The Yin and Yang of Sox proteins: Activation and repression in development and disease
Abstract
The general view of development consists of the acquisition of committed/differentiated phenotypes following a period of self-renewal and progenitor expansion. Lineage specification and progression are phenomena of antagonistic events, silencing tissue-specific gene expression in precursors to allow self-renewal and multipotentiality, and subsequently suppressing proliferation and embryonic gene expression to promote the restricted expression of tissue-specific genes during maturation. The high mobility group-containing Sox family of transcription factors constitutes one of the earliest classes of genes to be expressed during embryonic development. These proteins not only are indispensable for progenitor cell specification but also are critical for terminal differentiation of multiple cell types in a wide variety of lineages. Sox transcription factors are now known to induce or repress progenitor cell characteristics and cell proliferation or to activate the expression of tissue-specific genes. Sox proteins fulfill their diverse functions in developmental regulation by distinct molecular mechanisms. Not surprisingly, in addition to DNA binding and bending, Sox transcription factors also interact with different protein partners to function as coactivators or corepressors of downstream target genes. Here we seek to provide an overview of the current knowledge of Sox gene functional mechanisms, in an effort to understand their roles in both development and pathology. © 2009 Wiley-Liss, Inc.
The Sox protein family of transcription factors has been identified as one of the most important groups of developmental regulators both in vertebrates and in invertebrates (Bowles et al.,2000; Wegner,1999). The biological functions of these proteins during embryonic and postnatal development have been defined in a variety of tissues and cell types. Furthermore, the roles of Sox transcription factors are also being investigated in disease and in cell repair. The Sox family has no single biological function, and, among their multiple roles, Sox transcription factors induce or suppress progenitor cell properties, such as proliferation and multipotentiality, or initiate differentiation programs by activating the expression of tissue-specific genes. Despite overlapping functions within families, each Sox protein modulates a unique set of target genes in a specific cell type. There is already a large body of evidence supporting the generation of specific Sox-mediated cellular responses through mechanisms of DNA bending and binding and recruitment of different interacting protein partners to function as coactivators or corepressors of downstream target genes.
As is expected of developmental regulators with multiple functions, Sox transcription factors display very diverse tissue-specific expression patterns during embryonic and postnatal development. These expression patterns are modified not only during normal development but also in a variety of pathological states, to the extent that some members of the Sox family are considered important diagnostic markers of childhood brain tumors (de Bont et al.,2008).
All these findings continue to raise a number of important issues that are crucial to understanding the relationship between the molecular properties of Sox proteins and the extreme diversity of their functional roles. The molecular mechanisms of gene activation or gene repression mediated by Sox proteins and the functional redundancy of these transcription factors represent important areas for current and future research. In this Mini-Review, we highlight functional aspects of Sox proteins in the context of developmental events, with a particular focus on Sox proteins as transcriptional activators and repressors, in an effort to understand their roles in both development and disease.
THE Sox SUBFAMILY WITHIN THE HMG SUPERFAMILY: CLASSIFICATION INTO Sox SUBGROUPS AND SOME MOLECULAR PROPERTIES
The Sox transcription factors bind to the minor groove in DNA and were first identified based on molecular conservation of the 79-amino-acid HMG (high mobility group) DNA binding domain found in the gene for the mammalian testis-determining factor SRY (Gubbay et al.,1990). Almost all Sox (SRY-BOX) genes display at least 50% amino acid similarity with the Sry HMG box. To date, the total number of Sox transcription factors identified is approximately 30 (20 in humans and mice, and eight in Drosophila), subgrouped into 10 distinct families (A–J), based not only on homology within the HMG domain and other structural motifs but also on functional properties (Schepers et al.,2002). These groups include: A, Sry; B, Sox1, -2, -3, -14, and -21; C, Sox4, -11, and -12; D, Sox5, -6, and -13; E, Sox8, -9, and -10; F, Sox7, -17, and -18; G, Sox15 and -20; and H, Sox30. The existence of two new subgroups (I and J) has been postulated, including Xenopus Sox31 and C. elegans SoxJ, respectively (Bowles et al.,2000). A thorough molecular analysis of the evolutionary history of the SOX family was based on the entire number of available HMG domain sequences and full-length protein sequences (Wegner,1999; Bowles et al.,2000). This study demonstrated that the majority of the SOX subfamilies defined in vertebrates were also represented in invertebrates by a single Sox gene.
All Sox factors recognize a similar DNA binding motif A/TA/TCAAA/TG, but their amino acid sequences outside the HMG domain are highly variable and display domains that favor binding with other regulatory protein partners. Molecular interactions with other transcriptional regulators are crucial for each SOX factor to recognize a specific target gene, because each protein of this family is expressed in many different cellular contexts, and a specific cell type can coexpress many SOX factors (Wegner,1999). SOX proteins can form stable transcription factor complexes with a variety of coregulators to activate or repress gene transcription by modulating promoter activity. A typical example of this cooperative interaction is the demonstration that SOX2 and PAX6 activate transcription of the lens-specific δ-crystallin gene by acting on its DC5 enhancer (Kamachi et al.,2001). There are also several examples of SOX/OCT complexes that regulate expression of different genes. SOX2 and SOX3 interact with Oct3/4 to regulate Fgf4, UTF1, and osteopontin gene expression in the embryo (Yuan et al.,1995; Botquin et al.,1998; Nishimoto et al.,1999; Kamachi et al.,2000). SOX and Oct binding sites are necessary for enhancer activity and cell specificity of the Hoxb1 gene, which encodes a protein involved in hindbrain patterning. A single SOX protein can also simultaneously recruit more than one transcription factor to regulate gene expression from a single promoter or enhancer (Ma et al.,2000) or can promote binding with transcriptional repressors to promoters, as demonstrated by the SOX6/CtBP2 interaction (Murakami et al.,2001).
