Interaction of Notch and gp130 Signaling in the Maintenance of Neural Stem and Progenitor Cells

Abstract

Notch and gp130 signaling are involved in the regulation of multiple cellular processes across various tissues during animal ontogenesis. In the developing nervous system, both signaling pathways intervene at many stages to determine cell fate—from the first neural lineage commitment and generation of neuronal precursors, to the terminal specification of cells as neurons and glia. In most cases, the effects of Notch and gp130 signaling in these processes are similar. The aim of the current review was to summarize the knowledge regarding the roles of Notch and gp130 signaling in the maintenance of neural stem and progenitor cells during animal ontogenesis, from early embryo to adult. Recent data show a direct crosstalk between these signaling pathways that seems to be specific for a particular type of neural progenitors.

Introduction

Many intrinsic and extrinsic factors are integrated in ontogeny to specify cell fates and thus define the nervous system. In fact, these factors could influence a series of processes such as neuroectoderm specification, the generation and growth of the mitotically active progenitor pool, and finally the generation of mature nervous cells. A number of in vivo and in vitro studies found that Notch and gp130 signaling participate significantly in the regulation of these processes. The outcome of the neurosphere assay system is also considerably affected by Notch and gp130 signaling, just as the in vivo systems described below.

Both Notch and gp130 signaling are known to maintain the undifferentiated state of neural stem cells (NSCs); but on the other hand, they promote differentiation to glial cells (Fig. 1) (Grandbarbe et al. 2003; Chojnacki et al. 2003; Kageyama et al. 2009; Muller et al. 2009). This phenomenon seems to be contradictory; however, the effect is influenced by the differentiation status. Therefore, it is necessary to evaluate the effects of Notch and gp130 signaling separately in the early and terminal phases of differentiation.

Fig. 1.

Role of Notch and gp130 signaling pathways in the self-renewal and differentiation of neural progenitors. gp130 signaling supports self-renewal (red line) of primitive, LIF/gp130-dependent NSCs (PNSCs). Both Notch and gp130 pathways promote self-renewal (red line) of definitive, FGF2-/EGF-dependent NSCs (NSCs; mentioned both radial glia in embryo and radial glia-like in adult) and astrocytogenesis (blue line) of intermediate progenitors (IP). Active Notch also promotes transition (green line) from primitive to definitive NSCs. Points of inhibition mediated by particular signaling are marked by a dashed black line

Based on various published data, reciprocal successive activation and also direct crosstalk between Notch and gp130 signaling in neural progenitors are predicted. In the following chapters we summarize these findings into the integrated view.

Neural Stem and Early Progenitor Cells

The first step in the development of the nervous system is the differentiation of the ectoderm to the neuroectodermal neural plate. This process, called neural induction, starts with the suppression of bone morphogenic protein 4 (BMP4) signaling. Three secreted factors have been identified as BMP4 inhibitors and thus neural inducers: noggin, chordin, and follistatin (Levine and Brivanlou 2007). Ectoderm-derived neuroectodermal (neuroepithelial) cells are regarded as primitive neural stem cells (PNSCs), which are found in the E5.5–7 murine embryo. During the next step, the development of the neural tube, the PNSCs are reduced and substituted by definitive neural stem cells (NSCs, E7.5–8.5 in mice), which broadly persist throughout the entire lifespan. Initially they form the basis of the neural tube as radial glia, characterized by their potential, structure, long radial processes and expression of specific markers, mainly the glial fibrillary acidic protein (GFAP). After birth, when the central nervous system is definitively formed, radial glia are replaced by astrocyte-like/radial glia-like adult NSCs generated from radial glia. Adult NSCs share the majority of radial glia features and are dominantly localized in the subventricular zone (SVZ) of the telencephalon (Ihrie and Alvarez-Buylla 2008; Kriegstein and Alvarez-Buylla 2009; Yang et al. 2006). NSC self-renewal and stemness are dependent upon the presence of the fibroblast growth factor-2 (FGF2/basic FGF), and epidermal growth factor (EGF) (Bauer 2009; Hitoshi et al. 2004). However, NSC-like cells can also be isolated from another part of the central nervous system (CNS), and they differ from SVZ-derived NSCs in their multipotency and self-renewal properties (Giachino et al. 2013; Palmer et al. 1997; Seri et al. 2004; Temple and Alvarez-Buylla 1999). Therefore it is not clear whether all the NSC subpopulations derived from the CNS are pure stem cells, or whether they are only multipotent and short-time self-renewing neural progenitors (Gould 2007).

To study the features of NSCs in vitro the neurosphere assay system was developed. In this assay, both primitive and definitive NSCs isolated from the neural plate and embryonic brain, respectively, are cultured in non-adherent conditions at clonal density in the presence of responsible mitogens and form floating colonies called neurospheres (Reynolds and Weiss 1992). Neurospheres are heterogeneous colonies containing cells in various phases of differentiation (Bez et al. 2003). When dissociated to single cells and subcultured in the presence of growth factors, some of the cells give rise to new secondary, tertiary, etc., neurospheres. These proliferating cells are supposedly NSCs or more restricted precursor cells. The number of generated neurospheres corresponds with the number of NSCs in the source population. The size of the new neurosphere is also analyzed as it reflects the proliferation activity of the NSCs (Androutsellis-Theotokis et al. 2006; Molne et al. 2000).

