The mammalian CNS develops from a population of stem cells. Within the developing cortex (telencephalon), these cells first have the morphology of neuroepithelial cells within a pseudo-stratified epithelium, becoming radial glial cells that span the CNS from the pial basement membrane to the ventricle as the cortex thickens during neurogenesis (Fishell and Kriegstein,2003; Alvarez-Buylla and Lim,2004; Gotz and Huttner,2005). At the earliest stages (up to E11 in the mouse), the cells undergo symmetric divisions, expanding the stem cell pool (Temple,2001; Gotz and Huttner,2005). Subsequently, they initiate asymmetric divisions that generate a committed neuronal precursor cell as well as another daughter cell identical to the parent, hence fulfilling the key criteria for a stem cell of self-renewing asymmetrical divisions (Gotz and Huttner,2005). The committed neuronal cell, in contrast, may undergo one or more further symmetrical divisions before differentiation, and is therefore a precursor cell (analogous to the transit amplifying cells in many adult lineages) rather than a stem cell. Many cell culture studies do not distinguish between these two cell types as the precursor cells can divide repeatedly and form both neurons and glia in vitro, and the term stem/precursor cell is, therefore, often used in discussion. Throughout the stem cell divisions, and despite the ever-lengthening basal process attached to the pial surface, the nuclei of the radial glial NSC remains within 70 μm of the ventricular surface, a region termed the ventricular zone (VZ) (Takahashi et al.,1995). Within this zone, the nucleus shuttles up and down during the cell cycle such that mitosis occurs at the ventricular surface and S phase at the abventricular surface, a process termed interkinetic nuclear migration (IKNM) (Takahashi et al.,1996). In contrast, the committed precursor (or basal progenitor) cell created by these asymmetric divisions migrates out of the VZ into an overlying secondary proliferative region (the subventricular zone, SVZ) where it may undergo another round of symmetrical division before the neurons migrate into the developing cortical plate (Tabata and Nakajima,2003; Haubensak et al.,2004; Miyata et al.,2004; Noctor et al.,2004). As development proceeds, an additional zone of proliferation is therefore seen in the SVZ, populated by cells from the VZ that in turn begins to diminish in size. Unlike the stem cells in the VZ, these SVZ proliferative cells do not have an attachment to the basal lamina and their nuclei do not exhibit the classical interkinetic translocation. The SVZ increases in size until birth, mainly in the cranial regions of the rostrocaudal axis. It then begins to diminish, although remnants persist throughout life as the subependymal zone (SEZ) of the lateral ventricles and around the third ventricle.
Since the original population of VZ cells divide repeatedly, can undergo self-renewing asymmetrical divisions, and generate both neurons and glia, they show many properties that define stem cells even though only a very small number will persist postnatally and become slowly-dividing adult neural stem cells. They will, therefore, form the focus of this review. A better understanding of the signals and pathways regulating the properties of this embryonic neural stem cell population is clearly an important goal for developmental neuroscientists. However, many studies on neural stem cell biology are conducted on the slowly-dividing cells found in the adult CNS within the remaining SEZ and dentate gyrus. Here, the stem cells are found within specialized microenvironments or niches containing both the stem and other cell types that, as they do in other adult stem cell systems, provide signals that promote and maintain the stem cells throughout life. In contrast, the embryonic VZ is a transient structure that does not fulfill the criteria conventionally used to define niches. In this review, we will examine whether, despite these differences, the extracellular signals found in the adult (neural and non-neural) stem cell niches are present in the VZ. Using the similarities and differences we observe, we will re-consider whether the location of embryonic stem cell populations such as the VZ can be thought of as niches. Finally, we will ask how these lessons from the niche inform our understanding of neurodevelopmental diseases and cancers of the CNS.
LOCATION OF ADULT NEURAL STEM CELL NICHES
Adult neural stem cells are thought to reside in at least three structurally and anatomically distinct sites in the adult brain. The best described is the SEZ adjacent to the lateral ventricles. Here a slowly-dividing population of stem cells generates the neuronal precursor cells that migrate to the olfactory bulb through the rostral migratory stream and enable the turnover of the olfactory neurons therein (Goldman and Nottebohm,1983; Alvarez-Buylla et al.,1990; Doetsch et al.,1999). More recently, these stem cells have also been shown to generate oligodendrocytes in the adult CNS (Menn et al.,2006) Studies examining the regeneration of the SEZ niche following cytotoxic ablation of the precursor cell populations have revealed three cell types that can be distinguished by specific markers or by morphology in the electron microscope. These are type B cells (SEZ astrocytes) that represent the neural stem cells themselves, type C cells that are the transit amplifying cells, and the type A cells that are the neuroblasts that will give rise to newly born neurons (Doetsch et al.,1997,1999). This region also contains ependymal cells that line the ventricle. These separate the A, B, and C cells from the ventricle although type B (stem) cells may extend fine processes through the ependymal cell layer to contact the cerebrospinal fluid (CSF) within the ventricular space.
