From cradle to grave: neurogenesis, neuroregeneration and neurodegeneration in Alzheimer’s and Parkinson’s diseases

Neurogenesis in Alzheimer’s disease

As mentioned previously, neuropathological hallmarks of AD include the presence of extracellular Aβ plaques and intraneuronal NFTs. The accumulation of these plaques and tangles at synaptic sites eventually leads to synaptic degeneration and neuronal loss (DeTure and Dickson., 2019).

Interestingly, the analysis of post mortem brain tissues from AD patients reveals a significant reduction in NSCs in the SVZ (Ziabreva et al., 2006) and an increase of NSCs in the dentate gyrus of the hippocampus (Sung et al.,2020; Babcock et al., 2021). Through in vitro experiments using mouse SVZ-derived NSCs, researchers have demonstrated significantly increased neurogenesis when NSCs are exposed to Aβ_1–42 (Scopa et al., 2019). Transgenic mouse models expressing mutant APP exhibit significantly decreased neurogenesis in the SVZ and the dentate gyrus of the hippocampus (Wirths, 2017; Houben et al., 2021). Surprisingly, another study reported significantly increased neurogenesis in the SVZ of mice expressing both mutant APP and mutant Presenilin 1 (PSEN1). A study by Chevallier and colleagues demonstrated that the rate of neurogenesis is linked to each specific mutant form of Presenilin, as different mutant PSEN1 transgenic mice showed different variations of increased or decreased neurogenesis (Chevallier et al., 2005). However, the authors in this study only looked at the increasing BrdU numbers, whereas contradictory reports looked at different neuronal markers (Wen et al., 2004).

Similar observations have been reported using the triple transgenic mouse model (3xTg-AD) generated by Oddo et al., 2003. This widely used model mimics AD pathology through the expression of mutant APP, PSEN1, and MAPT genes. These transgenic mice demonstrate cognitive impairment due to the region-specific accumulation of Aβ and tau (Marlatt et al., 2015; Wirths et al., 2017). Moreover, both the SVZ and SGZ of 3xTg-AD mice exhibit impaired neurogenesis (Rodriguez et al., 2008). These mice also display an age-dependent decline in neurogenesis when compared with age-matched controls. The authors correlate this decline in neurogenesis with the accumulation of Aβ plaques (Rodriguez et al., 2009). It has been suggested that Aβ interferes with the balance between excitatory and inhibitory inputs in newly generated neurons, impairing neurogenesis (Mucke and Selkoe et al., 2012). Other AD mouse models like 5xFAD (Zalatel et al., 2018), Tg30 (Houben et al., 2019), Mapt^–/– (Hong et al., 2010) have all demonstrated a significant reduction in the rate of adult hippocampal neurogenesis compared with controls. However, a recent study of 14 month old Mapt^–/– mouse has demonstrated a significant increase in adult hippocampal neurogenesis, via the quantification of BrdU⁺ cells (Criado-Marrero et al., 2020).

As mentioned above, different AD rodent models demonstrate significant variation in regard to rate of adult neurogenesis. The presence of extracellular Aβ and intracellular tau makes it difficult to determine their individual impact on differentiating stem cells (Winner and Winkler., 2015), limiting our understanding of adult neurogenesis in the context of AD. The genotype-dependent mechanisms responsible for varying rates of neurogenesis in different AD mouse models require further investigation.

Neurogenesis in Parkinson’s disease

In the past decade, PD research has successfully demonstrated that early-stage neuronal loss originates in the hippocampus and olfactory bulb, rendering neurogenesis research in these brain regions particularly interesting (Weintraub and Burn, 2011; Carlesimo et al., 2012). However, similar to AD, the analysis of post mortem brain tissues from PD patients has revealed conflicting results. For instance, Höglinger and colleagues demonstrated a significant reduction in the number of proliferative progenitors in the SVZ of PD patients relative to healthy controls (Höglinger et al., 2004), whereas van der Berge and colleagues found no change (van der Berge et al., 2011). These conflicting results likely originate from a number of differences between these two studies, such as the significantly older population (10 years older) and longer post mortem interval (~20 hours) studied in Höglinger et al., 2004. The two studies also used different markers and analyzed different areas of the SVZ.

