Keywords

1 Numbers Matter and Two Cautions

During the long process of evolution the brain has—on average—increased in size, but most importantly in complexity of structure (emergence of neocortex and subsequently of gyrencephalia) and of connectivity, with the latter—among mammals—being correlated to the emergence of more complex patterns of behaviour (Herculano-Houzel 2009). The striking increase in the size of the brain has been underlined by the appearance of new neurogenic progenitor populations, such that the “ancient” neuroepithelial cells that (early in evolution and in embryogenesis) form the primitive embryonic neural tube are transformed to radial glial cells when the tube becomes thicker and these are subsequently complimented by subventricular zone progenitors (for the neocortex to appear) and by outer subventricular zone progenitors (for gyrencephalia to appear) (Fietz et al. 2010; Stancik et al. 2010). Interestingly, the emergence of larger and more complicated brains has been accompanied by a marked decrease in the number of neurogenic progenitors surviving within the adult, mature tissue. Not only certain adult neurogenic areas (or niches) disappeared during evolution but also the absolute numbers of neuronal progenitors per area has decreased (Ferretti 2011; Kazanis and ffrench-Constant 2012). This has resulted in a significant weakening of the neuroregenerative capacity of the adult brain; thus, to a higher susceptibility to injury. But why has that happened? Intuitively, we believe that the increased complexity of the brain, with the emergence of very specialised neuronal types that are interconnected via numerous axonal networks, is not compatible with efficient cell-replacement machineries. The additional observation that cell replacement of glial cells—cells considered to be only supportive to neurons and more homogeneous in morphology and function—has won the battle of survival during evolution turned this intuitive belief to a “dogma” that dominated developmental neurobiology to such a degree that for many decades the first indications of persistent neurogenic activity in the adult rodent brain by Altman (1969) were dismissed. On the other hand, why would a neuronal progenitor, located within a specialised and protective microenvironment in the brain, be worried about what happens to its progeny? How can an inefficient cell-replacement process impact directly on the progenitor cell, since the vast majority of stem cell progeny normally die before reaching maturity and before being incorporated in existing networks even in the adult neurogenic systems that survived the forces of evolution (Morshead and van der Kooy 1992; Lu et al. 2011)? A possible explanation could be that the enlargement of the brain required more space for mature tissue, leading to a gradual reduction in the size and number of niches, especially since this did not incur any serious functional defects. Another explanation could be sought in energy efficiency pressures, with brains spending less energy for maintaining a futile neurogenic process being positively selected. Finally, recently published data indicate that the cell cycle kinetics of progenitors depend on real time, rather than on developmental time (i.e. a progenitor can keep dividing for a certain number of days/weeks irrespective of whether this means old age in a rodent and infancy in humans) (Amrein et al. 2011). This—among many other possibilities—indicates that the decrease in the numbers of surviving progenitors is a function of time, with organisms with long life duration spending most of their adult lives with small pools of progenitors.

Nevertheless, one of the most important messages stemming from the discovery of persistent neurogenic activity in the adult mammalian brain is that replacement of neurons is not a process completely lost during evolution; and this has great implications in regenerative medicine. It is much easier to exogenously influence an existing cellular process rather than to attempt to re-introduce it after it has vanished. In that context, the discovery that neurogenesis has a direct role in the function of the hippocampus, involving memory and learning (Deng et al. 2010), added a new promising experimental target in the field of mental health disorders and of cognitive biology. Importantly, though, we should not forget that in regenerative and rehabilitation medicine numbers matter: not only the numbers of progenitors surviving in the adult brain, but also the numbers of cells that are affected during degeneration or disease. In this review, I will attempt to summarise the available information on persistent neurogenesis in the adult brain with a focus given, whenever possible, to numbers. Most of the evidence stems from experiments carried out in animals, especially rodents, but whenever appropriate I will discuss findings from the primate (including the human) brain.

Before proceeding further, though, two cautions have to be stated regarding “adult neurogenesis”. The first one has to do with the definition of the term “adult”. One possible way to define adulthood in the context of a tissue is as “the stage in which the tissue has reached a stable anatomical structure, with no further addition or elimination of subdivisions and which can only be altered by degenerative processes (injury or aging)”. In that sense, the brain—or at least some specific structures such as the corpus callosum (Sturrock 1980)—might be reaching adulthood much later than generally considered in rodents and much earlier in humans. In the same direction, recent evidence from the human and rodent brain suggest that the gradual decrease in postnatal neurogenic activity might not be a degenerative process of the adult brain but rather the end of a “prolonged developmental process” with the system in rodents reaching the adult steady state at around 6 months post natally (Ben Abdallah et al. 2010; Knoth et al. 2010; Amrein et al. 2011). In that case, a large volume of experimental work in “adult neurogenesis” in the rodents has been performed in a non-adult neurogenic system. Moreover, if the numbers of persisting progenitors are a function of time, then an additional caution has to be made regarding the extrapolation of experimental results from adult rodents (e.g. 3 month old) to the adult human brain (e.g. 30 years old) (Amrein et al. 2011; Sanai et al. 2011).