A number of other potential locations for neural stem cell niches in the adult CNS have been described. First, the generation of new neurons and glia in the hippocampus has been reported in rodents, primates, and humans (Altman and Das,1965; Kaplan and Bell1984; Cameron et al.,1993; Eriksson et al.,1998; Gould et al.,1999). The source of these new cells originating from the subgranular zone (SGZ) could potentially to be astrocyte stem cells (Seri et al.,2001), as has also been reported in the SEZ. By utilizing methods (electron microscopy, BrdU labeling, and regeneration and repopulation after anti-mitotic treatment to ablate precursor cells) similar to those used to study the SEZ, the cytoarchitecture of the putative SGZ stem cell compartment has been characterized (Seri et al.,2004). In this compartment, the astrocytes give rise to intermediate progenitor cells (type D), which eventually form newly born granular cells (Seri et al.,2004). However, there are contrary reports on the location, origin, and potency of these potential stem cells. While the SGZ in the dentate gyrus has been characterized as a niche in vitro (Seaberg and van der Kooy,2002; Becq et al.,2005; Bull and Bartlett,2005) and in vivo (Nakatomi et al.,2002), these studies suggest that the CA1 region contains multipotent stem cells while the SGZ contains only precursors. Second, there is evidence for neural stem cell populations in the embryonic and adult cerebellum (Laywell et al.,2000; Klein et al.,2005) and between the hippocampus and corpus callosum in the subcollosal zone (SCZ) (Seri et al.,2006), although the composition of these potential niches is unknown. Finally, it has been suggested that there is an additional adult neural stem cell population distributed widely throughout the CNS. Cell labeling studies using BrdU (Reynolds and Hardy,1997; Dawson et al.,2003) have revealed a proliferating cell population that expresses the oligodendrocyte precursor marker NG2. Using a CNPase-GFP transgenic mouse (to label oligodendroglial cells and their progeny), these cells were shown to contribute to the generation of both neurons and glia in the hippocampus (Belachew et al.,2003; Aguirre et al.,2004) and olfactory bulb (Aguirre and Gallo,2004). The exact biological role of this population has yet to be functionally determined.
SIGNALING IN ADULT NEURAL AND OTHER STEM CELL NICHES
Mammalian adult stem cell niches have also been described in many other tissues including the germinal (testis), hematopoietic, epidermal, and intestinal system. Despite the significant anatomical differences between these tissues, the extracellular mechanisms by which the stem cell population within the niche is regulated share many components (see Fig. 1). A common set of growth factors regulate stem cells, though their relative contribution differs from niche to niche. Short-range interactions such as ephrin and notch signaling also appear to be important. Finally, there is an ECM-rich basal lamina within many niches that may provide both structural support in orienting the cells and a regulatory role in stem cell behavior.
Growth factors that regulate stem cell behavior can originate from supporting cells within the niche or from stem cells themselves. They may regulate survival, cell cycle kinetics, cell proliferation, and differentiation. A key role of the BMPs was initially identified in the Drosophila ovary, where the BMP homologue decapentaplegic (dpp) secreted by neighboring cells maintains the germinal stem cell state (Xie and Spradling,1998; Yamashita et al.,2005). BMPs have diverse roles in several mammalian stem cell niches. Mice in which the BMP receptor type 1A was inactivated show an increase in the number of haematopoetic stem cells as a result of an increase in the osteoblast population that forms part of the niche (Zhang et al.,2003), and loss of BMP signaling due to receptor inactivation in the intestine increases the number of stem-cell containing crypts (He et al.,2004). In contrast, BMPs act in the hair follicle and epidermis by promoting differentiation (Kobielak et al.,2003; Andl et al.,2004). In the adult CNS, BMPs and corresponding receptors are expressed by the SEZ cells and inhibit neurogenesis by directing the cells towards glial differentiation (Lim et al.,2000). Neurogenesis in the SEZ results from the secretion of noggin, a BMP antagonist, by the neighboring ependymal cells (Lim et al.,2000). The effects of BMP are age dependent, as in embryonic stem cells BMPs act in combination with LIF to suppress differentiation and maintain self-renewal (Ying et al.,2003).
Like BMPs, the Wnt family of cytokines plays a critical role in self-renewal and differentiation. Wnts act as hematopoietic stem cell growth factors (Reya et al.,2003; Willert et al.,2003) and promote proliferation of intestinal stem cells (Korinek et al.,1998; Pinto et al.,2003). They can also act as regulators of differentiation; in epidermal stem cells high levels of Wnt signaling generated by expression of a truncated β-catenin promote hair follicle formation (Gat et al.,1998), while mutation or inhibition of β-catenin results in epidermal differentiation (Huelsken et al.,2001; Niemann et al.,2002).
Other growth factors have been implicated in stem cell proliferation. FGFs are expressed on the stromal cells that provide trophic support for the hematopoitic stem cell as well as by the stem and progenitors cells in the hematopoetic system, and can be used to expand haematopoetic cells in culture suggesting a role in maintenance of the stem cell population (Kashiwakura and Takahashi,2005). VEGF stimulates cell proliferation of adult rodent cortical cells in vitro and intraventricular administration of VEGF resulted in increased proliferation of SEZ cells (Jin et al.,2002), as it also does with FGF, EGF, or TGFα (Craig et al.,1996; Kuhn et al.,1997; Wagner et al.,1999). However these early studies did not distinguish between the stem and progenitor cell populations (including both transit amplifying cells and neuroblasts), and subsequent studies of the EGF-induced proliferation in the SEZ suggest that it is the transit amplifying cells that are responsive to the EGF (Doetsch et al.,2002). Shh is a well-established mitogen for granule cell precursors (Wechsler-Reya and Scott,1999), and also promotes stem/precursor cell proliferation and increases proliferation in the SEZ (Palma and Ruiz i Altaba,2004). Here, a recent cell marking study using expression of lacZ in the R26R reporter mouse, activated by expression of Cre recombinase driven by the Shh-target Gli1, showed directly that the Shh-responsive cell population includes both neural stem cells and transit amplifying cells (Ahn and Joyner,2005). SEZ stem cells also express the PDGFα receptor and the level of expression maintains the balance between neuron and oligodendrocyte production. Conditional ablation of the receptor in SEZ stem cells results in an arrest in oligodendrogenesis but not neurogenesis while infusion of PDGFα into the lateral ventricles increased B cell production and stopped neuroblast generation (Jackson et al.,2006).
