Protein seeding in Alzheimer’s disease and Parkinson’s disease: Similarities and differences

Abstract

Neurodegenerative pathology can be seeded by introduction of misfolded proteins and peptides into the nervous system. Models of Alzheimer’s disease (AD) and Parkinson’s disease (PD) have both demonstrated susceptibility to this seeding mechanism, emphasizing the role of misfolded conformations of disease-specific proteins and peptides in disease progression. Thinking of the amyloidogenic amyloid-beta peptide (Aβ) and alpha-synuclein (α-syn), of AD and PD, respectively, as prionoids requires a comparison of these molecules and the mechanisms underlying the progression of disease. Aβ and α-syn, despite their size differences, are both natively unstructured and misfold into β-structured conformers. Additionally, several studies implicate the significant role of membrane interactions, such as those with lipid rafts in the plasma membrane, in mediating protein aggregation and transfer of Aβ and α-syn between cells that may be common to both AD and PD. Examination of inter-neuronal transfer of proteins/peptides provides evidence into the core mechanism of neuropathological propagation. Specifically, uptake of aggregates likely occurs by the endocytic pathway, possibly in response to their formation of membrane pores via a mechanism shared with pore-forming toxins. Failure of cellular clearance machinery to degrade misfolded proteins favours their release into the extracellular space, where they can be taken up by directly connected, nearby neurons. Although similarities between AD and PD are frequent and include mechanistically similar transfer processes, what differentiates these diseases, in terms of temporal and spatial patterns of propagation, may be in part due to the differing kinetics of protein misfolding. Several examples of animal models demonstrating seeding and propagation by exogenous treatment with Aβ and α-syn highlight the importance of both the environment in which these seeds are formed as well as the environment into which the seeds are propagated. Although these studies suggest potent seeding effects by both Aβ and α-syn, they emphasize the need for future studies to thoroughly characterize “seeds” as well as analyze changes in the nervous system in response to exogenous insults.

Key Words: Alzheimer’s disease, Parkinson’s disease, Prionoid, Seeding, Propagation, Protein misfolding

Core tip: The disease-specific proteins of Alzheimer’s and Parkinson’s disease show many similarities as prion-like seeds in the brain. In addition to sharing structural features as misfolded proteins, the interactions and mechanisms that underlie the propagation of these proteins may also be shared, hijacking natural cellular responses in ways not unlike those of pore-forming toxins. Differences in temporal and spatial patterns of disease progression stems from the existence of conformational variants. Misfolded proteins that can be generated in vitro, can seed widespread pathology in non-transgenic animal models and question our understanding of disease progression in neurodegenerative diseases.

INTRODUCTION

The purpose of this review is to examine patterns that are present in the progression of Alzheimer’s disease (AD) and Parkinson’s disease (PD), with a focus on identifying parallel mechanisms that underlie disease as well as differences that characterize each disease classified as neurodegenerative diseases, both exhibit neuronal loss/dysfunction in various regions of the central nervous system (CNS). Observations made from analyses of tissue from patients as well as animal models continue to provide evidence that the pathological route taken by both AD and PD is related to their disease-specific proteins. The amyloidogenic proteins, amyloid beta peptide (Aβ) in AD and alpha-synuclein (α-syn) in PD[1], have recently been shown to propagate throughout cerebral networks as misfolded seeds. This has likened the spread of pathology to that of prion diseases, which exhibit templating[2], a process where misfolded proteins can induce endogenous protein to misfold.

AD and PD are the two most prevalent neurodegenerative diseases worldwide. Neither race nor gender is spared by either disease and cures for both have remained elusive. By 2010, 35.6 million people were living with AD, a number that is expected to double in the following two decades. PD has a worldwide prevalence estimated at 7 million patients.

