*P. J. Lucassen, Centre for Neuroscience, Swammerdam Institute for Life Sciences, University of Amsterdam, PO Box 94084, 1090 GB Amsterdam, the Netherlands. E-mail: firstname.lastname@example.org
This review addresses the role of adult hippocampal neurogenesis and stem cells in some of the most common neurodegenerative disorders and their related animal models. We discuss recent literature in relation to Alzheimer’s disease and dementia, Parkinson’s disease, Huntington’s disease, amyotrophic lateral sclerosis, alcoholism, ischemia, epilepsy and major depression.
With the advanced aging of the ‘baby boom’ generation in many Western countries, age-related neurodegenerative disorders like dementia and Parkinson’s disease (PD) are becoming a major health care problem. Approximately 10% of adults above 65 years and 47% of adults above 85 years suffer from dementia, 1–3% of the population above 60 years currently suffers from PD, while (partly) genetically determined neurodegenerative disorders like Huntington’s chorea or amyotrophic lateral sclerosis (ALS) affect a considerable number of subjects (Dorsey et al. 2007).
Recent developments, particularly in the field of adult neurogenesis, have raised the issue as to whether or not damaged or lost neurons could possibly be replaced by newly generated ones either using transplanted neurons grown from exogenous stem cells or through the recruitment of endogenous stem cells and whether this might help to restore the associated behavioral deficits.
Alternatively, it has been suggested that if such responses fail, they may induce cell cycle arrest or abortive exit through apoptosis and could even be involved in disease etiology. Therefore, it is of particular relevance to learn whether or not animal models for neurodegenerative disorders would show regenerative responses following the onset of the pathological changes. It is also unclear if damaged brain regions can incorporate newly generated neurons into existing circuits at all.
Hence, the aim of this article was to address these ‘cause-and-effect’ issues while specifically focusing on the functional relevance of adult hippocampal neurogenesis and stem cells for some of the most common neurodegenerative disorders and their related animal models. We will limit our review to dementia, PD, Huntington’s disease (HD), ALS, alcoholism, ischemia, epilepsy and, given the frequent comorbidity, also to major depression.
Adult neurogenesis in the hippocampus, SVZ and other brain regions
Although neurogenesis occurs throughout the brain during development, the adult brain only holds two true neurogenic niches where neurogenesis occurs endogenously, i.e. the subgranular zone (SGZ) of the hippocampal dentate gyrus (DG) and the subventricular zone (SVZ). In the former, progenitor cells undergo extensive proliferation before they migrate into the granular cell layer (GCL). In young rodents, approximately 9000 new cells are born per day per hippocampus, but about 50% of them die within the first few weeks (Cameron & McKay 2001; Dayer et al. 2003). Significant proportions of the remaining newborn cells eventually differentiate into fully functional and mature hippocampal neurons.
Besides the DG, neurogenesis occurs in the SVZ of the lateral ventricle. Progenitor cells migrate through the rostral migratory stream (RMS) into the olfactory bulb (OB) where they differentiate into interneurons involved in olfactory discrimination learning (Alonso et al. 2006; Curtis et al. 2007; Gheusi et al. 2000).
Regulation of adult neurogenesis occurs through a wide array of hormones, intrinsic growth factors and environmental conditions. Environmental factors such as enriched environmental housing, learning experiences or physical exercise stimulate neurogenesis, whereas stress or glucocorticoid overexposure potently inhibits neurogenesis. It is important to note that different regulatory stimuli can affect different stages of the neurogenic process differently (Kempermann et al. 2004). Which phase is detected is further influenced by the experimental design, the concentration and timing of 5-bromo-2′-deoxyuridine (BrdU) injections and the time until killing (Lucassen et al. 2008).
Consistent with the functional incorporation of adult-generated cells into the strategically located hippocampal circuit of the DG (van Praag et al. 2002) and with initial observations that hippocampal learning increases neurogenesis (Gould et al. 1999a), many studies have found adult hippocampal neurogenesis to be related to cognitive function and to adaptations in hippocampal functioning (Aimone et al. 2006; Leuner et al. 2006; Saxe et al. 2006). Conversely, pathological alterations within the trisynaptic hippocampal circuit may stimulate neurogenesis after acute insults to the hippocampus. Some of the regulatory factors involved could be specific growth factors that are upregulated after ischemic damage (Kuhn et al. 1997). Although it is not known whether neurogenesis is also upregulated after chronic lesions or during ‘slow’ neurodegenerative processes, it has been speculated that either aberrant or reduced neurogenetic responses are involved in the cellular pathology of chronic neurodegenerative disorders like dementia (Herrup & Yang 2007; Kuhn et al. 2007).
Dementia is a progressive neurodegenerative condition characterized by multiple cognitive deficits and generally associated with progressive memory loss or amnesia. The most well-known and prevalent form of dementia is AD; however, other types of dementia include dementia with Lewy bodies, vascular dementia and frontotemporal dementia (FTD). Characteristic pathologies in the AD brain include senile plaques composed of extracellular deposits of beta-amyloid (Aβ), intracellular neurofibrillary tangles composed of hyperphosphorylated microtubule-associated protein tau as well as dystrophic neurites, i.e. diminished synaptic densities, that together are thought to underlie an overall loss of neuronal function (Boeve 2006; Ritchie & Lovestone 2002).
According to the amyloid hypothesis, accumulation of Aβ fragments 1–40 and 1–42 is primarily responsible for AD pathology. An imbalance of Aβ production and Aβ clearance would enable neurofibrillary tangle formation resulting in cognitive impairment. Sporadic AD is the most prevalent type of AD but early onset (familial) forms of AD (FAD) exist as well. Familial AD is associated with specific mutations in genes including amyloid precursor protein (APP), presenilin-1 (PS-1) and presenilin-2 (PS-2). Although intraneuronal Aβ is also important (Wirths et al. 2004, 2007), mutant APP, PS-1 and PS-2 have been implicated in the degradation of APP into fibrillogenic Aβ plaques that can accumulate near the neural precursor cells (NPCs) in the hippocampal DG (Buxbaum et al. 1998; Morys et al. 1994).