It is currently thought that the HMG domain itself plays a crucial role not only in the DNA-binding of SOX factors but also in their interaction with other transcriptional coregulators and in nuclear import (Wilson and Koopman,2002). However, because HMG domains are interchangeable between SOX factors (Bergstrom et al.,2000), other parts of the protein outside the HMG domain are likely to play a major role in selecting specific protein partners and generate cell- and developmental stage-specific functions for SOX transcription factors (Wilson and Koopman,2002). A recent study of SoxE binding partners revealed specific interactions with Oct3/4, C/EBP, Olig2, Pax3, and SP1 but not the receptors for estrogen or thyroid hormone (Wibmuller et al.,2006), which supports the pleiotropic nature of SOX protein activity, although the manner in which cell type or temporal specificity is determined is not understood.
Sox PROTEINS IN MECHANISMS INVOLVING FUNCTIONAL ANTAGONISM DURING DIFFERENTIATION
SoxB1 (Sox1–3)
Although it is well established that neural progenitors in the CNS are prevented from premature differentiation, the mechanism by which this occurs is poorly understood. The expression of SoxB1 members in dividing CNS progenitor cells, and not in those expressing definitive neuronal markers such as NeuN, Tuj1, and Lim2 (Bylund et al.,2003), strongly suggests a role for these SoxB1 transcription factors in the maintenance of the progenitor state. Indeed, inhibition of Sox2 signaling resulted in enhanced neuronal differentiation (Graham et al.,2003). Consistently with this hypothesis, it was demonstrated that these Sox genes 1–3, with some measure of redundancy (Graham et al.,2003), prevented neurogenesis by antagonizing the function of proneural bHLH proteins, namely, neurogenin (Ngn), and were down-regulated once progenitor cells left the cell cycle (Bylund et al.,2003). SoxB1 members bear transactivation domains, and the conclusion that they function as transcriptional activators was revealed by a hybrid HMG domain containing the Engrailed protein repressor module. HMG-VP16 activator suppressed neuronal differentiation similarly to intact Sox1–3, whereas HMG-EnR generated postmitotic neurons (Bylund et al.,2003). These Sox genes collectively specify and maintain stem cells and progenitors, but differential functions for them are being discovered in more recent studies. Although Sox1–3 are down-regulated in mature neurons, the expression of Sox1, though not required for neurosphere generation, was shown to be necessary for the differentiation of neural progenitors from the ganglionic eminence and dorsal telencephalic wall into GAD65+ neuronal progenitors (Kan et al.,2007). Sox1-null mutant mice also lack telencephalic neurons that constitute the ventral striatum (Ekonomou et al.,2005). Indeed, only Sox1, not Sox2 or -3, was shown to be able to induce neuronal lineage commitment by repressing Hes1 gene expression (Kan et al.,2004). In addition to up-regulating the transcription of Ngn1, Sox1 suppresses beta-catenin-mediated TCF/LEF signaling to attenuate the Wnt pathway and promotes cell cycle exit, leading to neuronal cell differentiation (Kan et al.,2004). Like Sox1, Sox3 is transiently expressed by proliferating and differentiating neural progenitors in the olfactory bulb and dentate gyrus, while persisting in specific postmitotic neuronal populations (Wang et al.,2006), although the mechanism by which Sox3 induces neurogenesis is unclear.