The second caution has to do with the fact that experiments in the laboratory are performed in animals kept in captivity and in very controlled, stereotypic and stable conditions. Recent data suggest that in mice captured in the wild neurogenesis in the hippocampus is not directly dependent on exercise, as has been shown in laboratory-kept mice (Klaus and Amrein 2012; Klaus et al. 2012). A possible explanation for this discrepancy could be that in experimental conditions the levels of homeostatic neurogenesis are low due to the minimal exposure of animals to stimuli. This would therefore mean that it is “relatively easy” to induce increases in neurogenesis in the hypo-active laboratory animals. Other studies have also showed that specific features of progenitor behaviour (such as cell cycle kinetics) show a rather genetic regulation, while others (such as the rate of differentiation and the rate of survival) are more influenced by the environment (Amrein et al. 2011; Roth et al. 2012). In that sense, the caution regards the extrapolation of results regarding the effects of the administration of chemical substances or of the manipulation of the environment in the levels of neurogenesis in animals kept in captivity to animals living in the wild, such as humans (Hauser et al. 2009).

2 Cellular Plasticity in the Adult Brain; When New Cells are Required?

2.1 Gliogenesis in the Parenchyma of the Brain

Although the brain is widely perceived to be a structurally non-plastic organ, apart from the micro-level of the synapse and of the dendrite, the truth is that whenever appropriately stimulated, brain tissue can host robust proliferative and migratory events. If an area is injured by mechanical forces, cytotoxic substances or hypoxia, the widely spread astrocytes react immediately. They change their morphology by increasing in size and extending processes, they proliferate and they also migrate at the site of lesion. This is called astrogliosis (Fig. 1a, b) and in many cases results in the formation of permanent scars in the tissue (Fawcett and Asher 1999). In exactly the same instances another pool of cells becomes equally activated: microglial cells (Fig. 1c, d). Microglia comprise the innate immune system of the brain and in many cases their response is complimented by blood-borne macrophages that invade the tissue, often through the broken blood–brain barrier (Perry et al. 2010). More impressively, if the insulating and protective myelin sheaths that cover neuronal axons are destroyed due to chemotoxic or autoimmune reasons (a process called demyelination), then another pool of cells, the oligodendrocyte progenitors, are recruited: they proliferate (Fig.  1e), migrate and successfully regenerate myelin sheaths (remyelination) (Franklin and ffrench-Constant 2008). Importantly, in recent years evidence has appeared indicating that these processes, that were considered to be relevant only to the world of glial cells, might hide a more dynamic capacity. Astrocytes activated by a focal cortical injury were shown to be able to act as multipotent progenitors when cultured in vitro (Buffo et al. 2008), while oligodendrocyte progenitors responding to focal demyelination were found to generate cells of the peripheral nervous system (schwann cells) and neurons in vivo (Guo et al. 2010; Zawadzka et al. 2010).

Fig. 1
figure 1

Gliogenesis in the brain parenchyma. (Panels a, b) Microphotographs of adult rat brain tissue immunostained for glial fibrilary acidic protein (GFAP, in red), a marker of astrocytes. Numbers of astrocytes increase significantly in response to degeneration, in this case after an experimental model simulating stroke, a phenomenon called astrogliosis. (Panels c, d) Microphotographs of adult rat brain immunostained for isolectin B4 (IB-4, in green) a marker of microglial cells and Ki67 (in red) a marker of proliferating cells. A degenerative insult, in this case an experimental model simulating stroke, induces a massive proliferative response of microglial cells. (Panel e) Microphotograph of adult mouse brain immunostained for CNPase (in green) a marker of cells of the oligodendroglial cell lineage and PCNA (in red) a marker of proliferating cells. A demyelinating insult, in this case the focal infusion of lysolecithin (1 %) in the cortex, leads to the destruction of myelin sheaths (note the absence of CNPase positive processes), and to the mitotic activation of oligodendrocyte progenitors that will regenerate lost oligodendrocytes (remyelination). (ctx cortex, LV lateral ventricle, str striatum)

Full size image

2.2 Neurogenesis/Gliogenesis in Cytogenic Niches of the Brain

Surprisingly, and in contrast to the above-mentioned glial cell-replacement processes that operate in very low levels in the homeostatic adult brain and show a robust response after degeneration, neuron cell-replacement processes show a higher daily level of activity during homeostasis and serve specific functional needs. Adult neurogenesis is almost exclusively driven by pools of adult neural stem cells that survive in two specialized microenvironments in the adult brain: in the subependymal zone (SEZ/often called subventricular zone) of the lateral walls of the lateral ventricles and in the subgranular zone (SGZ) of the dentate gyrus in the hippocampus (Doetsch et al. 1997; Seri et al. 2004) (Fig. 2). The same cytogenic niches also generate glial cells (astrocytes and oligodendrocytes), though in much lower frequencies.

