Musashi Proteins in Neural Stem/Progenitor Cells

Many RNA-binding proteins are encoded in the genomes of various organisms and play a critical role in several life systems. The human genome contains thousands of RNA-binding proteins (Glisovic et al., 2008). The most important biological role for these proteins involves the post-transcriptional events in gene expression, e.g., splicing, export, stabilization, localization, and translation. Recent studies have shown that these post-transcriptional events are of similar importance to transcriptional and post-translational events and are highly orchestrated (Keene et al., 2007).


The discovery of Musashi proteins and their function
Musashi was originally identified in a Drosophila mutant with abnormal external sensory organs (Fig.1A&B). In the early '90s, Nakamura and co-workers showed that the musashi gene is responsible for the asymmetrical division of sensory organ precursor (SOP) cells in Drosophila, which are precursors for the ectodermal system common to both neural and nonneural cell lineages in loss-of-function experiments (Nakamura et al., 1994). In wild-type animals, the SOP cell divides into a non-neural precursor cell (green in Fig.1C) and a neural precursor cell (white in Fig.1C), whereas in musashi mutants, two non-neural precursor cells (white in Fig.1D) are produced instead. The symmetrically divided non-neural precursor cells differentiate to hair-forming cells (Sf and So in Fig.1A&B), leading to a double-bristle phenotype instead of the single-hair wild-type phenotype (Fig.1 A&B). Based on this double-hair shape, the gene was named "Musashi" after a famous Japanese swordsman who fought with two swords, Musashi Miyamoto (A.D. 1584-1645). Subsequent studies revealed that the Musashi protein, which has RNA-binding activity, introduces neural differentiation potential for one daughter cell of the SOP cell via selective translational repression of the mRNA of a neural differentiation inhibitory factor (a transcription repressor possessing a BTB domain and zinc-finger domain) called ttk69 www.intechopen.com Musashi Proteins in Neural Stem/Progenitor Cells 207 (Okabe et al., 2001). The ttk69 protein is located downstream in the Notch signaling pathway and acts as a determinant of non-neural identity.
All listed organisms are multi-cellular eukaryotes. Fig. 2. A dendrogram of the organisms bearing the musashi or musashi-related genes.
The Systematic Evolution of Ligands by EXponential enrichment (SELEX) assay (Tuerk & Gold, 1990), an in vitro selection method for RNA, was employed to identify the specific target RNA motifs of Musashi proteins from a synthesized random-sequence RNA library. Uridine-rich sequences containing two or three (GUU…UAG) or (GUU…UG) repeats were identified as Musashi binding targets (Okabe et al., 2001). Indeed, ttk69 mRNA contains 15 of these motif sequences in the 3'UTR, and it has been demonstrated that the Musashi protein binds to the 3'UTR of ttk69 mRNA and inhibits the translation of a reporter gene linked to the 3'UTR in vitro (Okabe et al., 2001).
Musashi and Musashi-like proteins have since been discovered in several multicellular organisms, but these genes have not been found in prokaryotes, plants, or monocellular organisms (Fig.2). This implies that the Musashi protein is specifically required for the development and evolution of multicellular animals.

The homologs of Musashi protein in mammals
Further studies in Drosophila have revealed that the Musashi protein is also expressed in the compound eye primordium (Hirota et al., 1999), CNS (Nakamura et al., 1994), and neural stem/precursor cells in the larval brain (Nakamura et al., 1994), which have many characteristics in common with mammalian neural stem cells (NSCs) (Ito & Hotta, 1992). Thus, to elucidate the functions of the Musashi gene family in mammals, a homolog search and immunohistochemical studies were performed in mice.