In contrast to Sox1 or Sox3, Sox2 has been more consistently associated with multipotentiality and progenitor cell proliferation (Rex et al.,1997). Sox2 expression is itself subject to repression by histone deacetylase activity (Lyssiotis et al.,2007), which is important for oligodendrocyte maturation (Shen et al.,2005,2008). Sox2 expression was restricted to proliferating progenitor cells, and its overexpression was observed to inhibit neurogenesis (Bani-Yaghoub et al.,2006). Sox2 is thought to maintain cellular proliferative potential through the up-regulation of Notch1 (Bani-Yaghoub et al.,2006) for the purpose of generating sufficient numbers of progenitors (Ellis et al.,2004). Sox2-deficient neural stem cells, however, remain multipotent, likely as a result of compensation by Sox3, but the capacity for generating sufficient neurons is greatly reduced (Miyagi et al.,2008). This notion of progenitor maintenance is in agreement with the finding of high levels of Sox2 expression in glial tumors of astroglial, oligodendroglial, and ependymal lineages (Phi et al.,2008), but not in neuronal tumors. Possible molecular mechanisms have been offered from studies in nonneural systems and in embryonic stem cells, implicating the regulation of cyclin D1 expression as a primary function of Sox2 activity. In breast cancer cells, modulation of Sox2 activity by overexpression or knockdown had little effect on other cell cycle-regulated genes, including cyclin D3, cyclin E, p21Cip, and p27Kip (Chen et al.,2008). The authors also demonstrated that Sox2 and beta-catenin interact directly to activate the cyclin D1 proximal promoter synergistically (Chen et al.,2008). In embryonic stem cells, Sox2 and Oct4 bound to the promoter of miR-302a—a cluster of miRNAs specifically expressed in stem cells and pluripotent cells—to stimulate its expression. miR-302a repressed the translation of cyclin D1 and promoted a decrease in G1 cells and an increase in S-phase cells (Greer Card et al.,2008). These observations are consistent with the notion of Sox2 as a promitotic regulator, and, although cell cycle regulation alone is insufficient to alter cell-fate decisions through the modulation of cyclin D1 expression (Lobjois et al.,2008), SoxB1 proteins function to maintain neural precursors by mechanisms that may involve the Notch pathway. Sox2 and Sox1, respectively, activate (Bani-Yaghoub et al.,2006) and inhibit (Kan et al.,2004) Notch signaling, but there is also evidence that SoxB1 and Notch inhibit neuronal maturation by distinct mechanisms (Holmberg et al.,2008).
SoxB2 (Sox14, Sox21)
The SoxB2 group possesses repression domains, and Sox21 and Sox14 inhibit the activation of the delta-crystallin DC5 enhancer to different extents (Uchikawa et al.,1999). Sox21 promotes neurogenesis by counteracting the activities of Sox1–3, and the ability of proneural proteins to promote lineage progression relies on their ability to up-regulate Sox21 expression (Sandberg et al.,2005). Ngn2 stimulates Sox21 expression, and the latter in turn is associated with increased cell cycle exit, although the detailed mechanism is presently unknown.
SoxC (Sox4, Sox11, Sox12)
As activators, Sox4 and Sox11 have been shown to induce a neuronal phenotype in immature neuronal precursors of the early chicken embryo (Bergsland et al.,2006) in a manner that is independent of cell cycle exit. In the oligodendrocyte lineage, however, prolonged expression of Sox4 in vivo under the control of the MBP promoter results in reduced and delayed myelin gene expression in the spinal cord (Potzner et al.,2007). High levels of Sox11 are also found in human gliomas, in accordance with the degree of cell dedifferentiation (Weigle et al.,2005). In further support of an inhibitory role of SoxC in oligodendrocyte development, the ablation of YY1, which resulted in hypomyelination, also increased the levels of oligodendroglial-lineage repressors, including Sox11 (He et al.,2007). These SoxC proteins thus appear to possess opposing effects on neuronal vs. glial differentiation, promoting one and repressing the other. The mechanism of SoxC protein repression is not understood, but, because SoxC proteins, unlike SoxD proteins, do not interfere with Sox10-dependent up-regulation of myelin genes (Potzner et al.,2007), it is believed that the SoxC-induced hypomyelination is an indirect effect on the MBP gene.
SoxD (Sox5, L-Sox5, Sox6, Sox13)
Members of this group possess an HMG domain but are missing transactivation domains, suggesting roles in structural organization in gene expression. Sox5 and Sox6 are essential for neuronal and chondrocyte development, and Sox6 is also an important enhancer of erythropoiesis (Dumitriu et al.,2008). Sox5 was recently found to prevent the premature development of later-born neurons and thus regulates the timing of neocortical neuronal subtypes (Lai et al.,2008). Sox5 promotes cell proliferation by repressing the expression of SPARC (Huang et al.,2008), a protein believed to be a tumor suppressor. In chondroblast differentiation, Sox5 and Sox6 promote the development of a proliferating pool of chondroblasts and act by down-regulating Indian hedgehog signaling and the expression of FGFR3 and Runx2 (Smits et al.,2004).
In neural crest development, Sox5 is expressed in early neural crest cells and continues to be expressed in melanocytes, but its loss does not affect melanoblast generation. However, the loss of Sox5 rescues melanoblast generation in Sox10-reduced mice. Sox5 binds to and represses melanocytic target genes of Sox10, such as Mitf and Dct (Stolt et al.,2008). Sox5 alone did not modulate these genes directly but did so only in the presence of Sox10 (see below under SoxE proteins). Competition for the same DNA binding site is possible, but it appears more likely that, in situations with multiple Sox recognition sites, the recruitment of corepressors, C-terminal binding protein 2 (CtBP2) and HDAC1 to the promoters of Mitf and Dct genes, could be more effective. Sox5 does not function exclusively as a corepressor but also recruits coactivators to Sox9 target genes in chondrocytes (Hattori et al.,2008).