Fig. 2
figure 2

The cytogenic niches of the adult mammalian brain. Graphic illustration of the cytoarchitecture of the subependymal zone (SEZ/centtre left) and the subgranular zone (SGZ/centre right) neurogenic niches. Note the common features: adult neural stem cells are of astroglial morphology (light blue cells), neurogenesis occurs through an intermediate precursor stage (green cells) leading to the generation of immature neurons (red cells) and is closely related to the presence of blood vessels (in black). Note also the ependymal cell layer that forms the wall of the lateral ventricle at the side of the SEZ and the process of the NSC that intercalates among ependymal cells and contacts the content of the ventricle. In the panels at the left side are shown microphotographs of adult mouse brain tissue immunostained for EYFP expressed via the human promoter of GFAP (in green, marking astrocytes—including adult NSCs—and their progeny) and either doublecortin (Dcx/in red, marking new neurons) or Olig2 (in red, marking oligodendrocyte progenitors). Double positive cells (yellow) within the olfactory bulbs are SEZ-generated new neurons and within the corpus callosum are SEZ-generated oligodendrocyte progenitors. In the panels, at the right side are shown microphotographs of adult mouse brain tissue immunostained for GFAP (in green, marking astrocytes) and Dcx (in red, marking new neurons). Note the coexistence of astrocytes (some of which are NSCs) and Dcx-positive cells within the thin SGZ at the base of the dentate gyrus. (This figure is an adaptation from Kazanis et al. 2008 published under the Creative Commons Attribution 3.0 Unported License http://creativecommons.org/licenses/by/3.0/)

Full size image

2.2.1 The Architecture of Cytogenic Niches

The process of cell generation is similar in the two neurogenic niches of the adult rodent brain; therefore, the main cell types located therein are also similar. NSCs of astroglial morphology that remain relatively quiescent generate precursors of neuronal commitment that are called neuroblasts and express doublecortin and the polysialiated form of NCAM, via a cell amplification step (Fig. 2). In the SEZ, transit amplifying precursors undergo many symmetric self-renewing divisions before producing committed progenitors (Morshead et al. 1998; Doetsch et al. 1999) and express the transcription factor Mash1 (Parras et al. 2004) and the receptor for epidermal growth factor (Doetsch et al. 2002). Apart from neuron-committed progenitors few recent reports have demonstrated the generation of oligodendrocyte progenitors from the SEZ that are subsequently migrating to the corpus callosum (Jackson et al. 2006; Menn et al. 2006; Etxeberria et al. 2010; Jablonska et al. 2010). Although the process of lineage commitment remains largely unexplored the available evidence indicates both the existence of separate populations of transit amplifying progenitors (with either neuronal or glial commitment, expressing transcription factors Pax6 and Olig2, respectively) (Hack et al. 2005) and that cells of neuronal fate retain the ability to transdifferentiate into oligodendroglial progenitors upon gliogenic stimulation (Jablonska et al. 2010). Approximately 15,000 cells are born every day in each SEZ (Lois and Alvarez-Buylla 1994) and the vast majority of these are neuroblasts. The average ratio of oligodendrogenesis versus neurogenesis has been estimated to 1:20, with approximately the same number of oligodendrocytes being generated in all rostrocaudal levels of the SEZ, but neurogenesis being markedly decreased caudally (Menn et al. 2006). In the SGZ, the volume of cells generated is smaller than that of the SEZ with estimates varying from 9,000 cells (Cameron and McKay 2001) to 4,000 cells per SGZ per day (Rao and Shetty 2004). In the hippocampal niche, the amplification process is limited to 1–2 divisions and transit amplifying progenitors express doublecortin, similar to neuroblasts and immature neurons (Seri et al. 2004), although there is evidence for Mash1 and Tbr2 expression (Yoshihara et al. 2007; Hodge et al. 2008).

Another common structural feature of the two niches is the close distance between proliferating neural precursors and the vasculature (Palmer et al. 2000; Mirzadeh et al. 2008; Shen et al. 2008; Tavazoie et al. 2008). The SEZ neurogenic niche—which has been more extensively described—is preferentially rich in blood vessels as compared to other periventricular domains (Kazanis et al. 2010) with precursors contacting blood vessels in microdomains void of astrocytic endfeet and pericytes (Tavazoie et al. 2008). Regarding the SGZ, in mice subjected to exercise increased neurogenesis was correlated with an increase in vascularisation (Clark et al. 2009). A structural characteristic that is specific for the SEZ is the proximity to the lateral ventricle, from which it is separated by the monolayer of ependymal cells. The importance of ependymal cells and of the regulated communication with the cerebrospinal fluid (that fills the ventricles) in the regulation of neurogenesis is underlined by several findings. Ependymal cells express regulating factors, such as several bone morphogenetic proteins (Colak et al. 2008; Gajera et al. 2010) and pigment epithelium-derived growth factor (Ramirez-Castillejo et al. 2006), while their cilia create a gradient of factors at the ventricular side that instructs the direction of migration to neuroblasts (Sawamoto et al. 2006). Moreover, ependymal cells create specific pinwheel structures in the ventricular wall, allowing the regulated intercalation of monociliated processes extended by astrocytes (potentially the NSCs) that contact the content of the ventricles (Doetsch et al. 2002; Mirzadeh et al. 2008) (Fig. 2). Interestingly, such cilium-bearing processes have also been identified on SGZ stem cells (that are not positioned near the ventricle) and are important for adult neurogenesis mainly through regulation of sonic hedgehog signalling (Han et al. 2008). Finally, when comparing the neurogenic and non-neurogenic parts of the ventricular walls two more—possibly important—features can be observed: first, the neurogenic side is rich in myelinated axons (Fig. 3) and second, at the non-neurogenic side an astrocytic layer appears below the ependymal cell layer as if insulating the subependymal area from the content of the ventricle (Kazanis et al. 2010) (Fig. 3). Another major element in the architecture of the niche is the extracellular matrix. The SEZ, more similar to the embryonic CNS than to mature neural tissue, is rich in laminins, tenascin-C and sulphated proteoglycans (Kazanis et al. 2007, 2010; Akita et al. 2008) with fractones (extensions of the blood vessel basal lamina) contacting almost all the different cellular elements of the niche (Mercier et al. 2002). Various components of this matrix have been shown to be important for the post-natal formation of the niche (Peretto et al. 2005; Kazanis et al. 2007), and the modulation of growth factor activity (von Holst et al. 2006; Kerever et al. 2007; Sirko et al. 2007). Importantly, the interaction between extracellular matrix molecules and their receptors on cells, such as the laminin-integrin interactions, are crucial for regulating the proliferation of progenitors (Shen et al. 2008) and possibly the activation of NSCs (Kazanis et al. 2010).

