Exploring potential developmental origins of common neurodegenerative disorders

Abstract

In the United States, it is now estimated that 6.7 million people over the age of 65 are afflicted by Alzheimer’s disease (AD), over 1 million people are living with Parkinson’s disease (PD), and over 200,000 have or are at risk for developing Huntington’s disease (HD). All three of these neurodegenerative diseases result in the ultimate death of distinct neuronal subtypes, and it is widely thought that age-related damage is the single biggest contributing factor to this neuronal death. However, recent studies are now suggesting that developmental defects during early neurogenesis could also play a role in the pathology of neurodegenerative diseases. Loss or overexpression of proteins associated with HD, PD, and AD also result in embryonic phenotypes but whether these developmental defects slowly unmask over time and contribute to age-related neurodegeneration remains highly debated. Here, we discuss known links between embryonic neurogenesis and neurodegenerative disorders (including common signaling pathways), potential compensatory mechanisms that could delay presentation of neurodegenerative disorders, and the types of model systems that could be used to study these links in vivo.

Introduction

Neurodegenerative disorders arise from depletion of neurons from distinct regions of the brain during aging. Huntington’s disease (HD) is caused by death of the medium spiny neurons (MSN) in the striatum [1]. Parkinson’s disease (PD) results from the death of the dopaminergic neurons of the substantia nigra pars compacta [2], and patients with Alzheimer’s disease (AD) exhibit massive loss of neurons in the cerebral cortex and dentate gyrus (DG) and first cornu ammonis (CA1) region of the hippocampus [3, 4]. It has long been thought that this progressive neuronal death stems from accumulated damage during the aging process [5]. However, recent data suggests that this death could originate from embryonic neurogenesis defects that are at first neutralized by compensatory mechanisms but become slowly unmasked over time [6–9].

It is now generally accepted that HD has a developmental component [7, 8], and data supporting developmental origins of AD and PD are also steadily accumulating [6, 9]. These data include identification of developmental roles for proteins known to be involved in PD and AD progression, common signaling pathways that play key roles in both neurogenesis and neurodegeneration, and compensatory mechanisms that mask developmental phenotypes until later in life [7–12]. Here, we will first summarize the data indicating developmental origins of HD, PD, and AD and discuss the signaling pathways that could participate both in setting the foundations of disease and promoting the onset of symptoms later in life. We will also discuss evidence of compensatory mechanisms that could allow for the delayed timeline between these two events and describe potential new mouse models for understanding the mechanistic links between abnormal neurogenesis and age-dependent neurodegeneration.

Indications of neurodevelopmental origins

HD is characterized by polyglutamine (polyQ) expansion of the huntingtin protein (HTT) that leads to progressive loss of motor and cognitive function. Of the common neurodegenerative disorders, HD is the first to have an undisputed developmental component [13]. Juvenile HD, seen in patients with greater than 54 CAG repeats in the HTT polyQ expansion, affects neurodevelopment, including bilateral underdevelopment of the putamen and caudate nucleus [14, 15]. Expression of Huntingtin CAG repeats in human embryonic stem cells (hESCs) results in misregulation of germ layer patterning [7], and increasing CAG repeat length shifts neural differentiation to giant multinucleated telencephalic neurons due to chromosomal instability and failed cytokinesis [16]. Tissues from human fetuses that express mutant HTT exhibit defects in HTT localization, neuro-progenitor cell polarity, differentiation, and cell cycle progression [13]. It has now also been shown that these localization and polarity defects are, at least in part, due to a disrupted cytoskeletal network and subsequent impaired axonal growth [8].

Though not as widely accepted as HD to have a developmental component, there are indications that PD and AD may as well. First, the nonmotor symptoms of PD, including sleep disorders, hallucinations, and depression, can precede the motor symptoms by years or decades [2]. Second, motor symptoms only materialize after 50–60% of the dopaminergic neurons in the substantia nigra have degenerated [17], indicating PD originated long before appearance of these symptoms [9]. Finally, of the 17 genes known to have PD-causative mutations [18], many are also involved in neurodevelopment. α-synuclein is required for development of a subset of dopaminergic neurons, as its loss results in mouse embryos that exhibit less dopaminergic neurons in the substantia nigra [10]. Vacuolar protein sorting 35 (VPS35) has an important role in hippocampal neurogenesis, specifically regulating axonal protein transport and promoting dendritic growth [19]. In Xenopus, knockdown of PTEN-induced putative kinase 1 (PINK1) results in a deceased amount of central dopaminergic neurons and alterations in mitochondrial function [20], and in mice, loss of leucine-rich repeat kinase 2 (LRRK2) in neural stem cells (NSCs) increases their differentiation rate and expression of the PD-associated LRRK2 mutant R1441G impairs their differentiation [21].