AD is the most common cause of dementia and is characterized by a progressive loss of cognitive function, most noticeably in the form of memory loss. This can be followed by further dysfunction in language, visuospatial and executive systems. Hallmarks of AD include the presence of Aβ-containing amyloid (senile) plaques in the extracellular milieu of the CNS and surrounding blood vessels as well as neurofibrillary tangles (NFT) produced by intracellular aggregation of the microtubule associated protein Tau. PD is the most common cause of movement disorders and exhibits clinical symptoms that can be divided into two groups: the more apparent motor symptoms such as resting tremors, bradykinesia and loss of postural reflexes; and the non-motor symptoms such as olfactory dysfunction, sleep disturbances and depression[3]. The pathology is characterized by the loss of dopaminergic neurons in the substantia nigra pars compacta (SNpc) of the midbrain and decreased levels of dopamine in regions of the striatum where SNpc neurons project. PD is also characterized by formation of α-syn-containing intracellular inclusions known as Lewy bodies and Lewy neurites[4]. The detection of a combination of amyloid plaques, neurofibrillary tangles and Lewy bodies in patients, signifying overlapping pathology of AD and PD, have led to the classification of the conditions Dementia with Lewy Bodies and Parkinson’s Disease Dementia; Dementia with Lewy Bodies patients demonstrate clinical dementia within 1 year of being diagnosed with PD while PD Dementia patients exhibit dementia at later stages of PD[5]. Both AD and PD are chronic illnesses in which clinical symptoms may manifest many years after pathology has commenced. In AD, the accumulation of Aβ protein in the brain primarily occurs before the appearance of the first clinical symptom[6,7]. Lewy bodies of PD also typically appear before the appearance of a classic motor symptom[8,9]. Despite linkage of both diseases to mutations in certain genes, the majority of cases are sporadic in nature.

AMYLOIDOGENIC PROTEINS

Protein folding is a fragile event that requires specific environmental conditions. Subcellular processes exist to regulate and to ensure the folding and stabilization of newly generated proteins into native conformations. Maintenance of pH, temperature and osmolarity as well as ensuring appropriate post-translational modifications of folded proteins ensure the stability required to perform the intended function. Disruption of these processes enhances the rate at which proteins adopt alternate secondary and tertiary structures, which in turn may favor the production of misfolded conformers. It is now thought that subpopulations of misfolded proteins can act as pro-aggregatory seeds that initiate a cascade of pathology in susceptible neuronal populations and may represent the catastrophic event from which many neurodegenerative pathologies progress[10]. Misfolded structures accumulate leading to their deposition into insoluble lesions that spread throughout the central nervous system over time. These lesions can appear both intracellularly and extracellularly and are implicated in the development of neurodegenerative disease states. Diseases such as AD, PD and Amyotrophic lateral sclerosis are characterized by the presence of these deposits which are thought to spread in a prion-like manner[11].

In AD and PD, supramolecular structures, amyloid fibrils, are formed through the accumulation of subunits rich in β-pleated sheet secondary structure (Figure 1). The disease-specific proteins, Aβ[12] and α-syn[13] have been shown to adopt secondary structures that allow for the formation of the cross β-structure characteristic of amyloid[1]. The monomeric proteins undergo a nucleation-dependent mechanism of aggregation to form oligomers, protofibrils and other intermediate structures. Oligomers formed on the amyloid aggregation pathway will eventually reach a fibrillar state while off-pathway aggregates typically reach equilibrium and terminate aggregation in a prefibrillar state. Physicochemical properties such as hydrophobicity, secondary structure propensities, charge and aromatic interactions underlie the propensity of these proteins and peptide fragments to aggregate[14]. For both Aβ and α-syn monomers, stabilization of a partially folded pre-molten globule-like state, either by lowering of pH or increase in temperature[15,16], is believed to be the critical first step in the formation of the fibrils[17]. Fibrils are metastable structures that, once formed, may represent lower energy states than the native protein structure. The stability is strengthened by dehydration of misfolded species, a process promoted by binding of ligands that render the protein less solvent-accessible[18]. Surfaces that promote clustering of Aβ and α-syn, then, likely promote fibrillization. Several events promote protein misfolding and must be actively countered.

Figure 1 Fibrillar amyloid-beta peptide and alpha-synuclein. A and B: Amyloid-beta (10 mg/mL) was aggregated at room temperature for 72 h with shaking; C and D: α-synuclein (2 mg/mL) was aggregated by incubation at 37 °C for 144 h with shaking. Aggregation leads to formation of fibrils with varying structures. Fibrils were negatively stained with 2% phosphotungstic acid and their formation was confirmed using Transmission Electron Microscopy at 75 kV. Images display 40000 × magnification.