Notch Signaling Pathway Overview

Activation of canonical Notch signaling requires ligand–receptor interactions through direct cell–cell contact. There are four mammalian Notch receptors (Notch1–Notch4). Notch ligands belong to the DSL family, which includes Delta and Serrate/Jagged in Drosophila as well as Jagged (Jagged1 and Jagged2) and Delta-like (Dll1, Dll3, and Dll4) in mammals (Bettenhausen et al. 1995; Del Amo et al. 1992; Dunwoodie et al. 1997; Ellisen et al. 1991; Lardelli et al. 1994; Lindsell et al. 1995; Oda et al. 1997; Shawber et al. 1996; Uyttendaele et al. 1996; Weinmaster et al. 1992). Individual Notch receptors differ in the number of tandem EGF-like repeats in the extracellular part of the receptor and in the presence or absence of the transactivating domain (TAD) in their intracellular part (Kurooka et al. 1998; Rebay et al. 1991). Notch signaling depends on the sequence of proteolytic events triggered by the interaction of Notches with membrane-anchored ligands. The first cleavage, occurring at the interface between the extracellular and membranal part, is catalyzed by ADAM-family metalloproteases and results in the separation of the extracellular domain (Lieber et al. 2002). The second cleavage is mediated by γ-secretase, an enzyme complex containing presenilin 1 or 2 (PS1, 2), which releases the intracellular part of the Notch receptor (NICD—Notch intracellular domain) for nuclear translocation (Handler et al. 2000; Struhl and Adachi 1998; Wong et al. 1997). In the nucleus, the RAM (Rbp-associated molecule) domain of NICD associates with the Recombining binding protein suppressor of the hairless (RBP-J also called CBF1 or CSL) DNA-binding protein to form a transcriptional activator and thus induce the expression of a set of targets. When signaling is inactive, RPB-J alone acts as a transcription repressor of target genes (Borggrefe and Oswald 2009; Ohtsuka et al. 1999). The DNA-binding ability of RBP-J/NICD is stabilized by the scaffold protein mastermind-like (MAML); therefore the dominant negative mutants of MAML completely block Notch1-4 induced transcription (Maillard et al. 2004).

In the nervous system the majority of Notch target genes are Hairy and Enhancer-of-split (Hes) genes in Drosophila and related Hes and Hey genes in mammals. Hes (in particular Hes1 and Hes5) and Hey genes encode transcriptional regulators of the basic helix–loop–helix (bHLH) class that mainly act as neuronal repressors (Kageyama et al. 2005). Hes proteins also mediate negative feedback in Notch signaling and are repressors of their own transcription. Accumulation of Notch-induced Hes thus leads to inhibition of Notch signaling, which results in periodical oscillation of Notch–Hes activity (Kageyama et al. 2005).

Recent evidence implies the existence of bi-directional Notch–Delta signaling mechanisms. Dll and Notch can be cleaved by the same enzymes (ADAM protease and γ-secretase), and the released intracellular domain of Dll mediates TGF-β/Activin signaling through binding to Smads (D’Souza et al. 2008; Hiratochi et al. 2007).

Non-canonical Notch signaling may be divided into an NICD-dependent pathway, where NICD interacts with other signaling pathways such as mTOR/Akt or Wnt/beta-catenin, and NICD-independent signaling leading to Hes/Hey expression mediated by Sonic hedgehog, FGF, and/or Wnt pathways (Andersen et al. 2012; Sanalkumar et al. 2010).

Notch Signalization Regulates Early Neurogenesis and Neural Progenitors

Self-renewal of NSCs seems to be Notch signaling-dependent and inhibition of Notch signaling leads to their differentiation to more mature cells (Androutsellis-Theotokis et al. 2006; Hitoshi et al. 2002; Chambers et al. 2001; Nakamura et al. 2000).The in vivo roles of Notch signaling were investigated mainly using Notch1-deficient mouse embryos, which die at embryonic day 11 (Conlon et al. 1995). In addition, mouse knockouts of Notch-related molecules (Hes1, RBP-J) showed reduced proliferation in the neuroepithelium of the neural tube (Ishibashi et al. 1995; Oka et al. 1995). Disruption of Notch pathway genes results in the reduction of the NSC pool size (Hitoshi et al. 2002; Nakamura et al. 2000). This effect is probably based on the symmetric division of existing NSCs, which in turn prevents the asymmetric division and thus differentiation of NSCs to neural and glial progenitors (Hitoshi et al. 2002, 2004). This implies the promoting role of Notch signaling for the self-renewal of NSCs and the expansion of neural tissue. The Notch role in NSCs is also confirmed through the detection of all the components of Notch signaling; mRNA for Jagged1, Dll1, Dll3, Notch1, PS1, PS2, RBP-J, Hes1, and Hes5 was detected in primary tissue cultures obtained from the E14.5 mouse forebrain and in neurospheres derived from it (Hitoshi et al. 2002; Tokunaga et al. 2004). Moreover, the use of Notch1 or Dll1 anti-sense mRNA, inhibition of γ-secretase or depletion of PS1 (PS1 knockout mouse has non-functional γ-secretase) as well as the depletion of Notch1 (Notch1^−/−) result in a reduced number of secondary neurospheres and increased frequency of differentiation, yet the cells were still able to differentiate in all the three types of neural cells (Donoviel et al. 1999; Handler et al. 2000; Hitoshi et al. 2002, 2004; Yoshimatsu et al. 2006).

The inverse approach with the constitutive activation of Notch1 (Notch1C), avoiding the γ-secretase step in the activation of Notch1, produced a correlating outcome: an increased number of NSCs (Hitoshi et al. 2002). The features of NSCs were evidenced by radial glia morphology, nestin positivity, and preferred self-renewal at the expense of further differentiation. The increased frequency of self-renewal induced by Notch is evidenced by the higher number of secondary or tertiary colonies (Gaiano et al. 2000; Nagao et al. 2007; Oishi et al. 2004). Specifically, the most important part of the Notch receptor, involved in the self-renewal of NSCs, is the intracellular RAM domain. Its deletion may completely abolish the positive effect of Notch1C on neurosphere formation (Nagao et al. 2007).

It is significant that during the neural plate stage, Hes1 and Hes3 are widely expressed in neuroepithelial cells along the neuroaxis, but Hes5 is not expressed there (Hatakeyama et al. 2004). This initial expression of Hes1 and Hes3 in neuroepithelial cells seems to be independent of the Notch signaling components. As neuroepithelial cells gradually change to radial glia cells, subsequent increase in Hes5 expression occurs together with the expression of Notch signaling components (Hitoshi et al. 2004; Ohtsuka et al. 2001). This corresponds with the Notch-dependent initiation of Hes5 expression in radial glia NSC. In addition, Hes1 expression later overlaps with the expression of Notch signaling components, which is consistent with the importance of Notch signaling for radial glia/NSCs summarized in (Kageyama et al. 2008).