For AD, studies on the temporal dynamics of known biomarkers indicate a preclinical stage of the disease that could last several decades [6]. This was supported by the finding that approximately 25% of autopsy patients under the age of 30 exhibit pre-tangle tau formations [22]. Also, similar to patients with juvenile HD, carriers of apolipoprotein E (APOE) susceptibility mutations, who develop an early form of AD, exhibit brain abnormalities many decades before the onset of typical dementia symptoms [23], including neurodevelopmental differences in gray and white matter volumes [24]. They also exhibit altered rates of cognitive development [25]. In addition, data suggest the two mainly proposed etiologic agents of AD, amyloid precursor protein (APP) and microtubule-associated protein tau (MAPT/tau) also have neurodevelopmental roles. APP begins to be expressed by embryonic day 9.5 (E9.5) in the mouse neural tube [26], and it is also found in the radial glial stem cells (RGCs) of the fetal mouse brain [27]. sAPPα, the product of APP cleaved by α-secretase, stimulates the proliferation of rat embryonic NSCs [28]. Stably overexpressed APP robustly drives neural differentiation of hESCs [11], while inhibiting α-secretase activity decreases neural progenitor cell proliferation [29]. In addition, amyloid beta peptide (Aβ), the amyloidogenic product of APP cleaved by both β- and γ-secretase, promotes differentiation of NPCs toward a neuronal fate [30]. Tau is also expressed throughout brain development, beginning in the RGCs and increasing as migrating neuronal precursors mature [31]. In mice, loss of tau alters neuronal migration and differentiation, where a portion of tau knockdown neurons reach the cortical plate but they have underdeveloped dendrites and reduced connectivity [32].

Signaling pathways common to neurogenesis and neurodegeneration

Another indication that there might be a developmental component to neurodegenerative disorders is that both embryonic neurogenesis and neurodegeneration utilize a number of similar pathways. Therefore, defects in a pathway later in life could denote that similar defects were present during development. For HD, it has long been known that mutant HTT (mHTT) alters TGFβ signaling [33]. Asymptomatic HD patients have low levels of serum TGFβ and post-mortem examination of HD brains also shows reduced TGFβ expression in cortical neurons [33]. Correspondingly, expression of mHTT reduces TGFβ mRNA and protein levels in cultured astrocytes and striatal cell lines [33], and LPS stimulation of wild type mast cells induces expression of TGFβ mRNA but is unable to do so in mast cells that express mHTT [34]. This repression of TGFβ signaling is hypothesized to contribute to the systematic pro-inflammatory phenotypes characteristic of HD [34]. Not surprisingly, it has also been subsequently found that mHTT affects TGFβ signaling during development. Early TGFβ signaling regulates density-dependent induction of apical/basolateral polarity, via basolateral concentration of TGFβ receptors [35]. Interestingly, while 20 CAG repeat HTT does not disrupt the basolateral localization of TGFβ, expression of 56 CAG HTT or HTT knockout results in a significant amount of TGFβ receptors exposed on the apical surface, leading to increased mesendodermal patterning and a contracted ectodermal compartment [7], suggesting early defects in the foundation of the nervous system.

For PD, it has been established that wingless-related integration site (Wnt) signaling is critically involved in the regulation of dopaminergic neuronal health and protection [36]. Specifically, Wnt signaling regulates neuronal differentiation, axonal extension, synapse formation, and neurogenesis, and its decrease in PD brains is thought to worsen the effects of age-dependent inflammation and oxidative stress [36, 37]. Multiple genes in the Wnt signaling pathway have been found hypermethylated in PD brains, which corresponded to reduction of their protein expression [38]. Also, multiple genes in the Wnt signaling pathway are mutated in PD. LRRK2, a scaffolding protein that inhibits the Wnt/β-catenin pathway but activates the Wnt/planar cell polarity pathway [39], is one of the most commonly mutated proteins found in PD, and expression of human LRRK2 PD mutants has been shown to drive α-synuclein deposition and formation of phosphor-tau-positive inclusions [40–42]. Similarly, human PD polymorphisms of glycogen synthase kinase-3beta (GSK3β), an inhibitor of Wnt/β-catenin signaling, correlate with increased Tau phosphorylation [43]. Increased α-synuclein deposition is thought to promote Lewy body formation, and abnormal hyperphosphorylation of tau protein is thought to promote neurofibrillary tangle (NFT) formation, both of which may contribute to the cell death and poor axonal transport observed in PD [44].