Current transgenic mouse models are incapable of accurately mirroring PD pathology, as none of them demonstrate the nigrostriatal degeneration observed in human patients. Instead, researchers mimic the loss of DA neurons in mice with the help of neurotoxic compounds such as 6-hydroxydopamine (6-OHDA; Hernandez-Baltazar et al., 2017; Zeng et al., 2018) or 1-methyl-4-phenyl-1, 2, 3, 6-tetrahydropyridine (MPTP; Meredith and Rademacher., 2011). Interestingly, mice treated with 6-OHDA or MPTP demonstrate significantly reduced numbers of proliferating cells in the hippocampus (Suzuki et al., 2010), but an increase in dopaminergic neurogenesis in the olfactory bulb (Yamada et al., 2004; Winner et al., 2006) similar to that observed in some PD patients (Huisman et al., 2004).

Researchers have also generated transgenic mice carrying human wild-type α-syn (Masliah et al., 2000). In these mice, the observable accumulation of α-syn in different brain regions leads to spatial memory deficits (Masliah et al., 2011). The overexpression of human wildtype α-syn contributes to a significant reduction in hippocampal neurogenesis that coincides with increased neuronal loss (Winner et al., 2004). Interestingly, a similar reduction in hippocampal neurogenesis was also observed in a conditional transgenic mouse model expressing human wildtype α-syn (Nuber et al., 2008) and in mouse models expressing the human mutant A53T α-syn (Koprich et al., 2017; Regensburger et al., 2020). Studies have reported the role of α-syn in dendritic outgrowth, branching, and spine density and maturation (Winner et al., 2012), specifically in regard to hippocampal neurons. These findings highlight the importance of α-syn in hippocampal neurogenesis (Winner et al., 2012). Interestingly, the authors have also demonstrated similar impairment in adult neurogenesis and neurite outgrowth in LRRK2 G2019S transgenic mice (Winner et al., 2011) and in PINK1 deficient zebrafish, where generation of dopaminergic neurons in adult brain was significantly affected (Brown et al., 2021). As in AD research, elucidating the mechanisms that regulate neurogenesis in PD is crucial for screening potential targets that might modulate adult neurogenesis.

Neuroregeneration in Alzheimer’s disease

Axonal degeneration is an early pathological sign of many neurodegenerative diseases, including AD. This observation is supported by the decreased white matter volumes identified in patients with mild cognitive impairment, a high-risk precursor to AD (Kalus et al., 2006; Rogalski et al., 2009; Ihara et al., 2010; Bozzali et al., 2011). The accumulation of NFTs and Aβ plaques in AD brains causes neurons to undergo a slow, “dying-back” process (Salvadores et al., 2020), where NFT and Aβ-accumulation drives degeneration from the axon terminals inward toward the cell body, eventually leading to cell death (Gilley et al., 2011; Nishioka et al., 2019). This Wallerian-like degeneration contributes to synaptic loss and interrupts axonal transport, causing connective deficits and driving cognitive decline over time (Blazquez-Llorca et al., 2017).

Interestingly, a study conducted by Blazquez-Llorca and colleagues suggests that the early prevention of Aβ plaque accumulation could stimulate axonal regeneration and prevent AD progression (Blazquez-Llorca et al., 2017). NFT and Aβ-accumulation disrupt axonal transport, reducing the concentration of the axon survival factor nicotinamide nucleotide adenylyltransferase-2 (Gilley and Coleman, 2010; Ljungberg et al., 2012; Ali et al., 2016) and contributing to mitochondrial dysfunction, oxidative stress, and the dysregulation of Ca²⁺ homeostasis (Cieri et al., 2018; Mata, 2018; Albensi, 2019).

AD is also characterized by a loss of dendrites and a reduction in dendritic spine density (Boros et al., 2019). Exploring methods by which to regenerate normal dendritic structure and synaptic function in hippocampal neurons of AD patients is necessary to “shift the balance from neurodegeneration to regeneration” and reverse cognitive decline (Iqbal et al., 2014).