Two highly conserved homolog genes, musashi1 (msi1) (Sakakibara et al., 1996) and musashi2 (msi2) (Sakakibara et al., 2001), were discovered in mice. While the length of these proteins are considerably shorter than that of Drosophila Musashi, the RNA-recognition motifs (RRMs) were highly conserved (~95% identical) between mammalian and insect systems (Fig.3). A high level of expression of Msi1 in NSCs of the periventricular area and undifferentiated neural precursor cells (Sakakibara et al., 1996;Kaneko et al., 2000;Sakakibara & Okano, 1997) was observed. Therefore, Msi1 is now widely used as a marker of NSCs and progenitor cells in the CNS of a variety of vertebrates. These cells, which can form neurospheres, were identified in the adult human brain using this approach (Pincus et al., 1998). Precise immunohistochemical analyses revealed that Msi1 is strongly expressed in the ventricular zone of the neural tube in embryos and in neurogenic sites within the postnatal brain, including the subventricular zone (SVZ), olfactory bulb, and rostral migratory stream (RMS) (Sakakibara & Okano, 1997). The Msi1 protein is expressed in neural stem/progenitor cells within these tissues and is rapidly down-regulated in post-mitotic neurons (Sakakibara et al., 1996).
The Msi2 protein in mice is a paralog of Msi1, displaying more than 90% homology with the Msi1 protein in the RRMs (Sakakibara et al., 2001) (Fig.3). Although the expression pattern in the CNS is very similar between the members of this family, Msi2 is also continuously expressed in a subset of neuronal lineage cells, such as parvalbumin-containing GABA neurons in the neocortex and neurons in several nuclei of the basal ganglia (Sakakibara et al., 2001). Other reports have also shown a differential expression pattern of these genes in uroepithelial cells (Nikpour et al., 2010). Although the functional properties (e.g., RNAbinding specificity) of the two proteins are similar, these differences in expression might explain the functional assignation of these proteins.
Although the partial 3D structures of RRMs of Msi1 protein have been solved by NMR, the full-length 3D structures of Msi1 and Msi2 remain to be elucidated (Nagata et al, 1999;Miyanoiri et al, 2003).
The human genome also contains both msi1 and msi2 genes (Good et al., 1998;Sakakibara et al., 2001). The structure and expression pattern of these proteins in the CNS highly resemble those in mice. As described below, Musashi proteins are related to several diseases. Functional studies in mice will thus contribute to therapeutic developments for Musashirelated conditions.
Additionally, a small Msi2-like gene, LOC100504473, has been found near the Msi2 locus on the mouse genome, but its expression and function are yet to be defined.

Molecular and physiological functions of Musashi proteins in stem/progenitor cells
To identify the target RNAs of Msi1 in mammals, a SELEX analysis from a randomsequence RNA library was performed, similar to those done in Drosophila. The selected consensus sequence revealed that the mouse Msi1 protein binds specifically to RNAs that possess a (G/A)UnAGU [n=1-3] sequence   (Fig.3).
A survey for the motif was performed in mRNAs expressed in the embryonic CNS. The 3'UTR region of m-numb mRNA (Zhong et al., 1996) was highlighted as a candidate target. Subsequent experiments found that m-munb mRNA is a specific binding target of the Msi1 protein in vitro and in vivo. Its translation is repressed by the Msi1 protein Kawahara et al., 2008).
The m-Numb protein binds to the intracellular domain of the Notch protein, which has nuclear translocation and transactivation activities, and inhibits the Notch signaling pathway (Berdnik et al., 2002), which positively regulates neural stem cell self-renewal Hitoshi et al., 2002;Tokunaga et al., 2004) (Fig.4). In agreement with this hypothesis, oscillation in the expression of the hes1 gene, a downstream target of Notch, controls the differentiation of embryonic stem cells to neural cells (Kobayashi et al., 2009). Indeed, the Msi1 protein induces the expression of the hes1 gene Yokota et al., 2004).
Musashi proteins in both mammalian species and Drosophila contribute to maintaining the stem/progenitor cell status via translational repression of target mRNA. However, the target mRNAs are, interestingly, not orthologous. This result implies a highly conserved function of Musashi proteins in maintaining the stemness of progenitor/stem cells and the probable presence of other unknown target RNAs of the Musashi proteins.