Fig. 3
figure 3

Structural hallmarks of the SEZ. Microphotographs of adult mouse brain tissue, at the area around the lateral ventricle (LV) immunostained either for GFAP (left panels, in green, marking astrocytes) or for myelin basic protein (MBP/right panels, in red, marking myelin sheaths surrounding axons). Note that the neurogenic side (the SEZ niche) is located at the lateral wall of the lateral ventricle (here at the right walls of the LV). Also, note the astrocytic layer that is found next to the ventricular wall at the non-neurogenic side (inset 1) and the high density of myelinated axons found adjacent to the ventricular wall only at the neurogenic side (inset 2, with arrowheads indicating the axon-poor non-neurogenic side of the LV). (cc corpus callosum. The figure is an adaptation from Kazanis et al. 2010)

Full size image

In addition to cells of the adult NSC lineage and blood vessels, experimental work has also revealed the presence of cells of the innate (in homeostasis) and the blood-borne (after injury) immune system within cytogenic niches in the brain. These cells have been shown to exert important regulatory roles, in vitro (Walton et al. 2006; Thored et al. 2009) but data from in vivo experiments still remain inconclusive. For example, after stroke, the activation of microglial cells can be both pro-neurogenic (Walton et al. 2006) and anti-neurogenic (Hoehn et al. 2005), while a recent study failed to find any significant impact caused by their ablation (Heldmann et al. 2011). An interesting finding has been generated from animal models of Huntington’s disease. Increased neurogenesis was observed only in animal models in which the disease was mimicked by the use of cytotoxic substances, inducing inflammatory reactions, while no change was observed in transgenic models, in which cell loss is minimal (Phillips et al. 2005). In addition, recent experimental work in an animal model of demyelination indicated impaired proliferation caused by the inflammatory microenvironment (Pluchino et al. 2008), while examination of tissue from patients suffering from multiple sclerosis indicated sustained activation of the SEZ within the inflammatory microenvironment created by the disease (Nait-Oumesmar et al. 2007). Nevertheless, locally residing microglial cells are likely to be important in the accurate buffering of cell numbers and local migratory events, as they participate in the removal of dying progenitors (Sierra et al. 2010), surprisingly aided by neuronal progenitors (Lu et al. 2011). Furthermore, niche microglia were shown to have distinct properties from those residing outside, such as the ability for indefinite expansion in vitro (Marshall et al. 2008).

The above presented description of the cytoarchitecture of the niches refers to the rodent brain (most data are derived from experiments in the mouse but they largely apply to the rat). Recent ultrastructural and immunohistochemical analyses of the human and non-human primate brains have revealed the existence of similar neurogenic areas at the SEZ of the lateral walls of the lateral ventricles (Jackson et al. 2006; Fancy et al. 2009) and in the hippocampus (Eriksson et al. 1998). The main differences between the rodent and the human SEZ are: the existence of a hypocellular (gap) area underneath the ependymal cell layer that is followed—deeper in the tissue—by an astrocyte-rich ribbon-like zone, the existence of displaced ependymal cells, the absence of chains of migrating neuroblasts and the absence of transit amplifying precursors (Sanai et al. 2004; Quinones-Hinojosa et al. 2006; Wang et al. 2011). Notably, the human SEZ acquires this distinct, mature architecture as soon as cytogenic activity is drastically reduced, at around 18 months after birth (Sanai et al. 2011). On the other hand, NSCs are of similar astroglial morphology—often in contact with the content of the ventricles—and clusters of neuroblasts are found within the hypocellular zone (Jackson et al. 2006; Fancy et al. 2009). In the Macaque brain, the structure of the SEZ neurogenic niche is similar to that of the human brain, albeit chains of migrating neuroblasts are observed similarly to the rodent SEZ (Jackson et al. 2006).