Together these studies on growth factors show that, while many of the same growth factors act in different stem cell compartments, their roles and relative contributions may differ. This likely reflects the combinatorial nature of signaling in the regulation of cell fate (Flores et al.,2000; Sommer and Rao,2002). Additionally, many other growth factor effects remain to be explored. For example, while Sonic hedgehog (Shh) is a mitogen for neural stem cells, the role of Shh in regulating stem cell behavior within non-neural niches remains poorly understood, although it has been shown to play a critical role in the patterning of the intestine, ensuring proper spacing of the villus and the crypt (Madison et al.,2005; Crosnier et al.,2006).
Cell–cell signaling also plays an important role within stem cell niches. Notch signaling regulates differentiation in many developmental systems (Artavanis-Tsakonas et al.,1999). Notch receptors and their ligands are highly expressed in stem cell niches, including those of the adult CNS (Stump et al.,2002), and, in many instances, notch signaling promotes stem cell maintenance at the expense of differentiation. In the intestinal stem cell niche, mice expressing constitutively active notch 1 show increased proliferation of stem cells at the expense of goblet and enteroendocrine cell differentiation (Fre et al.,2005). A complementary study removing a notch pathway transcription factor (CSL/RBP-J) from the intestinal niche in mice showed a rapid conversion of intestinal stem cells to goblet cells, which was also achieved by using a notch signaling inhibitor of γ-secretase (van Es et al.,2005). In the dental epithelial niche, notch is present in the stem cell compartment (Harada et al.,1999) but only found in tooth types that grow continuously (Tummers and Thesleff,2003), suggesting a role in regulating cell proliferation. In the hematopoietic niche, the interplay between the osteoblasts and stem cells is a key regulator of stem cell maintenance. In mice with osteoblast specific activated parathyroid hormone receptors, the notch receptor jagged was expressed at high levels. In turn, notch activation in the hematopoietic stem cells was increased, leading to an increase in the stem cell numbers (Calvi et al.,2003). This notch dependence of hematopoietic stem cells is also seen in development as the notch 1-null (Kumano et al.,2003), notch 1-chimeric (Hadland et al.,2004), and RBPjkappa (a transcription factor required for notch activity)-null mice (Robert-Moreno et al.,2005) all show abnormal hematopoietic stem cell development. In muscle stem cells (satellite cells), activated notch increases proliferation while inhibition of notch signaling by its antagonist, numb, results in commitment of progenitor cells to myoblast cell fate (Conboy and Rando,2002). Moreover, in response to injury, aged satellite cells do not proliferate and generate myoblasts sufficient for muscle regeneration due to the inability to upregulate the notch ligand delta. However, forced expression of activated notch ameliorated such age related defects (Conboy et al.,2003). Notch signaling can also promote stem cell maintenance indirectly; in the Drosophila ovary, the niche for the germline stem cell is created by contact with a support cell, the cap cell. Increased notch activation in the ovary leads to more cap cells, which in turn leads to more stem cells (Song et al.,2007). However, in some cases notch signaling can repress stem cell maintenance and promote differentiation. In stem cell population of the adult Drosophila midgut, analogous to the vertebrate intestine, loss of notch function results in increased stem cell proliferation at the expense of differentiation (Ohlstein and Spradling,2006). Moreover, the overall level of the notch ligand, delta, is a determining factor of cell fate specification in this system (Ohlstein and Spradling,2007). In human skin, high levels of delta1 are seen in the stem cell–containing (basal) regions, and this prevents these cells from responding to delta signaling from their neighbors suggesting that blocking notch signaling prevents stem cell differentiation (Lowell et al.,2000).
Ephrin ligands and their corresponding Eph family of tyrosine kinase receptors also participate in cell–cell interactions within the stem cell niche. In the mammalian intestine, terminally differentiated paneth cells are in the bottom of the crypts, immediately below the region containing stem and progenitor cells, with enteroendocrine and goblet cells in the villus above the stem/progenitor region. Paneth cells express EphB3, the progenitor cells express EphB2, and differentiated cells in the villus express high levels of ephrins-B1 and B2 (Batlle et al.,2002). Double mutant mice for EphB2/3 lose this elegant cellular organization with proliferating and non-proliferating cells intermingled (Batlle et al.,2002). This would be consistent with a role for eph/ephrin interactions in regulating migration and cell population sorting, as seen in other developmental systems. However, further studies show that eph/ephrin signaling also regulates proliferation in the intestine stem cell and adult neural niche directly. Inhibition of signaling by injection of monomeric ephrin B2 ectodomains or stimulation by overexpression of the ephrinB2 ligand or a constitutively active EpbB2 receptor increased or decreased intestine stem cell proliferation (Holmberg et al.,2006), while ephrins-B2/3, Eph B1-3, and EphA4 (Conover et al.,2000) and ephrin-A2 and Eph A7 (Holmberg et al.,2005) positively and negatively regulate adult neural stem cell proliferation, respectively. As EphB expression can be regulated by Wnts (Batlle et al.,2002), it appears that Eph/ephrin signaling provides a pathway by which the mitogenic effects of Wnts in the stem cell niche might be effected.
Cell adhesion molecules are also found in a variety of stem cell niches. An example of their role is seen in the Drosophila testis and ovary. Germ stem cells are located on the anterior end of the Drosophila ovary and are anchored to the cap cell by adherens junctions, consisting of DE-cadherin and Armadillo (β-catenin vertebrate homologue). Genetic disruption of this complex results in a loss of germ stem cells (Song et al.,2002). Adherens junctions are also present in the hematopoietic niche, where they anchor the HSC to the osteoblasts via N-cadherin and β-catenin (Zhang et al.,2003). Gap junctions, formed by connexins and responsible for chemical and electrophysiological communication between adjacent cells, play a role in the spermatogonial niches as revealed by germ cell defects in Cx 43 −/− mice, but not in Cx 32 or Cx 40 knock-in mice (Juneja et al.,1999; Plum et al.,2000).