Both LRRK2 and GSK2β also play roles in neurodevelopment, so it stands to reason that genetic mutations in these genes could additionally give rise to early neurogenic defects. In rats, expression of PD-associated LRRK2 mutations induces a reduction in neurite length and branching [42]. In mice, expression of the LRRK2 mutant R114G in NSCs leads to downregulation of neuronal-differentiation-inducing pathways, including let-7a, and impairs neuronal differentiation [21]. GSK3 signaling regulates the proliferation and differentiation of progenitor cells during brain development [45]. Inactivation of GSK3β, specifically, leads to increased nuclear transfer of β-catenin, decreased apoptosis rates, and increased neuronal differentiation in neurospheres [45], indicating patients that harbor gain- or loss-of-function mutations of GSK3β could begin life with significant neurogenic defects.

Notch signaling, important in AD, is another critical neurodevelopmental pathway, regulating quiescence and differentiation of NSCs [46]. Reduction of Notch signaling results in delayed or bypassed quiescence and rapid initiation of terminal differentiation [47, 48]. Both Notch protein and APP are cleaved by presenilin-1 (PS1), the catalytic subunit of γ-secretase [49], resulting in the formation of Notch intracellular domain (NICD) and Aβ, respectively. It is proposed that neuronal loss in AD is due to competition between APP and Notch-1 for cleavage by PS1 [50]. Mutation in the gene encoding PS1, PSEN1, is the most common cause of familial AD (FAD) [51], and the L166P PSEN1 mutation, found in a particularly aggressive form of FAD, impairs NICD production and Notch signaling [52]. As mutation in PSEN1 alters Notch signaling in the later stages of FAD, it was hypothesized that it might also impair Notch signaling during early neurogenesis [49]. Accordingly, expression of five FAD PSEN1 mutations (int4del, Y115H, M139V, M146I, and R278I) in induced pluripotent stem cells (iPSCs) results in decreased Notch signaling and a corresponding premature differentiation phenotype with earlier neural projections and a significantly reduced number of proliferative progenitor cells [49], indicating that patients with FAD could also be beginning life with significant neurogenic defects.

Compensatory Mechanisms

A major key to determining if AD and PD have neurodevelopmental components, is demonstrating if there are compensatory mechanisms present that can mask symptoms until later in life. Work in HD suggests that cytoskeletal defects during neurogenesis can establish neurons with weakened connectivity that are predisposed to later degeneration [8], and that weakened connections can be compensated for by re-routing neuronal circuitry [53]. For example, appearance of motor abnormalities in HD could be delayed by re-routing indirect pathway signaling through the cerebellum to the thalamus [54]. It has been proposed that neurons from the subthalamic nucleus are still functional in HD and can reroute signaling to the pons, the cerebellar cortex, the dentate nucleus, and finally back to the thalamus where the loop is completed (Figure 1A,B) [53]. Evidence to support this re-routing of indirect pathway signaling is shown by connectivity studies of resting state functional MRI (rs-fcMRI) performed on children ages 6–18 years with genetically expanded (≥ 40 CAG repeats) or genetically non-expanded (≤ 39 CAG repeats) HTT sequences [53]. In this study, children in the genetically expanded group had much stronger connections from the subthalamic nucleus to the anterior lobe of the cerebellum (aCB) and the pontine nuclei to the aCB [53]. It was also shown that individuals in the genetically expanded group had increased output from the dentate nucleus to the ventrolateral nucleus of the thalamus [53]. It is proposed that compensation from the cerebellum in the short-term can mask symptoms until the cerebellum also becomes pathologic [55].

Figure 1. Indirect pathway signaling with and without potential compensatory signaling from the cerebellum.

(A) Normally, the indirect pathway uses inhibitory signaling from striatal regions (shown in tan) to prevent the thalamus from signaling to the cortex and inhibit movement. (B) In Huntington’s disease, death of dopaminergic neurons in certain regions of the striatum prevents this signaling. However, it is proposed that this loss of indirect signaling can be temporarily compensated for by signaling from the subthalamic nucleus to cerebellar regions (shown in red). This compensatory mechanism could delay onset of motor phenotypes until cerebellar pathology presents. Created with BioRender.com.