2.2.2 Focusing on Adult NSCs: How Many—How Plastic?

Two of the factors that determine the regenerative capacity of adult cytogenic niches are the numbers of available NSCs and their differentiation potential. In the absence of an absolute marker for adult NSCs, it is very difficult to estimate their numbers in the adult brain. Three such attempts have been published, all focusing in the SEZ and producing a range of values. Two of the studies are more comparable because they are based on the analysis of the regenerating niche, i.e. the NSC-driven repopulation of the niche after the exogenously induced ablation of their downstream daughter cells (transit amplifying progenitors and committed progenitors are actively dividing cells, thus, very sensitive to cytostatic drugs such as AraC or to tritiated thymidine). According to these studies approximately 600 (Morshead et al. 1998), or 300 (Kazanis and ffrench-Constant 2012) NSCs reside within each SEZ. The third study (Golmohammadi et al. 2008) used a combination of in vitro colony-formation assays and in vivo labelling-retention experiments and estimated a much lower number of potential NSCs per SEZ (approximately 50), that is closer to the number of active NSCs at any random time estimated by Kazanis and ffrench-Constant (2012). Interestingly, when the mouse and rat SEZs were compared, the latter was found to contain approximately three times more NSCs, although its volume was five times bigger (Kazanis and ffrench-Constant 2012). Moreover, it was shown that the size of the neurogenic area was determined by the scale of the whole brain, while the number of NSCs was limited by the number of ependymal cells. This suggests that during evolution the enlargement of the brain (leading to a higher demand for cell-replacement events in the case of degeneration) is not isometrically followed by the enlargement of the NSC pool. This could potentially underline the emergence of the gap zone in the even larger primate brain although more comparative analyses have to be performed in order to test this hypothesis. Notably, both adult brain cytogenic niches are not populated by NSCs in their full (volumetric) capacity. Detailed analyses of the SEZ have revealed areas of high and low density of astrocytic endfeet projecting in the ventricle (interpreted as areas of high and low abundance of NSCs, respectively) (Mirzadeh et al. 2008) and similar results have been obtained by investigating the presence of NSCs in serial coronal vibratome-cut sections (Golmohammadi et al. 2008). In the rat hippocampus, quiescent domains have also been identified within the SGZ of the upper blade of the dorsal dentate gyrus (Gil-Mohapel et al. 2010). Nevertheless, the absolute number of NSCs is only one of the parameters potentially affecting the cytogenic capacity of adult brain niches. Another parameter is the ability of these NSCs to become activated when necessary. Again by investigating the regenerating SEZ, it was estimated that only the mitotic activation of almost the whole NSC population can explain the fast repopulation of the transit amplifying progenitor pool (Kazanis et al. 2007), but only limited data exist directly showing activation of NSCs after degeneration (Zhang et al. 2004).

Adult NSCs retain the cardinal property of all stem cells that is the capacity for inexhaustible self-renewal for the duration of the life span of the organism (Ahlenius et al. 2009). They also retain the multipotentiality of embryonic NSCs, as indicated by their ability to generate—in vivo and in vitro—the three main CNS cell types: neurons, astrocytes and oligodendrocytes (Suh et al. 2007; Jessberger et al. 2008; Scott et al. 2010) (Fig. 5). Amazingly, isolated adult NSCs not only can act as embryonic NSCs when re-introduced in a host neural tube (Neumeister et al. 2009) but can also contribute to all germ layers in chimeric chick and mouse embryos (Clarke et al. 2000). Additional experimental studies showed that SEZ-derived NSCs can differentiate into cells of the hematopoietic lineage when transplanted in the bone marrow of irradiated mice (Bjornson et al. 1999) as well as into muscle cells both in vitro and in vivo (Galli et al. 2000; Rietze et al. 2001). The differentiation potential of SGZ-derived NSCs has not been directly explored, although transplantation experiments have demonstrated that SGZ progenitors can behave similar to SEZ progenitors when grafted in this neurogenic system, suggesting that NSCs of the two distinct niches might be fundamentally analogous (Suhonen et al. 1996).

The differentiation repertoire of adult NSCs is normally restricted, although it is altered in abnormal conditions (such as after injury, or genetic and chemical manipulation) as will be discussed later in this review. NSCs of the SEZ give rise mainly to GABAergic periglomerular and granule cell interneurons (Doetsch et al. 2002) and possibly interneurons of the external plexiform layer (Yang 2008) and glutamatergic juxtaglomerular neurons (Colak et al. 2008). The limited plasticity of SEZ NSCs is even more dramatically revealed by the fact that they generate only specific subpopulations of periglomerular cells: those expressing calretinin and tyrosine hydroxylase (Peretto et al. 2004). The other major cell types homeostatically generated by SEZ progenitors are oligodendrocyte precursors and oligodendrocytes (Jackson et al. 2006; Menn et al. 2006; Etxeberria et al. 2010; Jablonska et al. 2010) that migrate short distances and then become stationary within the corpus callosum. SEZ progenitors have also been reported to generate astrocytes (Chmielnicki et al. 2004). However, most of the published cell fate experiments have not identified SEZ-derived astrocytes outside the niche (Merkle et al. 2007; Havrda et al. 2008; Scott et al. 2010); thus, the level of SEZ astrogliogenesis remains controversial. No data exist regarding the generation of microglial cells from adult NSCs. The plasticity potential of SGZ progenitors is also normally limited; they generate almost exclusively granule neurons, very low numbers of astrocytes (less than 10 % of generated cells) (Suh et al. 2007) and extremely low numbers of early oligodendrocyte progenitors (approximately 3 %) that do not mature efficiently (Jessberger et al. 2008).