Protein tau is a microtubule-associated protein involved in tangle pathology. Hyperphosphorylated tau is present in neurofibrillary tangles, and the extent of tangle pathology generally correlates well with the memory impairment and cognitive decline. As phosphorylation of Tau affects cytoskeletal stability, Tau has been implicated in cellular plasticity, migration, regulation of the cell cycle and neuronal differentiation and is as such also of interest regarding neurogenesis (Boekhoorn et al. 2006b; Sennvik et al. 2007).
In view of its limited availability, the considerable variation in medication, disease duration and post-mortem delay, and its end-stage nature, only correlative questions can be addressed in human post-mortem brain tissue. To better understand disease progression and the underlying mechanism, various transgenic mouse lines have been developed that overexpress AD-related proteins like APP, PS or tau (for reviews, see Gotz et al. 2004; Muyllaert et al. 2006; Spires & Hyman 2005) that have led to modifications of the original Aβ hypothesis (Schmitz et al. 2004; Wirths et al. 2004). Although aspects of redundancy and indirect effects are important putative confounders, various models accurately recapitulate aspects of dementia that allow to address cause-and-effect issues, e.g. by monitoring the temporal onset of changes in neurogenesis in relation to the onset of neuropathology using timed BrdU injections and stereological quantification, impossible approaches in human brain.
There is a remarkable discrepancy between the deficits in morphological markers of synaptic integrity, long-term potentiation (LTP) and behavior that occur in these mice models, even at young ages. Many APP transgenic mouse lines display cognitive deficits and phenotypic traits early in life that are temporally dissociated from the formation of amyloid plaques occurring later in life. For example, in 3-month-old PDAPP mice, a 12% overall reduction in total hippocampal volume was found. Subsequent stereological analysis of the hippocampal subfields localized the volumetric reduction (28%) mainly to the molecular layer and the GCL of the DG. Also, in many models, the rise of insoluble Aβ 42 levels and amyloid plaque deposition is not detectable until middle and old age, i.e. 12–18 months (Jacobsen et al. 2006; Moechars et al. 1999; Redwine et al. 2003).
This supports the notion that cognitive impairment and functional deficits may appear well before classical neuropathological hallmarks such as rising amyloid levels or plaque and tangle deposition. The selectivity of the hippocampus, particularly the DG, in these models is poorly understood but may relate to the presence of stem cells that continue to generate new functional neurons in adulthood in this region. So far, there is not yet a clear-cut consensus about whether the neurogenic response may contribute to the cause or follow the onset of AD pathology.
Hippocampal neurogenesis in AD
5-Bromo-2′-deoxyuridine cannot be used in humans, and only a few studies have thus far focused on neurogenesis in human brain tissue obtained from AD patients. Those that did reported changes mainly in the expression of immature neuronal markers. One report has shown increases in various immature neuronal markers in a cohort of senile AD cases (Jin et al. 2004). In a younger cohort of presenile patients with a faster and more severe disease course, these results could not be replicated (Boekhoorn et al. 2006a). Although a significant increase in the number of Ki-67-positive proliferating cells was found, consistent with earlier studies (Busser et al. 1998; Smith & Lippa 1995), quantification showed that this did not reflect neurogenesis as the increase was mostly because of nonneuronal components such as glia and the vasculature. In addition, methodological considerations regarding patient age, post-mortem delay and fixation time may have been responsible for the discrepancy between these studies.
One example concerns doublecortin (DCX)-expressing young neurons: DCX accurately reflects adult neurogenesis in rodents (Brown et al. 2003; Couillard-Despres et al. 2005; Oomen et al. 2007; Rao & Shetty 2004). Unlike BrdU, detection of DCX does not require injections into live subjects, thus making DCX a promising putative marker for neurogenesis in the post-mortem brain. Quantification of DCX in the Jin et al. (2004) paper was, however, based solely on Western blot analyses of a small selection of their patients, and no anatomical quantification was available for the DCX-positive cells. Furthermore, DCX, like many other microtubule-associated proteins (Swaab & Uylings 1988), was found to be very sensitive to degradation during post-mortem delay (Boekhoorn et al. 2006a). Additional studies indicate that DCX may also be expressed by astrocytes under pathological conditions (Verwer et al. in press). As such, results from DCX studies require careful interpretation when used for neurogenesis analysis in humans.
Changes in neurogenesis in Alzheimer mouse models
The hippocampus is important for learning and memory and is also one of the first areas affected in AD (and secondary to frontal cortical loss in FTD). In view of the early vulnerability of the DG, it is important to clarify whether or not changes in adult neurogenesis are somehow involved in AD pathology in a spatiotemporal manner. Mice with either mutations in APP, PS-1 or combinations of the two typically have age-related disease progression, Aβ plaque distribution and cognitive deficits similar to the human condition. In general, most studies using PS-1 and APP mice have shown a decrease, rather than an increase (Jin et al. 2004), in hippocampal neurogenesis (Chevallier et al. 2005; Donovan et al. 2006; Verret et al. 2007; Wen et al. 2004a,b; Zhang et al. 2007), already at young ages.
In a PS-1 mouse model (A246E), Chevallier et al. (2005) found a transient increase in BrdU-labeled NPCs in the DG. Four weeks after BrdU injection, however, there was no difference in NPC number between the PS-1 mice and the wild-type (WT) mice, which implies increased cell death of the progenitor cells in the AD mice. As such, the authors postulated that even if there is an initial signal for increased cell birth in the AD mice, other compensatory mechanisms such as growth factors may control the long-term survival of new neurons. These results were complementary to an earlier study by Wen et al. (2004b) who discovered a decrease in adult neurogenesis in the DG of the PS-1 P117L mouse model despite enriched environmental housing conditions, a paradigm that has been repeatedly shown to increase neurogenesis in adult rodents.