Aβ ranges in length from 36-43 amino acid residues and is formed by the cleavage of the single transmembrane protein amyloid precursor protein (APP). The actions of two proteolytic events, the enzyme beta (β) secretase and the complex gamma secretase, generate Aβ_40/42. These sequential cleavage events are localized to lipid bilayers where APP and the proteases are present; the plasma membrane as well as the membranes of golgi apparatus and endoplasmic reticulum have been shown to generate Aβ peptides[19-21]. As Aβ₄₂ is more hydrophobic and, as such, more fibrillogenic[22] than Aβ₄₀, processes that lead to higher ratios of Aβ₄₂:Aβ₄₀ promote the development of amyloid. Mutations in several genes encoding proteins involved in the production of Aβ, including APP, PS1 and PS2, promote this pathological ratio and are linked to familial, early-onset AD. Aβ fibrils cluster to form the extracellular cerebral amyloid plaques that are observed in AD[23].

α-Syn differs from Aβ in that it is a full length protein and is significantly larger (14 KDa vs 4 KDa for Aβ). It consists of 140 amino acid residues and is localized to the presynaptic terminals of neurons in most brain regions[24]. Missense mutations in the α-syn gene, SNCA, (A53T, A30P and E46K) are linked to autosomal dominantly-inherited early-onset PD[25-27]. Furthermore, overexpression of α-syn by multiplication of SNCA, showed accelerated formation of Lewy-like structures in SHSY-5Y neuroblastoma cells, of which fibrillar α-syn is the main component[28]. In vivo mouse models of α-syn overexpression have been developed. Mice primarily display an increase in Lewy body formation in neuronal populations that are dictated by the nature of the transgene incorporation[29]. Models with higher levels of α-syn overexpression may also display dopaminergic cell loss in the SNpc and motor deficits associated with this loss[29]. Models of transgenic mice expressing human α-syn with the A53T mutation display α-syn aggregation and inclusion formation; however, these mice do not show any degeneration in dopaminergic populations in the SNpc[30]. Although the native functional role of α-syn is currently unknown, many studies have pointed to the involvement of α-syn in synaptic transmission by regulating neurotransmitter release[31]. Involvement in synaptic transmission has also been suggested for Aβ where it may be involved in feedback regulation of neuronal activity and modulation of glutamatergic and cholinergic inputs[32,33].

In addition to Aβ and α-syn, aggregation of tau protein is implicated in neurodegeneration seen in AD and other tauopathies. Hyperphosphorylation of tau leads to formation of intracellular insoluble neurofibrillary tangles, the presence of which correlates with the clinical manifestation of dementia in AD[6]. The colocalization of α-syn inclusions and tau neurofibrillary tangles in some neurodegenerative diseases may be due to a common mechanism. In vivo studies have suggested seeding of both proteopathic species occurs by specific strains of fibrillar α-syn[34]. Mouse models have also shown acceleration of neurofibrillary tangles formation by injections of fibrillar Aβ[35] and overexpression of mutant APP[36,37]. Interestingly, prior studies have also established an interaction between α-syn and Aβ. Both Aβ and α-syn have been shown to seed each other’s aggregation in vitro[38] and a stretch of hydrophobic amino acid residues (61-95) found in the central portion of α-syn has been termed the non-Aβ component of AD amyloid (NAC) after discovery in Aβ plaques[39]. Both are thought to be natively unstructured and flexible monomers with brief but residual secondary structure[40-42]. Aβ has a central hydrophobic region (17-21) that permits strong intermolecular interactions between monomers and provides the peptides with an enhanced propensity to aggregate[40]. α-Syn contains the central hydrophobic NAC region, which forms intermolecular interactions[43]. Despite a lack of homology in amino acid sequence, several amyloidogenic proteins and peptide fragments are thought to share common structural features as oligomers[44]. This commonality is supported by the inhibition of amyloid formation of many peptides/proteins by a single compound. Inositol stereoisomers such as scyllo-inositol have been shown to stabilize non-toxic forms of Aβ₄₂[45,46] while also attenuating Aβ[47] and α-syn[48] mediated toxicity in vitro. Polyphenols have also been shown to inhibit fibrillogenesis through aromatic interactions[14] that prevent conversion of large oligomer into fibrils or further elongation of pre-existing fibrils[14]. Many additional classes of inhibitors have been shown to promote oligomeric forms of both Aβ and α-syn that do not effect cell viability[49].

