Oncogene (2003) 22, 3024–3034 & 2003 Nature Publishing Group All rights reserved 0950-9232/03 $25.00 www.nature.com/onc

The importance of having your SOX on: role of SOX10 in the development of -derived and

Ramin Mollaaghababa1 and William J Pavan*,1

1National Research Institute, National Institutes of Health, 49 Convent Drive, Bethesda, MD 20892-4472, USA

SOX10w is a member of the high-mobility group-domain differentiate to form melanocytes of the skin, hair, and SOX family of transcription factors, which are ubiqui- inner ear while others move ventrally, either through the tously found in the animal kingdom. Disruption of neural or in the space between the somites and the crest development in the Dominant megacolon (Dom) neural tube, and contribute to the formation of mice is associated with a Sox10 . in additional distinct lineage. These include sensory human Sox10 w have also been linked with the neurons and glia, neurons and glia of cranial ganglia, occurrence of neurocristopathies in the Waardenburg– cartilage and bone, connective tissue, and neuroendo- Shah syndrome type IV (WS-IV), for which the Sox10Dom crine cells (Le Douarin and Kalcheim, 1999). mice serve as a murine model. The neural crest disorders The specification of neural crest to distinct lineage in the Sox10Dom mice and WS-IV patients consist of and their proper differentiation is dependent on both hypopigmentation, cochlear neurosensory deafness, and intrinsic factors and environmental interactions (La- enteric aganglionosis. Consistent with these observations, Bonne and Bronner-Fraser, 1998). The use of mouse a critical role for SOX10 in the proper differentiation of neural crest mutants has been instrumental in the neural crest-derived melanocytes and glia has been identification and analysis of essential for proper demonstrated. Emerging data also show an important neural crest function (Le Douarin and Kalcheim, 1999). role for SOX10 in promoting the survival of neural crest In particular, the critical role of a number of transcrip- precursor cells prior to lineage commitment. Several genes tion factors in differentiation and function of neural whose regulation is dependent on SOX10 function have crest cell (NCC) lineage was established using mouse been identified in the peripheral nervous system and in mutants. This review article focuses on one such melanocytes, helping to begin the identification of the called SOX10 (acronym for Sry-like multiple pathways that appear to be modulated by SOX10 HMG bOX), which has a significant function in proper activity. In this review, we will discuss the biological development of the NCCs. We will present a synthesis of relevance of these target genes to neural crest development the current state of knowledge about SOX10 and its and the properties of Sox10 as a transcription factor. developmental role, discussing the implications of these Oncogene (2003) 22, 3024–3034. doi:10.1038/sj.onc.1206442 findings and their relevance to future work.

Keywords: SOX10w; neural crest; glia; melanocytes; differentiation; transcription factor Discovery of SOX10

Sox10 was first discovered by RT–PCR analysis of total Introduction RNA from 11.5-day-old mouse embryos (E11.5), using degenerate oligonucleotide primers directed against the The health and survival of all organisms depends on high-mobility group (HMG) domain (Wright et al., proper differentiation and function of diverse cell types. 1993). Members of the HMG-domain family are DNA- During vertebrate development, a variety of distinct cell binding that fall into a vast array of functional types are derived from a special precursor population categories (Bianchi and Beltrame, 2000). Their HMG called the neural crest (Le Douarin and Kalcheim, 1999; Knecht and Bronner-Fraser, 2002). Neural crest cells designation relates to the observation that the family members initially analysed showed fast migration on originate dorsally at the top, or ‘crest’, of the neural tube SDS–PAGE. Sequence comparison showed that within in early development, and travel in well-defined migra- the HMG group, SOX10 belongs to the SOX family of tory paths following epithelial-to-mesenchymal conver- proteins, which are related to the mammalian testis sion and detachment from the neural tube (Erickson and determination factor SRY (Gubbay et al., 1990). More Reedy, 1998). Some follow a dorso-lateral pathway and than 20 vertebrate SOX members have been identified to date, and related genes have been found in Drosophila *Correspondence: WJ Pavan; E-mail: [email protected] melanogaster and Caenorhabditis elegans (Wright et al., wGene symbols and proteins are respectively designated by italic and 1993; Pevny and Lovell-Badge, 1997; Wegner, 1999). upper-case letters throughout the text. SOX proteins show restricted patterns of tissue-specific SOX10 function in neural crest development R Mollaaghababa and WJ Pavan 3025 expression and are essential for a multitude of develop- (HSCR) (Shah et al., 1981; Badner and Chakravarti, mental processes, including nervous system develop- 1990; Parisi and Kapur, 2000). These include cochlear ment, bone morphogenesis, pigment cell formation, deafness and pigmentary defects (heterochromia iridis, formation of the germ layer, development of the white forelock and eyelashes, leukoderma) characteristic immune system, sex determination, and eye develop- of the Waardenburg disorder, and enteric aganglionosis ment (for reviews, refer to Pevny and Lovell-Badge, in the distal region of colon characteristic of the HSCR. 1997; Wegner, 1999). Based on pairwise comparisons of The illustration of the genetic defect in the Dom mice led partial HMG-domain sequences, the SOX members to the identification of Sox10 mutations in a number of were originally classified into six separate groups (A–E), individuals with WS-IV features (Pingault et al., 1998; with SOX8, SOX9, and SOX10 belonging to group E Bondurand et al., 1999; Inoue et al., 1999; Southard- (Wright et al., 1993). This was later expanded to seven Smith et al., 1999b; Touraine et al., 2000; Sham et al., (van de Wetering and Clevers, 1993; Meyer et al., 1996) 2001a). A Sox10 mutation (795delG) that does not give and eight groups (A–H) (Osaki et al., 1999), owing to the established features of WS-IV but presents with the discovery of additional SOX genes. The latest phenotypes associated with abnormal neural crest refinements of these classifications are based on full- development has also been identified in an individual length sequences comparisons. They consist of (Pingault et al., 2000). A list of these mutations, along the addition of two new groups, I and J, in a study that with those identified in mice and zebrafish, is presented included the invertebrate SOX sequences (Bowles et al., in Figure 2. They include frameshift mutations, inser- 2000), and subdivision of group B into B1 and B2 tions, deletions, extensions, and point mutations, and (Uchikawa et al., 1999; Bowles et al., 2000; Kamachi span different domains of the SOX10 protein. However, et al., 2000). Since its discovery in mouse, Sox10 no mutations within the regulatory region of Sox10 have homologues have been identified in several other been linked to neurocristopathy phenotypes to date. vertebrates. In rat, the Sox10 gene was cloned from Identification of such mutations in the future should rat glial cells (Kulhbrodt et al., 1998a). Human Sox10 help with discovery of transactivating factors that was identified by characterization of a cosmid from a modulate Sox10 expression. human -22-specific library and by com- parative mapping studies (Pusch et al., 1998; Southard- Smith et al., 1999a). Molecular analysis of the zebrafish SOX10 structure mutation colourless (cls) led to the discovery of a Sox10 et al homologue in this organism (Dutton ., 2001). In HMG domain chick, a Sox10 homologue was identified by screening of a chick embryo cDNA library with HMG-domain- A better appreciation of the functional properties of specific degenerate primers (Cheng et al., 2000). A SOX10 requires an understanding of the structural Sox10-related homologue has also been identified in the mechanisms by which SOX10 binds its target DNA and invertebrate species D. melanogaster. The Drosophila regulates transcription. Similar to other HMG proteins, homologue (Sox100B) was discovered by sequence SOX10 binds its target DNA sequences via its HMG homology within the HMG domain and is equally domain (Figure 2), and mutations in the HMG box related to the Sox9 gene (Loh and Russell, 2000). Its disrupt SOX10 structure and compromise its binding to function in the fruit fly remains to be elucidated. A DNA (Southard-Smith et al., 1999b; Bondurand et al., sequence comparison of SOX10 across these different 2000). Model structures of the SOX10 HMG domain species is presented in Figure 1. This alignment shows have been generated from human, rat, and mouse the high degree of sequence conservation for SOX10 sequences by threading each sequence through the during evolution, underscoring its functional signifi- NMR coordinates of the second HMG-1 box from cance and suggesting that many of the mechanisms rat, followed by energy minimization studies to deter- governing neural crest development may be conserved mine the most favorable conformation (Southard-Smith throughout vertebrate evolution. et al., 1999b). The predicted structure is composed of three alpha-helices forming an L shape, which is proposed to facilitate binding to the DNA. The HMG box is highly conserved within the SOX family, with Identification of Sox10 mutation in Waardenburg–Shah X90% amino-acid sequence identity within an indivi- type IV mouse model dual SOX group and about 60% identity between distant groups (Kamachi et al., 2000). Interest in the functional role of SOX10 intensified with The evolutionary conserved HMG motif consists of the discovery that the neurocristopathy phenotypes in about 80 amino acids and binds in the minor groove of the Dominant megacolon (Dom) mice are linked to a DNA (Weiss, 2001). In addition to conferring DNA- Sox10 mutation (Herbarth et al. , 1998; Southard-Smith binding capacity, the L-shaped structure of HMG et al., 1998). The Dom mice arose spontaneously at the domain is known to allow DNA bending for several Jackson Laboratory and have served as a murine model SOX proteins (Pevny and Lovell-Badge, 1997; Wegner, for the Waardenburg–Shah syndrome type IV (WS-IV), 1999). In vitro studies with either the HMG domain of a disorder that combines the features of the Waarden- SOX10 alone, or with a truncated form that lacks burg syndrome (Read, 2000) and Hirschsprung’s disease sequences C-terminal to the HMG domain, have