On the other hand, the function of the Msi2 protein in neural stem/progenitor cells is still unclear, though it is known that Msi1 and Msi2 have similar RNA-binding specificity (Sakakibara et al., 2001). The results of an Msi1 and Msi2 double knockout experiment suggested that these proteins have mutually complementary functions (Sakakibara et al., 2002).
Recently, the molecular mechanism of translational repression by Msi1 has been uncovered. Kawahara et al. identified the poly(A)-binding protein (PABP) as an Msi1-binding protein and found that Msi1 competes with elF4G for PABP binding on its target mRNAs (Fig.5) (Kawahara et al., 2008).
However, the molecular machinery of other functions of Musashi protein (described below) remains to be elucidated. A survey for the co-factors of Musashi proteins is an important future task in order to fully understand the functions of these proteins.

Musashi-related diseases -Cancers and neural disorders
Msi1 has been shown to play a role in a variety of cancers and neural disorders.
Although the exact function of Msi1 in these cancer cells remains unclear, the knockdown of Msi1 via siRNA resulted in arrested tumor growth in colon adenocarcinoma xenografts transplanted in athymic nude mice, reduced cancer cell proliferation, and increased apoptosis (Sureban et al., 2008). These results suggest an important potential role for Msi1 in tumorigenesis and tumor proliferation.
It has also been shown that some tumors express the Msi2 protein in addition to Msi1 (Seigel et al., 2007). This may indicate a complementary role of the two proteins in tumors.
Msi1 is also hypothesized to be a key player in neuronal disorders. Several reports indicate that Msi1 is relevant to neurodegenerative disorders, such as Alzheimer's disease (AD). Ectopic expression of Msi1 was observed in the hippocampus of AD patients (Lovell & Markesbery, 2005), while a significant decrease in Msi1-expressing cells was observed in the SVZ (Ziabreva et al., 2006). Although it is difficult to explain these phenomena at present, the function of Msi1 in maintaining the ste m n e s s o f N S C s m i g h t p l a y a r o l e i n t h e pathogenesis of the disease.
Msi1 may also play a role in Parkinson's disease (PD), an another type of neurodegenerative disorder. A clinical experiment found that chronic treatment with an anti-PD drug increased Msi1-positive cells in the SVZ of PD patients. The authors suggested that impaired neurogenesis may contribute to the decline in this neurodegenerative disease (O'Sullivan et al., 2011). Oki et al. (2010) reported an up-regulation in Msi1 expression in collapsed nervous system tissue arising from a blood circulation defect. In the ischemic striatum induced by middle cerebral artery occlusion (MCAO), an increase in Msi1-immunoreactivity was observed in reactive astrocytes beginning at 2 days after MCAO and persisting until 14 days after MCAO. The proliferation of Msi1-positive cells was observed beginning at 4 days after MCAO and reached a peak at 7 days after MCAO (Oki et al., 2010) Nakayama et al. (2010 also observed an induction of Msi1-positive cells at the site of ischemic lesions beginning on day 1 after stroke in humans. This result indicates the presence of a regional regenerative response in the human cerebral cortex and the importance of Msi1 in this phenomenon (Nakayama et al., 2010).
Interestingly, Msi1 protein-expressing cells are increased in the hippocampus of mesial temporal lobe epilepsy (MTLE) patients (Crespel et al., 2005). Large numbers of Msi1positive cells were also observed in the SVZ in these patients (Crespel et al., 2005). Increased neurogenesis has been reported in animal models of MTLE (Crespel et al., 2005). The abnormal proliferation of such Msi1-expressing neural progenitors in the hippocampus might cause epilepsy.