Sox6 promotes cardiomyocyte differentiation, an event associated with the down-regulation of an L-type Ca2+ alpha1c gene. There are eight consensus Sox recognition sites in 1.6 kb of the Ca2+ alpha1c promoter, and overexpression of Sox6 suppresses reporter expression in constructs bearing this promoter region (Cohen-Barak et al.,2003). Prtb (proline-rich transcript of the brain), which interacts with Sox6, represses the activity of the alpha1c gene promoter alone, as does Sox6 alone, but, interestingly, repression of reporter expression is abolished by coexpression of Prtb with Sox6 (Cohen-Barak et al.,2003). This suggests that Prtb and Sox6 antagonize each other as repressors. Skeletal muscle development consists of two successive waves of myogenesis, embryonic and fetal, generating slow and fast fibers, respectively. Sox6 mutant mice show increased slow fiber type-specific gene expression and reduced fast fiber gene expression (Hagiwara et al.,2005). Sox6 has been demonstrated to repress the transcription of slow fiber type-specific genes typified by the myosin heavy chain-beta (MyHCβ), slow isoform (Hagiwara et al.,2007).
During erythropoiesis, definite erythroid cells in the fetal liver express adult beta globins, while the epsilon gene is silenced. Sox6 represses embryonic epsilon gene expression by directly binding its promoter (Yi et al.,2006). Sox6 has been shown to silence other targets in various systems: 1) the Fgf-3 promoter through the recruitment of a corepressor, CtBP2 (Murakami et al.,2001); 2) the insulin II gene by serving as an interacting corepressor of the homeobox factor PDX1 (Iguchi et al.,2005); and 3) the cyclin D1 promoter by interacting with beta-catenin and histone deacetylase (Iguchi et al.,2007).
Sox13 is expressed at high levels in a subset of neural progenitor cells as they exit from the cell cycle and migrate away from the ventricular zone (Wang et al.,2005). In T-lymphocyte differentiation, Sox13 promotes the development of gammaDelta lineage from progenitors while suppressing differentiation to the alphaBeta lineage (Melichar et al.,2007). This fate determination was mediated through inhibition of Wnt signaling, not by interacting with beta-catenin per se but instead by antagonizing its coactivator T-cell factor 1 (TCF1). Sox13 also down-regulated a known TCF1 target gene, Ly49a, in EL4 T cells. Finally, Sox13-null mice showed enhanced rates of thymocyte proliferation, indicating that Sox13 antagonism of Wnt/TCF signaling mediated both lineage commitment and proliferative arrest (Melichar et al.,2007).
SoxE (Sox8, Sox9, Sox10)
SoxE proteins promote neural crest progenitor formation as well as differentiation of neural crest-derived melanoblasts and glia (Bondurand et al.,2000; Britsch et al.,2001; Kim et al.,2003; Hong and Saint-Jeannet,2005; Taylor and LaBonne,2005), oligodendrocyte differentiation (Stolt et al.,2002,2003,2004), and cartilage development (Lefebvre et al.,1998; Sekiya et al.,2000; Akiyama et al.,2002).
SoxE, in particular Sox9, acts in two phases of neural crest development: promotion of neural crest properties and of glial and melanocyte fate, while suppressing neuronal fates (Cheung and Briscoe,2003; Kim et al.,2003). Sox9 suppresses the expression of neuronal markers Pax6, Pax7, Nkx6, and Irx3, and its mode of action appears to be independent of BMP or Wnt signaling (Cheung and Briscoe,2003). In the neural crest, Sox10 acts to maintain characteristics of multipotentiality in neural crest stem cells (Kim et al.,2003) and modulates neural crest progenitor properties by promoting glial development (Britsch et al.,2001) and by overcoming the antigliogenic and antineurogenic activities of BMP2 and TGFbeta, respectively (Kim et al.,2003). The loss of Sox10 from Schwannoma cells promotes differentiation into myofibroblasts with carbachol-stimulated contraction and calcium transients, indicating the role and bias of this SoxE protein in repressing myofibroblast-specific genes in a cell system known to transdifferentiate into several derivatives (Roh et al.,2006).
SoxE proteins are responsible for oligodendrocyte specification and maturation, with Sox9 and Sox10 playing critical roles in determining progenitor fate commitment and terminal differentiation. Initially, both proteins appear to function redundantly in oligodendrocyte precursors, but Sox9, whose expression precedes that of Sox10 and Sox8, has been shown to be essential for the neuron-glia fate switch, suppressing the formation of spinal cord motoneurons, V2 interneurons, and gray matter astrocytes. On the other hand, Sox10, which continues to be expressed in developing oligodendrocytes and astrocytes, is essential for myelination (Stolt et al.,2002,2003,2004). More recently, simultaneous deletion of Sox9 and Sox10 revealed a joint function in the maintenance of PDGFRα+ oligodendrocyte progenitor cells (Finzsch et al.,2008), which underscores the importance of multiplicity of progenitor function of Sox9. SoxD family members Sox5 and Sox6 are now known to modulate SoxE protein function in oligodendrocyte development by preventing precocious oligodendrocyte specification (Stolt et al.,2006).