Partially redundant signaling pathways could also provide a means for early compensation of neurogenic defects. In PD, ventral midbrain (VM) development is crucial for production of the DA neurons whose death ultimately drives disease progression [56]. In the developing VM, Wnt1 signaling from the isthmic organizer initiates expression of LIM homeobox transcription factor 1 alpha (LMX1A), a transcription factor seen in the early development of DA neurons that is important for the expression of late developmental genes of the DA lineage (including nuclear receptor related 1 (NURR1) and neurogenin 2 (NGN2)) [57]. However, in the case of altered Wnt signaling, sonic hedgehog (SHH) signaling in the floorplate can induce forkhead box A2 (FOXA2) expression, which can also promote expression of NURR1 and NGN2 (Figure 2) [58]. Therefore, one potential way of compensating for misregulated Wnt signaling is through upregulated SHH signaling [59]. Activation of SHH in a mouse model of PD helps protect adult dopaminergic neurons from cell death [60], indicating a similar compensatory mechanism could also be happening during early development.

Figure 2. Mutations that affect Wnt signaling could be compensated for by enhanced SHH signaling.

Wnt signaling through LMXA1 is necessary for driving transcription of late developmental genes for the DA lineage, including NURR1 and NGN2 (via Msh homeobox1 (MSX1)). However, disruption in Wnt signaling could be compensated for by activation of FOXA2 through the SHH pathway, which can also activate NURR1 and NGN2. Created with BioRender.com.

Early compensation of AD phenotypes could be regulated similarly to disease relapse, remission, and progression in multiple sclerosis (MS). MS is characterized by inflammatory lesions of damaged myelin in the brain [61]. Interestingly, microglia can both initiate formation of these lesions by promoting inflammatory signaling and initiate repair of these lesions by promoting myelin repair [61]. What ultimately shifts the balance toward inflammatory signaling over myelin repair once MS becomes progressive remains unknown [61], but it has been proposed that a similar balance of positive and negative microglial actions is occurring in AD [62]. An important role of microglia early in AD progression is phagocytic clearance of Aβ through APOE/TREM2/ITAM/SYK-mediated signaling (Figure 3A) [62–64]. However, increased Aβ levels can alternatively activate TLR4/NF-κB/complement signaling in the microglia, which leads to release of pro-inflammatory cytokines and neurotoxic factors that promote synapse engulfment and neuronal loss (Figure 3B) [62, 65–68]. In this way, phenotype presentation could be delayed through microglial clearance of Aβ, but once a threshold of Aβ burden is reached, complement signaling and neuronal death would predominate. A more comprehensive understanding of these potential compensatory pathways will help best establish the full etiology of neurodegenerative diseases.

Figure 3. Changes in microglia function from low to high Aβ burden is a potential compensatory mechanism in AD.

(A) In times of low burden, Aβ binds APOE, which can activate triggering receptor expressed on myeloid cells 2 (TREM2)/immunoreceptor tyrosine-based activation motif (ITAM)/spleen tyrosine kinase (SYK)-mediated signaling. This then activates calcium (Ca2+) and mitogen-activated protein kinase (MAPK) signaling and actin remodeling to promote phagocytic uptake of Aβ by the microglia. (B) In times of increased burden, Aβ can activate toll-like receptor 4 (TRL4)/nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) signaling, which in turn activates complement signaling (including release of complement component 1q (C1q)) and release of pro-inflammatory cytokines (including tumor necrosis factor alpha (TNFα), interleukin 6 (IL-6), and interleukin 1 beta (IL-1β)) and neurotoxic factors, such as inducible nitric oxide synthase (iNOS). Created with BioRender.com.

Alternative model systems

While established mouse models with defects in the above pathways can phenocopy many aspects of human neurodegenerative disease, the majority of these models involve expression of mutant proteins later in life. However, determining the neurodevelopmental components of each disease requires thinking outside the box and identifying alternative models that present both neurogenic and neurodegenerative defects. These models could be used to demonstrate mechanistically how early neurogenic defects can be translated into late-onset neurodegeneration and to help identify new compensatory mechanisms. Interestingly, a number of them could come from the methyltransferase-like (METTL) family of methyltransferases, whose members have recently emerged as key players in both NSC regulation and age-related neurodegeneration [69–73].

METTL3 and METTL14, are the main enzymes responsible for N6-methyladenosine (m⁶A) formation. m⁶A is a one of the most common mRNA modifications and is involved in alternative splicing, nuclear export, transcript stability, and translation [74]. m⁶A is highly abundant in the brain, and plays roles in both NSC regulation and neurodegenerative pathways [75–77]. METTL3 and METTL14 complex together, where METTL3 serves as the catalytically active subunit and METTL14 serves as a catalytically inactive regulator [78]. METTL14 enhances METTL3 catalytic activity and provides a binding site for substrates [78, 79]. Both METTL3 and METTL14 are important for RGC development in the embryonic brain, where they prevent the expression of late-lineage specific genes that would otherwise push the cell towards premature differentiation and deplete the stem cell pool [76]. Accordingly, their loss of function in NSCs leads to developmental defects, especially in the cerebellum, where severe hypoplasia is seen, specifically loss of the granule cell and internal granule layers [80].