The potentially overlapping response caused by the redundancy of Notch family members is resolved in experiments with RBP-J mutants. RBP-J mediates the downstream response to all four Notches. A nestin-positive neuroepithelium is found in RBP-J-null embryos but it is reported to be thinner and RBP-J^−/− E8.5 neuroepithelial cells are not able to create neurosphere colonies (Hitoshi et al. 2002). This is in accordance with Notch1 depletion as mentioned above. However, the influence of other Notch receptors is not yet fully understood. In vitro effects similar to Notch1 on transactivation of Hes1 and Hes5 were also seen with Notch3 signaling. On the other hand, Notch2 activation resulted in decreased transcription of Hes1 and Hes5. Moreover, co-activation of Notch2 and Notch1/3 even reduced Notch1/3-mediated transcription (Shimizu et al. 2002).

The precise mechanism of how Notch supports the formation of a higher number of neurospheres is still elusive. The logical explanation of Notch1 effects would be the modulation of target genes, which balance the regulation between asymmetric/symmetric and non-differentiating/differentiating type of division of NSC (summarized, for example, in Fortini 2009). Moreover, Notch signaling also regulates the ratio between cell proliferation and apoptosis. A double knockout of Hes1 and Hes5 resulted in the generation of a lower number of smaller neurospheres, i.e., decreased proliferation of a smaller number of progenitors (Ohtsuka et al. 2001). First, the reason could be the increased frequency of division due to Notch-induced promotion of the cell cycle (Ronchini and Capobianco 2001). Second, Notch signaling could inhibit apoptosis as a low level of constitutively active Notch1-IC (0.4-fold to the endogenous level) was shown to decrease apoptosis of neural progenitors cultivated in the absence of the growth factor. Interestingly, the relatively high level of Notch1-IC (1.2-fold to the endogenous level) seems to be toxic and initiate apoptosis (Oishi et al. 2004). This dose paradox could also explain the contradictory outcome of Lardelli et al. and Yang et al., who described increased apoptosis after constitutive activation of Notch (Lardelli et al. 1996; Yang et al. 2004). To sum up, we neither exclude the role of Notch1 in cell cycle control nor apoptosis.

The survival of NSCs was promoted by the presence of the RAM domain of the Notch1 receptor, which is accompanied by an increased expression of anti-apoptotic molecules Bcl-2 and Mcl-1. Hes proteins alone are probably not vital for the maintenance of NSCs since dominantly negative RBP-J mutants (lacking Hes proteins) do not affect the survival supported by the constitutively active Notch1C, which confirms the role of the non-canonical pathway (Oishi et al. 2004).

The pro-survival effect of Notch was also described using Dll4, another ligand of Notch, which supports the survival of mouse NSCs derived from E13.5. This effect seems to be mediated by non-canonical Notch signaling via S473 and T308 phosphorylations of Akt kinase, accompanied by the increase of Hes3 protein and reverted by DAPT (γ-secretase inhibitor) (Androutsellis-Theotokis et al. 2006).

Importantly, this effect was mediated by soluble Dll4. However, the potential role of soluble Notch ligands in the regulation of neurogenesis in mammals is more complicated, and their role in in vivo is far from clear. All hitherto known mammalian Notch ligands of the DSL family belong to the membrane-anchored proteins (Komatsu et al. 2008). Moreover, soluble Notch ligands were described to inhibit the canonical Notch signaling by blocking of Notch (Hicks et al. 2002; Imayoshi et al. 2010).

Regarding the activation of Notch via direct cell-to-cell contacts we can expect a correlation between increased cell density and Notch activity. Indeed, a higher density of cells supports the survival of NSCs, and this effect is mediated by active Notch signaling (Oishi et al. 2004).

Surprisingly, Notch signaling does not affect the generation of PNSCs as unchanged numbers of PNSCs were present in the neural plate of E7.5 Notch1^−/− embryos. This also fits the observation that in contrast to definitive NSCs, PNSCs derived from E7.5 brain express less Hes5, but still retained neural multipotentiality. The importance of Notch is demonstrated as soon as the definitive NSCs appeared. In mice, the majority of the early FGF2-responsive definitive NSCs from the E8.5 Notch1^−/− embryos were significantly lower in number and self-renewal capability (Hitoshi et al. 2004). This implies that Notch signaling is important for the transition from primitive to definitive NSCs as well as for the self-renewal of NSCs and thus it affects the size of the definitive NSC pool in murine embryos lacking Notch1. Certainly, the occurrence of some definitive NSCs in the Notch^−/− embryos suggests that other signaling pathways may permit some transition from primitive to definitive NSCs (Lui et al. 2011).

The role of the bi-directional process of Notch signaling in neurogenesis is still not well known. Over-expression of the Dll1 intracellular domain also induces neuronal differentiation in the pluripotent embryonal carcinoma cell line P19. Whether this mechanism represents the inhibition of NICD activity by the cleavage of the intracellular Delta domain and/or by the activation of TGF-β signaling is not clear (Hiratochi et al. 2007). However, this may play a key role during the asymmetrical division of NSCs, where a small imbalance in Notch signaling decides the fate of daughter cells (Fortini 2009). This implies the further detailed research is required to solve this issue (Fig. 2).

Fig. 2.

Multiple interactions between Notch and gp130 signaling pathway. Notch signaling targets Hes association with JAK2/STAT3, which is followed by phosphorylation and activation of STAT3. Active STAT3 positively regulates the expression of Notch ligand Delta along with SOCS proteins (inhibitors of JAK activity) and leads to the destabilization and proteasome degradation of Hes. Hes itself has a negative loop effect on Notch signaling. The intracellular domain of Notch ligand (Delta) inhibits Notch expression. Reciprocal positive and negative regulation of the Hes and STAT3 signaling pathways is the basic element of their mutual signaling oscillation. For details see text