Oncogene SOX10 function in neural crest development R Mollaaghababa and WJ Pavan 3026

Figure 1 Sequence comparison of SOX10 protein across different species. Identical residues are shown in black and conserved residues are in gray. Dashes (-) represent sequence gaps introduced to obtain optimal alignment. The HMG domain is boxed in red. Protein lengths (residue number) are shown on the left

Oncogene SOX10 function in neural crest development R Mollaaghababa and WJ Pavan 3027

Figure 2 Schematic representation of the SOX10 protein and location of all Sox10 mutations identified to date. The ‘X’ symbol is placed next to the mouse mutation 929insGDom (green), and the zebrafish mutation N85ins (red), to indicate that these frameshift mutations cause the introduction of heterologous amino acids before terminating the protein prematurely. The SOX10 diagram is drawn to scale, with the corresponding amino-acid numbers marked below the figure in 50-residue blocks suggested that SOX10 also possesses DNA-bending extraneous residues (Figure 2; Herbarth et al., 1998; properties (Peirano and Wegner, 2000). The degree of Southard-Smith et al., 1998). In vivo studies have shown relevance of these observations to the in vivo chromatin that the ability of SOX10 to induce its target genes is environment within which SOX10 operates remains to considerably compromised in Sox10Dom heterozygous be established. Nevertheless, experimental results for mice (Potterf et al., 2000, 2001). The TA function may other SOX proteins suggest that this mode of DNA also be compromised without removal of any TA binding allows them to play an architectural role in domain residues. In an individual with WS-IV and with forming transcriptionally active chromatin (Prior and associated deficiencies, a Sox10 mutation was Walter, 1996; Pevny and Lovell-Badge, 1997). identified in which the TA domain is extended by 82 amino acids without any disruption of the coding region Transcriptional activation domain (Inoue et al., 1999). It is postulated that the TA domain activity in this mutation is compromised through Similar to most SOX proteins analysed to date, SOX10 interference from a proline-rich region that occurs carries a classical transactivation domain (TA) at its C- within the extension and has homology to Wilms’ tumor terminus, rich in serine, proline, and glutamine residues (WT1) proline-rich represser motif. These studies (Figure 2, Pusch et al., 1998). of this region in together demonstrate the ability of the TA domain of entirety abolishes the ability of SOX10 to induce SOX10 to activate transcription. However, it should be promoter activity of its target genes in vitro (Bondurand noted that they do not formally exclude the possibility et al., 2000; Lee et al., 2000; Potterf et al., 2000; that SOX10 may serve as a repressor in certain cellular Verastegui et al., 2000; Potterf et al., 2001). Further- contexts. more, fusion of various segments of human SOX10 with the DNA-binding domain of yeast GAL4 shows that only the fusion construct carrying the TA domain is capable of CAT induction following cotransfection into SOX10 binding sites and monomeric versus heteromeric cells with a GAL4-dependent CAT reporter plasmid mode of binding (Pusch et al., 1998). In vivo, several Sox10 mutations that lead to either a partial or full deletion of the TA Characterization of SOX10 binding sequences facilitates domain have been identified (Figure 2). The Sox10Dom the identification of those genes that are directly mutation belongs to this group and entirely lacks the TA targeted by SOX10 and further elucidates the mechan- domain. Thus, a translational frameshift downstream of isms by which SOX10 mediates gene activation. In vitro the HMG box removes the TA domain and adds 99 studies using a collection of random sequences identified

