Untold New Beginnings: Adult Hippocampal Neurogenesis and Alzheimer’s Disease

Regulation of the rate of AHN

In the DG, GSK-3β increases the proliferation of neuron precursors and prevents them from acquiring a neuronal fate [34]. Moreover, GSK-3β overactivation increases the apoptosis of mature granule neurons and blocks the differentiation of neuroblasts [35]. Interestingly, an increased number of Doublecortin (DCX)⁺ neuroblasts is observed in GSK-3β-oe mice [35]. In this regard, the successive stages of AHN are featured by the expression of specific markers, and a stereotyped pattern of expression is thought to drive the maturation of these cells [25, 36]. For instance, DCX expression is switched off at 3-4 weeks of cell age under physiological conditions. However, we found that its expression was aberrantly prolonged until the sixth week of cell age in GSK-3β-oe mice [35], thus rendering an increased number of neuroblasts whose maturational progression was blocked. As will be further discussed, this blockade has important consequences for NGN functionality in GSK-3β-oe mice.

As previously mentioned, tau is one of the main downstream targets of GSK-3β and is considered a capital regulator of AHN [37, 38]. In this regard, both the expression and the post-transcriptional modifications of tau are tightly regulated during the maturation of NGNs [37–39]. For instance, the tau isoform featured by the presence of three-repeat microtubule-binding domains (namely, 3R-tau) is transiently expressed during the NGN neuroblast stage [38]. In rodents, individual NGNs show the highest expression of 3R-tau at 2 weeks of cell age [39], and the expression of this molecule is maintained until 4 weeks, a time point at which 3R-tau is replaced by the four-repeat microtubule-binding domain form of the protein (4R-tau) [38]. By using an antigen retrieval protocol, we showed that 3R-tau is a transient marker of NGN axons [39]. Noteworthy, this isoform of tau confers the cytoskeleton with greater plasticity than the 4R-tau isoform and the expression of 3R-tau is coincident with the period of time in which NGNs exhibit highest plasticity [40]. Although the 3R-tau/4R-tau ratio in humans and rodents appears to differ [41], the absence of reliable human 3R-tau markers for immunohistochemistry determinations has hindered the detailed study of these regulatory mechanisms in the context of human AHN to date. However, it has been proposed that an imbalance in the aforementioned ratio underlies certain neurodegenerative diseases [42]. This notion deserves further exploration in the context of AD.

Despite the predominant role that tau is assumed to play in AHN regulation, tau knock-down does not cause alterations in the proliferation, differentiation, or survival rates of neural precursors in the DG in vivo [43]. In this regard, compensatory mechanisms exerted by other microtubule-associated proteins (MAPs), such as MAP-1A and MAP-1B [44, 45], occur in the absence of tau. In contrast to the apparent absence of alterations showed by tau knock-out (tau-/-) mice regarding the basal rate of AHN, we and others have demonstrated that various animal models of tauopathies in which pathogenic forms of tau are overexpressed exhibit marked alterations in AHN [37, 46]. In this regard, the overexpression of a pathogenic form of the protein is not only characterized by its lack of physiological functions but also by the gain of toxic functions. A recent study by our group also revealed that stereotaxic injection of a soluble form of tau has devastating effects on the structural plasticity of granule neurons and impairs pattern separation ability [47]. These data might be relevant for the field of neurodegenerative disorders, since they contribute to shedding light on novel pathological roles played by distinct tau species in vivo.

The “V-shape” phenotype of newborn granule neurons in AD

As previously mentioned, NGNs go through a multi-stage development process before they reach maturity [25]. Although fully mature newly and developmentally generated granule neurons were initially described to be undistinguishable [48], growing evidence indicates that they show both functional and anatomical differences [49]. Regarding the latter, these two types of granule neurons differ in their positioning with respect to the GL [49, 50]. In this respect, NGNs are located in the inner third of the GL, whereas developmentally generated granule neurons are found in the two outer thirds of this layer [50]. Moreover, both types of cell show morphological particularities worthy of further discussion, the most remarkable difference being the number of primary apical dendrites. Under physiological conditions, most newly generated granule neurons show a single primary apical dendrite, which is extensively branched in the ML, thus resembling a “Y- shape” [49]. In contrast, developmentally generated granule neurons show several primary apical dendrites emerging from the soma [51]. In contrast to the scenario observed under physiological conditions, we showed that GSK-3β OE causes a dramatic change in the morphology of NGNs by triggering the presence of several primary apical dendrites, thereby conferring them a “V-shape” [16]. The relevance of this observation lies in the fact that the same morphology is observed in the granule neurons of AD patients [16]. Representative neuronal tracings corresponding to Golgi-stained granule neurons belonging to control subjects and AD patients are shown in Fig. 1. To the best of our knowledge, our work was the first to describe the occurrence of this phenomenon in AD patients; however, subsequent studies by other authors have revealed a similar “V-shape” phenotype, including the appearance of several primary apical dendrites in various pathological conditions [52–55]. Further efforts should address the molecular mechanisms involved in the appearance of this particular phenotype. In this regard, several molecules associated with the cytoskeleton may drive the development of dendritic branches and should be further explored in the context of neurodegenerative diseases [56].