gp130 Signaling Pathway Overview

The gp130 receptor constitutes a part of the receptor complex shared by the IL6 family cytokines such as interleukin (IL-) 6 and 11, leukemia inhibitory factor (LIF), cardiotrophin-1, oncostatin M, ciliary neurotrophic factor (CNTF), cardiotrophin-like cytokine (CLC), and neuropoietin (NP) (Hirano et al. 1997). The gp130-mediated signaling pathways are initiated by interactions between the ligand and the receptor complex containing a single gp130 chain and a cytokine-specific receptor α chain (Taga 1997). Each cytokine generates the formation of a different receptor complex; however, the gp130 component is vital for further signal transduction (reviewed in Heinrich et al. 2003). Changes of receptor conformation lead to the phosphorylation and activation of the Janus kinases (JAK1, JAK2), which are constitutively associated with gp130 (Lutticken et al. 1994; Stahl et al. 1994). These kinases then phosphorylate the signal transducer and activator of transcription (STATs). Typically, signaling through gp130 can activate STAT1 and STAT3 (Ernst and Jenkins 2004). Phosphorylated STATs dimerize and translocate into the nucleus, where they transactivate the expression of genes with STAT recognition sites (Schindler and Darnell 1995). Activation of gp130 also leads to the recruitment of the cytoplasmic protein tyrosine phosphatase SHP2, which activates the MAPK and PI3K/AKT pathways (Burdon et al. 2002; Feng et al. 1994; Li et al. 1994; Takahashi-Tezuka et al. 1998). Vice versa, the activation of MAPK and PI3K/Akt signaling can in turn modulate the activity of the STAT3 pathway (Ernst and Jenkins 2004; Ohtani et al. 2000). Overall, the most important pathway under the control of gp130 signaling is the activation of STAT3, and this is valid not only in the nervous system (Betz et al. 1998; Heinrich et al. 2003).

The key negative regulators of gp130 signaling are the suppressors of cytokine signaling (SOCS) and the protein inhibitor of activated STAT (PIAS). SOCSs are under direct control of STATs-regulated transcription and inhibit JAK activation and STAT dimerization. The induction of expression of SOCSs represents a negative feedback mechanism in the gp130 activation of STAT and is responsive to cyclic oscillation of this signaling (Alexander and Hilton 2004; Yoshiura et al. 2007). PIAS proteins are capable of associating with STATs and thus block their function. The expression of PIAS proteins seems to be cell-specific and independent of gp130 signaling itself (Shuai 2006).

gp130 Signalization Regulates Early Neurogenesis and Neural Progenitors

The first evidence about the role of gp130 signaling in neurogenesis appeared in FGF2/EGF-independent PNSCs derived from neuroepithelium. Their in vitro self-renewal and stemness maintenance are dependent on LIF/gp130 signaling (Hitoshi et al. 2004). However, the role of LIF is most likely specific for in vitro cultures, since the neuroepithelium is formed in vivo in the absence of LIFR or gp130 and the role of gp130 activation in vivo was not established in this case (Escary et al. 1993; Yoshida et al. 1996). On the other hand, the necessity of the STAT3, the most important downstream molecule of gp130 signaling in this process, is indisputable (Yoshimatsu et al. 2006). The STAT3-deficient embryo stops to developing and dies during implantation in the cylinder-stage of embryonic development, which takes place prior to the formation of neuroepithelium (Takeda et al. 1997).

Later in development during organogenesis, the presence of gp130 and LIFR was discovered in more than 90 % of NSCs isolated from the murine E14 brain (Rodriguez-Rivera et al. 2009). This finding confirms the importance of gp130 signaling in the modulation of NSCs during embryonic development. LIF- and CNTF-induced dimerization of gp130 with the LIF-receptor (LIFR) and LIFR–CNTF receptor, respectively, is vital for the nervous system (Port et al. 2007). CNTF/LIF/gp130 signaling also promotes the self-renewal and proliferation of definitive NSCs in vivo and it was also demonstrated in vitro by an increase in both the number and size of primary, secondary, and tertiary neurospheres (Bauer 2009; Carter et al. 2009; Gregg and Weiss 2005; Pitman et al. 2004; Represa et al. 2001; Shimazaki et al. 2001). Support for the clonogenicity effect was illustrated by the cultivation of primary neurospheres in the presence of LIF, which resulted in a higher number of secondary neurospheres (Bauer and Patterson 2006; Pitman et al. 2004). A LIFR knockout inhibited this phenomenon (Chojnacki et al. 2003). However, LIF signaling is not necessary for the generation of neurospheres because they develop even in LIFR deficient models. Such neurospheres contain cells that are able to differentiate into all three types of neural cells, but their clonogenic ability is impaired (Koblar et al. 1998; Pitman et al. 2004). Moreover, as was mentioned above, the key effector of gp130 signaling is STAT3, which may also be activated by other, although not fully known signals. And importantly, definitive NSCs also have a high level of STAT3 proteins which are in their active form (phosphorylated Y705) both in vivo and in vitro (Yoshimatsu et al. 2006). This is interesting since in standard in vitro cultures of definitive NSCs, ligands of gp130 signaling are not employed as well as any other strong STAT3 activators (Hitoshi et al. 2002, 2004). It was also observed that STAT3 directs pluripotent cells to neuronal lineages through the up-regulation of Sox2 (Foshay and Gallicano 2008). LIF and CNTF both signal through a receptor dimer consisting of gp130 and LIFR subunits, though CNTF requires an additional soluble CNTF receptor subunit to trigger a signal (Heinrich et al. 2003; Hirano et al. 1997). The absence of LIFR, causing the depletion of CTNF and LIF signaling, leads to precocious differentiation of precursor cells in the ventricular zone (VZ) of the embryonic brain. On the contrary, treatment of embryonic spinal cells (derived from area C-5) with CNTF, resulted in a decreased number of secondary neurospheres (Gregg and Weiss 2005). This contradictory finding is interesting; however, it was observed only in the case of LIF/LIFR signaling and this clearly shown difference between particular neural progenitors and NSCs in the CNS. Similar experiments, involving NSC-supportive pathways FGF2 and Notch, resulted in the same outcome in brain and spinal tissues (Hammerle and Tejedor 2007; Hitoshi et al. 2002; Chojnacki et al. 2003; Martens et al. 2000; Represa et al. 2001; Tropepe et al. 1999).

Support of the self-renewal of NSCs by gp130 is mediated by the JAK/STAT pathway as inhibition of this pathway by the AG490 inhibitor decreases the number of formed primary and secondary neurospheres in E11 mouse brain (Yoshimatsu et al. 2006). Further, NSCs (Nestin- and Sox2-positive cells) isolated from the cortex of E14 mice treated with LIF and CNTF express GFAP (Rodriguez-Rivera et al. 2009). This expression is probably mediated by the transcriptional activation of the GFAP promoter by phosphorylated STAT3 (Bartlett et al. 1998; Bonni et al. 1997; Nakashima et al. 1999). So similarly to Notch, gp130 signaling supports the self-renewal of NSCs, their survival and thus also the generation of neurospheres in vitro (Fig. 1).