Stem cell contact with extracellular matrix molecules within a basement membrane is a common feature in most adult epithelial niches. This basement membrane is often rich in laminins, a family of heterotrimeric ECM molecules recognized by at least three major classes of cell surface receptors, integrins, syndecans, and dystroglycan. Intriguingly, the expression of integrins in many stem cell systems is elevated and this property has been used to enrich a stem cell population from a pool of tissue by fluorescent activated cell sorting (FACS) in the following systems: colon (β1, Fujimoto et al.,2002), cornea (β1, β4, Chen et al.,2004; Pajoohesh-Ganji et al.,2006), epidermis (α6, β1, Jones and Watt,1993; A. Li et al.,1998; Blanpain et al.,2004), hematopoietic (α2, αIIb, α4, Scott et al.,2003; Corbel et al.,2005; Priestley et al., 2005; Wagers and Weissman, 2005), hepatic (α6, β1, Suzuki et al.,2000), muscle (α6, β1, Deb et al.,2004; Bajanca et al.,2006), prostate (α6, Lawson et al.,2007), and testes (α6, β1, Shinohara et al.,1999). Furthermore, integrin expression may be a key feature of the “stem cell molecular signature” (Eckfeldt et al.,2005). Integrins are heterodimers of an α and a β chain, and a comparison of three independent cross system stem cell transcriptional profiling studies (Ivanova et al.,2002; Ramalho-Santos et al.,2002; Fortunel et al.,2003), found that the α chain of the principal integrin laminin receptor α6β1 was the only shared gene expressed in all stem cells. Disruption of integrin activity in the epidermal and hematopoietic niches resulted in compromised self-renewal (Frye et al.,2003; Scott et al.,2003; Hidalgo et al.,2004; Priestley et al., 2005). There could also be a role in HSC homing and engraftment as a recent report on mice with a targeted cdc42 deletion in hematopoietic cells showed a decrease in β1 integrin expression, with excess HSCs throughout the periphery and associated defects of HSC engrafting into bone marrow (Yang et al.,2007). Although it is unclear how integrins regulate stem cell self-renewal, loss of their interaction with the basal lamina may result in loss of integrin-mediated signaling (Hynes,2002) or alterations in growth factor response (ffrench-Constant and Colognato,2004) resulting in compromised biological activity.
The blood vessels of the niche may also provide signals in addition to their role in perfusion. In the SGZ of the adult CNS, mitotic cells are closely associated with blood vessels (Palmer et al.,2000). Endothelial cells in co-culture with neural stem cells were found to release soluble factors that increase self-renewal of the stem cells, an effect that is likely to be mediated through VEGF-C/VEGFR-3 and/or pigment epithelium derived factor (PEDF) (Shen et al.,2004; Le Bras et al.,2006; Ramirez-Castillejo et al.,2006) Additionally, basal-lamina-like extensions (called fractones) extend from blood vessels in the SEZ and make contact with all cells in the niche providing a source of ECM containing collagen-IV, laminins, perlecan, nidogen and, in some cases, heparan sulfate proteoglycans (HSPG). These likely send important signals to the stem cells that may include the presentation of FGF-2, which binds to fractone HSPG (Mercier et al.,2002; Kerever et al.,2007).
In addition to these classes of signals seen in other stem cell niches, a key difference in the adult CNS is that the neural stem cells of the SEZ are also in contact with another potential source of signals, the CSF, via the processes that project through the ependymal cell layer. The role of CSF in the regulation of the adult neural stem cell is unknown, but recently it has been shown that the flow of the CSF generated by the ciliated ependymal cells in the adult CNS is critical in establishing a gradient of molecules (such as slit), which play a role in the directional migration of their daughter neuroblasts (Sawamoto et al.,2006).
SIMILARITIES AND DIFFERENCES BETWEEN ADULT AND FETAL NEURAL STEM CELL NICHES
The embryonic VZ has a much more simple cellular composition than the adult SEZ. The cells within this region appear homogenous, all retaining expression of the Sox2 protein as seen in neuroepithelial cells, and dividing only on the ventricular wall as illustrated in Figure 2. Before considering the role of extrinsic signals within the VZ in the regulation of these cells and comparing them with the adult stem cells in the SEZ and elsewhere, it is important to note that the behavior of embryonic neural stem cells is regulated at least in part by a cell intrinsic developmental timer. Single neural stem/precursor cells derived from embryonic cortex at E10 and grown in microculture show an intrinsic developmental program that generates first neurons and then glia, as seen in vivo (Qian et al.,1998,2000). Interestingly, this program extends to the ordered generation within these microcultures of layer specific neurons prior to gliogenesis (Shen et al.,2006), showing that many of the major transitions in the fate of daughter cells can occur on schedule and without any requirement for extrinsic signals.