To this end, many studies have focused on the neuroregenerative potential of neurotrophin therapy in the otherwise regeneration-adverse CNS. For example, BDNF has been identified as an important facilitator of axon regeneration, synaptic plasticity, and brain injury recovery (McGregor and English, 2019), and reduced levels of BDNF and its TrkB receptor have been observed in the AD brain (Sampaio et al., 2017). This reduced BDNF/TrkB activity has been shown to upregulate inflammatory pathways that facilitate the cleavage of tau and APP in an AD mouse model. Cognitive decline can be reversed via the inhibition of these BDNF-linked inflammatory pathways (Wang et al., 2019). Similarly, the administration of BDNF into the EC of mouse and primate AD models has also been shown to improve cognition (Nagahara et al., 2009). However, as described later, several obstacles preclude the administration of neurotrophins in vitro (Kazim and Iqbal, 2016; Uliassi et al., 2017), and the potential neurorestorative functions of BDNF and other neurotrophins require further study before therapeutic applications can be considered.

As previously mentioned, CSPGs inhibit the regenerative potential of neurons and have been found to be upregulated in AD brains (Howell et al., 2015). CSPGs bind and signal via tyrosine phosphatase sigma (PTPσ), which restricts neuronal growth (Tran et al., 2018). Researchers have observed reduced neuroinflammation, decreased synaptic loss, and enhanced cognition in AD mouse models following PTPσ inhibition (Gu et al., 2016). These results suggest that modulating PTPσ might prove an effective strategy for improving neuronal regeneration in AD brains.

Neuroregeneration in Parkinson’s disease

Like in AD, the neuronal death involved in PD disease begins with axonal degradation (Tagliaferio and Burke, 2016). Many of the debilitating motor symptoms associated with PD stem from the degeneration of nigrostriatal DA neurons, whose axons bridge the substantia nigra pars compacta and caudate putamen regions of the human brain (Sidorova et al., 2019). PD pathogenesis is driven by the aggregation of α-syn into Lewy bodies and neurites and the eventual loss of DA neurons (Dickson, 2017). As a result, much of the research related to achieving neuroprotection in PD has focused on preventing neuronal death rather than reversing axonal degeneration (Tagliaferro et al., 2016). However, in animal studies involving induced α-syn overexpression, researchers have identified swollen, dystrophic neurites consistent with Wallerian-like degeneration, a loss of striatal dopaminergic terminals in DA neurons, and crippled axonal transport prior to neuronal death (Chung et al., 2009; Decressac et al., 2012). Although Lewy bodies are generally identified in the neuronal soma, studies have described significant α-syn accumulation in the axons as Lewy neurites as well (Volpicelli-Daley et al., 2016).

The importance of axonal degradation over programmed cell death in early PD pathology may explain the failure of anti-apoptotic kinase inhibitors to prevent disease progression, as these neuroprotective agents do not confer any sort of axonal protection or regenerative influence (Cheng et al., 2010). Instead, research suggests that upregulation of the Akt-Rheb-mTor signaling pathway via constitutive Rheb activation is sufficient to induce sprouting in 6-OHDA-damaged DA neurons (Kim et al., 2012). This is particularly promising in that other studies have illustrated similar projection renewal in CNS axons by manipulating different upstream targets in the mTor pathway (Park et al., 2008; Cheng et al., 2011).

As in AD research, neurotrophin therapy offers another potential method for restoring the function of diseased DA neurons. Reduced serum and brain neurotrophin levels have been observed in both PD patients and rodent models (Khalil et al., 2016; Huang et al., 2018). Notably, BDNF, neurotrophin 3 , and neurotrophin 4 have been shown to influence the differentiation and structural development of DA neurons in vitro (Studer et al., 1995). In particular, BDNF has been shown to protect DA neurons against neurotoxic lesion (Hyman et al., 1991; Spina et al., 1992), to promote neurite outgrowth and arborization (Studer et al., 1995), and to reduce motor issues and protect DA neurons in animal models of PD (Altar et al., 1994; Hagg et al., 1995). Studies also suggest that α-syn interferes with BDNF receptor TrkB, contributing to the degeneration of DA neurons by effectively eliminating BDNF’s pro-survival influence in vitro and in vivo (Zhang et al., 2018).