Embryonic Stem Cells as a Model for Studies of the Role of Notch and gp130 Signaling During Early Steps of Neurogenesis

Embryonic stem cells (ESCs) are capable of giving rise to derivatives of each of the three primary germ layers (Evans and Kaufman 1981; Martin 1981). Their ability to differentiate into cells of the nervous system permits them to be used for studying the molecular mechanism of neural induction and following differentiation into various neural cells (Abranches et al. 2009; Keller 2005). Although various strategies for the induction of neurogenesis are employed in those studies, in the end early neural progenitors, PNSCs and NSCs derived from ES cells share the same features as their counterparts derived from neurogenic tissue during neurogenesis (Gregg and Weiss 2005; Pitman et al. 2004).

Similarly to the isolation of primitive and definitive NSCs from the neuronal plate and embryonic brain, respectively, the neurosphere technique is used for the isolation of ESC-derived NSCs. If ESCs are cultured in low density (<20,000 cells/ml) and in a neurogenic medium under the presence of LIF, they form neurospheres enriched by PNSCs that may be further passaged through the dissociation and reformation of new, secondary and tertiary neurospheres (self-renewal) (Smukler et al. 2006). If released cells from these neurospheres are transferred to neurogenic media for definitive NSCs containing FGF2 and EGF, they begin to form neurospheres enriched by definitive NSCs (Hitoshi et al. 2004; Tropepe et al. 2001). The most important role of gp130 signaling is usually connected with the formation and self-renewal of PNSCs, including those derived from ESCs (Hitoshi et al. 2004). This importance is evidenced by the LIF-dependent formation of PNSCs. The mechanism of the gp130 effect is most probably associated with STAT3 phosphorylation both at T705 and S727 (Androutsellis-Theotokis et al. 2006). The ES cell-derived LIF-dependent spheres arise from the proliferation of NSCs equivalent to epiblast-derived PNSCs as it was documented by their practically identical gene expression profile with primary E6.5 epiblast-derived spheres (Hitoshi et al. 2004). The generation of ESC-derived PNSCs in vitro was unaltered in ES cells deficient in the common Notch downstream signaling factor RBP-J (Hitoshi et al. 2002), suggesting that the formation of ESC-derived PNSCs is not dependent on Notch activity. Also Hes5 expression in the LIF-dependent, primary E6.5 spheres, as well as in the primary ESC-derived NSF, was much weaker compared to the substantial expression in E14.5 definitive neurospheres (Hitoshi et al. 2004) but Hes1 and Hes3 are still expressed (see above). ES cells, from which are derived LIF-dependent, clonal PNSC spheres then give rise directly to clonal FGF2-dependent spheres in vitro (Tropepe et al. 2001).

The experiments of Kobayashi et al. partially disprove that Notch does not affect the generation of PNSCs, showing that Hes1 is the most influential determinant of the ESC differentiation fate (Kobayashi et al. 2009). Notch activation and Hes1 expressions oscillate within a period of about 3–5 h; Hes1 expression levels at the time of induction of differentiation affect the preference in the cell fate choice. Hes1^high ES cells are prone to the mesodermal fate and Hes1^low ES cells are prone to the neural fate (Kobayashi et al. 2009). In contrast, the inactivation of Notch signaling by treatment with γ-secretase inhibitors or by genetic inactivation of Notch1 or RBP-J promotes ESC differentiation into cardiac mesodermal cells while constitutive activation of Notch directs ESCs into neuroectodermal Sox1-positive progenitor cells (Lowell et al. 2006; Nemir et al. 2006; Ramasamy and Lenka 2010; Schroeder et al. 2003). These experiments show that Hes1 does not mimic but rather antagonizes Notch signaling, probably through the direct repression of the expression of Notch ligands (Kobayashi and Kageyama 2010). However, these experiments do not actually show the real necessity of Notch signaling for the formation of PNSCs, they rather describe the influence of the preconditioned Notch status. Experiments which are more important for precise evaluation are those performed on Notch^−/− or RBP-J^−/− genotypes as described above; however, further detailed studies are still required.

Notch activation is important for the transition from ESC-derived PNSCs to the NSCs and for the maintenance of the NSC state (Hitoshi et al. 2004). Exposure of high-density ESC-derived NSCs to DAPT, a pharmacological inhibitor of the Notch pathway, was sufficient to induce premature neuronal differentiation, significantly disrupt sphere morphology and enhance maintenance of the NSC state along with proliferation in the presence of Dll4 and Jag1 (Elkabetz et al. 2008). Notch supports the self-renewal of ESC-derived NSCs most probably by modulating the cell cycle as DAPT-treated cells accumulate in the G1 phase (Alexson et al. 2006; Borghese et al. 2010). A detailed analysis showed that Notch activates the cyclin D1, which may also participate in the direction of the neural phenotype in both ESC- and ex vivo-derived NSC (Bryja et al. 2005, 2008; Das et al. 2010).

Similarly to the embryonic phase, the dual effect of Notch signaling is also observed during ESC differentiation. Activated Notch promotes the generation and expansion of neural precursors and suppresses terminal neuronal differentiation without promoting astrocyte differentiation. Thus, the Notch pathway directs lineage commitment in ES cells but restrains differentiation in neural stem/progenitor cells (Lowell et al. 2006). Notch signaling affects the clonality of NSCs derived from ESCs; the knockout of RBP-J reduces the ability to generate secondary and tertiary neurospheres (Hitoshi et al. 2002). The way Notch supports the formation of NSCs is also worth studying; the high cultivation density associated with the increased possibility of cell–cell contacts and thus also increased Notch activation, results in a lower production of Nestin-positive ESC-derived cells (Tropepe et al. 2001).

Importantly, Notch plays a role in the promotion of neural differentiation in human ES cells similar to mouse ES cells. In hESCs it was also discovered that Notch signaling contributed to their commitment towards the neural lineage (Lowell et al. 2006). The inhibition of Notch by DAPT during neural differentiation of hES results in the increase of MAP2-positive cells at the expense of the GFAP-positive population. These experiments also evidence that Notch supports the elongation and branching of neurites (Ramasamy and Lenka 2010).