Oncogene SOX10 function in neural crest development R Mollaaghababa and WJ Pavan 3028 Table 1 Summary of direct SOX10 target genes and the identified SOX10 binding sites SOX10 target Binding site Gene function Synergistic factors References c-ret CTGTGCT Development of PAX3 Lang et al. (2000) enteric ganglia nACh b4 and CCACCCC Neuronal signaling SP1, SP3, PURa, Liu et al. (1999) a3 subunits within PNS hnRNPK and Melnikova et al. (2000a, b) a Protein zero (P0) Site B: TACAATG Myelin compaction Peirano et al. (2000) Site C: CACAAAGa in Schwann cells and Peirano and Site C0: CTGTGTA Wegner (2000) Connexin 32 (Cx32) Site S1: CACAAAG Formation of Schwann EGR2, PAX3, Bondurand Site S2: CATTGTGa cell gap junctions POU3F1 et al. (2001) Microphthalmia Site SX2: TGAAAGAGAAA differentiation PAX3 Potterf et al. (2000) transcription Site SX3: CATTGTC factor (Mitf) Site A: TATTGCTGA Lee et al. (2000) Site B: CATTGTCTATa Site C+D: TTGGCCTTGATCTGACAa Site E: ACCTTGTTTA Site F: ACATTGTTA Site 4 (S4): TATTGCT PAX3 Bondurand et al. (2000) Site 5(S5): CATTGTCa Site 6 (S6): AATTGGCT Site 7/8 (S7/8): GTTTGACTTTGAT Site Sx: CATTGTC Verastegui et al. (2000)

aHigher affinity binding sites Consensus SOX binding sequence: 50-A/TA/TCAAT/A-30 and complementary sequence 50-T/ATTGT/AT/A-30

50-AACAAT-30 as the core binding sequence for the formation and binding strength. In addition, the HMG domains of SRY, SOX5, SOX9, and SOX17, sequences flanking the sites greatly influence SOX10 with subtle differences in the flanking nucleotides binding. The dimer formation also improves the DNA- (Denny et al., 1992; Harley et al., 1994; Kanai et al., bending ability of SOX10 in vitro (Peirano and Wegner, 1996; Mertin et al., 1999). A degree of degeneracy exists 2000). It is of note that a similar arrangement of within this core sequence as both adenosine and adjacent SOX10 binding sites, separated by only a few thymidine are allowed at three of the positions (50-A/ base pairs, can be found in the promoter regions of TA/TCAAT/A-30). Using this core sequence as refer- other direct targets of SOX10, either in the glial or in the ence, SOX10 binding sites have been identified within melanocyte lineage (Table 1). The recurrence of adjacent promoter regions of several SOX10 target genes SOX10 binding sites in promoters of targets other than (Table 1). A considerable degree of sequence variability P0 provides a rationale for future analysis of these sites exists among these sites, suggesting that other determi- to see if the dimeric mode of binding by SOX10 is used nants such as flanking sequences, protein–protein as a general mechanism for improving binding specifi- interactions, or cooperative interactions with other city and DNA bending. bound factors may improve specificity of SOX10 binding within the cell. Indeed, an analysis of the promoter region for the protein zero (P0) gene, a myelin glycoprotein that is Sox10 mutant phenotypes targeted by SOX10, has shown that flanking sequences and cooperative binding of SOX10 dimers to adjacent The crucial role of SOX10 in the proper development of target sites are important determinants of binding NCCs is manifested by severe neurocristopathy pheno- specificity for SOX10 (Peirano and Wegner, 2000). types that are observed in humans, mice, and fish with Analysis of the proximal region of the P0 gene promoter either a complete lack of SOX10 function or a has identified two types of DNA response elements significant reduction in SOX10 activity (Figure 2). The (Peirano et al., 2000; Peirano and Wegner, 2000). One occurrence of the phenotypes observed in Sox10Dom/+ allows binding of SOX10 in monomeric form (desig- heterozygous mice and in Sox10Dom/Sox1Dom homozy- nated site B), whereas the second type promotes gous embryos is consistent with the observed expression cooperative binding of two SOX10 molecules as a dimer patterns of Sox10 in human and mouse. They encom- (designated site C), with dimer formation improving pass expression in emerging NCCs and neural crest binding specificity. Site C is actually composed of two derivatives (Figure 3; Bondurand et al., 1998; Herbarth adjacent binding sites that are separated by only four et al., 1998; Kuhlbrodt et al., 1998a; Pusch et al., 1998; base pairs and are oriented toward one another (C and Southard-Smith et al., 1998; Kapur, 1999), including the C0). The spacing between these binding sites and their enteric ganglia of HSCR patients (Sham et al., 2001b) orientations are important determinants of dimer and mouse embryos (Figure 3), and expression in the

Oncogene SOX10 function in neural crest development R Mollaaghababa and WJ Pavan 3029