New research has also shown a role for m⁶A in the development of HD, PD, and AD. In HD, expression of mHTT in mice results in alteration of m⁶A patterns, including demethylation of important components of synapse organization [71]. It was also found that METTL14 mRNA levels modulate in the hippocampus of WT mice after training exercises, but not with mHTT expression, and increasing m⁶A levels in the hippocampus of mutant mice restored memory deficits [71]. Global m⁶A is down-regulated in the striatum of rat PD models and decreasing m⁶A levels in dopaminergic cells elevates oxidative stress and promotes apoptosis [72]. Soluble Aβ causes reduced METTL3 expression, and accordingly, METTL3 expression is significantly downregulated in human AD brains [81]. METTL3 knockdown in the hippocampus of adult mice leads to memory defects, gliosis, spine loss, and neurodegeneration due to increases in oxidative stress and premature cell cycle activation, and overexpression of METTL3 rescues synaptic damage and cognitive impairment [77]. Altogether, METTL3, METTL14 and the m⁶A modification are critical during early development for proper NSC function and also play critical roles during neurodegenerative diseases, indicating they may be useful models for tracing potential neurodevelopmental origins of neurodegenerative diseases.

Another family member, METTL11A (also called NRMT1/NTMT1) is highly expressed in the brain and also plays important roles in neurogenesis and neurodegeneration [73, 82]. METTL11A is an N-terminal methyltransferase that is predicted to trimethylate over 300 substrates, including RB, SET, RCC1, DDB2, CENPA, and CENPB, based on their N-terminal consensus sequence [83, 84]. N-terminal regulation regulates protein-DNA interactions, protein-protein interactions, and protein stability [85–87]. METTL11A knockout (Mettl11a^−/−) mice display phenotypes associated with premature aging, such as kyphosis, premature graying, dermal fibrosis, and polycystic ovaries [82]. METTL11A is also part of a spinal-specific super enhancer that regulates response to injury, and its knockdown alleviates behaviors associated with neuropathic pain [88].

In the brain, Mettl11a^−/− mice display both developmental and neurodegenerative phenotypes. Early in development, Mettl11a^−/− mice exhibit significant misregulation of NSC quiescence. First, there is an increase in the RGC pool at P0, likely resulting from abnormal embryonic proliferation [73]. This is followed by a robust expansion of the intermediate progenitor cell (IPC) pool around postnatal day 14 (P14) at the expense of the quiescent RGC pool, which becomes depleted [73]. The increased number of IPCs leads to a concomitant increase in the number of migrating neuroblasts, which appear to fully differentiate into neurons [73]. However, these neurons slowly begin to undergo apoptosis by 6 weeks [73]. Striatal neurodegeneration becomes detectable at this time, but significant neurodegeneration in the dentate gyrus (DG) and CA3 regions of the hippocampus is not detectable until three months, and this worsens with age [73]. Mettl11a^−/− mice also have corresponding behavioral deficits such as hyperactivity and impairments in short- and long-term memory [73], indicating they too could be a useful model for linking neurodevelopmental defects with later neurodegeneration.

Conclusions

In summary, while only HD has been definitively shown to have a neurodevelopmental component, data suggest other neurodegenerative diseases, such as PD and AD, may also have a developmental origin. Key proteins from each disease have both developmental and neurodegenerative roles and many similar signaling pathways are utilized during both processes. Compensatory mechanisms that could delay onset of phenotype presentation are beginning to be discovered, but to truly understand if PD and AD have developmental components, new model systems need to be developed for determining the different mechanistic links between early neurogenesis and neurodegeneration and identifying new compensatory mechanisms.

Perspectives.

• Though age-related damage is thought to be the single biggest contributing factor to neurodegenerative disorders, it has recently been demonstrated that Huntington’s disease has a neurodevelopmental component, and this could be true for Parkinson’s and Alzheimer’s diseases as well.

• The delayed onset of phenotype presentation in age-related neurodegenerative disorders could be due to the ability to mask neurodevelopmental phenotypes through different compensatory mechanisms.

• Utilization of new mouse model systems that exhibit both misregulation of neural stem cell development and play roles in neurodegeneration will be useful for understanding the mechanistic links between the two and identifying new compensatory mechanisms.