It is interesting that in vitro experiments with ESCs have also shown that the effect of Notch signaling is highly dependent on its ligands. Jag1 increases Nestin expression (neural progenitor marker); on the other hand, Dll4 increases the expression of brachyury (mesoderm marker). This difference is connected with the different expression of target Notch genes. Jag1 increases Hes5, and Dll4 potentiates Hey1 expression (Ramasamy and Lenka 2010).

Both Notch and gp130 Signaling Promote Astrogenesis In Vivo and In Vitro

During the formation of the neural tube, the proliferating NSCs stay in close proximity to the developing lumen and form the germinal ventricular layer. These multipotent stem cells can undergo self-renewal, enter a quiescent state, die, or differentiate to precursors which migrate peripherally and mature to establish layers of neocortex with fully differentiated neurons, astrocytes, and oligodendrocytes (Kriegstein and Alvarez-Buylla 2009; Lui et al. 2011). A series of recent studies has demonstrated that the stem cells of the developing nervous system are radial glia (elongated form of the stem cell of the early neuroepithelium), which are extended from the ventricular lumen to the pial surface and serve as guides that progenitors migrate upon to reach their final destination (Lui et al. 2011; Malatesta et al. 2008).

NSCs isolated from the embryonic brain differentiate in vitro in the absence of growth factors (e.g., LIF, FGF2, EGF) in adhesion culture conditions into various types of neurons and glias in these neural progenitors. The activation of Notch signaling induces astrogliogenesis and inhibits neuronal differentiation in such neural precursors (Ge et al. 2002; Nye et al. 1994; Tanigaki et al. 2001). However, it is necessary to notice that this effect concerns the late phase of neurogenesis involving the differentiation of intermediate progenitors and that this effect differs from the early phases (Fig. 1).

In this case Notch signaling is independent of FGF2 and EGF signaling. Although the Notch effect seems to be mediated by the induction of expression of Hes proteins, it is important that it is independent of the RAM domain. This fact may be crucial for cell response to Notch signaling in view of Notch-mediated induction of astrogenesis and Notch-mediated self-renewal of NSC, which is RAM-dependent, see above (Nagao et al. 2007).

The question is whether these pathways induce astrogenesis directly by the activation of proastrocytic genes or whether they rather block the differentiation in neurons and oligodendrocytes. Mizutani et al. favor the view of Notch inhibiting the neurogenesis (Mizutani et al. 2007). This view is supported by the use of Notch1C that results in increased Hes1 transcription and thus decreased differentiation of the neural progenitor cell (NPC) into neurons (decreasing of Mash1 transcription) (Ishibashi et al. 1994; Ohtsuka et al. 1999, 2001). It is important to notice that both Hes1 and Hes5 mediate this effect; the expression of any one of them is crucial for the Notch-dependent regulation of neurogenesis. The other approach, the ectopic expression of Hes proteins also results in a decrease of neuronal differentiation (Ohtsuka et al. 1999, 2001).

The opposite view, presented by Rodriguez-Rivera et al., suggests that activation of Notch does not affect the number of Tuj positive neurons but that it triggers only the differentiation to GFAP-positive cells (Rodriguez-Rivera et al. 2009). This opinion is supported by another observation, in which further differentiation of cells derived from neurospheres with inhibited Notch signaling by DAPT resulted mainly in the generation of Tuj, Mash1 and Prox1 positive neurons and only few GFAP-positive astrocytes (Nagao et al. 2007). Some studies suggest that Notch signaling promotes the generation of radial glia and can even bias cells toward a glial fate rather than promote self-renewal (Gaiano et al. 2000). However, because both embryonic radial glia and later adult SVZ radial glia-like/astrocyte-like possess stem cell properties, the reported bias toward a glial fate (in part defined by the expression of GFAP) may be equivalent to the maintenance of the stem cell fate (Doetsch et al. 1999; Malatesta et al. 2003; Noctor et al. 2001). On the other hand, other factors also drive the induction of radial glia/astrocyte-like NSCs and/or definitive astrocytes (Lui et al. 2011).

Similar to the self-renewal of NSCs, members of the gp130 family also promote the gliogenic effect of Notch. Stimulation of the gp130 pathway with CNTF or LIF simultaneously with Notch1 in neurospheres synergically down-regulates the neurogenesis (expression of Tuj and MAP2) and oligodendrogenesis (expression of O4), and up-regulates astrogenesis (expression of GFAP) (Bartlett et al. 1998; Bonni et al. 1997; Cao et al. 2006; Johe et al. 1996; Nagao et al. 2007; Rodriguez-Rivera et al. 2009).

CNTF and LIF are secreted by the choroid plexus cell in the E14.5 mouse brain (Gregg and Weiss 2005). The addition of LIF or CNTF in the early phase of the mouse cortex (E12) supports the self-renewal of NPCs; however, in the late phase (E14, E16), if the overall number of NSCs decreases (Haydar et al. 2003) it promotes astrogenesis (Molne et al. 2000). Surprisingly, it seems that the effect of LIF is also dependent on EGFR signaling (Viti et al. 2003).