During skeletogenesis, cartilage and bone progenitors arise from a common precursor, which is specified by Sox9 (Akiyama et al.,2004,2005). Osteoblast and chondrocyte development are mutually exclusive events: Wnt/beta-catenin signaling determines fate decisions between osteoblast and chondrocyte lineages (de Crombrugghe et al.,2001; Day et al.,2005; Hill et al.,2005). Sox8 expression is down-regulated upon differentiation of osteoblasts; it specifically prevents untimely osteoblast maturation (Schmidt et al.,2005) and premature expression of the run-related transcription factor 2 (Runx2), a master regulator of osteoblast differentiation and bone formation. Sox9 inhibits osteoblast development and osteoblast developmental regulators such as beta-catenin (Day et al.,2005), and overexpression of Sox9 drives cartilage formation, a process that is mimicked by beta-catenin deletion. The reverse is also observed (Akiyama et al.,2004). Furthermore, the commitment of progenitors to produce cartilage by the process of chondrogenesis is also effected through Sox9 by activating extracellular matrix genes aggrecan, Col11a2, and Col2a1 (Bridgewater et al.,1998; Sekiya et al.,2000; Kawakami et al.,2005). Interestingly, the recruitment of Sox9 to the aggrecan gene enhancer element is itself facilitated by SoxD members L-Sox5 and Sox6 (Han and Lefebvre,2008).
In promoting chondrogenesis, Sox9 induces cell cycle arrest or exit and reduces the levels of the mitogen-activated protein kinase ERK1 (Panda et al.,2001), whose inhibition had previously been found to promote chondrogenesis (Chang et al.,1998). Mechanistically, LEF/TCF transcription factors, which are beta-catenin coactivators, bind the same domain of beta-catenin as does Sox9. The binding of beta-catenin by Sox9 prevents LEF/TCF binding, inhibiting the expression of Wnt target genes. Interaction with beta-catenin could also inhibit the transactivating functions of Sox9, leading to the eventual degradation of the Sox9/beta-catenin complex. In addition to beta-catenin, Sox9 also represses Runx2 function in osteoblasts (Zhou et al.,2006). The loss of Sox9 derepresses the expression of Runx1, Runx2, and Runx3 as well as target genes of Runx2 such as Col10a1, indicating that Sox9-Runx2 antagonism regulates skeletal fate determination.
In the intestinal epithelium, Sox9 ablation results in increased cell proliferation and hyperplasia (Bastide et al.,2007). The hyperplastic crypts of the Sox9-deficient intestine were found to overexpress Wnt pathway-related genes, such as c-Myc and cyclin D1, as well as increase the number of cells expressing these genes. Direct interaction between Sox9 and beta-catenin could not be demonstrated by coimmunoprecipitation in this study; however, a mutated DNA binding domain of Sox9 ablated the ability of Sox9 to modify beta-catenin-TCF activity. Interestingly, when the transactivating and beta-catenin interacting domains of Sox9 were replaced with a VP16 transactivating domain, inhibition of beta-catenin-TCF activity surpassed that of the native Sox9. This observation indicated that the action of Sox9 on beta-catenin function is likely to be mediated by the induced expression of an inhibitor of beta-catenin-TCF (Bastide et al.,2007), which would agree well with the findings that many Sox proteins of different classes, which lack homology outside of their HMG domains, possess the conserved property of inhibiting beta-catenin activity.
In myogenesis, Sox8 and Sox9 inhibit the MyoD-induced conversion of C3H 10T1/2 cells into myoblasts and repress the expression of myogenin by MyoD. This supports the notion that Sox8 and Sox9 are responsible for maintaining myoblasts in an undifferentiated state to prevent precocious differentiation into myotubes (Schmidt et al.,2003).
In all three systems of melanocyte, oligodendrocyte, and chondrocyte development, SoxD and SoxE proteins are coexpressed. Whereas Sox5 and Sox6 costimulate Sox9 target genes in chondrocytes (Han and Lefebvre,2008), Sox5 was recently shown to repress the transcriptional activity of Sox10 during melanocyte development from the neural crest (Stolt et al.,2008) by binding to Sox10 response elements in melanocytic target genes, Mitf and Dct, in which multiple binding sites for Sox10 have been mapped.
In the oligodendrocyte lineage, Sox5 and Sox6 are coexpressed with Sox10 in progenitor cells that have not yet begun to express MBP, and ablation of Sox5 and Sox6 results in precocious specification of ventricular zone cells to oligodendrocyte progenitor cells (Stolt et al.,2006). Both SoxD proteins repress terminal differentiation of oligodendrocytes by interfering with SoxE-dependent stimulation of myelin gene expression (Stolt et al.,2006), so that binding of either class of protein to known Sox10-responsive elements in vitro was found in gel-shift assays to be mutually exclusive. With chromatin immunoprecipitation, however, both Sox6 and Sox10 were found at the Sox10-responsive MBP promoter at E16.5, but only Sox10 was present on these sites in postnatal spinal cord (Stolt et al.,2006). It would appear that, in myelin regulation, the nature of repression by SoxD includes competition for interacting proteins, insofar as both Sox5 and Sox6 interacted with Olig2 and Neurogenin 2 in GST-pulldown assays. These observations attest to the mode of SoxD protein activity: repression by competition for DNA or coactivator binding, or selective recruitment of corepressors at specific promoters. The last possibility is believed to be the likely mechanism for Sox5 in the neural crest (Stolt et al.,2008). Indeed, Sox9 promotes Sox10 expression in Neuro2A cells, and coexpression of either Sox5 or Sox6 with Sox9 ablated Sox9-dependent Sox10 expression (Stolt et al.,2003), indicating that Sox5 and Sox6 could interfere with the ability of Sox9 to regulate the Sox10 promoter. The precise mechanisms regarding Sox9-dependent Sox10 expression await elucidation.