Conversely, a recent study on cognitive and physical activity in an APP-23 model (Wolf et al. 2006) showed that environmental enrichment does increase neurogenesis in symptomatic APP-23 mice, in addition to showing improvements in water maze performance and increased levels of developmental growth factors. However, mice that received no enrichment or only had unrestricted access to a running wheel had no improvement in neurogenesis and a downregulation of hippocampal growth factors. These results are not necessarily in opposition to those of Wen et al. (2004b) as baseline neurogenesis was still decreased in all the AD models, but these disparities do highlight the fact that mice of different genotypes can have dissimilar cognitive abilities and neurogenic responses to environmental stimuli (Kuhn et al. 2007).
Another example of inconsistency in the literature is the discrepancy between the results of Donovan et al. (2006) and Jin et al. (2004) in two different APP models. Donovan et al. (2006) reported an age-dependent decrease in SGZ neurogenesis and apoptosis in PDAPP (V7171F) mice, while they also found a rare yet apparent increase in proliferation of ectopic immature neurons in the outer GCL of the DG. Nevertheless, despite this surprising discovery, their results showed an overall decrease in neurogenesis. On the other hand, Jin et al. (2004) found an increase in neurogenesis in APPSw,Ind mice both presymptomatically and after the occurrence of many amyloid plaques.
One of the newest studies on APP/PS-1 mice by Verret et al. (2007) took a different approach to studying AD lesions by comparing mice only expressing the APPSw mutation (which are known to develop late-onset Aβ pathology) with mice carrying both the APPSw and PS-1 (PS1dE9) mutations, which hastened the pace of Aβ deposition. This allowed for a direct comparison of the impact of Aβ on neurogenesis in intermediate-aged mice with different plaque loads. Verret et al. found that although early Aβ had no effect on hippocampal neurogenesis, the overall survival of hippocampal NPCs was diminished [similar to the results seen by Chevallier et al. (2005) in the PS-1 model], and those cells that did remain were less likely to develop into new neurons. Therefore, the long-term effect was decreased neurogenesis in the hippocampus, thus repeating the theme that amyloid pathology decreases neurogenesis and is a function of the rate of amyloid deposition. These results further beg the question of whether or not removing the amyloid would increase neurogenesis or at least stop the progressive decline of neurogenesis.
One of the potential ways of removing Aβ would be to recruit the immune system using antibody targeted opsonization and clearance of Aβ through microglial cells. This approach was attempted by Becker et al. (2007) using anti-EFRH immunization. EFRH is a sequence found on Aβ that controls the solubilization and disaggregation of the peptide. Using a standard PDAPP transgenic model, the authors immunized the mice starting at 4 weeks of age. The overall results of this experiment showed that the number of BrdU+ cells in treated animals led to a threefold increase in NPC proliferation vs. control, and neurogenesis itself was inversely correlated with amyloid load and was increased 2.5 times over control, as shown by BrdU+/NeuN+ double staining. These results suggest the possibility that under certain controlled circumstances, anti-Aβ therapy could potentially promote recovery of neurogenesis and/or alleviate the progression of amyloid deposition in AD and other diseases.
One of the clear pitfalls of the models mentioned above, however, is that the artificial insertion of exogenous genes and (sometimes) promoters into the host organism ultimately leads to overexpression of these genes above levels that would normally be seen in the real patient population. In an attempt to overcome the complication of ectopic overexpression of exogenous genes in transgenic mice, Zhang et al. (2007) reports on the creation of a APP/PS-1 double knock-in (KI) mutant mouse with FAD-causing mutations with the intention of studying the integrity of APP, PS-1 and APP/PS-1 on neurogenesis in the adult brain. This study showed clear differences between the double KI mice, PS-1 KI and APP KI, and control animals. The APP/PS-1 double KI mice have a significant and lasting decrease in hippocampal neurogenesis that was not found in either of the single KI models. This was not particularly unexpected, however, as neither of these single KI mutants had the typical amyloid pathology seen in most FAD mouse models. Yet, this may be of particular relevance to interpretation of the previous studies using overexpression systems: without the overexpressed proteins no significant effect on neurogenesis was found. Hence, only the expression of both the mutated form of PS-1 and APP reduced proliferation in the SGZ.
Regarding tangle pathology, considerably fewer mouse models are around and only in a few cases was neurogenesis studied. tau transgenic mice bearing the FTD mutation P301L recapitulate many of the features of FTD including axonopathy and memory impairments that are paralleled by tau hyperphosphorylation at later ages. Memory was tested at a young age, prior to hyperphosphorylation or the onset of axonopathy, and surprisingly, object recognition task performance was improved in P301L mice, which was associated with increased LTP. However, no changes were noted in several neurogenesis parameters or in individual structures of the dendritic tree. Thus, in young mice bearing the P301L tau mutation, hippocampal functioning is improved before the onset of tau phosphorylation. These results show that tau itself plays an important role in normal processes underlying hippocampal memory. Second, hyperphosphorylation of tau must be critical for the cognitive decline in tauopathies like FTD as opposed to the tau mutations themselves (Boekhoorn et al. 2006b).
Tau hyperphosphorylation seems to be a critical event in the progression of disease-related neurodegeneration. Results from earlier work on nonmutant human tau (htau) knockout (KO) mice show that the NFT load is not directly proportional to neurodegeneration; however, NFT aggregation is probably mediating re-entry of the cell into the cell cycle without division, an apoptosis stimulating event. This process occurs after tau hyperphosphorylation in the mice around 9 months of age (Andorfer et al. 2005). Furthermore, data from Tanemura et al. (2006) show that the deposition of tau and its consequent hyperphosphorylation may be mediated by the presence of typical mutations in the PS-1 gene (mPS1 mice), thus potentially linking changes in both APP and tau, i.e. Aβ plaques and NFT formation, to the neurogenerative process as well as potentially mediating an age-dependent and disease burden-related decline in neurogenesis.