Knockouts of the LIF gene result in the reduction of the number of astrocytes in the murine hippocampus (Koblar et al. 1998). LIF and CNTF, as well as all members of the IL-6 family of cytokines so far examined, promote the astrocyte differentiation of neuroepithelial cells via the gp130-JAK/STAT signaling pathway, namely via binding of STAT3 to its cognate sequence in the astrocyte-specific promoter (Bonni et al. 1997; Ohno et al. 2006; Takizawa et al. 2001; Uemura et al. 2002; Yanagisawa et al. 1999, 2000). Consistent with these findings, cultures of neural precursors from LIFR-null or gp130-null embryos display delayed generation of GFAP-positive astrocytes (Koblar et al. 1998), while S100b positive astrocytes developed normally (Nakashima et al. 1999; Pitman et al. 2004). Conversely, the inhibition of STAT3 supports neurogenesis and inhibits astrogenesis (Cao et al. 2010; Gu et al. 2005; Rajan and McKay 1998; Yoshimatsu et al. 2006). The use of the dominantly negative STAT3 mutant (STAT3F) and inhibition of the STAT3 pathway by SOCS3 also confirmed the pro-astrogenic effect of STAT3 (Zhu et al. 2008). Sun et al. suggest that the decision between neuronal or glial differentiation depends on the relation between STAT and Neurogenin expression (Sun et al. 2001). The Ras/ERK pathway also has an effect opposite to JAK/STAT3, which can be explained by the reciprocal competition of both pathways (Bonni et al. 1997; Ernst and Jenkins 2004). Finally, GFAP expression in Sox- and Nestin-positive cells could be a direct consequence of STAT3 binding to its promoter and not the marker of differentiation (Rodriguez-Rivera et al. 2009). Moreover, a population of neural progenitors treated with LIF and CNTF resulted in increased GFAP expression and co-expression with Nestin and Sox2 positive cells without any changes in morphology. In the same set of experiments, activation of Notch did not result in co-expression of GFAP with Sox2 and Nestin, even though the Sox- and Nestin-positive cells were larger and they morphologically resembled astrocytes. It is interesting that the presence of CNTF in E13.5 mice leads to the negative correlation of Notch1 expression with Mash1 in separate populations of cells but the absence of CNTF results in protein co-expression (Gregg and Weiss 2005). So both Notch and gp130 induce GFAP expression and thus increase the population of GFAP-positive, astrocyte-like cells, but detailed inspection reveals two distinct cell populations.

Crosstalk Between the Notch and gp130 Signaling Pathways in Neurogenesis

Based on the aforementioned data it can be assumed that Notch and gp130 pathways should crosstalk (Fig. 2). However, to date the mechanisms of this interaction have not been elucidated and this chapter only summarizes plausible hypotheses.

The constitutive activation of Notch1 in the presence of gp130 ligand CNTF results in the synergic activation of STAT3 and increases transcription of its target genes (Kamakura et al. 2004; Lee et al. 2009; Nagao et al. 2007). Interaction between gp130 and Notch signaling was also evidenced in vivo when plasmid carrying constitutively active Notch1 was injected into the VZ of the E13 murine neocortex. Successfully transfected cells stayed in the VZ and acquired the morphology of GFAP-positive radial glia. Co-transfection with a dominant negative mutation of STAT3 resulted in reversion of the Notch effect (Kamakura et al. 2004).

However, there are discrepancies regarding which phosphorylation site of STAT3 is activated by Notch signaling. Nagao et al. observed increased phosphorylation of S727 but not of Y705 (Nagao et al. 2007). On the contrary, Lee et al. described increased phosphorylation of Y705 only (Lee et al. 2009). Moreover, inhibition of Notch signaling by γ-secretase inhibitor DAPT decreased phosphorylation of both S727 and Y705 as well as the the Hes1 level (Gouti and Gavalas 2008; Kamakura et al. 2004). In their study Lee et al. also observed that this activation is Hes1-dependent, but the mechanisms are not connected with the transactivation potency of Hes1 as it was not affected by the Hes1 mutant’s inability to form dimers and activate transcription (Lee et al. 2009). It is interesting that the Notch targets Hes1 and Hes5 both directly bind to STAT3 and thus enable the induction of phosphorylation on Y705 by JAK2, probably independent of gp130 activation (Kamakura et al. 2004). This reaction was demonstrated only in lysates of the nuclear fraction. It is possible that phosphorylation occurs only in the nucleus or that the translocation of phosphorylated STAT3 is faster than the detection limits. It is remarkable that in this study was the presence of the Hes1 and JAK2 complex was observed in the nuclear fraction alone. The mechanism of Notch-induced phosphorylation of STAT3 likely also involves the RAM domain since cells with constitutively active Notch bearing a deletion in the RAM domain (ca-Notch1ΔRAM) do not increase any phosphorylation of STAT3 (Nagao et al. 2007).

On the other hand, inhibition of Dll1 by siRNA in NSCs/NPCs derived from the E12.5 murine brain resulted in a decreased number of forming neurospheres which was not reversed even by the transfection of constitutively active STAT3 (STAT3-C) (Yoshimatsu et al. 2006). This data indicate a possible superior role of Notch to gp130 signaling in the maintenance of NSCs. Thus involvement of another signaling cascade downstream of gp130 may be required to explain this effect, just as other Notch/gp130 crossing signaling pathways.

Further, it was found out that Notch ligand Dll4 and/or Jag1 also induce the phosphorylation of STAT3 at S727 in a dose-dependent manner; phosphorylation of Y705 was not observed. This was assumed to be a pro-survival signal mediated by non-canonical Notch signaling-induced PI3K-mTOR activity (Androutsellis-Theotokis et al. 2006). However, for this case it was important that a soluble Notch ligand was used which could complicate the interpretation of such data in an in vivo system (see above the chapter about Notch signaling). Notch signaling thus seems to regulate both Tyr705 STAT3 phosphorylation through the canonical NICD/Hes-dependent pathway and Ser727 STAT3 phosphorylation through the non-canonical PI3K-dependent pathway. Both these regulations of STAT3 activity may play an important role in animal ontogenesis but the details are still not clear.

Besides this Notch-induced effect on STAT3, reciprocal effects of gp130 on Notch signaling were observed. Down-regulation of STAT3 resulted in decreased expression of Dll1. This is not surprising considering the promoter of Dll1 contains the STAT3 binding site (Yoshimatsu et al. 2006). Furthermore, cultivation of ES cells in LIF-containing media results in the increase of Notch1 and Notch4 expression (Ramasamy and Lenka 2010). It is interesting that the absence of LIF resulted in an increased expression of Notch 2 and 3. CNTF also increases Notch1 activity in the forebrain EGF-responsive NSCs (Chojnacki et al. 2003). Infusion of EGF with CNTF into adult forebrain lateral ventricles also increases periventricular Notch1 activity compared to EGF alone. It is remarkable that the gp130-enhanced Notch1 signaling that regulates NSC maintenance appears to be Hes1/5 independent. What is really interesting is that the addition of CNTF does not increase the expression of Hes1 and Hes5 even though it increases the mRNA and protein level of Notch1 (Chojnacki et al. 2003).