The different classes of extrinsic signaling systems discussed above (growth factors, cell-to-cell contact, cell-to-ECM contact) have also been shown to instruct the behavior of cells in the embryonic VZ, and it is striking to what extent the same factors regulate both embryonic neural and adult stem cells (Table 1). BMPs (BMP2/4) are present in the VZ and in vitro studies have shown that overexpression of BMP2/4 leads to decreased proliferation and neuronal differentiation (W. Li et al.,1998), which can be blocked by the addition of the BMP signaling inhibitor noggin (Li and LoTurco,2000). In addition, in vivo overexpression and BMP receptor truncation studies have shown a similar role for BMP in promoting neuronal differentiation at the expense of cell proliferation in the VZ (W. Li et al.,1998). From early work on VZ stem/precursor cell cultures from the telencephalon (Murphy et al.,1990; Kilpatrick and Bartlett,1993) and neural tube (Kalyani et al.,1997), FGF-2 was found to be critical in maintaining proliferation and an undifferentiated state. Although there is expression in the VZ of all 4 FGF receptors in vivo (Bansal et al.,2003; Reid and Ferretti,2003), neural stem/precursor cells derived from the rat neural tube uniquely express FGFR4 (Kalyani et al.,1999). In contrast, a recent study found that FGFR1 and 3 are highly expressed in the VZ of dorsal telencephalon of the rat and activation of both is required for self-renewing symmetric division (Maric et al.,2007). Consistent with this role for supporting early neural stem/precursor cell proliferation, the phenotype of the FGF-2 deficient mouse has a smaller brain and fewer proliferating cells (Vaccarino et al.,1999). There is weak expression of Shh and Gli 1 and stronger expression of Gli 2 and Gli 3 in the VZ (Hui et al.,1994; Dahmane et al.,2001), and the reduced brain size of Shh mutant mouse embryos (Chiang et al.,1996) and decreased cell proliferation in both the VZ and SVZ of Shh- (Dahmane et al.,2001) and Gli 2- (Palma and Ruiz i Altaba,2004) deficient embryos shows that the Shh signaling pathway also plays a pivotal role in neural stem cell development. The VZ also has high levels of Wnt receptors, Frizzled 5, 8, 9, and secreted frizzled protein 1 (Kim et al.,2001; Van Raay et al.,2001). Wnt signaling may have diverse roles in neural development (Ille and Sommer,2005) including regulation of stem cell proliferation, as expression of stabilized β-catenin induced in vivo resulted in a dramatic increase in overall brain mass. This effect results from increased VZ stem/precursor cell expansion by a reduction in cells exiting the cell cycle (Chenn and Walsh,2002). Interestingly, targeted inhibition of β-catenin signaling results in the opposite effect, with stem cells prematurely exiting the cell cycle and differentiating into neurons (Woodhead et al.,2006).
Table 1. Similarities and Differences Between Nichc Components in the Embryonic and Adult CNSa
T.B.D., to be determined.
Growth factor signaling
BMP2, 4, Noggin
BMP2, 4, R1A, RIΔ, Noggin
FGF2, FGFR1, 3, 4
Shh, Gli1, Gli2, Gli3
SFRP-1, Fz5, 8, 9, β-catenin
Ephrin B1, A5, Eph A4, A7
Ephrin B2, B3, Eph B1, B3, A2, A4, A7
Notch 1, 3, Delta 1, Jagged
Notch 1, Jagged 1
Cx 26, 43, 45
Laminin α2, α4, α5, β2
α6, α7, β1
Syndecan 1, 4
Developing, but present
Present and in contact with all cells in nichc
Present and flow controlled by ependymal cells
Notch family members and their ligands jagged and delta are expressed in the stem cell-containing regions of the developing CNS. In situ hybridization shows high levels of notch1 and delta1 in the murine neural tube (Lindsell et al.,1996) and immunostaining shows high levels of notch-3 in the VZ at E.12 (Dang et al.,2006). LacZ transgenic reporter mice exhibit high expression of delta 1 in the VZ of the forebrain at E.10 (Beckers et al.,2000). Immunostaining analyses of human VZ show strong expression of notch-1, and weak expression of notch-3 as well as strong expression of delta-1 and sporadic expression of jagged-1 (Kostyszyn et al.,2004). Key roles for notch signaling in maintaining the VZ stem cell population have been revealed in a number of ways. First, neural stem cells differentiated from ES cells lacking RBJkappa show a failure of maintenance (Hitoshi et al.,2002). Second, depletion of stem cells is seen in the brains of mice lacking notch 1 or RBJkappa (Hitoshi et al.,2002). Third, there are increased numbers of newly-generated precursor cells following intraventricular injection of notch ligands (Androutsellis-Theotokis et al.,2006). Fourth, activation of either notch-1 or notch-3 in the forebrain results in increased numbers of (putative stem) cells with a radial glial phenotype (Gaiano et al.,2000; Dang et al.,2006). Finally, delta-like 1 mutant mice show defects in the VZ, with a thinning of the VZ and premature neuronal differentiation (Yun et al.,2002).
Ephrins B1 and A5 and Eph A4 and A7 receptors are expressed in the VZ (Mackarehtschian et al.,1999; Stuckmann et al.,2001; Greferath et al.,2002; Depaepe et al.,2005). With its apical to basal expression gradient in the VZ and diminished expression in the cortex, Ephrin B1 is thought to facilitate cell migration outside of the VZ (Stuckmann et al.,2001) while ephrin A5/Eph A7 promotes apoptosis of the stem/precursor population (Depaepe et al.,2005). This provides a mechanism in addition to proliferation to control the size of the stem cell pool (Depaepe et al.,2005).
Cadherins are expressed in adherence junctions (AJ) at the apico-lateral membrane of neuroepithelial cells (Aaku-Saraste et al.,1996). Neuroepithelial cells switch their expression from E- (epithelial) to N- (neural) cadherin prior to the onset of neurogenesis (Hatta et al.,1987; Duband et al.,1988). Inactivation of mouse N-cadherin at early developmental stages results in a distortion of the neural tube due to the disruption of cadherin-based adherence junctions (AJ) (Radice et al.,1997), and N-cadherin blocking antibodies injected directly into the developing chick CNS result in disruption of neuroepithelium organization within the VZ (Ganzler-Odenthal and Redies,1998). In contrast, inactivation of other cadherins (R-cadherin and cadherin-6) does not cause major proliferation or morphological defects in the CNS (Stoykova et al.,1997; Inoue et al.,2001). N-cadherin, therefore, appears to be the key cadherin at early stages of brain formation (Hirano et al.,2003). In support of this, N-cadherin-deficient Zebrafish show hyperproliferation of neural stem/precursor cells (Lele et al.,2002) and a very recent study using cre/lox technology to remove N-cadherin from the developing CNS showed almost complete randomization of cortical structure, with disruption of AJ, mixing of mitotic and post-mitotic cells, and a failure of the radial glial processes to span the cortex (Kadowaki et al.,2007).