HISTOLOGICAL SPREAD

Histological studies of tissue from AD and PD patients have identified patterns of pathology that can be classified by a staging scheme[6,8]. Despite notable inter-individual variation at early stages of AD, a three stage scheme was established following the consistent spread of amyloid deposition from basal isocortical areas to the hippocampal formation[6]. Also developed was a six stage scheme that correlates the preclinical and clinical progression of dementia with the increasing presence of NFTs[6]. Current AD diagnoses take measures of amyloid load into account along with NFT spread and the extent of dementia symptoms exhibited. These measures are compared to reference ratios that are predicted by age to diagnose individuals as having a high, intermediate or low likelihood of AD[7]. For PD, Braak et al[8] developed a six stage scheme for disease progression starting with the presence of intraneuronal inclusions in the dorsal (X) motor nucleus and proceeding upward through the midbrain culminating in the cortex. There is no standard diagnostic test for PD, with diagnoses typically relying on a combination of neurological examination to confirm α-syn pathology and SNpc dopaminergic neuron loss along with examination of motor symptoms. Currently there is intensive research on identifying accurate biomarkers for early detection of AD and PD[3,4,50-52].

Both AD and PD pathology follows a pattern of progression that is spatiotemporally regulated with pathological spread into interconnected regions. As pathological hallmarks of AD and PD increase within the brain, patterns of spread for each disease rarely differ in origin and direction, especially in familial forms of disease. This has lent credence to the recent hypothesis that Aβ and α-syn pathology spreads by transfer of aggregates to interconnected regions. Newly taken up aggregates, whether oligomeric or fibrillar, can then induce templating of endogenous peptide/protein and lead to elevated levels of pathological conformers that alter cellular homeostasis. This is in line with the proposal that Aβ and α-syn should be termed “prionoids”[53,54]; amyloids that exhibit prion-like transmission from cell-to-cell but have not yet proven transmissible between individuals and are not tractable by microbiological techniques[54].

MOLECULAR INTERACTIONS

The lipid bilayer of the plasma membrane exhibits dynamic features that are responsible for interaction with the extracellular environment. The composition of the plasma membrane is heterogeneous and is known to contain specialized domains that play a significant role in neuropathological mechanisms due to protein-lipid interactions. Protein-lipid interactions can modulate aggregation by recruiting proteins to increase local concentration, favoring conversion into a partially-folded state (aggregation prone) and by modulation of the depth that the protein/peptide penetrates the membrane affecting nucleation propensity. Anionic charges on surface head-groups first attract these peptides and act as conformational catalysts for amyloidogenic fibrillization. Polyunsaturated fatty acids concomitantly enhance interactions by increasing membrane fluidity, favoring association and insertion of peptides/proteins such as Aβ/α-syn[55] into the transmembrane. Initially, Aβ[46] and α-syn[56,57] were shown to interact with phosphatidylinositols, aiding in the binding to the membrane and enhancing aggregation[57]. More recent studies highlighted the importance of lipid rafts in the aggregation pathway, specialized micro-domains that are rich in cholesterol and sphingolipids such as gangliosides[58]. Gangliosides-Aβ interaction drives aggregation into toxic oligomers[59] by promotion of α-helical secondary structure that eventually converts to β-sheets[60]. The ganglioside GM1, the most abundant ganglioside in lipid rafts, is enriched in neuronal membranes, and compared to other gangliosides, demonstrated the greatest seeding of Aβ oligomers[60]. Seeding was potentiated by clustering of the lipid moiety, which was further stabilized by increasing cholesterol concentrations in the membrane[61,62]. Surprisingly, the more recent literature argues for the protective role of sterols through inhibition of Aβ fibrillization and toxicity[63,64] (reviewed by Bucciantini)[65]. As cholesterol decreases membrane fluidity, it likely hinders insertion and nucleation of Aβ and α-syn at the surface[65]; although these effects are not sufficient to counter GM1 interactions[66]. α-Syn interacts with GM1, which promotes conversion into an oligomeric α-helix rich structure, but in contrast to Aβ, prevents α-syn from aggregation into fibrils[67]. GM1 is also involved in endocytotic internalization of monomeric α-syn into microglial cells similar to previously described internalization of bacterial toxins[68]. Together, it is apparent that membranes play an important role in the misfolding of amyloidogenic proteins.

CELL TO CELL TRANSFER

Consistent evidence of spread of AD and PD pathology to interconnected regions of the CNS suggested cell-to-cell transfer of disease-related proteins. The presence of Aβ and α-syn in extracellular fluids supports the existence of mechanisms that release protein/peptides extracellularly. In vivo dialysis of human CNS revealed the presence of Aβ in the CSF[69]. Furthermore, Aβ is present in blood plasma[70] and recent studies suggest that pathological forms of Aβ (Aβ₄₂) in the CSF have predictive value as biomarkers[51]. α-Syn monomers are also detected in extracellular biological fluids like the cerebrospinal fluid[71] and blood plasma and oligomeric forms show elevated levels in the fluids of PD patients[50]. Careful understanding of the mechanisms that underlie how or why cells react to pathological forms of Aβ and α-syn may explain this spread and identify methods to limit it.