Normal tau is also important for neuronal development, particularly in the DG, where the 3R to 4R tau isoform switch occurs around the first two weeks of life. In vitro studies carried out in the absence of tau showed that deletion of the tau gene increases cell birth, decreases differentiation and decreases neuritic outgrowth. All these effects could be reversed by expression of tau-4R, whereas expression of tau-3R failed to affect cell birth or differentiation and could only partially reverse neuritic outgrowth. The relevance of these functions of tau for hippocampal development was studied in a transgenic mouse tau knockout, human 4R tau knock-in (KOKI) mouse model. These mice lack all endogenous mouse tau isoforms; only htau-4R is expressed from the second postnatal week onward at reduced levels in the hippocampus. These mice show a transient increase in cell birth starting at the second postnatal week. At 2 months of age, cell birth reduced to levels comparable to those in wild types. This increased cell birth was reflected by enhanced DG neurogenesis and an eventual increase in the cell number and volume of the adult hippocampus. Furthermore, these changes were paralleled by significantly improved memory function in KOKI mice with respect to the object recognition task; however, the improved memory function was not associated with altered LTP (Sennvik et al. 2007).
In conclusion, with the exception of the conclusions reported by Jin et al. (2004), the common theme between these studies is that basal neurogenesis is most likely decreased in adult FAD mice with amyloid pathology, even across different genotypes. This effect does not seem to be because of impaired cell birth but rather because of decreased survival of maturing neurons. Neurogenesis in tau transgenic mice has so far only been studied in very young animals where so far little or only transient stimulatory effects on neurogenesis were found. Nevertheless, questions about the differences in the age of the animals, the extent of AD pathology, hippocampal subregional specificity and unintended side-effects of the artificial overexpression of exogenous genes must be taken into account when comparing AD-linked effects on neurogenesis in transgenic animals with the human AD population.
PD and neurogenesis
Another major neurodegenerative disorder is PD where the dopaminergic nigrostriatal projection is severely affected. The loss of dopamine causes characteristic motor symptoms like akinesia, rigidity and tremor. These clinical symptoms are late manifestations of the disease as they usually become overt only when over 70% of the dopamine cells are lost. Interestingly, nonmotor symptoms such as olfactory dysfunction (Berendse et al. 2001), impaired spatial memory (Pillon et al. 1997) and depression (Oertel et al. 2001) typically precede the motor symptoms. Although the function of adult neurogenesis and its involvement in the SN is still a matter of debate, it is striking to note that experimental inhibition of neurogenesis impairs olfaction (Enwere et al. 2004), spatial memory (Nilsson et al. 1999) and may be implicated in depression (Czeh & Lucassen 2007; Santarelli et al. 2003). Moreover, the main neurotransmitter involved, dopamine, contributes to brain ontogenesis and adult neurogenesis by regulating NPC proliferation in both hippocampus and SVZ (Borta & Hoglinger 2007; Hoglinger et al. 2004a,b).
Consequently, PD has received much attention with respect to endogenous changes in neurogenesis in various adult brain regions, especially in regard to possible therapeutic strategies using recruitment or transplantation of stem cells aiming to replace lost neurons and overcome the dopamine deficit (Correia et al. 2005). An initial report (Zhao et al. 2003) suggests that dopaminergic neurons, albeit at a low pace, are continuously generated in the adult SN pars reticulata, whereas most degenerating dopamine neurons reside in the adjoining SN pars compacta. However, later studies failed to find any indication for either nigral neurogenesis or migration of new cells emanating from other regions into the nigra (Frielingsdorf et al. 2004). Several methodological issues may have been responsible for these differences (Borta & Hoglinger 2007). Despite the current consensus that it is unlikely that dopaminergic neurons are generated in relevant numbers in the adult SN, it has been shown that the progenitors who reside there have the potential to differentiate into astrocytes, oligodendrocytes and neurons (Lie et al. 2002). Hence, it is important to understand the mechanisms and local environmental conditions regulating neurogenesis in brain regions like the OB and hippocampus prior to designing endogenous or exogenous stem cell therapies.
Although they do not fully recapitulate the pathology and progression seen in PD patients, various animal models have been developed based on specific neurotoxins like methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) and 6-hydroxydopamine (6-OHDA) or on transgenic approaches, e.g. alpha-synuclein transgenic mice (Borta & Hoglinger 2007). While alpha-synuclein impaired neurogenesis (Winner et al. 2004), 6-OHDA lesioning in rats or MPTP lesioning in mice resulted in cell proliferation in the SN without the cells actually differentiating into a dopaminergic phenotype. Interestingly, when these proliferating cells were isolated and implanted in a neurogenic environment, 20% of the cells do differentiate into mature neurons. However, transplantation of the same cells into the SN of healthy rats resulted only in glial differentiation. Thus, local environmental factors appear to control differentiation to distinct neural lineages.
Experimental depletion of dopamine in MPTP-intoxicated mice and 6-OHDA-lesioned rats and mice resulted in a decreased proliferation of progenitors in SVZ and SGZ (Hoglinger et al. 2004), while striatal deafferentiation increased dopaminergic neurogenesis in the OB (Winner et al. 2006; see also Fig. 1). Detailed analysis showed a direct decrease in the proliferation of type C cells in the mouse SVZ, which resulted in reduced migration and neuronal differentiation in the OB. In contrast, levodopa significantly restored the cell proliferation in the SVZ of 6-OHDA-lesioned rats. Endogenous cell proliferation was also impaired in the SVZ, OB and SGZ of PD patients (Hoglinger et al. 2004), once again suggestive of an important role for dopamine in increasing adult neurogenesis. Additionally, factors known to stimulate neurogenesis, e.g. enriched environmental housing, affect cellular plasticity in the adult SN, reduce astrogliosis and improve motor function in the 6-OHDA lesion model of PD (Steiner et al. 2006). This implies that structural and microenvironmental changes are involved in the improved motor behavior.
