stem Cells

Knockdown of Rhov, a Ras homolog expressed in early NC and essential for NC induction, disrupts expression of Twist1, Slug, and Sox9 (Guemar et al., 2007). Twist1 is also expressed in human dental NC-derived progenitor cells (Degistirici et al., 2008), porcine NC skin-derived precursors (pSKPs) (Zhao et al., 2009), and multipotent rat periodontal ligament-derived NC stem cells (Techawattanawisal et al., 2007). Twist1 and Sox9 are overexpressed in NC Schwann cell-derived malignant peripheral nerve sheath tumors (MPNSTs). Reduction of Twist1 (implicated in metastasis, chemotherapy resistance, and inhibition of apoptosis) in MPNST cells has no effect on apoptosis or chemotherapy resistance but results in inhibition of MPNST cell chemotaxis (Miller et al., 2006). During carcinoma of the parathyroid glands, which have a dense outer mesenchymal NC-derived component, expression of Twist1 (and Snail) changes from a homogenous distribution among cancer cells to localization in the cells at the invasive front of the cancer with a corresponding loss of membranous Cdh1 (E-cadherin) (Fendrich et al., 2009).