2.2.3 Neurogenesis in Homeostasis and the Effects of Ageing

In rodents, the immature neurons generated in the SEZ migrate a long distance in order to reach the olfactory bulb (Lois and Alvarez-Buylla 1994) where they differentiate mainly in interneurons (Fig. 2). Augmenting evidence indicates that the addition of these newborn neurons is essential for odour recognition (Doetsch et al. 2002; Breton-Provencher et al. 2009; Mouret et al. 2009). What still remains unclear is whether these cells replace old neurons (having been generated during embryonic development) or only other neurons previously born in the SEZ (Lemasson et al. 2005; Ninkovic et al. 2007). To have a sense of magnitude, approximately 170,000 cells arrive in the OB from the SEZ in a period of 10 days and 40% of these survive for more than 3 months (Winner et al. 2002). Recent experimental work has shown that the newly incorporated cells behave differently to the old neurons, by for example being more sensitive to plasticity with the expression of long-term potentiation (LTP) (Nissant et al. 2009) and are necessary for short-term olfactory memory (Breton-Provencher et al. 2009). Notably, proper olfactory functionality depends not only to the arrival and integration of new cells, but also to the efficient removal of older cells, as it is disturbed upon inhibition of cell death (Mouret et al. 2009). The NSCs that reside in the SGZ generate new granule neurons; thus, progenitors migrate only short distances (Fig. 2) from the niche until their final destination. As in the olfactory bulb, newly born neurons behave differently from already established “old” cells. Importantly, it takes a few weeks for SGZ-generated neurons to mature; initially, they are not responsive to neuronal activity (Snyder et al. 2009) and gradually start to receive only GABAergic inputs from local interneurons (Esposito et al. 2005; Ge et al. 2006). Subsequently—similar to what happens during development—this GABAergic input becomes inhibitory and glutamatergic input starts to appear. Although these adult-born neurons gradually become morphologically identical to older cells, they exhibit lower threshold for the induction of LTP, thus are more plastic (Ge et al. 2006, 2007) and seem to be more responsive to stimuli generated by animal’s experiences (Deng et al. 2010). Many computational models have indicated that the integration of new neurons is compatible—and even necessary—for the efficient functioning of the dentate gurus and the hippocampus (Kempermann et al. 2004); however, the issue of whether new neurons replace older ones (Becker 2005) or are added to the network (Aimone et al. 2009; Weisz and Argibay 2009) remains unresolved. A recent quantitative analysis of post-natal neurogenesis in the macaque monkey dentate gyrus revealed that approximately 40% of granule neurons in a mature adult macaque are born post natally, with the 25% within the first 3 months after birth. In the same study, it was found that the size of the dentate gyrus keeps increasing even in mature animals (Jabes et al. 2010). Finally, two interesting aspects of the SGZ are that its activity is affected by gender, with females having higher basal levels of neurogenesis and showing fluctuations depending on the reproductive state (Galea and McEwen 1999; Westenbroek et al. 2004; Barker and Galea 2008) and by corticoid rhythms (Pinnock et al. 2007; Pinnock and Herbert 2008).

Homeostatic neurogenesis is significantly altered in the aged brain. In terms of structure, the ventral domains of the SEZ are gradually disappearing with ageing, leading to a smaller niche and lower numbers of generated cells (Luo et al. 2006; Blackmore et al. 2009). Similarly, the volume of the human hippocampus is reduced in the elderly (Small et al. 2002) and levels of neurogenesis in the SGZ are markedly decreased in the aged rodent and human brain (Heine et al. 2004; Lucassen et al. 2010). The decline in normal neurogenesis has been correlated with deficits in fine olfactory discrimination (Enwere et al. 2004) and cognitive deficits (Drapeau et al. 2003; Bizon et al. 2004). Very little is known about the mechanisms that underline this age-related decline in neurogenesis. It could be caused by a gradual exhaustion of the intrinsic self-renewal potential of NSCs (Amrein et al. 2011), to aberrant NSC maintenance signalling (affecting molecules such as leukaemia-inhibitory factor, wnt and notch signalling) (Lie et al. 2005; Bauer and Patterson 2006; Ferron et al. 2011), or to defects in cell cycle regulation (Kippin et al. 2005; Molofsky et al. 2006). Surprisingly, recent experimental work has revealed that NSCs within the aged SEZ retain their cardinal properties and behave similar to young NSCs in vitro (Ahlenius et al. 2009), that exercise can partially reverse the decrease in neurogenesis (van Praag et al. 2005; Blackmore et al. 2009) and that exposure to a young milieu (such as in parabiosis experiments) can rejuvenate adult progenitors either located within cytogenic niches (Villeda et al. 2011), or in the parenchyma (Ruckh et al. 2012).