SoxF (Sox7, Sox17, Sox18)
This group has been shown to be critical for endoderm and hair development (Kanai-Azuma et al.,2002; Irrthum et al.,2003), cardiogenesis (Liu et al.,2007), and angiogenesis (Matsui et al.,2006; Young et al.,2006). The gene targets of SoxF proteins identified to date, such as FoxA and HNF1beta (Sinner et al.,2004), laminin alpha1 (Niimi et al.,2004), VCAM1 (Matsui et al.,2006), and FGF-3 (Sinner et al.,2004) are dependent on their transactivation functions, and few specific repressed targets have been described. In respiratory epithelial cells, Sox17 promotes transdifferentiation of alveolar type II cells into epithelial cells and stimulates promoter activities of mFoxj1 while repressing those of mSftpc, an alveolus-specific gene (Park et al.,2006). However, repression of gene expression by Sox17 may also be indirect. In the endoderm, Sox17 induces the expression of a zinc-finger protein, Zfp202, which represses the expression of the transcription factor HNF4a (Patterson et al.,2008); thus gene targets of Sox17 transactivation have turned out to be transcriptional repressors themselves. Currently, the best established mechanism of repression by SoxF proteins is the interference of Wnt signaling, as is frequently observed with Sox proteins of other subfamilies. In the endoderm, Sox17 cooperates with beta-catenin to induce endodermal genes, and there is a shared sequence motif between Sox17 and Sox7 that mediates the physical interaction between beta-catenin and Sox17 (Sinner et al.,2004). Deletion of this motif abolished binding to beta-catenin and partially reduced beta-catenin-mediated transcriptional activity in lung cancer cells (Guo et al.,2008). The Sox7 promoter was found to be hypermethylated in prostate cancer cell lines and colorectal cancers, and demethylation of the Sox7 promoter resulted in a dramatic increase in Sox7 expression. Ectopic expression of Sox7 in Sox7-null cells inhibited the proliferation of colon cancer cells, indicating that Sox7 inactivation is involved in cell proliferation and tumor progression (Guo et al.,2008). Sox17 interacts with both TCF and beta-catenin (Sinner et al.,2007), and both domains onSox17 are required to antagonize Wnt signaling by promoting their proteasomal degradation.
Dual roles for Sox17 in promoting cell cycle exit and progenitor maturation were also revealed in the oligodendrocyte lineage, where reduced progenitor cell proliferation and increased morphological differentiation were simultaneously induced by the overexpression of Sox17 (Sohn et al.,2006). Understanding Wnt pathway modulation by Sox proteins could help to demystify the complex relationship between cell cycle control and lineage progression. The ability of SoxF members to regulate Wnt pathway signaling was further highlighted in a recent study comparing the repression of beta-catenin/TCF activity by various Sox family members (Sinner et al.,2007). Both Sox10 and Sox17 are potent inhibitors of beta-catenin-mediated transcription (Sinner et al.,2007) and are activators of myelin gene expression in the oligodendrocyte lineage (Stolt et al.,2002; Sohn et al.,2006), whereas members of SoxC (Sox4 and -11) were potent activators of Wnt signaling (Sinner et al.,2007), with Sox4 recently shown to be an inhibitor of oligodendrocyte maturation (Potzner et al.,2007). The activities of Sox proteins as activators and repressors are summarized in Tables I and II.
| Family | Function | References |
|---|---|---|
| SoxB1 | SoxB1 members activate differentiation of telencephalic precursors into neurons | Ekonomou et al.,2005 |
| Sox1 enhances neurogenin expression | Kan et al.,2004 | |
| Sox2 activates cyclinD1 expression with β-catenin in breast cancer | Chen et al.,2008 | |
| Sox1 and Sox2 activate crystallin genes | Uchikawa et al.,1999 | |
| SoxC | Sox4 and Sox11 induce neuronal phenotype, and activate neuronal class IIIβ-tubulin (Tubb3) gene promoter | Bergsland et al.,2006 |
| Sox4 supports glucokinase expression | Goldsworthy et al., 2008 | |
| Sox4 enhances β-catenin signaling | Sinner et al.,2007 | |
| SoxD | Sox5 recruits coactivators to Sox9 target genes | Hattori et al.,2008 |
| Sox5 and Sox6 costimulate Sox9 target genes in chondrocytes | Han and Lefebvre,2008 | |
| SoxE | Promote neural crest multipotentiality | Kim et al.,2003 |
| Promote glial development of neural crest progenitor cells | Britsch et al.,2001 | |
| Sox9 promotes progenitor commitment, promotes expression of Aggrecan, Col11a2, Col2a1 | Bridgewater et al.,1998; Sekiya et al.,2000; Kawakami et al.,2005 | |
| Sox9 promotes Sox10 expression in Neuro2A cells | Stolt et al.,2003 | |
| Sox9 and Sox10 maintain PDGFRalpha+ cells | Finzsch et al.,2008 | |
| SoxF | Promote FoxA and HNF1beta expression | Sinner et al.,2004 |
| Promote laminin alpha 1 expression | Niimi et al.,2004 | |
| Promote VCAM1 expression | Matsui et al.,2006 | |
| Promote FGF-3 expression | Sinner et al.,2004 | |