Figure 3 Expression of SOX10 protein in the neural crest-derived structures of mouse embryos. A purified anti-SOX10 antibody, directed against a stretch of amino-acid residues unique to SOX10, was used for staining of endogenous SOX10 (brown) in sections of E13.5 mouse embryos (R Mollaaghababa and William J Pavan, unpublished). To improve contrast, the sections were counter stained with hematoxylin. SOX10 expression is clearly observed in the DRG, the ventral spinal nerves, the vagus nerve, the esophagus (enteric nervous system), and in dermis. Arrowheads show the location of SOX10+ cells within the neural tube (NT), presumably corresponding to oligodendrocyte precursors. The appearance of SOX10+ cells in dermis is consistent with SOX10 expression in melanoblasts inner ear during mouse development (Watanabe et al., the Sox10Dom mutant protein may act in a similar 2000). The Sox10Dom heterozygous mice show enteric manner. However, a careful comparison of the Sox10Dom aganglionosis (megacolon) and hypopigmentation phe- and null alleles for penetrance and expressivity of the notypes characteristic of WS-IV. Similar to individuals Sox10 phenotypes needs to be performed using similar with HSCR, the hypopigmentation and megacolon genetic backgrounds. In the next section, we discuss the phenotypes show variable penetrance and expressivity current findings in regard to the cellular basis of the in these mice, and the severity of the phenotypes is Sox10 mutant phenotypes. modulated by the genetic background on which the mice are kept (Lane and Liu, 1984; Kapur et al., 1996; Southard-Smith et al., 1999a). The homozygous Sox10Dom mice die during gestation, and the primary Role of SOX10 in neural crest development reason for the observed embryonic lethality is not established. Recently, impaired function of the auton- NCC formation appears normal in the absence of omous nervous system of the heart was detected in an SOX10 function. Neural crest-derived embryonic struc- individual with WS-IV who carried a truncated form of tures are present in Sox10Dom/Sox10Dom or SoxLacZ/ SOX10 (Korsch et al., 2001). This observation suggests Sox10LacZ homozygous embryos (Herbarth et al., 1998; the possibility that embryonic lethality in homozygous Southard-Smith et al., 1998; Britsch et al., 2001), and Sox10Dom embryos may be a result of abnormal P75+ (NCC marker) cells can be isolated from the autonomic regulation of the heart. The homozygous DRG of Sox10À/À embryos at E13 (Paratore et al., Sox10Dom embryos are either missing or severely deficient 2001). Therefore, Sox10 is not required for initial neural for several neural crest derivatives such as dorsal root crest specification. The appearance of NC-derived ganglia (DRG), several cranial ganglia, and sympathetic embryonic structures in Sox10À/À embryos indicates and enteric ganglia (Southard-Smith et al., 1998; Kapur, that the initial migration of at least some of the specified 1999). NCCs is also undisturbed in the absence of SOX10. This More recently, knockin mice in which the Sox10 suggests that SOX10 is not required for NCC migration coding region is replaced by the LacZ gene have been either, at least in the earlier stages. However, the engineered, and they exhibit similar deficiencies as the multipotent postmigratory NCCs that form and migrate Sox10Dom mice (Britsch et al., 2001). While the Sox10lacz in the absence of SOX10 undergo apoptosis before knockin is a loss-of-function mutation, the Sox10Dom reaching maturation stage. TUNEL analysis of Sox10- mutation may act as a dominant-negative allele. This Dom/Sox10Dom or Sox10LacZ/Sox10LacZ homozygous em- prediction is based on the observation that the TA- bryos at XE9 shows significant increase in cell death deleted forms of Sox10 show dominant negative proper- compared to their age-matched wild-type littermates ties in vitro (Potterf et al., 2000; 2001), suggesting that (Southard-Smith et al., 1998; Kapur, 1999; Paratore

Oncogene SOX10 function in neural crest development R Mollaaghababa and WJ Pavan 3030 et al., 2001; Sonnenberg-Riethmacher et al., 2001), (Paratore et al., 2001). This cell–cell signaling effect implicating SOX10 as essential in the survival of might either be mediated via a pathway that involves undifferentiated NCCs. These results are consistent with SOX10 or might be part of an independent pathway for the observed lack of Schwann cells and satellite glia gliogenesis. The ability of SOX10 to influence melano- within the DRG of Sox10LacZ/Sox10LacZ and Sox10Dom/ cyte differentiation is also documented. As it will be Sox10Dom homozygous embryos by E11.5 (Britsch et al., further discussed in the next section, SOX10 directly 2001). Sensory and motoneurons form in these homo- regulates the expression of the Microphthalmia (Mitf) zygous mutants but subsequently degenerate, presum- transcription factor, which plays a critical role in ably because of the absence of a surrounding glial cell promoting melanocyte differentiation (for a review, see population as SOX10 expression is not observed in Goding, 2000). Therefore, SOX10 function appears to marked neurons. Recent results show that is be critical for both the survival of NCC progenitors and involved in the mediation of apoptosis in mouse the proper differentiation of NC-derived glia and embryos that lack the paired homeodomain transcrip- melanocytes. Future studies with Sox10 transgenic mice tion factor Pax3 (Pani et al., 2002). Since PAX3 will provide a powerful tool for examining the lineage shows synergistic interaction with SOX10 in certain determination role of SOX10 in vivo. cellular contexts (Bondurand et al., 2000; Lang et al., Sox10 is also expressed in mouse oligodendrocyte 2000; Potterf et al., 2000), the programmed cell progenitors within both the spinal cord and the brain death in Sox10-deficient cells may also involve the P53 (Stolt et al., 2002; Figure 3), with expression in these pathway. Based on Brdu analysis, it has been suggested regions continuing into adulthood (Kuhlbrodt et al., that SOX10 is also important for cell proliferation in 1998a; Southard-Smith et al., 1999b). The spinal cord DRG between E9 and E11, with the most significant oligodendrocyte precursors are specified ventrally at effect observed at E10 (Sonnenberg-Riethmacher et al., E11.5 within the ventricular zone. As they commence 2001). However, the interpretation of these results is their terminal differentiation several days later, they complicated by the fact that at E10 a significant number express a number of different myelin proteins such as of DRG cells are undergoing apoptosis in Sox10- myelin-associated glycoprotein (MAG), proteolipid deficient embryos (Sonnenberg-Riethmacher et al., protein (PLP), and myelin basic protein (MBP), and 2001), raising the possibility that the observed effect show significant increase in numbers until birth. Unlike on proliferation is an indirect consequence of pro- the Schwann cell progenitors within the PNS, these grammed cell death. precursors survive in the absence of SOX10 function. Is the absence of specific NC derivatives within the However, they fail to differentiate into myelinating PNS of Sox10-deficient embryos an indirect conse- oligodendrocytes later in development, as indicated by quence of the death of undifferentiated NCCs? Or, is lack of MAG, PLP, and MBP expressions (Stolt et al., SOX10 also crucial in promoting the differentiation of 2002). These observations are consistent with the such lineage in addition to regulating cell survival? occurrence of PNS and CNS myelin deficiencies in an Under culture conditions in which the fate of individual individual with WS-IV who carries a presumed domi- NCCs can be followed, the surviving Sox10À/À NCCs nant-negative Sox10 mutation (Inoue et al., 1999), and fail to produce any glia in a differentiation medium that the occurrence of neurological features attributable to promotes gliogenesis in wild-type progenitors of NCCs myelin deficiency of CNS and PNS in several other WS- (Paratore et al., 2001). Instead, they give rise to P75- IV individuals (Pingault et al., 1998; Southard-Smith, (NCC marker) and NF160– (neuronal marker) cells that et al., 1999b; Touraine et al., 2000). The differential show non-neural morphology and often express the importance of Sox10 only for survival of glial precursors smooth muscle cell marker actin (SMA). Under similar in the PNS and not the CNS may be because of differentiation conditions, cultured haploinsufficient functional redundancies from other SOX proteins that Sox10+/À NCC clones also show altered fate determi- are present in oligodendrocyte progenitors. While nation compared to wild-type NCCs, without showing SOX10 appears to be significantly more abundant than any difference in survivability (Paratore et al., 2001). In other SOX proteins in Schawnn cells, in oligodendrocyte these haploinsufficient cultures prepared at clonal precursors SOX4 and SOX11 are also robustly ex- density, gliogenesis remains drastically reduced even in pressed prior to their downregulation at the time of the presence of neuregulin-1 (NRG1), a ligand for the differentiation (Kuhlbrodt et al., 1998b). Therefore, the tyrosine kinase receptor ErbB3, which significantly survival of Sox10-deficient oligodendrocyte precursors increases glial fate acquisition in wild-type NCC cultures may be a result of compensatory functions from SOX4 (Shah et al., 1994; Shah and Anderson, 1997; Adlkofer and/or SOX11 proteins. and Lai, 2000). These observations demonstrate the importance of SOX10 in glial fate determination within the PNS in addition to its role in promoting the survival of undifferentiated NCC precursors. The instructive SOX10 targets and synergistic interactions of SOX10 capacity of SOX10 for glial formation appears to be in with other transcription factors part dependent on cell–cell interactions as, in contrast to clonal cultures, haploinsufficient Sox10+/À NCCs that What are the genes that are regulated by SOX10 to are plated at high density are capable of generating promote glial and melanocyte differentiation and what significant numbers of glial cells upon NGR1 addition other factors are involved in activation of these genes?