2.2.4 Plasticity of Cytogenic Niches: Exploring the Limits

Under abnormal conditions, such as in response to degeneration or to external stimuli (as during episodes of stress or exercise), cell production in the SEZ and the SGZ can be significantly altered. This is manifested as changes in the numbers of cells generated and in the balance between neurogenesis and gliogenesis. The plasticity of these systems has also been tested—probably to the extreme—by exogenous manipulations, such as genetic interference or administration of growth factors and morphogens. Intracerebroventricular (i.cv.) infusion of epidermal growth factor increases astrogliogenesis not only in the SEZ and the adjacent striatum (Doetsch et al. 2002), but also within the olfactory bulb (Kuhn et al. 1997), while i.cv. infusion of fibroblast growth factor-2 enhances neurogenesis. Both growth factors lead to significant increases in the size of the SEZ (Kuhn et al. 1997) an effect also elicited by i.cv. infusions of vessel endothelial growth factor (Jin et al. 2002). Pigment epithelium-derived growth factor and leukaemia inhibitory factor induce the self-renewal activity of NSCs (Bauer and Patterson 2006; Ramirez-Castillejo et al. 2006) while treatment with platelet-derived growth factor leads in increased oligodendrogenesis (Jackson et al. 2006). Interesting, albeit in many cases conflicting, results have also been generated by manipulating availability of brain-derived neurotrophic factor in the SEZ (reviewed in (Bath et al. 2012) and nitric oxide can be acting both to enhance or restrict neurogenesis depending on the context (reviewed in (Estrada and Murillo-Carretero 2005). The balance between neuro- and gliogenesis can also be altered by interfering with the molecular machinery of progenitors. Overexpression of transcription factors Pax6 and Olig2 enhances the generation of neurons and oligodendrocytes, respectively (Hack et al. 2005). Disruption of BMP signalling favours oligodendroglial over neuronal fate (Colak et al. 2008), while the transcription factor Sox9 was recently shown to promote self-renewal of NSCs and to regulate the balance in the generation of divergent cell fates (Scott et al. 2010). Moreover, disruption of microRNAs and of epigenetic modifications also result in changes in the level of neurogenesis (Liu et al. 2010; Szulwach et al. 2010).

In the SGZ neurogenic niche, even though the primary output is neurogenic, gliogenesis can be significantly enhanced by overexpressing the transcription factor Mash1 (Asc1) (Jessberger et al. 2008) and astrogliogenesis by knocking out reelin (Zhao et al. 2007). Infusion (i.cv.) of either fibroblast or epidermal growth factors does not induce proliferation in the hippocampus, but the latter results in a bias in differentiation in favour of gliogenesis (Kuhn et al. 1997). On the other hand, i.cv. administration of vessel endothelial growth factor leads to strong increase in proliferative activity within the SGZ (Jin et al. 2002).

2.2.5 Plasticity of Cytogenic Niches: Disease

Importantly, degenerative phenomena in the brain have been shown to induce the generation of neuronal subtypes not normally produced by SEZ progenitors, such as striatal spiny interneurons ectopically found in the striatum after stroke (Chmielnicki et al. 2004) and glutamatergic neurons in the injured cortex (Colak et al. 2008). In addition, astrogliogenesis has been reported to be enhanced after stroke (Li et al. 2010) and recent experimental work showed that migrating neuroblasts are diverted toward oligodendrogenic fate in areas of demyelination (Jablonska et al. 2010), thus widening the plasticity potential of SEZ progenitors to cells having exited the microenvironment of the niche that were previously thought to be almost irreversibly committed (Hack et al. 2005) (Figs. 4, 5). Another novel and surprising potential of NSCs residing in the SEZ that was recently demonstrated is the generation of ependymal cells either after their controlled chemotoxic ablation, or after their ageing-related destruction (Luo et al. 2008) (Fig. 5). This property of adult NSCs becomes even more interesting in light of the evidence that after ischaemic injury, ependymal cells behave as progenitors; they migrate out of the niche and into the penumbra where they generate neurons and glia (Fushiki et al. 2003), a property that has also been attributed to ependymal cells lining the wall of the central canal in the spinal cord (Barnabe-Heider et al. 2010).

Fig. 4
figure 4

Activation of the SEZ after degeneration. (Panel a) Microphotographs of adult rat brain tissue immunostained for ED1 (in red, marking activated macrophages) and Dcx (in green, marking newborn neurons). The tissue is taken from a rat subjected to a focal ischaemic insult (unilateral middle cerebral artery occlusion for 1 h). Note that neurons generated in the SEZ form chains (white arrows) with a direction towards the core of the lesion (asterisk) which is rich in activated macrophages and immature neurons. Yellow arrows indicate the possible route of SEZ-generated cells towards the lesion. (Panels b, c) Microphotographs of adult mouse brain tissue immunostained for Dcx (in red, marking neuronal progenitors). Note that although the normal migratory route of SEZ-born progenitors is the rostral migratory stream (rms/indicated with white arrows in c), which drives them to the olfactory bulb, after a focal demyelinating insult (induced by 1 % lysolecithin, the stars mark the site of demyelination) progenitors are diverted towards the lesion (in b). (cc corpus callosum, LV lateral ventricle, str striatum)

Full size image
Fig. 5
figure 5

Differentiation potential of NSCs and of their progeny. (Top panel) A collage of microphotographs illustarting the main cellular elements of the cytogenic niche of the SEZ. Ependymal cells (E, blue) form the wall of the ventricle (which is at the bottom of the image), adjacent are found the astrocyte-like NSCs (white), the transit amplifying progenitors (T, green), the neuroblasts (NB, red) and multiple blood vessels (BV, yellow). (Lower panel) Illustration depicting the known differentiation potential of the different cells of the adult NSC lineage (following the colours of the top panel). Note that the only cell type not shown in the top panel is the SEZ-generated oligodendrocyte progenitor (shown as a brown circle in the lower panel) that has not been clearly identified so far. The differentiation potential of progenitors follows the shape of a pyramid, with NSCs being at the base (thus exhibiting the widest differentiation potential). The red arrows indicate the normal routes of generation of the main three cell types of the central nervous system (neurons, astrocytes, oligodendrocytes) from adult NSCs. Interrupted lines indicate differentiation processes that occur only after degeneration