| Promote endoderm development | Kanai-Azuma et al.,2002 | |
| Promote hair development | Irrthum et al.,2003 | |
| Promote cardiogenesis | Liu et al.,2007 | |
| Promote angiogenesis | Young et al.,2006 | |
| Promote transdifferentiation to epithelial cells | Park et al.,2006 | |
| Stimulate activity of mFoxj1 promoter | Park et al.,2006 | |
| Sox17 induces expression of Zfp202 zinc finger protein | Patterson et al.,2008 | |
| Sox17 and Sox7 coactivate β-catenin target genes in endoderm | Sinner et al.,2004 | |
| Sox17 promotes myelin gene expression | Sohn et al.,2006 |
| Family | Function | References |
|---|---|---|
| SoxB1 | Repress neuronal differentiation and maintain progenitor-specific expression | Bylund et al.,2003; Graham et al.,2003; Holmberg et al.,2008 |
| Bind β-catenin and inhibits β-catenin signaling | Kan et al.,2004 | |
| SoxB2 | Inhibit activation of delta-crystallin DC5 enhancer | Sandberg et al.,2005 |
| Counteract activities of SoxB1 to promote neurogenesis | Sandberg et al.,2005 | |
| SoxC | Sox4 represses oligodendrocyte development | Potzner et al.,2007 |
| Sox11 represses oligodendrocyte development | He et al.,2007 | |
| SoxD | Sox5 prevents premature development of neurons | Lai et al.,2008 |
| Sox5 represses SPARC | Huang et al.,2008 | |
| Sox5 and Sox6 repress IHH signaling, FGFR3 and Runx2 | Smits et al.,2004 | |
| Sox5 represses Sox10 targets Mitf and Dct | Stolt et al.,2008 | |
| Sox5 and Sox6 suppress premature oligodendrocyte generation. Sox6 corecruited to Sox10 site on MBP promoter at E16.5 | Stolt et al.,2006 | |
| Sox5 and Sox6 repress Sox9-induced Sox10 expression in Neuro2A | Stolt et al.,2006 | |
| Sox5 and Sox6 interact with Olig2 and neurogenin 2 | Stolt et al.,2006 | |
| Sox6 represses Ca2+ alpha1c promoter | Cohen-Barak et al.,2003 | |
| Sox6 represses epsilon globin expression | Yi et al.,2006 | |
| Sox6 represses insulin II gene expression | Iguchi et al.,2005 | |
| Sox6 suppresses cyclin D1 promoter activity | Iguchi et al.,2007 | |
| Sox13 suppresses alphaBeta T-lymphocyte differentiation | Melichar et al.,2007 | |
| Sox13 antagonizes TCF1 function and cell proliferation | Melichar et al.,2007 | |
| SoxE | Sox8 prevents premature osteoblast maturation | Schmidt et al.,2005 |
| Sox9 suppresses neuronal markers Pax6, Pax7, Nkx6, Irx3 | Cheung and Briscoe,2003 | |
| Sox8 and Sox9 prevent precocious myotube differentiation | Schmidt et al.,2003 | |
| Sox10 overcomes BMP2 and TGFbeta signaling to maintain gliogenic potential | Kim et al.,2003 | |
| Sox10 prevents Schwannoma from differentiating into myofibroblasts | Roh et al.,2006 | |
| Sox9 inhibits osteoblast development | de Crombrugghe et al.,2001; Hill et al.,2005 | |
| Sox9 inhibits Wnt signaling inhibiting β-catenin transcriptional activity directly and by affecting its stability | Akiyama et al.,2004; Topol et al., 2009 | |
| Sox9 reduces ERK1 levels | Panda et al.,2001 | |
| Sox9 inhibits Runx2 function | Zhou et al.,2006 | |
| Sox9 suppresses proliferation in intestinal epithelium | Bastide et al.,2007 | |
| Sox9 suppresses generation of spinal cord motorneurons, V2 interneurons and astrocytes | Stolt et al.,2002,2003,2004 | |
| Sox10 inhibits β-catenin-mediated transcription | Sinner et al.,2007 | |
| SoxF | Repress mSftpc gene expression in alveolus | Park et al.,2006 |
| Sox17 represses HNFa expression | Patterson et al.,2008 | |
| Sox7 inhibits cell proliferation in colon cancer | Guo et al.,2008 | |
| Sox17 antagonizes Wnt signaling | Sinner et al.,2007; Zorn et al., 1999 | |
| Sox17 represses cell proliferation in oligodendrocyte progenitor cells | Sohn et al.,2006 |
Sox TRANSCRIPTION FACTORS AND DISEASE
Sox factor expression and function are modified in a variety of tumors and neurological disorders. Here we discuss recent findings that suggest a role for these proteins in the formation and maintenance of some brain tumors. Sox2 expression was analyzed in different types of brain tumors, based on the hypothesis that cancer stem cells are responsible for their propagation (Phi et al.,2008). Sox2 was expressed in a variety of glial tumors of astroglial, oligodendroglial, and ependymal lineages. Sox2 was also found in tumors of embryonal origin (supratentorial primitive neuroectodermal tumors) but not in medulloblastomas and pineoblastomas. Because most of the Sox2-expressing cells in the tumors coexpressed GFAP, and Sox2-negative cells in medulloblastomas and pineoblastomas expressed neuronal markers, it was concluded that Sox2 might be a marker of tumors of glial lineages (Phi et al.,2008). Conversely, heterozygous loss-of-function mutations in SOX2 causes anophthalmia, microphthalmia, and coloboma. These patients often present with extraocular abnormalities, such as learning disability, seizures, motor dysfunction, and postnatal growth failure (Ragge et al.,2005). Additionally, a number of patients with SOX2 mutations show hypogonadotropic hypogonadism, with anterior pituitary hypoplasia and gonadotropin