Oncogene SOX10 function in neural crest development R Mollaaghababa and WJ Pavan 3031 Although these questions have not yet been fully SOX10 also regulates the expression of the Dopa- answered and continue to motivate research in this chrome tautomerase gene (Dct), which functions in area, a number of direct targets of SOX10 and melanin biosynthesis (Britsch et al., 2001; Potterf et al., transcription factors that show synergism with SOX10 2001). In Sox10Dom/+ heterozygous embryos at E11.5, have been identified (Table 1). In the melanocyte Dct expression is severely reduced in the trunk and lineage, SOX10 directly regulates the expression cranial regions (Potterf et al., 2001), and similar results of the Mitf gene (Bondurand et al., 2000; Lee et al., are obtained from comparison of Sox10LacZ/+ and wild- 2000; Potterf et al., 2000; Verastegui et al., 2000), a type embryos at E12.5 (Britsch et al., 2001). Consistent bHLH-LZ transcription factor that promotes differen- with these in vivo results, DCT+ cells are absent in age- tiation of NC-derived melanocytes (Hodgkinson matched neural tube explants obtained from Sox10Dom/ et al., 1993). Initially, it was observed that Sox10Dom/ +; Dct promoter-LacZ embryos and double stained for +; Mitfmi/+ double heterozygous mice show consider- the SOX10 protein and the DCT transgene (Potterf able increase in loss of pigmentation compared to et al., 2001). Intriguingly, Dct expression is significantly heterozygous mice that carry only one of the two restored later in development. Sox10Dom/+ heterozygous mutations (Potterf et al., 2000), prompting further embryos show a partial but significant restoration of Dct investigation of the link between these two transcription expression by E14.5, and DCT+ cells are observed in factors. Transient transfection assays showed that neural tube explants obtained from these embryos and SOX10 acts as a potent of the melanocyte- matched in age to E14.5 (Potterf et al., 2001). This specific promoter of the Mitf gene (M promoter) suggests that SOX10 regulation of Dct expression may (Bondurand et al., 2000; Lee et al., 2000; Potterf et al., be transient, with separate regulatory elements fulfilling 2000; Verastegui et al., 2000). Evidence suggests that this any need for Dct activation later in development. transactivation is direct as electrophoretic mobility shift Alternatively, it is possible that a gradual accumulation assays (EMSA) have demonstrated that SOX10 binds of the wild-type SOX10 protein partially overcomes the two of the identified target sites within this promoter effect of the Sox10Dom allele, allowing significant region with high affinity (Bondurand et al., 2000; Lee restoration of Dct expression with the passage of time. et al., 2000; Potterf et al., 2000; Verastegui et al., 2000). Further analysis of the Dct promoter region is needed to More recent results suggest the existence of an determine whether SOX10 directly transactivates the additional group of putative SOX10 binding sites in a Dct gene. more distal regulatory region of the M promoter Several direct targets of SOX10 have also been (Watanabe et al., 2002). Consistent with these observa- identified in the nervous system. SOX10 transactivation tions in mice, expression of the zebrafish Mitf homo- of the neuronal nicotinic acetylcholine receptor b4 logue, nacre, is absent in NCCs of colourless mutants and a3 subunit genes in vitro has been reported, (Dutton et al., 2001). with direct interaction of SOX10 with the b4 Two independent studies of Mitf activation by SOX10 gene promoter (Liu et al., 1999). The b4, a3, and a5 in HELA cells reported synergistic interaction of SOX10 subunits are the predominant receptor subtypes within with PAX3 to activate Mitf expression, consistent with the PNS (Conroy et al., 1993; Conroy and Berg, 1995). the presence of a PAX3 binding site in the Mitf Based on in vitro protein binding assays, it has been promoter (Bondurand et al., 2000; Potterf et al., 2000). postulated that SOX10 activation of the b4 gene This synergy is also observed in NIH3T3 cells (Ver- involves direct interactions between SOX10 and SP1, astegui et al., 2000). However, studies of Mitf activation SP3, Pura, and hnRNPK, all of which are binding by SOX10 in cell lines did not provide factors that affect b4 gene regulation (Melnikova et al., evidence for any SOX10–PAX3 synergy (Lee et al., 2000a, b). To test this hypothesis further, pull-down 2000; Verastegui et al., 2000). Aside from variables assays of whole-cell extracts need to be performed, using such as transfection efficiencies and protein levels, a specific anti-SOX10 antibody to demonstrate the these apparently contradictory results may simply presence of SOX10 in the putative complex. Also, in vivo reflect somewhat distinct mechanisms of action for regulation of b4 and a3 genes by SOX10 needs to be SOX10 and PAX3 in different cellular contexts (HELA tested. and NIH3T3 versus melanoma cell lines). In this regard, The SOX10 protein also directly regulates the c-ret it is noteworthy that SOX10–PAX3 synergism has also tyrosine receptor kinase gene, which plays an important been reported in activation studies of the c-ret gene role in the proper development of the enteric nervous using NIH3T3 and 293T cells (Lang et al., 2000). Since system (Lang et al., 2000). Interestingly, as with the PAX3 binding site within the Mitf promoter is regulation of Mitf, SOX10 shows synergistic interaction adjacent to the identified SOX10 binding sites, one with PAX3 to activate c-ret expression. Also, similar to might speculate that the observed synergism between the Mitf promoter, the identified PAX3 and SOX10 SOX10 and PAX3 involves direct protein–protein binding sites within the c-ret promoter are in close interactions. However, no evidence supporting this proximity (Lang et al., 2000). Mutations in c-ret and its model has been presented to date, and it is possible ligand, glial cell-derived neurotrophic factor (Gdnf), and that the SOX10–PAX3 synergy instead involves a in endothelin receptor B (Ednrb) and its ligand reconfiguration of the DNA at and around their endothelin 3 (Edn3), have also been linked to long- adjacent binding sites to allow an increased rate of segment HSCR (Gabriel et al., 2002). Activation of c-ret transcription. by SOX10 suggests that the same pathway may be