Interestingly, nigrostriatal degeneration by MPTP or 6-OHDA intoxication in monkeys, rats and mice causes a severalfold increase in tyrosine hydroxylase (TH)-immunoreactive cells in the dorsal striatum, a phenomenon also seen in the same region in PD patients (Borta & Hoglinger 2007). It was not clear whether this compensatory increase in TH-positive cells originates from newly generated precursor cells in the striatum or from cells that have migrated from the SVZ. No colabeling of BrdU-positive and TH-positive cells was observed in adult mice or in aged macaques, thus suggesting that the new TH-positive cells in the striatum of lesioned animals or in PD patients most likely represent mature striatal cells undergoing a phenotypic switch aimed to compensate the dopamine loss. In various lesion models, the number of granular GABAergic neurons decreases, whereas the number of dopaminergic neurons in the glomerular layer of the OB increases. Similarly, in PD patients, the number of dopaminergic cells in the OB is doubled, whereas the number of nestin precursors is reduced (Borta & Hoglinger 2007; Hoglinger et al. 2004 and references therein). Together, these results suggest a shift in the ratio of newly generated interneurons induced by changes in the neurotransmitter dopamine.
HD and neurogenesis
Huntington’s disease is an autosomal dominant neurodegenerative disease caused by an overexpansion of the CAG trinucleotide repeat sequence in the huntingtin gene. Individuals with over 39 copies of this sequence produce a mutated form of the huntingtin protein that accumulates intracellularly throughout the brain and leads to progressive neuronal cell death, particularly in the cortex and striatum. Patients who suffer from HD have problems with inappropriate movement, speech, swallowing and executive functioning, and nearly 25% suffer from dementia or depression in the course of the disease (Spires & Hannan 2005; Walker 2007). Neurogenesis in the context of HD is still an emerging research topic; however, there have been several studies over the past few years that suggest that HD may be tied to changes in NPC proliferation in the DG.
Thus far, two different transgenic mice models have been used to study neurogenesis in HD: R6/1 and R6/2 (Mangiarini et al. 1996). R6/1 and R6/2 mice carry between 112 and 148 copies of the human CAG repeat under an endogenous murine promoter, display a complex phenotype similar to that of the human HD population, acquire intracellular accumulations of mutant huntingtin and have a typical disease onset of either 9–11 or 15–21 weeks for R6/1 and R6/2, respectively. Yet, unlike human HD patients, the R6 lines do not show significant neurodegeneration or hippocampal atrophy, which may be because of the rapid disease progression leading to an early death within several weeks (Lazic et al. 2006; Mangiarini et al. 1996).
Nevertheless, the R6 mouse line is an appropriate model for early stage HD with strong phenotypic evidence for hippocampal involvement. Like the majority of FAD mouse model studies, the HD R6/1 and R6/2 mice both have dramatically decreased proliferation in the DG. In R6/2 mice, Gil et al. (2005) reported a 70% reduction in DG proliferation using several cell cycle and neuronal markers including BrdU, Ki-67 and NeuN. In addition, the R6/1 mice (Lazic et al. 2004) showed a significant decrease in the number of NPCs vs. WT using BrdU and DCX staining. Neither study showed an increase in SVZ proliferation, which suggests that the decreased proliferation in the DG is a specific effect of HD models and may possibly be a common characteristic of murine neurodegenerative disease models (Gil et al. 2005; Lazic et al. 2004, 2006; Phillips et al. 2006). However, the DG-specific reduction in neurogenesis may not necessarily translate directly to the human HD population (similar to human studies of AD) because data from a human HD population showed increased SVZ neurogenesis (Curtis et al. 2003).
Methodological consideration must again be taken into account when comparing data across studies. For example, in the human study, only one marker, proliferating cell nuclear antigen (PCNA), was used for immunohistochemistry. PCNA is a general cell cycle marker, but it is difficult to predict whether, under disease conditions, it is expressed in mature neurons undergoing cell cycle-related changes prior to neurodegeneration or whether it reflects an upregulation of cytogenesis and possibly neurogenesis. Further, differences between the exact number of CAG mutations and pathological changes in the animal vs. human disease should be considered (Curtis et al. 2003; Gil et al. 2005).
In recent studies that have attempted to stimulate neurogenesis, enriched environmental housing increased the number of BrdU+ cells in the DG and improved migration of DCX+ cells in a cohort of 21-week-old R6/2 mice (Lazic et al. 2006). These effects were not significant in a younger group of mice; therefore, the authors speculated that this result was probably because of longer exposure to the enriched environment. Another study using R6/2 mice attempted to clarify the effect of the microenvironment on neurogenesis by examining the effects of neurogenesis in vitro using extracted NPCs (Phillips et al. 2005). Their work failed to show a difference in neurogenesis and cell survival between WT and R6/2 mice, thus lending support to the theory that changes in the hippocampal microenvironment affect neurogenesis and that the NPCs themselves are not intrinsically defective (Phillips et al. 2005).
Finally, in an experiment testing the role of fibroblast growth factor-2 (FGF-2) in HD mice, Jin et al. (2005) discovered that daily injection of FGF-2 into R6/2 mice from 8 weeks until death stimulated neurogenesis in the SVZ of afflicted mice by approximately 150%, but this effect was not seen in the DG. The authors also provided evidence that these new SVZ neurons were being recruited into the cortex and striatum, i.e. areas of neurodegeneration in HD patients, as medium spiny neurons. Also, FGF-2 was neuroprotective, reduced intracellular protein accumulation and extended the life span of the R6/2 mice by nearly 20%. Although the DG NPC proliferation was not enhanced by FGF-2, these results may be relevant to the treatment of HD, especially if biological or pharmacological compounds are to be used to increase neurogenesis in neurodegenerative zones and potentially decrease the disease severity or extend the life span of afflicted patients.
Overall, none of the studies on R6 mice provides evidence for whether or not changes in neurogenesis are a cause or effect of HD; however, they do show a trend for decreased neurogenesis in HD models, particularly in the DG, that can possibly be extrapolated to humans pending further examination and methodological clarifications in human HD tissue. Another suggestion would be to examine neurogenesis in the 140 CAG-repeat KI model (Menalled et al. 2003). So far, these mice have been characterized as having a longer disease course that is more similar to humans as well as having typical huntingtin protein accumulations, cortical neurodegeneration and motor impairments such as reduced gait stride and decreased climbing ability (Dorner et al. 2007; Menalled 2005). Neurogenesis has not been analyzed yet.