deficiency (Tziaferi et al.,2008), lending support to a multisystem function of SOX2 in human development. Surveys of gene expression in medulloblastomas have found an overexpression of SOX4 and SOX11 (Lee et al.,2002; Yokota et al.,2004), with strong nuclear staining in a majority of these tumors (de Bont et al.,2008). SOX11 is expressed in fetal progenitors (Weigle et al.,2005) and absent from normal mature cerebellum (de Bont et al.,2008), so the expression of this marker suggests an embryonic origin of the tumor. SOX6 is expressed at higher levels in human gliomas and fetal brain than in normal adult tissue, and it is considered a potential diagnostic marker for these tumors (Ueda et al.,2004). In an independent study, SOX6 was found to be up-regulated together with SOX8 and SOX13 in oligodendrogliomas (Schlierf et al.,2007). SOX10 is also ubiquitously expressed in gliomas (Bannykh et al.,2006). Altogether, these findings are consistent with the hypothesis that glioma cells are less differentiated than normal adult glia. Interestingly, Sox genes were found to be up-regulated in tumors of both oligodendroglial and astrocytic lineages, consistent with the notion that gliomas might share a common origin from stem cells or glial progenitors. This hypothesis has also been confirmed in pheochromocytomas, in which both Sox9 and Hey1—genes that are involved in the maintenance of a progenitor state—are highly expressed (Powers et al.,2007).
Different mutations in the SOX10 gene are associated with two distinct groups of neurocristopathies: a milder form comprising Waardenburg syndrome (WS) and Hirschsprung disease (HSCR) and a more severe trait, peripheral demyelinating neuropathy, central dysmyelinating leukodystrophy, Waardenburg syndrome and Hirschsprung disease (PCWH). PCWH is characterized by deficiencies in Schwann cells, oligodendrocytes, melanocytes, and enteric ganglia neurons (Inoue et al.,1999,2002). WS and HSCR are autosomal dominant disorders caused by cellular defects in the embryonic neural crest. WS and HSCR occur simultaneously in patients with Waardenburg-Shah syndrome (WS4). Sox10 mutations were identified in families with WS4 but not in patients with HSCR alone (Pingault et al.,1998). Importantly, these mutations are likely to cause haploinsufficiency of the SOX10 protein. Analysis of these disease-causing SOX10 mutations has revealed premature terminating codons and a relationship between the position of the mutation and the severity of disease (Inoue,2004). Efforts to characterize the various mutant Sox10 proteins have demonstrated some dominant negative properties (Inoue,2004) but have been unable to fully explain the toxic effects of extending translation into the 3′ untranslated region (Inoue,2007).
Recent studies point to a role of SOX10 and oligodendrocytes in the pathophysiology of schizophrenia. In patients with schizophrenia, the Sox10 gene is highly methylated in the CpG island of the sex-determining region Y-box, and its expression is significantly decreased (Iwamoto et al.,2005). Importantly, the methylation status of Sox10 correlated with its reduced expression and with modified expression of other oligodendrocyte genes. The CpG island of the oligodendrocyte transcription factor Olig2 was rarely methylated, and the methylation status of myelin basic protein was not associated with changes in other oligodendrocyte gene expression (Iwamoto et al.,2005). The authors of the study concluded that, in schizophrenia, Sox10 methylation is likely to be an epigenetic indicator of oligodendrocyte dysfunction.
CONCLUSIONS
It is apparent that oligodendrocyte dysfunction found in schizophrenia could be attributed to the activating functions of Sox10, but, in many tumors, the silencing function of Sox proteins on the proliferative effects of Wnt signaling has been attenuated (Guo et al.,2008). Given the structural features of the known Sox groups and the diversity of protein interactions documented to date, the mechanism of target gene selection by Sox proteins in specific cell types has become a subject of intense study. Current models indicate that the dynamic patterns of tissue-specific Sox gene expression, coupled with partner protein availability and selectivity, are mechanisms likely to underlie the specific changes in gene expression associated with fate decisions and important transitional phases in development (Kamachi et al.,2000). Selective DNA binding of promoters that are activated by Sox proteins and their cofactors is enhancer context-dependent (Kamachi et al.,1999), and it would be reasonable to expect that transcriptional repression by Sox proteins would be no less complex and more difficult to predict, in that neither activation domains nor DNA binding would be required. Elucidating mechanisms of target specificity in developmental regulation by Sox proteins will be enhanced by continued efforts in the identification and characterization of Sox protein function.