OPEN IN VIEWER

Fig.1

Tracings of Golgi-stained granule neurons of control subjects (A) and patients with Alzheimer’s disease (AD) (B). It should be noted how granule neurons of control subjects display a single primary apical dendrite emerging from the soma, thus resembling a “Y-shape”, whereas granule neurons of AD patients show several primary apical dendrites that are poorly branched in the distal domains, thus acquiring a “V-shape”.

Despite the observed marked effects of GSK-3β OE on the morphology of mouse NGNs, a critical question for the AD research field is whether these effects are a cell-autonomous consequence of GSK-3β OE or whether they represent a non-cell-autonomous indirect effect derived from massive cell death or neuroinflammation. This question is particularly relevant in the context of AHN, given that the promoter used to drive GSK-3β OE, namely CamKII, is active only in mature cells [16]. In order to address this point, we developed an innovative system to selectively drive GSK-3β OE in our target cells, namely NGNs [57, 58]. This system takes advantage of the capacity of retroviruses to exclusively transduce proliferating cells and of the fact that proliferation is restricted mostly to NGNs in the DG [30]. This novel system is based on the stereotaxic injection of a retrovirus encoding the reverse Tetracycline activator (rtTA) element into the hippocampus of tetR-GSK-3β mice. These animals carry a bi-directional Tetracycline repressor (TetR) promoter followed by a GSK-3β cDNA in one direction and a cDNA encoding β-Galactosidase (β-Gal) fused to a nuclear localization signal in the other. By using this methodology, GSK-3β OE occurs only in those NGNs infected by the retrovirus and after the administration of Doxycycline. Thus, our novel methodology has three major advantages over traditional transgenic mouse systems. The first is the use of an rtTA element instead of the classical tTA element. This approach renders a tet-ON system in which GSK-3β OE is selectively triggered by the administration of Doxycycline—a feature that allows precise temporal control of GSK-3β OE. The second advantage is that the system allows tight regulation of the age and type of cell populations that overexpress this protein by means of Doxycycline administration during the desired periods. Finally, the system allows the double monitoring of GSK-3β OE by means of two reporter proteins, namely EGFP (encoded by the retroviral genome) and β-Gal (encoded by the murine genome). As previously mentioned, this methodology allowed us to achieve rapid and selective GSK-3β OE in the NGNs infected by the retrovirus [57]. Using this strategy, we demonstrated the cell-autonomous character of the alterations triggered by GSK-3β OE in NGNs, which showed a “V-shape” phenotype only under tet-ON conditions.

This system shows great versatility and potential utility in the field of neurodegenerative diseases, since it has proved highly suitable for the study of the cell-autonomous effects of other pathogenic proteins involved in these conditions [58].

Synaptic integration of newborn granule neurons

Although the functional consequences of the morphological alterations previously described remain to be fully elucidated, we have proposed that a reduced dendritic mass in the ML decreases the afferent connectivity of NGNs by reducing the likelihood of distal dendrites receiving afferent contacts from the EC. This hypothesis is strongly supported by the clear detrimental effect of GSK-3β at the synapse [59]. In this regard, long-term depression downregulates long-term potentiation through GSK-3β activation [60, 61], a phenomenon which is followed by synapse elimination.

By using a PSD95:GFP-expressing retrovirus [62], we demonstrated that the NGNs of GSK-3β-oe mice exhibit a marked reduction in the number and size of postsynaptic densities (PSDs), thus revealing impaired afferent connectivity [16, 63]. Special mention should be given to the important consequences of selective impairment of the synaptic integration of NGNs on the whole trisynaptic circuit. In this regard, during the period in which NGNs are young and excitable [40], they are thought to play a key role in information processing along the hippocampal circuit [64, 65]. In fact, it has been proposed that newly and developmentally generated granule neurons cooperate to create an accurate and complex representation of new memories. Developmentally generated granule neurons are believed to be involved mostly in pattern completion (an ability based on generalization) [66], whereas the lower activation threshold and higher excitability of NGNs favor their involvement in hippocampal pattern separation (a phenomenon consisting of the production of two differentiated outputs in response to very similar inputs) [67, 68]. The continuous re-definition of this elegant hypothesis proposed by Sahay [68] and McHugh [67] is continuously revealing the existence of additional complex orchestration mechanisms. For instance, competition phenomena between newborn and developmentally generated neurons to establish synaptic contacts with afferent fibers from the EC may also drive the intricate communication between this structure and the DG [69, 70]. This notion has been further supported by an elegant study by Sahay et al., in which the authors observed that reducing the connectivity of mature granule neurons dramatically increases the survival, maturation, and number of NGNs and maturity of synaptic connections of these cells [71]. These observations may have important consequences in the context of pathological conditions involving an imbalance between the synaptic integration of newly and developmentally generated granule neurons. In this regard, we found a similar imbalance in a murine model of AD overexpressing GSK-3β (GSK-3β-oe mice) in the hippocampus [51]. Granule neurons of these animals show a generalized reduction in the number of dendritic spines and synaptic contacts [51]. However, detailed morphometric analysis of these structures revealed that NGNs show a reduction in PSD volume (which in turn revealed a decrease in the synaptic strength of their afferent connections), whereas the dendritic spines of developmentally generated neurons show a marked enlargement (thereby suggesting a strengthening of the afferent connections of these cells). According to the competition hypothesis, these findings suggest the selectively impaired synaptic integration of NGNs in this animal model of AD. On the basis of the higher plasticity conferred by these cells to the hippocampal circuit, we have proposed that their reduced synaptic integration favors greater stability of old synaptic connections and a generalized lack of plasticity at the hippocampal level. Noteworthy, GSK-3β-oe mice exhibit alterations in hippocampal-dependent behaviors requiring high behavioral flexibility, such as in pattern separation [55].