Neuronal stem cell populations in ALS
Amyotrophic lateral sclerosis ranks as the third most common neurodegenerative disorder after PD and AD. During the course of the disease, neurodegeneration occurs in motor neurons mainly in the brainstem and spinal cord (Liu & Martin 2006). Liu and Martin (2006) examined the integrity of NPCs throughout the classic neurogenic niches in a transgenic model of ALS, the hemizygous manganese superoxide dismutase (mSOD-1) mouse, whose phenotype resembles human familial ALS and includes motor neuron degeneration throughout the spinal cord. No significant changes were found in the NPCs in any of the neurogenic regions in either the presymptomatic or the symptomatic mice. However, they did find that cells in the SVZ had a somewhat reduced response to mitogens. Regardless of this, the integrity and proliferation of NPCs in the SVZ and DG are essentially intact despite neurodegenerative changes elsewhere in the brain and spinal cord. Thus, they may have the potential for recruitment in therapies attempting to replace lost or damaged motor neurons, as performed in a new study by Corti et al. (2007).
Corti et al. (2007) harvested a subset of neural stem cells (NSCs) double positive for specific antigens LeX and CXCR4 from the SVZ of three different mouse strains, exposed them to growth factors and morphogenic stimuli to induce neuronal differentiation and transplanted them into the spinal cord of mSOD-1 mice. They showed that the NSCs were able to generate partially functional neuron and motor neuron-like cells in the mSOD-1 mice, reduce motor neuron loss and improve the neurological difficulties associated with ALS degeneration, which led to an approximately 3-week longer life span vs. untreated mice.
Generally speaking, endogenous neurogenesis in the DG and SVZ is normal in the ALS mouse model of mSOD-1, and there seems to be the potential to manipulate and transplant the NPCs and NSCs to replace lost motor neurons, especially in spinal cord, which is more accessible for surgery than the brain itself. Even so, there are still serious ethical and technical challenges facing the practical use and recruitment of these cells in treating human ALS patients. For example, obvious complications arise concerning access to the endogenous NPC/NSC populations in the hippocampus and SVZ, directing their neuron-specific differentiation and functional synaptic integration of new neurons within the existing circuitry.
Alcoholism and neurogenesis
Alcoholism starts principally as a behavioral disorder, but chronic alcohol consumption, especially in toxic quantities, is a progressive neurodegenerative disease that can eventually lead to cognitive impairment and is an important cause of amnesia and dementia known as Korsakoff’s syndrome (Aberg et al. 2005; Nixon 2006; Tranel et al. 1994). However, clinical and imaging studies show that some functional recovery may occur with abstinence, suggesting the possible involvement of neurogenesis. Recent research on the effects of ethanol on NPCs has shown that chronic alcohol exposure can inhibit NPC proliferation and decrease dendritic outgrowth of newly born neurons in the hippocampus of young rats (He et al. 2005). Fortunately, it seems that only chronic alcohol ingestion has a permanent effect on hippocampal neurogenesis, although moderate consumption of alcohol may even increase neurogenesis (Aberg et al. 2005). Also, extensive neurodegeneration because of acute exposure of embryonic NPCs to alcohol can be functionally overcome in adulthood. However, recovery was speculated to be because of synaptic reorganization rather than increased neurogenesis (Wozniak et al. 2004). Abstinence from alcohol may partially reverse the effects of hippocampal neurodegeneration and even improve cognitive performance (Nixon 2006; Nixon & Crews 2004; Wozniak et al. 2004). The mechanisms of functional recovery are still unknown, but there is speculation that glial changes and structural plasticity are involved (Nixon 2006).
Neurogenesis outside the neurogenic niche areas
Given the intense public interest in the potential use of NSCs and progenitor cells for the treatment of neurodegenerative diseases, the idea of neurogenesis occurring outside the classic neurogenic niches is appealing. Several years ago, Gould et al. (1999c) published a controversial study claiming to have found new neurons moving into the adult macaque neocortex under normal conditions; similarly, Bernier et al. (2002) claimed to have found neurogenesis in the amygdala, inferior temporal cortex and piriform cortex of both Old World and New World primates. However, the scientific community has been quite critical of these claims as neuronal development in mammals was believed to occur only during specific and tightly regulated stages early in development (Rakic 2002a,b).
In particular, the extent of the number of newly generated cells claimed to have been found in the prefrontal cortex was the most contentious aspect of the report by Gould et al. (1999c; Rakic 2002a,b). Consequently, Kornack and Rakic (2001) attempted to replicate these results, but they were unable to find any cells in the neocortex that were double labeled with both a proliferative and a neuronal marker. However, they did find cells that were single labeled for BrdU, a marker of DNA synthesis, throughout the neocortex. The authors proposed that the labeled cells were probably glial or endothelial cells that are both known to divide in the adult brain throughout life and that methodological problems may be at the core of the problem with the data presented by Gould et al. (1999c). The recent work by Rakic (2002a,b) agrees with his earlier publications advocating the view that cortical neurogenesis does not occur after development in primates under normal conditions (Rakic 1985), as recently confirmed in humans using different methodologies (Bhardwaj et al. 2006).
As with the data on cortical neurogenesis, the possible occurrence of neurogenesis in the adult SN has generated a similar debate (Frielingsdorf et al. 2004; Zhao et al. 2003) (see above). Nevertheless, the possibility that rare and/or slow neurogenesis can take place outside the SVZ and DG in the course of disease or brain injury has not been ruled out, and recent studies suggest that this may be a possibility in ischemia, for example (Jin et al. 2006; Leker et al. 2007).
Neurogenesis and ischemia
Upregulation of neurogenesis after ischemia, more commonly known as stroke, is a well-studied phenomenon (Carmichael 2006; Wiltrout et al. 2007; Zhang et al. 2005). Ischemia is caused by compromised blood flow in the brain, ultimately leading to blood vessel occlusion followed by a massive loss of neurons and glial cells in the infarcted areas. Stroke affects a large number of people every year, and many strokes are fatal. Of the survivors, nearly 60% suffer from a permanent disability (Carmichael 2006). As such, understanding the function of increased neurogenesis after stroke is a worthwhile endeavor for possible future medical treatments as there are currently few options other than post-stroke rehabilitation available to patients.