In vitro studies have shown that small amounts of Aβ can be transported outside cells via fusion of peptide containing multivesicular bodies (MVBs) with the plasma membrane followed by the release of Aβ-carrying exosomes[72]. More recent evidence showed that soluble oligomeric forms of Aβ undergo cell-to-cell transfer between directly connected rat primary neurons and human SHSY-5Y cells in donor-acceptor co-culture models[73]. Transfer failed when cells didn’t form direct connections, identifying the importance of connectivity and supporting observations that prion-like transfer in vivo occurs between interconnected neuronal networks[74-76]. Although differing oligomeric Aβ isoforms (Aβ₄₀/Aβ_11-42/Aβ₄₂/Aβ_3(pE)-40), including amino-terminally truncated pyroglutamylated Aβ (pE-Aβ) were transferred between cells, the more cytotoxic Aβ₄₂ was transferred to a greater extent[77]. This transfer appeared to be an early response to stress in degradation and clearance pathways and occurred more robustly with increased accumulation of non-degradable forms of the peptide[77]. Interestingly, protofibrillar/fibrillar forms of Aβ were more easily degraded and cleared by the lysosomal machinery and thus undergo less transfer than oligomeric forms[77]. Earlier studies also showed that the majority of internalized Aβ was found in compartments that compose the endocytic pathway, specifically late endosomes and lysosomes, and that uptake is dependent on endocytosis[19]. Despite these results, there remains many questions in regard to the mechanisms that underlie Aβ transfer and release into the extracellular milieu.

More progress has been accomplished in elucidating mechanisms of α-syn transfer between neurons. Several studies have shown that exocytotic/endocytotic mechanisms[78-80] as well as exosomes[81,82] are used by cells to externalize and internalize α-syn. Lee et al[80] showed that despite α-syn generally being localized within the cytosol, small amounts of α-syn are constitutively released extracellularly, in both native and pathological forms. Exocytosis of α-syn is via a non-classical vesicular transport system independent of the endoplasmic reticulum and Golgi apparatus. Prior to release, α-syn was detected in the lumen of intracellular large core dense vesicles where the intravesicular α-syn fraction was more aggregation-prone than the cytosolic fraction[80]. Another study by Desplats et al[79] demonstrated α-syn transfer between co-cultured neuronal populations in the absence of membrane leakage, although, transfer was more robust in the presence of neuronal degeneration. Furthermore, α-syn uptake was mitigated in dominant negative mutant acceptor cells that cannot form endocytotic vesicles[79]. Co-localization of internalized α-syn with endosomal markers was detected in vitro[79] and in vivo[78] and provided support for the reliance of cells on endocytosis for uptake of extracellular α-syn. Vesicular export of α-syn was further studied in vitro and demonstrated that, similar to Aβ, exosomes were responsible for the release of a portion of the α-syn detected extracellularly[82,83]. Exosomes were found to be contain oligomeric α-syn both bound to the outer membrane and within lumenal fractions. Exosomes containing α-syn were more readily taken up by cells than lipid-free α-syn oligomers[83]. Additionally, induction of lysosomal dysfunction, using lysosomal inhibitors, potentiated the release of exosomes associated α-syn oligomers[81,83] while enhancement of autophagy diminished α-syn release[83]. Cellular conditions that lead to accumulation of α-syn, particularly pathological forms, favors extracellular release. Conversely, packaging in vesicles may explain how misfolded proteins enter naïve cells undetected.

Since the discovery of Lewy bodies in long-term fetal nigral tissue grafted into the striatum of PD patients[84-86], studies have subsequently shown α-syn transfer in rat models of PD. Angot et al[78] injected a viral vector into the SNpc of Sprague-Dawley rats to overexpress human α-syn in local neurons. Three weeks post-transfection, dopaminergic neurons were grafted intrastriatally and showed uptake of human α-syn one week later. Uptake was more robust in rats sacrificed at later time points (2 and 4 wk post-graft)[78]. The study also demonstrated, in vivo, the presence of human α-syn in endosomal compartments as well as core regions of cytoplasmic inclusions[78], supporting previous in vitro results that exogenous α-syn recruits endogenous protein to aggregate[87]. It has become apparent that trans-synaptic interneuronal transfer of Aβ and α-syn occurs in various conformations and structures. This identifies a relatively similar mechanism by which amyloidogenic proteins may propagate through the nervous system in AD and PD. Further investigation into whether this process is exclusive to pathological states, as in, whether this transfer is a cellular response to aberrant structures, is important for the development of therapeutic strategies targeting amyloid transfer.