There is intense debate in the field as to whether competition phenomena established between newly and developmentally generated NGNs are a physiological mechanism aimed to compensate the reduced capacity of the oldest neurons to cope with insults, or whether the natural purpose of NGNs is to gradually replace the oldest cells in order to favor their withdrawal from the hippocampal circuit [72, 73]. Nevertheless, the observation of these competition phenomena acquire particular relevance in the context of neurodegenerative diseases, given that impairments in pattern separation and in AHN have been found in AD patients and in animal models of this disease [32]. These observations and the extraordinary regenerative potential of AHN have placed this process in the spotlight of therapeutic strategies aimed to tackle neurodegenerative diseases.

By using the same retroviral approach previously mentioned, we demonstrated that GSK-3β OE increases tau phosphorylation in single NGNs in a cell-autonomous manner, thus revealing tau as a key mediator of the alterations in NGN functionality caused by GSK-3β OE. In this regard, despite the classical association of tau with the axonal compartment, growing evidence strongly supports the notion that this protein is present and involved in various functions of the somatodendritic compartment. In 2011, Ittner et al. [74] elegantly demonstrated the participation of tau in cortical synapses. In that study, the authors showed that tau depletion is neuroprotective against synapse destabilization caused by the presence of Aβ. More recently, these authors have demonstrated that tau phosphorylation finely tunes the synaptic roles played by tau [75]. It has been proposed that tau modulates the synaptic localization of Fyn kinase, which favors synapse stabilization [74]. On the other hand, tau phosphorylation by GSK-3β increases tau aggregation and impedes several of its actions at the synapse. A detailed schematic diagram showing these regulatory mechanisms is shown in Fig. 2. In agreement with previous observations in other neuronal populations [74], our data reveal that the absence of tau impairs the synaptic integration of NGNs [43]. These alterations are particularly apparent in the most distal part of the dendritic tree, where the NGNs of tau-/- mice show the most drastic reduction in the number of PSDs, accompanied by a marked decrease in the volume of these structures [43]. Importantly, these data indicate that the synaptic integration of NGNs in tau-/- mice is impaired, as a result of a reduction in the connectivity of the outer parts of their dendritic trees with afferent fibers from the EC.

OPEN IN VIEWER

Fig.2

Schematic model proposed to explain the mechanisms through which GSK-3β and tau regulate the synaptic integration of granule neurons. A) Under basal conditions, tau plays a central role at the synapses through its interaction with Fyn kinase. On the other hand, tau stabilizes microtubules, thus allowing the dendritic transport of synaptic receptors to the synapses, where they are stabilized through their interactions with the scaffold protein PSD95. GSK-3β exerts tau-dependent and -independent regulatory actions at the synapse. B) Under pathological conditions, GSK-3β overexpression increases tau phosphorylation. This phenomenon decreases tau affinity to bind microtubules, which are destabilized. On the one hand, this compromises dendritic transport. On the other hand, tau phosphorylation triggers its aggregation, which further impairs cell function. In addition, synapses are destabilized by tau-dependent and tau-independent mechanisms. C) The chronic maintenance of these pathological conditions leads to the disappearance of the synapse.

Nevertheless, the general consensus in the field is that the absence of tau confers a certain degree of neuroprotection in response to exposure to various toxic agents. In this regard, we demonstrated, for the first time, that the absence of tau conferred NGNs with extraordinary resistance to stress [43]. However, our results also showed that the absence of tau completely blocked the stimulatory actions exerted by environmental enrichment (EE) on NGNs. Of note, EE is one of the most potent positive regulators of AHN [76]. It increases the survival, maturation, and plasticity of NGNs in WT mice [77]. However, we did not observe these stimulatory effects in tau-/- NGNs [43]. Our results therefore indicate that tau depletion has a negative impact on neurodegenerative diseases by further decreasing the plasticity of especially sensitive neuronal populations. These data should be taken into account during the development of strategies aimed to ameliorate neurodegeneration caused by aberrant forms of tau. Special emphasis should be placed on identifying the physiological roles played by the various species of tau in specific neuronal populations, given that not all of these functions can be compensated by other MAPs.