Taken as a whole, literature shows that there is an increase in the proliferation of NPCs in the SVZ and DG after an ischemic insult as well as induction of NPC migration from the SVZ and RMS into the corpus callosum and striatum near the compromised region(s). Various animal models of experimental stroke have consistently revalidated these findings, but only recently did Jin et al. (2006) report evidence for upregulation of neurogenesis in post-mortem human brain issue using various cell cycle proliferation and neuronal lineage markers. Their experiments show an increase in new cells near blood vessels in the area around the central ischemic lesion, suggesting that new cells may be either migrating into the affected areas from the SVZ, as shown in the animal models, or neurogenesis is being induced through endothelial growth factors in neurogenic vascular niches. Moreover, this study once again highlights the question of whether or not cortical neurogenesis can take place after early development, particularly in response to damage or disease pathology. So far, there has been only limited evidence for proliferation of NPCs in the cortex after lesions or insults, although NPCs from the SVZ or through glial intermediates may be able to undergo long-distance migration along the border of the cortex near the lesion site and possibly even into the cortex itself (Jin et al. 2003; Magavi & Macklis 2002; Magavi et al. 2000, 2005).
Recent investigations into the role of neurogenesis after stroke have focused heavily on the role of growth factors, neurotransmitters and hormones, which have been shown to modulate the proliferation, survival and/or migration of post-stroke neurogenesis (Wiltrout et al. 2007). One frequently studied and important growth factor in post-stroke neurogenesis is FGF-2 that may also play an important role in AD pathology as FGF-2 is elevated in AD and can induce the upregulation of tau in a dose-dependent manner (Chen et al. 2007). Like so many other factors involved in neurogenesis, FGF-2 has a complex role in the development and proliferation of new neurons. For example, Nelson and Svendson (2006) have shown that elevated FGF-2 can increase (but is not essentially required for) neurogenesis at low concentrations, whereas a previous study by Chen et al. (2007) provided evidence to the contrary. They showed that elevated FGF-2 keeps NPCs in an immature proliferative state vs. helping NPCs mature into new neurons; however, the suppression of neurogenesis by FGF-2 could be overcome by optimal concentrations of other key growth factors such as insulin growth factors 1 and 2. Furthermore, new evidence by Zhao et al. (2003) shows that FGF-2 is required for NPC proliferation and neuronal generation in the DG using a FGF receptor KO model, i.e. Fgfr1. Taken as a whole, the functions of FGF-2 in neurogenesis are likely dose dependent and differ based on the cell and culture models used.
To complicate matters further, a new study on post-stroke neurogenesis by Leker et al. (2007) using a model of focal ischemia in rats has provided evidence that FGF-2 can increase neurogenesis for up to 90 days after an ischemic insult. Their data show that post-stroke neurogenesis is a long, ongoing process where newborn cells migrate from the SVZ and eventually accumulate in the cortex and subcortical matter around the infarct, thus providing further evidence for the premise that endogenous ‘local’ cortical neurogenesis is not possible.
Another important growth factor in ischemic neurogenesis is brain-derived neurotrophic factor (BDNF). Schabitz et al. (2007) described the effects of i.v. injection of BDNF into the rat parietal cortex using a photothrombotic stroke protocol. Peripheral BDNF application led to increased neurogenesis in both the hippocampus and the SVZ as well as enhanced SVZ cell migration into the ipsilateral striatum. Most importantly, the rats showed an improvement in their functional sensorimotor skills as analyzed by rotarod testing. Although these improvements may in theory be tied to increased levels of neurogenesis, especially in the hippocampus, a more likely scenario, however, is that improved motor skills are a result of decreased apoptosis thanks to the neuroprotective effects of BDNF. Also worth mentioning are other factors recently noted for their various roles in increased post-stroke neurogenesis, particularly in the SVZ; these factors include estradiol (Suzuki et al. 2007), vascular-associated endothelial growth factor (VEGF) (Suzuki et al. 2007; Wang et al. 2007a,b) and erythropoietin (Tsai et al. 2006), an especially important chemokine released at the site of injury that may help NPCs differentiate into new neurons (Carmichael 2006). Other promising new research being conducted into post-stroke neurogenesis concerns the effects of voluntary exercise (Luo et al. 2007) and vascular influences (Ohab et al. 2006) on NPC migration and differentiation.
Epilepsy and neurogenesis
To some extent, similar to stroke, epilepsy is another example of a neurodegenerative disorder that has a major impact often in the hippocampus where it elicits profound changes in neurogenesis (Fig. 2). Epileptic seizures initially trigger cell death of selective neuronal populations (Gorter et al. 2003), and animal studies indicate that seizures may subsequently stimulate DG neurogenesis (Bengzon et al. 1997; Gray & Sundstrom 1998; Parent et al. 1997, 1998; Scott et al. 1998) as well as angiogenesis (Hellsten et al. 2005). A still unresolved issue is whether this response reflects pathology or is part of an endogenous regenerative response, and if so, whether the contribution of the newly generated cells is sufficient.
In the adult rodent kainate and pilocarpine models of temporal lobe epilepsy, status epilepticus is followed by a latent period of at least several days before severalfold increases in DG cell proliferation occur. Seizures stimulate not only proliferation but also induce dispersion of at least some of the neurogenic cells to ectopic locations. Kainic acid-induced seizures in mice, e.g. did not affect nestin-expressing early precursor cells but stimulated the division of DCX-positive migratory neurons, leading to an ectopic dispersion within the DG GCL, as has also been observed in human hippocampal tissue (Jessberger et al. 2005; Parent et al. 2006a), and aberrant physiological properties (Scharfman et al. 2000). Others found that the functional maturation of new neurons was accelerated under seizure conditions resulting in persistent hyperexcitability (Overstreet-Wadiche et al. 2006). Ectopic precursor cell proliferation following the initial seizure may thus result in a subsequent dispersion of migratory neurons within the adult GCL. Part of these new cells may be recruited from SVZ-derived gliogenesis, and seizures may attract newly generated glia to regions of hippocampal damage (Parent et al. 2006b).