Unlike Msi1, the relevance of Msi2 to various diseases has not yet been elucidated. However, it was recently demonstrated that Msi2 triggers the acute transformation of chronic myelogenous leukaemia (CML) through translational control of the Numb protein in humans (Ito et al., 2010;Kharas et al., 2010;Nishimoto & Okano, 2010). Byers et al. (2011) reported that Msi2 protein expression can be a clinical prognostic biomarker of human myeloid leukaemia (Byers et al., 2011)

Novel finding for the functions of Musashi proteins
Although Musashi proteins are thought to act as translational suppressors, MacNicol and co-workers found a novel function of Musashi in Xenopus oocytes. In this system, it activates the translation of mos mRNA (Charlesworth et al., 2006), a gene that is related to the meiotic cell cycle progression (Sagata et al., 1988). This is an opposite result from previous findings on the translational effect of Msi1. Interestingly, in human oocytes, a parallel physiological phenomenon is controlled by factors other than the Musashi homolog proteins (Prasad et al., 2008).
A similar translation-activating effect of Msi1 was also observed in mammals. Kuwako et al. (2010) found that the Msi1 protein up-regulates the translation of the Robo3 protein and controls midline crossing in precerebellar neurons. While previous studies reported that the Msi1-binding sites are in the 3'UTR of mRNAs, the Msi1-binding region in Robo3 mRNA is in the protein-coding region and does not bear the Msi1-binding consensus sequence (Kuwako et al., 2010). This result implies that the discovery of novel Msi1 co-factors will be an important task for understanding the molecular mechanism of the Msi1 proteins in translational control.
Until recently, only a small number of Msi1-targeted mRNAs Battelli et al., 2006) had been reported. De Sousa Abreu et al. (2009) performed an RNA immnoprecipitation (RIP)-Chip assay in HEK293T cells, a cell line derived from human embryonic kidney, to comprehensively identify the target mRNAs (de Sousa Abreu, 2009). They identified a group of 64 mRNAs whose genes belong to two main functional categories pertinent to tumorigenesis and protein modification. A subsequent proteomics study also revealed that Msi1 can have not only negative but also positive effects on gene expression for some of the targets (de Sousa Abreu et al., 2009). This result is consistent with other recent findings.
Our group also performed in vitro screening analysis to detect specific binding targets of Msi1 controlling the stem cell status of NSCs. We succeeded in identifying a novel target mRNA of Msi1, doublecortin (dcx), from an mRNA library of embryonic mouse brain tissue (Horisawa et al., 2009).
The dcx is a gene related to the migration of newborn neurons and neural development. Mutations in this gene cause an X-linked dominant disorder characterized by classic lissencephaly with severe mental retardation and epilepsy in hemizygous males and subcortical laminar heterotopia, also known as double cortex syndrome, associated with milder mental retardation and epilepsy in heterozygous females (Gleeson et al., 1998;des Portes et al., 1998;Sossey-Alaoui et al., 1998).
The Msi1 protein specifically bound to the 3'UTR region of the mRNA in vitro, which contains an Msi1 binding motif, and repressed translation of a reporter gene linked to the mRNA fragment (Horisawa et al., 2009).
We hypothesize that the Msi1 protein prevents inappropriate migration of NSCs through translational inhibition of the dcx gene. Several findings support our hypothesis. First, the Dcx protein is expressed only in neuronal precursors just differentiated from NSCs (Couillard-Despres et al., 2005). Secondly, mutually exclusive protein expression of Msi1 and Dcx in human brains was observed (Crespel et al., 2005). Finally, a knock-out of the Musashi family genes reduced the number of neurospheres isolated from embryonic mouse brains, while the knock-down of dcx prevented migration of the cells from neurospheres, leaving their structure intact (Ocbina et al., 2006). In addition, we also identified another candidate Msi1-binding mRNA that is related to neuronal migration and axon outgrowth (unpublished data). Thus, Msi1 might repress the maturation of neural stem/progenitor cells to neurons through direct translational inhibition of the genes that influence neuronal maturation and migration.
All of the known target RNAs of Musashi proteins, which have been validated in previous studies, are listed in Table 1.
The precise mechanism through which the function of Msi1 is controlled remains unclear.