A number of other studies also reveal the importance of the AJ between neural stem cells in the VZ. Interference with the AJ-associated afadin/AF-6 protein in mice leads to the disruption of neuroepithelial cell polarity, as evidenced by mislocalization of prominin-1 to the baso-lateral plasma membrane (Zhadanov et al.,1999; Manabe et al.,2002). In addition, disruption of AJs by inactivation of cadherin cytoplasmic partners β-catenin (Machon et al.,2003; Junghans et al.,2005) and α E-catenin (Lien et al.,2006) leads to impairment in the apical distribution of Par complex proteins and centrosomes. The importance of apical-basal polarity has been directly demonstrated by analyzing the phenotype of the hyh (hydrocephalus with hop gait) mouse (Chae et al.,2004). In this mouse, the apical delivery of some proteins, including N-cadherin and aPKC, is impaired resulting in a premature depletion of the stem/precursor pool within the VZ and a consequent reduction of cortical thickness. In addition, disruption of cell polarity and associated loss of AJs in Lgl1 knockout mice result in the appearance of rosettes of mislocalized proliferating stem/precursor cells in areas normally occupied by more differentiated cells (Klezovitch et al.,2004). Such rosettes are also seen in a Kap3 knock-out mouse (Kap3 being part of the Kif3 molecular motor complex), where the delivery of N-cadherin protein at the plasma membrane is impaired (Teng et al.,2005).
Gap junctions may also play a significant role in the regulation of embryonic neural stem/precursor cells. Cx 26 and 43 are expressed in the VZ (Bittman and LoTurco,1999). In vitro, gap junction communication plays a role in survival and proliferation of VZ cell cultures derived from the neural tube of rats (Cai et al.,2004; Cheng et al.,2004), which express both Cx43 and Cx 45. In addition, Cx43 is closely linked with the ability of FGF-2 to maintain VZ cells in an undifferentiated state, with inhibition resulting in differentiation and cell death even in the presence of FGF-2 (Cheng et al.,2004). A detailed analysis of mice with knockouts of gap junctions expressed in the VZ (Cx 26 and 43) will provide further insights into the role of gap junctions in neural stem/precursor cells, just as the analysis of the Cx43 null mice showing precursor cells accumulating in the intermediate zone and unable to migrate upward to participate in cortical patterning (Fushiki et al.,2003) reveals a role in migration. The exencephaly associated with a thickened and distorted neuroepithelium suggestive of stem/precursor cell proliferation defects in the Cx40/Cx43 double knock-out (Simon et al.,2004) emphasizes that other connexins are likely expressed in the VZ and that double knock-outs may be required to reveal the roles of gap junctions in NSC biology.
As in adult epithelial and neural stem cells, embryonic neural stem cells directly contact the extracellular matrix. However, while in the adult SEZ the fractone extensions from blood vessel basement membranes contact each stem cell in the embryo, in the embryo the basal process of the radially-orientated stem cells makes contact with the pial basement membrane directly. Interestingly, matrix molecules are also found in the VZ adjacent to the cell body at the other end of the stem cell, with both laminin β1 and α2 chains present in this region (Hunter et al.,1992; Campos et al.,2004). Like stem cells in many other tissues, neural stem cells from the embryonic VZ also express high levels of β1 integrin (Graus-Porta et al.,2001; Campos et al.,2004; Nagato et al.,2005; Hall et al.,2006), suggesting a role for integrin/extracellular matrix interactions in their regulation. In vitro studies using neurosphere formation (aggregates of stem, precursor, and more differentiated cells that form from a single neural stem cell when grown in non-adherent culture conditions) (Reynolds and Weiss,1992) to assay neural stem cell behavior have given conflicting results. Blocking antibody experiments suggested a role for integrins in maintenance (Campos et al.,2004), but more definitive experiments using neurospheres grown from cells lacking β1 integrin showed no defects in maintenance, although significant decreases in proliferation and migration capacities and an increase in apoptosis were seen (Leone et al.,2005). Critically, the studies of knock-out mice for β1 and α6 integrin (that together make the principal integrin laminin receptor in the CNS) have revealed a cortical, rather than a VZ, phenotype. In the α6 −/− mouse, there is no morphological difference in the VZ while ectopic neuroblasts are present in the pia in embryonic brains (Georges-Labouesse et al.,1998; Haubst et al.,2006). The perinatal lethality of this knock-out prevents postnatal analysis. The effect of removing β1 from neural stem cells has been examined using a cre/lox approach using the promoter of nestin (an intermediate filament expressed in neural stem cells) to drive cre expression. These mice also have ectopic growths in the cortical marginal zone along with retraction of the endfeet of the radial glia at the pial surface and a poorly formed, discontinuous pia (Graus-Porta et al.,2001). However, just as in the α6 −/− mice, there is no morphological abnormality in the VZ (Graus-Porta et al.,2001). As similar abnormalities of pial discontinuities and cortical ectopias are seen in the mice with deletion of the laminin γ1 chain (so preventing laminin trimer formation) (Halfter et al.,2002), we can conclude that the interaction between the stem cell basal end foot and the pia is important for neuronal migration, as expected given that these cells use the basal process as a guidepost for migration (Rakic,1995). However, the lack of a VZ abnormality is surprising given the levels of integrin expression in this region and recent evidence in ES cells showing that laminin 111 (comprising the α1, β1, and γ1 chains) can efficiently facilitate transition of cells into neuronal precursors and neurons (Goetz et al.,2006), further suggesting a role for the integrin/extracellular matrix interaction in mediating cell proliferation and fate choices. However, given the lack of definitive in vivo evidence for such a regulatory role in the VZ, experimental approaches that specifically target the VZ while not perturbing the same interactions between the pial membrane and basal process may be required.