Oncogene SOX10 function in neural crest development R Mollaaghababa and WJ Pavan 3032 affected in HSCR mutations of these genes. Given the Future directions overlapping expression patterns of Ednrb and Sox10 in embryos (Southard-Smith et al., 1998) and similarity of Despite the continuous accumulation of knowledge their mouse mutant phenotypes (hypopigmentation and about SOX10 over the last few years, several lines of megacolon) (Hosoda et al., 1994; Pavan and Tilghman, investigation merit pursuit. While the phenotypes 1994), SOX10 and EDNRB might also function as parts associated with loss of Sox10 function have been of the same pathway. characterized, Sox10 gain-of-function phenotypes have In glia, the mouse P0 gene and human connexin 32 not been reported. Such phenotypes can be studied by (Cx32) genes have been shown to be direct targets of construction of Sox10 transgenic mice, addressing the SOX10 (Peirano et al., 2000; Bondurand et al., 2001). question of whether SOX10 can function as a lineage The P0 gene product is a highly abundant structural determination factor in vivo. Sox10 conditional knock- protein of the peripheral myelin sheath (Suter and out mice also need to be engineered and will provide a Snipes, 1995), and Cx32 is a gap junction protein powerful tool to study SOX10 function in later associated with myelin in the PNS and CNS (Scherer embryogenesis and during the postnatal period. Another et al., 1995). The direct regulation of these genes by area for further exploration is the discovery of SOX10 shows that, in addition to being important for additional genes that are regulated by SOX10. Although CNS myelination, SOX10 also has a critical role for several PNS targets of SOX10 have been determined, myelination within the PNS. Similar to the P0 promoter, the genes that are regulated by SOX10 in the CNS two adjacent SOX10 binding sites were found within the remain to be identified and other targets may also exist promoter of Cx32 gene, and in addition, target sites for in the melanocyte and PNS lineage. Furthermore, little the transcription factor EGR2 (human homologue of is known about the transactivating factors that regulate Krox20) were identified in the vicinity (Bondurand et al., expression of Sox10 itself. A full characterization of the 2001). Transfection assays show synergistic interaction Sox10 promoter region and other upstream regulatory between EGR2 and SOX10 in activating Cx32 gene, elements will help with their identification. One current although as with SOX10-PAX3 synergism, the mechan- candidate is the bHLH transcription factor Olig2, since ism of this cooperative interaction remains to be it is expressed prior to SOX10 in oligodendrocyte elucidated. In the PNS, expression of the erbB3 gene is progenitors and plays an essential role in oligodendro- also influenced by SOX10 (Britsch et al., 2001), although cyte differentiation (Lu et al., 2000; Zhou et al., 2000; Lu the presence of SOX10 binding sites within erbb3 gene et al., 2002; Zhou and Anderson, 2002). The possible promoter has not been reported and therefore it is not connection between SOX10 and the Olig genes needs to clear whether SOX10 regulation of this gene is direct. be investigated and any additional upstream regulators ErbB3 is an EGF receptor tyrosine kinase that binds need to be identified. Microarray analysis of cell-type- neuregulins and, similar to SOX10, has a critical specific libraries, such as the one characterized for the function in the development of neural crest derivatives melanocyte lineage (Loftus et al., 1999), can serve as a including Schwann cells (Riethmacher et al., 1997; powerful tool in the discovery of genes that act either Britsch et al., 2001). The upregulation of erbB3 by upstream or downstream of Sox10 in particular cell SOX10 was demonstrated in vivo by comparing wild- types. In addition, the use of genetic modifier screens in type and Sox10Dom/Sox10Dom embryos at E9.0 and E10.5, vivo and the NCC culture and the neural tube explant and in vitro by RT–PCR analysis (Britsch et al., 2001). culture systems in vitro (Stemple and Anderson, 1992; While erbB3 regulation is affected by SOX10 function in Dunn et al., 2000) should also prove invaluable tools in the PNS glia, control of erbB3 expression in the CNS these endeavors. Current data strongly suggest that, as oligodendrocytes does not require SOX10 function anticipated, SOX10 does not function in isolation and (Stolt et al., 2002). This suggests that the effects of has synergistic interactions with several other transcrip- SOX10 in peripheral and central glia are mediated tion factors. Elucidation of the underlying mechanisms through distinct mechanisms. This specificity may be of these synergistic interactions, including identification determined by a combinatorial function of SOX10 and of the factors that may directly bind to SOX10, will other transcription factors with which it shows syner- greatly enrich our understanding of how SOX10 gistic interaction. Besides PAX3 and ERG2, this list regulates transcription. Biochemical approaches will be may include the glial POU-domain protein Tst-1/Oct6/ indispensable tools for addressing these questions in the SCIP with which SOX10 shows synergistic interaction future. In short, we anticipate that the area of research on artificial promoters (Kuhlbrodt et al., 1998a, c; on SOX10 function and the mechanisms of neural crest Bondurand et al., 1999), and additional transcription development in general will remain very active and factors may also be involved. vibrant in the foreseeable future.