While shortly, e.g. 16 days, after the induction of temporal lobe epilepsy, the injured hippocampus exhibited increased dentate neurogenesis, the chronically epileptic hippocampus, e.g. 5 months after the first seizure, showed severely declined neurogenesis, which was associated with considerable spontaneous recurrent motor seizures (Hattiangady et al. 2004). As a fraction of the newly born neurons become GABAergic interneurons, a decline in neurogenesis may contribute to the increased seizure susceptibility of the DG during chronic epilepsy. Seizure-generated new neurons may thus be involved in the recurrent continuation of seizure activity in rodent models. However, in human temporal lobe epilepsy, increases rather than decreases in neural progenitors have generally been reported (Crespel et al. 2005), raising the issue of which models actually mimic the characteristic pathological features in human temporal lobe epilepsy (Kempermann 2006). Future research will have to address whether or not epilepsy is the cause or consequence of an initial seizure-related disturbance in adult neurogenesis and angiogenesis.
Neurogenesis in relation to hippocampal volume reductions during stress and major depression
Over 90% of the AD patients exhibit depressive symptoms during the course of the disease process (Grossberg 2003). In fact, depressive symptoms can be early manifestations of AD and a considerable overlap exists in their symptoms: loss of interest, difficulty in concentration, decreased energy, etc. One of most well-studied hypotheses in depression focuses on hyperactivity of the hypothalamus–pituitary–adrenal (HPA) axis that co-ordinates the stress response (Lucassen et al. 2001; Swaab et al. 2005). Of note, HPA activation and hippocampal volume changes are also apparent in many other (psychiatric) disorders like schizophrenia, autism and panic disorder as well as in AD, albeit less robustly, and are hence, not specific for AD.
Initial rodent studies reported hippocampal neurodegeneration after chronic stress exposure, suggesting that stress may affect hippocampal viability. Later studies, however, failed to find indications for massive cell loss in animal models for stress or in hippocampal tissue from depressed patients (Fuchs et al. 2004a,b; Lucassen et al. 2001; Muller et al. 2001; Swaab et al. 2005). Preclinical and clinical studies have shown that stress-induced reductions of hippocampal volume are reversible. Hence a possible explanation may be that stress reduces DG turnover and hippocampal neurogenesis rather than inflicting damage (Czeh & Lucassen 2007). Together with data showing that almost all clinically effective antidepressant drugs stimulate neurogenesis and data showing that antidepressive drugs require neurogenesis for their behavioral effects (Santarelli et al. 2003), this premise has given strong support to the concept of failing neurogenesis in stress-related disorders like depression.
Taken together, depression is not a true neurodegenerative disorder despite the frequent comorbidity of depression in PD and AD. Rather, recent studies suggest the involvement of impaired structural plasticity and cellular resilience in depression. Failing neurogenesis as well as other factors (Czeh & Lucassen 2007; Fuchs et al. 2004b; Lucassen et al. 2008) might contribute to hippocampal disorders by compromising the hippocampus’ ability to initiate cellular plasticity and long-term structural adaptations.
Functional consequences: cause or consequence
Given the estimated number of newborn cells per day (approximately 9000 cells in young rats of which only 50% survive the first week) and the fact that less than 75% of these surviving newborn cells will eventually differentiate into a neuronal phenotype, it seems unlikely that acute changes in the newborn cell population will give rise to more than 5% of the granule cells. In the long run, however, it takes several weeks to months before the newborn adult neurons are fully incorporated into the correct network. Reductions in the rate of neurogenesis and/or DG turnover because of neuropathological processes will affect the overall composition, average age and identity of DG hippocampal neurons (or glia). This will have considerable consequences for the connectivity, input and output properties of the entire hippocampal circuit and hence cognition. Other factors such as alterations in the dendritic, axonal and synaptic components will also contribute to the network’s possibilities for synaptic contacts and for further changes in network function and behavior (Joels et al. 2007).
Neurogenesis is a well-defined and well-regulated phenomenon that occurs during early development and in special neurogenic niche areas in the adult. Regulation of neurogenesis involves a wide range of environmental factors, hormones and growth factors, and the intricate nature of the system makes it vulnerable to aging and disease. As such, there are a diverse range of neurodegenerative diseases and conditions that may involve alterations in normal neurogenesis as either a cause, comorbidity, or a consequence of the disease. In particular, AD and HD show an overall trend for decreasing neurogenesis in most of the available animal models, whereas neurogenesis has not been shown to occur in the SN of PD models or outside the normal neurogenic niche areas, except for changes in the SVZ/olfactory system. Furthermore, the only neurodegenerative condition that conclusively shows an increase in neurogenesis is ischemia where neurogenesis is upregulated in both the SVZ and the DG; the difference from HD and AD may be because of the acute nature of ischemia vs. the more chronic nature of the other diseases. Neurogenic changes in epilepsy depend strongly on temporal aspects of the model involved.
In conclusion, the study of neurogenesis in neurodegenerative diseases is a relatively young field. Nevertheless, understanding how normal adult neurogenesis responds to chronic and acute diseases, particularly in regard to whether or not changes in neurogenesis are a cause or consequence of the disease, is important for possible future medical treatments, stem cell therapies and disease management.
Conflicts of interest
This article was presented at a symposium on Alzheimer’s disease – new insights from animal models and molecular pathways, to be translated into human pathology, which took place at the Genes, Brain and Behavior 2007 Society Annual Meeting, May 21–25, 2007, Doorwerth, the Netherlands. The symposium was sponsored by the European Commission [Marie Curie Early Stage Training, MEST-CT-2005-020013 (NEURAD), Alzheimer Ph.D. Graduate School].
The authors declare no conflicts of interest.
We thank the European Union (EU) for financial support. A.T. is supported by a Marie Curie Training grant from the EU (NEURAD consortium). P.J.L. is supported by the Nederlandse HersenStichting. P.J.L. and K.B. are supported by the Internationale Stichting Alzheimer Onderzoek (ISAO).