Two other potential sources of regulatory signals are CSF and blood vessels, which, as discussed above, both contribute to the structure of the adult SEZ niche. Evidence that the emergence of blood vessels in the CNS (at E10 in the mouse neuroepithelium) (Herken et al.,1978,1989) might play a role in regulating neural stem cell behavior comes from studies showing that E10 cortical neural stem cells cocultured with endothelial cells generated larger clones and fewer neurons (Shen et al.,2004). Moreover, lineage tracing showed colonies co-cultured with endothelial cells continued to undergo expansive symmetric divisions and lacked asymmetric neuron-generating divisions seen in neural stem cell colonies alone (Shen et al.,2004). CSF has been shown to maintain the survival, proliferation, and differentiation of neuroepithelial cells in vitro (Gato et al.,2005; Miyan et al.,2006). Furthermore, a recent in vivo study that drained the CSF and administered BrdU found a marked decrease in proliferation 3 days after analysis and suggests a role of the CSF in regulating neural stem/precursor cell proliferation (Mashayekhi and Salehi,2006).
IS THE EMBRYONIC VENTRICULAR ZONE A NICHE?
It is clear from the comparison above that there is much in common between signals regulating embryonic neural stem cells and those known to control adult stem cell types within their niches. This, therefore, raises the interesting question as to whether the embryonic VZ constitutes a niche. As described above, a conventional definition of a niche would be a microenvironment containing both the stem and other cell types, which provides signals that promote and maintain slowly-dividing stem cells throughout life. At first sight, the embryonic VZ fails to meet this definition on three counts. First, prior to the development of ependymal cells later in embryonic development, the VZ contains only one cell lineage (the stem cells themselves and their daughters). Second, the structure is transient, being lost in postnatal development. Third, embryonic neural stem cells divide rapidly. However, it can easily be countered that the definition is over-simplistic and that exceptions have already been described. Stem cells in the midgut of adult Drosophila also have only their daughter cells as neighbors, and do not interact with any other cell lineage (Micchelli and Perrimon,2006; Ohlstein and Spradling,2006). Transient niches during development are also seen in haematopoiesis, where the yolk sac and aortic wall provide temporary niches prior to definitive haematopoieses being established in the liver (Medvinsky et al.,1993). Similarly, the rate of stem cell division in the embryo would be expected to be greater than in the adult, given the differing requirements for growth and maintenance. That said, two sets of experiments do suggest that any VZ niche is neither necessary nor sufficient to promote stem cell behavior in the VZ and argue against the VZ being considered a conventional stem cell niche. First, isolated single neural stem cells recapitulate the timing of neuronal and glial production (as described above) even when removed from the VZ and grown in microculture (Qian et al.,1998). At least this aspect of neural stem cell behavior does, therefore, not require any niche. Second, the basal progenitor daughter cells generated by asymmetrical divisions of the neural stem cells may migrate back out of the SVZ into the VZ and establish contact with the ventricular surface before exiting again and completing their neuronal differentiation (Noctor et al.,2004). The reason for this is unknown, but the fact that this does not lead to the re-entrant cells becoming stem cells again argues that the VZ does not have an instructive role in cell programming as has been shown for the ovarian niche in Drosophila (Kai and Spradling,2004). Whatever the extent to which the VZ represents a niche, it is however clear from the review of signaling mechanisms above that many of the regulatory pathways present in bona fide stem cell niches operate in the VZ. Lessons learnt from these other stem cells are, therefore, likely to have much to tell us about embryonic neural stem cell biology.
DYSREGULATION OF THE NICHE AND ASSOCIATED DISEASES
It is reasonable to assume that since the specialized environment provided by the niche regulates multiple aspects of stem cell behavior, abnormalities in this environment as well as in the intrinsic properties of the NSC will lead to abnormal development. These abnormalities might be manifested by changes in NSC behavior such as rate of proliferation, cell cycle time, spindle regulation, apical or basal adhesion, or migration as summarized in Figure 3. Abnormalities of embryonic neural stem cell function, therefore, represent a potentially important cause of neurodevelopmental diseases. One major class of such diseases are the microcephalies, in which an architecturally normal brain is abnormally small (Bond and Woods,2006) as would be expected if the number of symmetrical expansive divisions prior to the onset of neurogenesis was reduced. A number of causative genes have been identified, and three of these encode proteins linked to the centrosome: ASPM, CDK5AP2, and CENJP (Bond et al.,2002,2005). This is an interesting observation, as the centrosome is a key organelle in determining the angle of cell division by orientation of the mitotic spindle (Hyman and White,1987; Hyman,1989; Grill et al.,2001). In the case of ASPM, a direct role in the regulation of division angle and the subsequent plane of cell cleavage has been shown in neural stem cells within the VZ (Fish et al.,2006). Alterations in the angle of division would be predicted to have a major effect on the balance between symmetrical and asymmetrical divisions. The current model of asymmetrical divisions, based on work in invertebrates, is that putative fate determinant molecules such as notch and numb are localized to one pole of the cell and then the angle of cleavage during division is regulated such that the determinants are inherited by only one daughter cell, thus resulting in daughter cells adopting different fates (Betschinger and Knoblich,2004). Initial studies in the ferret CNS supported such a model; notch was shown to be located apically and stem cells dividing with a vertical cleavage plane (which will split apical determinants equally between the daughter cells) generated two stem cells while horizontal cleavage planes generated a neuronal precursor and one stem cell (Chenn and McConnell,1995). While subsequent work has shown the number of horizontal divisions to be far too few to account for the degree of neurogenesis (Smart,1973; Landrieu and Goffinet,1979; Chenn and McConnell,1995; Heins et al.,2001; Haydar et al.,2003) elegant studies from Huttner and colleagues have shown that it is the distribution of the very small apical plasma membrane which is the determinant in the control of symmetrical versus asymmetrical division (Kosodo et al.,2004; Gotz and Huttner,2005). A cell whose cleavage plane bissects the apical plasma membrane undergoes symmetric division, while a cell whose cleavage plane still appears to be vertical but bypasses the apical plasma membrane undergoes asymmetric division. Consequently, the model in which cleavage plane regulates fate appears to be correct; it is just that in the vertebrate CNS, the putative fate determinants are so restricted in location that very small changes in cleavage plane are sufficient to change fate (Kosodo et al.,2004; Gotz and Huttner,2005). Small perturbations in cleavage plane secondary to centrosome abnormalities could, therefore, dramatically alter the ratio of symmetric and asymmetric divisions, thus generating large changes in final cell number as seen in the disorders of brain size.