References

Adlkofer K and Lai C. (2000). Glia, 29, 104–111. Bondurand N, Girard M, Pingault V, Lemort N, Dubourg O Badner JA and Chakravarti A. (1990). Am. J. Med. Genet., 35, and Goossens M. (2001). Hum. Mol. Genet., 10, 2783–2795. 100–104. Bondurand NK, Kobetz A, Pingault V, Lemort N, Encha- Bianchi ME and Beltrame M. (2000). EMBO Rep., 1, Razavi F, Couly G, Goerich DE, Wegner M, Abitbol M and 109–114. Goossens M. (1998). FEBS Lett., 432, 168–172.

Oncogene SOX10 function in neural crest development R Mollaaghababa and WJ Pavan 3033 Bondurand N, Kuhlbrodt K, Pingault V, Enderich J, Sajus M, Kuhlbrodt K, Schmidt C, Sock E, Pingault V, Bondurand N, Tommerup N, Warburg M, Hennekam RCM, Read AP, Goossens M and Wegner M. (1998c). J. Biol. Chem., 273, Wegner M and Goossens M. (1999). Hum. Mol. Genet., 8, 23033–23038. 1785–1789. LaBonne C and Bronner-Fraser M. (1998). J. Neurobiol., 36, Bondurand N, Pingault V, Goerich DE, Lemort N, Sock E, 175–189. Caignec CL, Wegner M and Goossens M. (2000). Hum. Mol. Lane PW and Liu HM. (1984). J. Hered., 75, 435–439. Genet., 9, 1907–1917. Lang D, Chen F, Milewski R, Li J, Lu MM and Epstein JA. Bowles J, Schepers G and Koopman P. (2000). Dev. Biol., 227, (2000). J. Clin. Invest., 106, 963–971. 239–255. Le Douarin NM and Kalcheim C. (1999). The Neural Crest, Britsch S, Goerich DE, Riethmacher D, Peirano RI, Rossner 2nd edn. Cambridge University Press: Cambridge, UK. M, Nave K-A, Birchmeier C and Wegner M. (2001). Genes Lee M, Goodall J, Verastegui C, Ballotti R and Goding CR. Dev., 15, 66–78. (2000). J. Biol. Chem., 275, 37978–37983. Cheng Y-C, Cheung M, Abu-Elmagd MM, Orme A and Liu Q, Melnikova IN, Hu M and Gardner PD. (1999). J. Scotting PJ. (2000). Dev. Brain Res., 121, 233–241. Neurosci., 19, 9747–9755. Conroy WG and Berg DK. (1995). J. Biol. Chem., 270, 4424– Loftus SK, Chen Y, Gooden G, Ryan JF, Birznieks G, 4431. Hilliard M, Baxevanis AD, Bittner M, Meltzer P, Trent J Conroy WG, Vernallis AB and Berg DK. (1993). Neuron, 9, and Pavan WJ. (1999). Proc. Natl. Acad. Sci. USA, 96, 9277– 679–691. 9280. Denny P, Swift S, Connor F and Ashworth A. (1992). EMBO Loh SHY and Russell S. (2000). Mech. Dev., 93, 185–188. J., 11, 3705–3712. Lu QR, Sun T, Zhu Z, Ma N, Garcia M, Stiles CD and Dunn KJ, Williams BO, Li Y and Pavan WJ. (2000). Proc. Rowitch DH. (2002). Cell, 109, 75–86. Natl. Acad. Sci. USA, 97, 10050–10055. Lu QR, Yuk D, Alberta JA, Zhu Z, Pawlitzky I, Chan J, Dutton KA, Pauliny A, Lopes SS, Elworthy S, Carney TJ, McMahon AP, Stiles CD and Rowitch DH (2000). Neuron, Rauch J, Geisler R, Haffter P and Kelsh RN. (2001). 25, 317–329. Development, 128, 4113–4125. Melnikova IN, Lin H-R, Blanchette AR and Gardner PD. Erickson CA and Reedy MV. (1998). Curr. Top. Dev. Biol., 40, (2000a). Neuropharmacology, 39, 2615–2623. 177–209. Melnikova IN, Yang Y and Gardner PD. (2000b). Eur. J. Gabriel SB, Salomon R, Pelet A, Angrist M, Amiel J, Fornage Pharm., 393, 75–83. M, Attie-Bitach T, Olson JM, Hofstra R, Buys C, Steffann J, Mertin S, McDowall SG and Harley VR. (1999). Nucleic Acids Munnich A, Lyonnet S and Chakravarti A. (2002). Nat. Res., 27, 1359–1364. Genet., 31, 89–93. Meyer J, Wirth J, Held M, Schempp W and Scherer G. (1996). Goding CR. (2000). Genes Dev., 14, 1712–1728. Cytogenet. Cell Genet., 72, 246–249. Gubbay J, Collignon J, Koopman P, Capel B, Economou A, Osaki E, Nishina Y, Inazawa J, Copeland NG, Gilbert DJ, Munsterberg A, Vivian N, Goodfellow P and Lovell-Badge Jenkins NA, Ohsugi M, Tezuka T, Yoshida M and Semba R. (1990). Nature, 346, 245–250. K. (1999). Nucleic Acids Res., 27, 2503–2510. Harley VR, Lovell-Badge R and Goodfellow PN. (1994). Pani L, Horal M and Loeken MR. (2002). Genes Dev., 16, 676– Nucleic Acids Res., 22, 1500–1501. 680. Herbarth B, Pingault V, Bondurand N, Kuhlbrodt K, Paratore C, Goerich DE, Suter U, Wegner M and Sommer L. Hermans-Borgmeyer I, Puliti A, Lemort N, Goossens M (2001). Development, 128, 3949–3961. and Wegner M. (1998). Proc. Natl. Acad. Sci. USA, 95, Parisi MA and Kapur RP. (2000). Curr. Opin. Pediatr., 12, 5161–5165. 610–617. Hodgkinson CA, Moore KJ, Nakayama A, Steingrimsson E, Pavan WJ, and Tilghman SM. (1994). Proc. Natl. Acad. Sci. Copeland NG, Jenkins NA and Arnheiter H. (1993). Cell, USA, 91, 7159–7163. 74, 395–404. Peirano RI, Goerich DE, Riethmacher D and Wegner M. Hosoda K, Hammer RE, Richardson JA, Greenstein Baynash (2000). Mol. Cell. Biol., 20, 3198–3209. A, Cheung JC, Giaid A and Yanagisawa M. (1994). Cell,79, Peirano RI and Wegner M. (2000). Nucleic Acids Res., 28, 1267–1276. 3047–3055. Inoue K, Tanabe Y and Lupski JR. (1999). Ann. Neurol., 46, Pevny LH and Lovell-Badge R. (1997). Curr. Opin. Genet. 313–318. Dev., 7, 338–344. Kamachi Y, Uchikawa M and Kondoh H. (2000). Trends Pingault V, Bondurand N, Kuhlbrodt K, Goerich DE, Prehu Genet., 16, 182–187. M-O, Puliti A, Herbarth B, Hermans-Borgmeyer I, Legius Kanai Y, Kanai-Azuma M, Noce T, Saido TC, Shiroishi T, E, Matthijs G, Amiel J, Lyonnet S, Ceccherini I, Romeo G, Hayashi Y and Yazaki K. (1996). J. Cell. Biol., 133, 667– Smith JC, Read AP, Wegner M and Goossens M. (1998). 681. Nat. Genet., 18, 171–173. Kapur RP. (1999). Pediatr. Dev. Pathol., 2, 559–569. Pingault V, Guiochon-Mantel A, Bondurand N, Faure C, Kapur RP, Livingston R, Doggett B, Sweetser DA, Siebert JR Lacroix C, Lyonnet S, Goossens M and Landrieu P. (2000). and Palmiter RD. (1996). Dev. Biol., 174, 360–369. Ann. Neurol., 48, 671–676. Knecht AK and Bronner-Fraser M. (2002). Nat. Rev., 3, 453– Potterf SB, Furumura M, Dunn KJ, Arnheiter H and Pavan 461. WJ. (2000). Hum. Genet., 107, 1–6. Korsch E, Steinkuhle J, Massin M, Lyonnet S and Touraine Potterf SB, Mollaaghababa R, Hou L, Southard-Smith EM, RL. (2001). Eur. J. Pediatr., 160, 68–69. Hornyak TJ, Arnheiter H and Pavan WJ. (2001). Dev. Biol., Kuhlbrodt K, Herbarth B, Sock E, Enderich J, Hermans- 237, 245–257. Borgmeyer I and Wegner W. (1998b). J. Biol. Chem., 273, Prior HM and Walter MA. (1996). Mol. Med., 2, 405–412. 16050–16057. Pusch C, Hustert E, Pfeifer D, Sudbeck P, Kist R, Roe B, Kuhlbrodt K, Herbarth B, Sock E, Hermans-Borgmeyer I and Wang Z, Balling R, Blin N and Scherer G. (1998). Hum. Wegner M. (1998a). J. Neurosci., 18, 237–250. Genet., 103, 115–123.