However, while a range of studies have documented the neurotrophin-boosting effects of exercise and food-borne polyphenols (Hirsch et al., 2018), the blood-brain barrier (BBB) presents a significant obstacle to the direct administration of such neurotrophin-based treatments (Kazim and Iqbal, 2016; Uliassi et al., 2017). Issues related to BBB permeability and molecular specificity have been exemplified in a range of clinical trials involving neurotrophic factors like glial-derived neurotrophic factor and neurturin (Sidorova et al., 2019). Neurotrophic factor small molecule mimetics represent a popular solution due to their ability to pass through the BBB (Kazim and Iqbal, 2016). Various neurotrophic factor small molecule mimetics have been shown to elicit neuroprotective and neuroregenerative responses, but issues related to receptor specificity and dosing indicate a need for further investigation (Kazim and Iqbal, 2016). While gene and stem cell therapies also represent promising options for overcoming these obstacles and generating a regeneration-friendly microenvironment for neurons, additional research is required in this area as well (Figure 2; Glavaski-Joksimovic and Bohn, 2013; Ghosh et al., 2014; Reddy et al., 2021).

Neurodegeneration in Alzheimer’s disease

Several reports speculate that Aβ plaques appear years before the onset of AD symptoms and could trigger the accumulation of hyper-phosphorylated tau in tangles (Bloom, 2014). AD progression has been previously linked to synaptic failure (Chen et al., 2019), which has been suggested as a better marker of cognitive decline than the accumulation of NTFs and Aβ plaques (Bereczki et al., 2018). Although aging is considered to be one of the most important risk factors for AD (Hebert et al., 2013), multiple genetic mutations have been implicated in the disease, including PSEN1, PSEN2 and APP (Lin et al., 2020). Mutations in the APP gene, such as E693Q (Petersen et al., 2010; Petersen, 2018; Roehr et al., 2020), A673T (Peacock et al., 1993; Jonsson et al., 2012), and KM670/671NL (Oksanen et al., 2018) have given to the amyloid cascade hypothesis, which states that an imbalance in APP metabolism leads to altered Aβ homeostasis, triggering AD-type neurodegeneration (Uddin et al., 2021). Almost all patients suffering from an inherited form of AD have mutations in either APP or PSEN1/2 (Haass et al., 2012).

Neurodegeneration in Parkinson’s disease

As previously described, PD is characterized by the loss of DA neurons in the substantia nigra pars compacta in the human midbrain. Previous research estimates that 30–70% of DA neurons are lost prior to PD symptom manifestation (Cheng et al., 2010), highlighting the need to identify biomarkers for early disease diagnosis (Pan et al., 2019). Several genetic mutations are associated with the loss of DA neurons, serving as a template for the study of neurodegeneration. Well-researched genes include SNCA (α-syn; Sato et al., 2011; Winner et al., 2011; Mahul-Mellier et al., 2020), PTEN-induced putative kinase 1 (PINK1; Cooper et al., 2017), Glucocerebrosidase (GBA1; Murphy et al., 2014; Mazzulli et al., 2016), Parkin (Sanyal et al., 2015) and Leucine-rich repeat kinase 2 (LRRK2; Volpicelli-Daley et al., 2016; Ferreira and Massano, 2017). Specific mutations in these genes have been extensively studied. For example, the A53T mutation in SNCA induces mitochondrial dysfunction in rodent models (Bido et al., 2017), the G2019S mutation in LRRK2 gene increases α-syn accumulation and causes autophagy dysregulation (Su et al., 2015; Volpicelli-Daley et al., 2016), and the G411S mutation in PINK1 gene causes mitochondrial dysfunction (Puschmann et al., 2017).

Recent reports have also highlighted the importance of β-syn and the V70M and P123H mutations in the neurodegeneration of DA neurons in rodent models of PD and dementia with lewy bodies, respectively (Psol et al., 2021; Raina et al., 2021). This study demonstrates that overexpression of β-syn and its mutants are toxic to hiPSC-derived DA neurons and rat primary cortical neurons in a manner similar to α-syn-induced toxicity. This β-syn induced neuronal toxicity was preceded by mitochondrial and synaptic dysfunction in these cultured neurons (Psol et al., 2021). These results indicate that rodent models using viral vectors to overexpress proteins to study neurodegeneration remain attractive research tools.

It is well known that the upregulation or overexpression of α-syn causes a loss of DA neurons in rodent models (Taschenberger et al., 2012) and in hiPSC-derived neurons (Mahajani et al., 2019). Dopaminergic and glutamatergic neurons patterned from the same hiPSC line demonstrate significantly different vulnerability to toxicity induced by α-syn overexpression with differentiated DA neurons proving to be more susceptible to neuronal loss than differentiated glutamatergic neurons (Mahajani et al., 2019). Researchers are working on transdifferentiating rat primary cortical neurons to generate DA neurons and evaluate the effect of α-syn-induced toxicity in DA neurons (Raina et al., 2020).