Can extracellular components of the niche also regulate the plane of cell division? A recent study of epidermal stem cells showed that integrins and cadherins are both required for the regulation of mitotic spindle angle that ensures the division plane is either perpendicular to the underlying basement membrane, generating two stem cells in a symmetrical division, or parallel to the underlying basement membrane, generating one stem cell and a committed keratinocyte (Lechler and Fuchs,2005). These adhesion molecules, therefore, regulate the axis of division in at least one class of stem cells, and may do so in neural stem cells. In light of this, it is interesting to note that some children with mutations in the laminin α2 chain (causing the disease merosin-deficient congenital muscular dystrophy) show a failure of occipital lobe development (agyria) (Philpot et al.,1999). This phenotype would be consistent with a local failure of stem cell expansion analogous to the more global failure seen in microcephaly. Further work asking whether mutations in genes encoding other regulators of stem cell division cause microcephaly is required, as are studies examining other developmental diseases such as macrocephaly and autism where cell numbers may also be abnormal (McCaffery and Deutsch,2005).
Another group of diseases that could result from dysregulated signalling within the niche are the brain tumors arising from either persistent growth of an embryonic neural stem cell population (as shown in Fig. 3) or a transformation of an adult neural stem cell population. Recent work has confirmed the hypothesis that gliomas are driven by a cancer stem cell population (Hemmati et al.,2003; Singh et al.,2003; Singh et al.,2004b; Bao et al.,2006), expressing the same marker (prominin-1/CD 133; Singh et al.,2003; Singh et al.,2004b) used to select for human neural stem cells (Weigmann et al.,1997; Uchida et al.,2000) As few as 100 of these cancer stem cells were sufficient to efficiently form tumors in a mouse model (Singh et al.,2004b). It has recently been proposed that while these cancer stem cells and their progeny are randomly distributed, they still lie within a specialized niche whose ECM may facilitate paracrine signaling and allow tumor progression (Polyak and Hahn,2006). Interestingly, BMP signaling can inhibit cancer stem cell proliferation (Piccirillo et al.,2006), just as it does with embryonic VZ, SVZ, and adult SEZ neural stem cells. This highlights the point that a better understanding of normal stem cell regulation may lead to effective therapies for stem cell tumours such as gliomas.
While the formation of new ectopic microenvironments might contribute to tumour development, it should be noted that in other scenarios this might represent a normal regenerative response. Examination of the response to experimental ischaemic lesions in the mouse showed that vascular remodeling in the peri-infarct area was associated with the production of the chemokine SDF-1 and the growth factor Ang1 that promote the migration of neighbouring neuroblasts (Ohab et al., 2007). While this “neurovascular niche” is distinct from conventionally defined stem cell niches such as those in the adult SEZ, the results do emphasize that development and repair will likely induce changes in multiple different microenvironments and suggest that examination of stem cell niches during repair is an important area for further study.
Despite the variety of adult stem cell types, and the varying degrees of anatomical complexity of their immediate microenvironments, it is clear that common mechanisms regulate proliferation and differentiation. We have shown here that many of the lessons learnt from studies of these adult stem cells can be applied to questions about the regulation of embryonic neural stem cells in the VZ. Equally, we have argued that the VZ lacks some of the properties expected in a stem cell niche. One role of a stem cell niche is to keep the division of a potentially immortal cell under control for the life of the animal, the importance of which is emphasized by the observation that gliomas and other tumors may arise from cancer stem cell populations (Hemmati et al.,2003; Singh et al.,2003; Singh et al.,2004a,b). However, given the transient nature of the embryonic neural stem cell population and the need for a series of rapid divisions to generate the cell numbers required for growth, the lack of such controls may not be as disadvantageous as would be the case for an adult cell population. That said, the significant frequency of CNS stem cell tumors in children (Pallapies,2006) may, at least in part, be a consequence of the dysregulation of niche-related signals for the neural stem cells. A better understanding of the signaling mechanisms may not only inform studies on the normal and abnormal growth of the brain but also point to novel therapies for cancers of the developing CNS.
We thank K.C. Alexander (NIA) for assistance in the illustrations and the members of our laboratories for stimulating and insightful discussions. This work was supported by the NIA Intramural Research Program. J.D.L. is supported by the NIH-Cambridge Graduate Partnership Program. M.S.R. is supported by the Packard Center. C.ff-C. is supported by the Wellcome Trust, the MRC, and the BBSRC.