Full size image

Experimental studies have shown that proliferation is significantly increased in the SEZ in response to traumatic brain injury (Gotts and Chesselet 2005; Ramaswamy et al. 2005) and focal ischaemic lesions that model stroke in humans (Li et al. 2001, 2010; Zhang et al. 2004) with many of these newly generated cells migrating towards the infracted areas (Arvidsson et al. 2002; Yamashita et al. 2006; Thored et al. 2007; Pluchino et al. 2008; Jin et al. 2010; Li et al. 2010). Stereological analysis of the rat SEZ after a focal cortical injury revealed that cell numbers in the niche increased from a total of 300,000 cells to a total of 450,000 cells in a 2 days time frame (Gotts and Chesselet 2005). Enhanced proliferation in the SEZ has also been observed in patients suffering from epileptic seizures (Grote and Hannan 2007) and multiple sclerosis (Picard-Riera et al. 2002; Nait-Oumesmar et al. 2007). Neurogenesis is increased in human cases and animal models of Huntington’s disease (reviewed in (Curtis et al. 2011) while it is decreased in patients and animal models of Alzheimer’s and Parkinson’s disease (Hoglinger et al. 2004; Elder et al. 2006; Ziabreva et al. 2006; Curtis et al. 2007). The effects of exercise or environmental enrichment in SEZ activity are not still clear (Komitova et al. 2005a, b; Blackmore et al. 2009) but these external conditions seem to influence its response to ischaemia (Komitova et al. 2005a, b).

Status epilepticus strongly increases SGZ proliferation in short term, while leading to decreased activity long term (reviewed by (Parent et al. 2007) and ischaemia enhances proliferation but only of neurogenic progenitors (Tureyen et al. 2004). Importantly, the plasticity of the SGZ neurogenic system is demonstrated either as changes in proliferation, or in cell survival. Traumatic injury and enhanced physical activity significantly induce the proliferation of progenitors (van Praag et al. 2005; Urbach et al. 2008), but hippocampus-dependent learning promotes the survival of newly born cells (Leuner et al. 2004; Epp et al. 2007). A critical finding stemming from experimental work in animals was that antidepressants increase neurogenesis in the SGZ and that by either inhibiting this effect, or by depleting neurogenesis, the efficacy of the drugs is markedly decreased (Santarelli et al. 2003). Nevertheless, a recent study failed to reveal any effect of antidepressants in SGZ neurogenesis in elderly human patients (Lucassen et al. 2010). This observation highlights the difficulty in extrapolating experimental results to human conditions; difficulty that might be even bigger in the case of SGZ neurogenesis due to the effects of the endocrine system, as reflected in the role of the hypothalamic–pituitary–adrenal axis (Snyder et al. 2011) and of the gender (Barker and Galea 2008). Importantly, augmenting evidence indicates that the plasticity of hippocampal neurogenesis is tightly regulated by epigenetic modifications (Zhao et al. 2003; Parent et al. 2007; Ma et al. 2009).

3 Conclusions: Requirements—Limitations—Opportunities

Experimental animal work and descriptive studies in the human brain performed during the last two decades have clearly demonstrated that persistent neurogenesis in the adult brain is essential for normal functions, such as olfactory discrimination, learning and memory. What is still largely unknown is whether endogenous neurogenesis is or can be an important contributor to any therapeutic strategies regarding mental and cognitive disorders as well as to the response of the brain to degeneration. The fact is that adult neurogenesis in mammals is restrained within specific microenvironments and furnishes new neurons only in very specific target areas and neuronal networks. It also seems that numbers of surviving NSCs are significantly reduced with ageing, resulting in depleted cell reserves at the time when they are wanted the most. Moreover, as the brain became bigger and more complicated, during evolution, comparatively less NSCs survived in adulthood. On the other hand, recent analysis revealed that the capacity of adult and even aged NSCs to generate new cells can be markedly restored experimentally (in rejuvenation experiments, or with exercise and administration of growth factors). Even more, the observation that the majority of adult-born neuronal progenitors and neurons normally die before reaching their target and before maturation indicates that with the appropriate treatments, the overall output of adult cytogenic niches can be significantly improved. In the case of cognitive and mental disorders, small level manipulations might prove to be highly efficient since it would not be necessary to exogenously divert migratory routes and to re-specify commitment. In the case of tissue degeneration, it will be necessary to recruit small numbers of progenitors in novel (for them) directions (such as towards the substantia nigra in patients with Parkinson’s), or to induce both higher levels of cell generation and a redirection of cells (such as after stroke, or in Huntington’s disease) (Batista et al. 2006). Again, in order to have a measure of requirements and capacity, the loss of cortical neurons in a mouse model of Alzheimer’s disease has been estimated at approximately 50,000 cells (Lemmens et al. 2011), while the enlargement of the SEZ cell population in response to stroke has been documented to reach even 150,000 cells (Gotts and Chesselet 2005). Nevertheless, in the end it might be proven that in certain degenerative conditions the most important contribution of adult brain cytogenic niches is not neuronal cell replacement, but rather the creation of a neuroprotective environment (Jin et al. 2010) possibly via interactions with the immune system (Cusimano et al. 2012).