Although variance and a lack of reproducibility between different hiPSC lines make it difficult to compare significant findings (Mahajani et al., 2021), studies with hiPSC-derived neuronal models have contributed immensely to our understanding of the mechanisms impaired in PD. Impaired cellular mechanisms that contribute to the loss of DA neurons in the midbrain include oxidative phosphorylation (Protter et al., 2012), mitochondrial dysfunction (Ryan et al., 2015), mRNA translation (Kim et al., 2020), autophagy dysfunction (Sanchez-Danes et al., 2012), and the degeneration of axons and dendrites (Czaniecki et al., 2019), which has been demonstrated in other disorders as well (Giacomini et al., 2014; Cortelli et al., 2015; Giacomini et al., 2016). Some of these impaired cellular mechanisms are briefly summarized below.

Alzheimer’s disease

Recent advances made in the field of single-cell RNA sequencing seem poised to improve our understanding of the molecular targets responsible for Aβ/NFT-mediated neuronal toxicity in AD. At the time of writing, 73 datasets containing more than 700,000 cells from different regions of the human and mouse brain have been generated in the study of AD. Analysis has been performed extensively on single cells from the prefrontal cortex, EC, superior parietal lobe, and superior frontal gyrus from AD and control human brains. The same is true of the hippocampus, cerebral cortex, prefrontal cortex, SVZ and the cerebellum of the mouse brain (Jiang et al., 2020). These datasets are free and publicly available for other researchers to use in their own analyses. AD is characterized by a slow neurodegeneration beginning in the EC and eventually progressing to the limbic and neocortical structures, making the EC and hippocampus the earliest affected brain regions. When Grubman and colleagues performed scRNA-seq on the EC region from AD patients, they demonstrated that the AD risk gene APOE was specifically suppressed in oligodendrocyte precursors cells and astrocytes, but was surprisingly upregulated in microglial cells (Grubman et al., 2019). Another study sequenced more than 80,000 nuclei from prefrontal cortex of 48 AD patients at different disease stages. The authors observed that the most significant AD- specific changes typically occur in the early stages of disease progression (Mathys et al., 2019). Otero-Garcia and colleagues isolated and profiled neuronal somas with or without NTFs from the prefrontal cortex of AD patients and controls, demonstrating that there exists a selective susceptibility of different neuronal subtypes to form NFTs throughout AD progression (Otero-Garcia et al., 2020).

Parkinson’s disease

Significantly more single-cell RNA sequencing has been used in AD research than in studies related to PD. In the past five years, multiple analyses have been performed on single cells isolated from hiPSC-derived neurons (Lang et al., 2019; Fernandes et al., 2020), rodent tissues (Hook et al., 2018; Tiklova et al., 2019, 2020; Bryois et al., 2020), and human post mortem samples from different brain regions (Welch et al., 2019; Agarwal et al., 2020; Bryois et al., 2020; Smajic et al., 2020). An interesting study by Fernandes and colleagues has revealed six transcriptionally distinct cell clusters including two dopaminergic progenitor clusters and four mature dopaminergic neuronal clusters, each of which demonstrate a differential sensitivity to stress (Fernandes et al., 2020). Using rodent brain tissues and differential gene expression, researchers have obtained a set of genes downregulated in mouse DA neurons (Bryois et al., 2020). But single-cell RNA sequencing provides the most insight when human post mortem tissues from PD patients are analyzed. The profiling of cells from the substantia nigra of PD patients and control individuals reveals multiple distinct cell types, including neurons, astrocytes, oligodendrocytes, microglia, oligodendrocyte progenitor cells and endothelial cells (Agarwal et al., 2020; Welch et al., 2020). These studies illustrate that PD-related upregulation of microglia and astrocytes correlate with an increase in cytokine signaling and stress response to other unfolded proteins (Agarwal et al., 2020; Welch et al., 2020). Along with these analyses, various computational and machine learning tools have been developed for researchers to better understand single-cell RNA sequencing data and visualize their results. Researchers are working toward compiling comprehensive single-cell atlases that others can use freely as reference datasets.