Oncogene SOX10 function in neural crest development R Mollaaghababa and WJ Pavan 3034 Read AP. (2000). Adv. Otorhinolaryngol., 56, 32–38. Stolt CC, Rehberg S, Ader, M, Lommes P, Riethmacher D, Riethmacher D, Sonnenberg- Riethmacher E, Brinkmann V, Schachner M, Bartsch U and Wegner M. (2002). Genes Dev., Yamaai T, Lewin GR and Birchmeier C. (1997). Nature, 16, 165–170. 389, 725–730. Suter U and Snipes GJ. (1995). Annu. Rev. Neurosci., 18, 45–75. Scherer SS, Deschenes SM, Xu YT, Grinspan JB, Fischbeck Touraine RL, Attie-Bitach T, Manceau E, Korsch E, Sarda P, KH and Paul DL. (1995). J. Neurosci., 15, 8281–8294. Pingault V, Encha-Razavi F. Pelet A, Auge J, Nivelon- Shah NM and Anderson DJ. (1997). Proc. Natl. Acad. Sci. Chevallier A, Holschneider AM, Munnes M, Doerfler W, USA, 94, 11369–11374. Goossens M, Munnich A, Vekemans M and Lyonnet S. Shah KN, Dalal SJ, Desai MP, Sheth PN, Joshi NC and (2000). Am. J. Hum. Genet., 66, 1496–1503. Ambani LM. (1981). J. Pediatr., 99, 432–435. Uchikawa M, Kamachi Y and Kondoh H. (1999). Mech. Dev., Shah NM, Marchionni MA, Isaacs I, Stroobant P and 84, 103–120. Anderson DJ. (1994). Cell, 77, 349–360. van de Wetering M and Clevers H. (1993). Nucleic Acids Res., Sham MH, Lui VCH, Chen BLS, Fu M and Tam PKH. 21, 1669. (2001a). J. Med. Genet., 38(9), E30. Verastegui C, Bille K, Ortonne J-P and Ballotti R. (2000). J. Sham MH, Lui VCH, Fu M, Chen B and Tam PKH. (2001b). Biol. Chem., 275, 30757–30760. Gut, 49, 220–226. Watanabe K-I, Takeda K, Katori Y, Ikeda K, Oshima T, Sonnenberg-Riethmacher E, Miehe M, Stolt CC, Goerich DE, Yasumoto K-I, Saito H, Takasaka T and Shibahara S. Wegner M and Riethmacher D. (2001). Mech. Dev., 109, (2000). Mol. Brain Res., 84, 141–145. 253–265. Watanabe K-I, Takeda K, Yasumoto K-I, Udono T, Saito H, Southard-Smith EM, Angrist M, Ellison JS, Agarwala R, Ikeda K, Takasaka T, Takahashi K, Kobayashi T, Baxevanis AD, Chakravarti A and Pavan WJ. (1999b). Tachibana M and Shibahara S. (2002). Pigm. Cell Res., 15, Genome Res., 9, 215–225. 201–211. Southard-Smith EM, Collins JE, Ellison JS, Smith KJ, Wegner M. (1999). Nucleic Acids Res., 27, 1409–1420. Baxevanis AD, Touchman JW, Green ED, Dunham I and Weiss MA. (2001). Mol. Endocrinol., 15, 353–362. Pavan WJ. (1999a). Mamm. Genome., 10, 744–749. Wright EM, Snopek B and Koopman R. (1993). Nucleic Acids Southard-Smith EM, Kos L and Pavan WJ. (1998). Nat. Res., 21, 744. Genet., 18, 60–64. Zhou Q and Anderson DJ. (2002). Cell, 109, 61–73. Stemple DL and Anderson DJ. (1992). Cell, 71, 973–985. Zhou Q, Wang S and Anderson DJ. (2000). Neuron, 25, 331–343.

Oncogene