gp130 ligands regulate Notch expression via the JAK2 phosphorylation of STAT3 as seen by the inhibition of JAK2 by AG490 or over-expression of dominantly negative mutant STAT3 (STAT3F) which decreased the mRNA level of Notch1, 2, 3, and the Hes5 target gene (Zhu et al. 2008). Neuronal markers such as Mash1, Neurogenin3, NeuroD1 and D2 (Gu et al. 2005) were up-regulated simultaneously with the suppression of STAT3 signaling. Even the deletion in the STAT3 gene (cre-lox system) resulted in the decrease of Notch1, 2, and Hes5 levels but not those of Hes1 (Cao et al. 2010). At this point it should be reminded that Hes1 maintenance could also be Notch-independent; as is mentioned above in PNSCs. It is remarkable that over-expression of SOCS3 (negative regulator of STAT3 activity) increased the mRNA level for Notch1 and Hes5 (Cao et al. 2006). These facts support the requirement for full STAT3 signaling for the induction of Notch expression. SOCS3 is a direct target of active STAT3 and its expression blocks STAT3 activation. This loop is responsible for the periodical oscillation of gp130/STAT3 signaling, which is connected with Notch-Hes signaling oscillation (Kobayashi and Kageyama 2010; Shimojo et al. 2008). The necessity for the completeness of the signaling pathway is similarily demonstrated in the developing placenta where SOCS3 deficiency has a phenotype comparable to STAT3 depletion (Boyle and Robb 2008).

The interconnection of the gp130 and Notch pathways was also confirmed by Shen et al., who discovered that neurotrophin-4 (Ntf4) added to murine E14 NSCs supported the degradation of LIFR, gp130 and also reduced the NICD level. This is most probably caused by the decrease of STAT3-dependent expression of Notch ligands, and in turn results in strengthened differentiation in beta III tubulin-positive neurons. This neurogenesis is mediated by the inhibition of Y705 STAT3 phosphorylation, which is made possible by the induction of Shp2 phosphatase (Shen et al. 2010).

However, it was observed that the activation of JAK2 by CNTF resulted in a decrease of Hes3 expression, which is the downstream messenger of Notch-PI3K-dependent neural progenitor/stem cell survival (Androutsellis-Theotokis et al. 2006). The inhibition of JAK2 and the Y705 phosphorylation of STAT3 increases Hes3 expression in neural precursors. On the other hand, Ser727 STAT3 phosphorylation mediated by soluble Dll4/Jag1 increases Hes3 activity and promotes survival of NSCs. So STAT3 site-specific phosphorylation may also play an important role in Notch–gp130 signaling interaction (Pacioni et al. 2012). This fact does not confirm the previous observation regarding the role of gp130/JAK2/STAT3 signaling in the self-renewal of neural precursors/stem cells and thus requires further analysis. The crucial question which should be addressed concerns the identity of Hes3+/Sox2+ Dll4-responsive neural precursors/stem cells (Pacioni et al. 2012). The other important question relates to the natural role of soluble Notch ligands during neurogenesis as has been mentioned previously.

In summary, gp130 and Notch signaling are in direct interaction mainly through Hes association with JAK2-mediated phosphorylation and the activation of STAT3 proteins. Activated STAT3 positively regulates the expression of Notch and its ligand Delta, STAT3 itself and SOCS proteins (inhibitors of JAK/STAT activity). On the other hand, active STAT3 leads to the destabilization and proteasome degradation of Hes (Yoshiura et al. 2007). Thus Hes/STAT3 interaction and turnover of other components of this signaling cascade lead to the cyclic oscillation of Notch/Hes signaling together with STAT3 oscillation, but this oscillation may also occur independently of gp130 ligand and receptor activity (Shimojo et al. 2011; Yoshiura et al. 2007). Therefore Notch may keep a high STAT3 expression in NSCs, even though they grow independently of the presence of the gp130 ligand. Notch activity leading to Y705 phosphorylation of STAT3 could thus also explain the intensity of this phosphorylation in neurosphere cultures without gp130 ligands (Yoshimatsu et al. 2006) (and our unpublished data).

Conclusion

Self-renewal of neural progenitors and astrogliogenesis are indispensable processes for neural system development and homeostasis. Both Notch and gp130 signaling promote the self-renewal of neural stem and early progenitors, as well as the astrogliogenesis of neural progenitors both in vivo and in vitro. This ambiguity is mediated by the direct Hes or STAT3-induced expression of GFAP. GFAP is a dominant marker of astrocytes, but also a marker of radial glia, which represent self-renewing NSCs during embryonic development and radial glia-like NSCs in adults. Signaling overlap is mediated mainly by the direct interaction of Hes and JAK2/STAT3 proteins, which in turn maintain the signaling oscillation. However, in spite of this immediate interaction, the final phenotype of neural progenitors is slightly different after treatment with specific Notch or gp130 ligand (see above).

However, it has recently become clear that further factors as well as the developmental status of targeted cells decide the final fate of induced cells. For example, Notch signaling may be blocked by active GSK3, which connects Notch with Wnt signaling (Hayward et al. 2008; Katoh 2009; Lui et al. 2011). On the contrary, a lot of different tissue stem cells seem to be exposed to hypoxic environments and express a high level of HIF, which directly stabilizes NICD and thus increases expression of Notch target genes such as Hes proteins, etc. (Gustafsson et al. 2005). Hypoxia may thus increase STAT3 expression and activity, which is favorable for the self-renewal and expansion of NSCs and/or ESCs (Jung et al. 2008; Ong et al. 2005; Yang et al. 2006). Further, BMP signaling in the first step of neurogenesis inhibits the formation of neuroepithelium; in the next step of neurogenesis it induces the differentiation of neural progenitors to neurons and further glia progenitors to astrocytes (Deverman and Patterson 2009). All these interactions are sources of variability that lead to the precise regulation of neurogenesis and NSCs both in vivo and in vitro. Moreover, for the proper interpretation of experimental results a precise characterization of the identity of the studied cells seems to be crucial.

So although the complexity of regulation has not yet been fully understood, many previously presented, and in this review mentioned, observations help to clarify the regulation of normal nervous system development and homeostasis. Such knowledge enables us to expand NSCs both in vivo and in vitro, and modulate their differentiation, which could be therapeutically beneficial in cellular therapy and tissue replacement, etc. (Imitola 2007; Raedt and Boon 2005).