Although Wang et al. (2008) proposed that the Msi1 protein is involved in both Notch and Wnt signaling pathways as a novel autocrine process (Nagata et al., 2006;Glazer et al., 2008), details of the mechanism remain unclear, and direct regulators of Msi1 have not been identified.
On the other hand, Ratti et al. (2006) reported post-transcriptional regulation of Msi1 mRNA by embryonic lethal abnormal vision (ELAV), an RNA-binding protein of Drosophila (Ratti et al., 2006). This is an interesting result because it may imply that some kind of cascade of post-transcriptional regulation contributes to neurogenesis, in addition to other machinery, i.e., signal transduction, transcriptional regulation, and post-translational modification.
Upstream mechanisms regulating Msi1 transcription have also been studied. Kawase et al. (2011) found that the sixth intron of the msi1 gene has a regulatory element for msi1 transcription in neural stem/progenitor cells. The identification of transcription factors for the msi1 gene will help elucidate the role of the Msi1 protein in stem cells (Kawase et al., 2011).
More recently, Kawahara et al. (2011) discovered a novel function of the Msi1 protein. They showed that Msi1 works in concert with Lin28 to regulate post-transcriptional microRNA (miRNA) biogenesis in the cropping step, which occurs in the nucleus. This indicates that Msi1 can influence stem cell maintenance and differentiation by controlling the subcellular localization of proteins involved in miRNA synthesis, as well as by regulating the translation of its target mRNA (Kawahara et al., 2011).

Conclusion
Somatic stem/precursor cells, including neural stem cells, are promising targets for regenerative medicine. Because these cells are derived from the patients themselves, they are not subject to the ethical questions and possible immunological rejection that are common problems in regenerative therapies using embryonic stem (ES) cell.
The Musashi family proteins are key factors for the understanding and application of somatic stem cells. Musashi proteins control the stem cell state through the translational regulation of target mRNAs, and the Musashi family is a highly conserved RNA-binding protein group expressed in undifferentiated stem/precursor cells at both embryonic and adult stages.
Although several studies have revealed that a Notch signal inhibitor, m-numb, and a cell cycle regulator, p21 WAF1 , are direct targets of Msi1, a mouse homolog of Musashi and that the machinery involved is located in the context of both the Notch and Wnt signaling pathways , the full picture of Msi1 function in neural stem/precursor cells remains to be uncovered. Recently, de Sousa Abreu et al. employed an RNA immunoprecipitation (RIP)-chip technique to comprehensively detect Msi1 targeted mRNAs, and they identified a group of 64 mRNAs from HEK293T cells, whose genes belong to two main functional categories pertinent to tumorigenesis: 1) cell cycle, proliferation, differentiation, and apoptosis and 2) protein modification (de Sousa Abreu et al., 2009). However, a more specific survey is necessary to identify the key factors that regulate stemness. We speculate that Msi1 might have specific targets in each cell type or site, such as dcx, in addition to the previously discovered targets, which are related to the cell cycle, proliferation, and self-renewal.
While Musashi proteins were originally thought to act as translational suppressors, recent findings indicate that these proteins can activate the translation of specific targets (Charlesworth et al., 2006;de Sousa Abreu et al., 2009;Kuwako et al., 2010). This result indicates that currently unknown molecular machinery may exist that differs from the translational suppression machinery (Kawahara et al., 2008). Components of this machinery, e.g., binding proteins of Msi1, need to be comprehensively clarified using high-throughput techniques (Rigaut et al., 1999;Horisawa et al., 2004Horisawa et al., , 2008.
Many researches indicate that Musashi proteins have strong associations with some diseases, such as cancers and neuronal disorders. A more complete understanding of the Musashi proteins will also contribute to the development of therapies for these diseases.

Acknowledgment
We thank Dr. Takao Imai and Prof. Hideyuki Okano (Department of Physiology, Keio University School of Medicine) for valuable advice and discussion.