MECHANISM OF TRANSFER

Transfer of proteins between cells is common however the mechanism that underlies transfer of pathologic species is not fully elucidated. Notably, certain forms of both Aβ and α-syn have been identified as stressors of the lysosomal degradation pathway. Emerging data implicate this stress in the extracellular release of aggregates. Lysosomal stress by Aβ is not fully understood but recent observations that oligomeric Aβ accumulated in lysosomes[88-90], induced abnormal lysosomal morphology and increased the size of the lysosomes suggests a similar mechanism as that reported for α-syn[77,91]. Internalized α-syn was shown to induce lysosomal rupture and it was suggested that this rupture is a mechanism that contributes to the release of α-syn[91]. Small fractions of these proteins in the cell may be diverted from the lysosomal degradation pathway to MVBs and exosomes for transport out of the cell. This diversion is likely a cellular attempt to rid itself of potentially toxic accumulations and is supported by the highly cytotoxic fractions of α-syn and Aβ located within exosomes[83]. Recent reports have shown that exosomes contain only 1.5% of the released α-syn¹⁴ and as such, it is not the primary route for α-syn release from cells. These results point to the hijacking of toxin-induced cellular responses as a method for the further spread of amyloidogenic toxic species via a mechanism similar to the pore-forming toxins released by the bacterium Staphylococcus aureus (S. aureus)[92] and Bacillus anthracis[93]. S. aureus releases α-hemolysin to form heptameric pores in the membrane, and cellular survival depends on internalization of these pores. Once internalized, non-degraded fractions are sorted into MVBs to be released as exosomes, toxin-carrying exosomes[92]. Anthrax toxin is an AB toxin that relies on its two components, lethal factor and protective antigen, to infiltrate cells and compromise their function[93]. Protective antigen and lethal factor interact at plasma membranes, are endocytosed and interact with the limiting and intraluminal membranes of endosomes, delivering lethal factor to the cytosol and lumen of intraluminal vesicles. This loading partially allows lethal factor to be released within exosomes to deliver the toxin to nearby cells where it enters and rapidly attacks its targets. These mechanisms are analogous to aggregated forms of Aβ and α-syn interacting with plasma membranes, entering cells, and being loaded into exosomes for extracellular release.

Furthermore, pores formed by amyloidogenic proteins are similar in structure to the β-barrels formed by pore-forming toxins[94]. Antibodies that recognize the structure of Aβ and α-syn annular protofibrils also bind heptameric β-barrels of α-hemolysin, indicating that Aβ and α-syn protofibrils share a conformational state with this pore-forming toxin[95]. Formation of these small pores by bacterial toxins elicits endocytosis in an attempt by the cell to maintain survival. Endocytosed toxins are targeted for autophagy but as these pores are resistant to degradation, cells then exocytose vesicles as a protection method. Removal of these membrane interactions is intended to restore homeostasis and prevent cytotoxicity. Studies of Aβ transfer suggest that propagation is not secondary to cytotoxicity and instead occurs as an early response to failure of the cell’s clearance machinery[77]. Endocytosis of membrane disrupting agents, mainly those that form small (approximately 2 nm in diameter) pores, occurs on the scale of hours to remove the pores from the surface[96]. The similarity by which many non-endogenous interactions manipulate a cell’s natural response systems suggests that the propagation of amyloidogenic proteins is aided by lipid membrane interactions that enhance association with the endocytic pathway (Figure 2).

Figure 2 Several mechanisms are proposed to underlie the extracellular release of amyloidogenic proteins from neurons. Lysosomal dysfunction due to uptake and accumulation of proteopathic species that are not degraded promotes the release of misfolded species by exosomes and exocytosis. Uptake by nearby neurons likely occurs via an endocytic mechanism and allows proteopathic species (oligomers and fibrils) to seed the templating of endogenous protein into a misfolded conformation. Pore formation by insertion of protofibrillar amyloid-beta and alpha-synuclein prompts a cellular response to undergo endocytosis and degradation. Inhibition of autophagosome formation may contribute to amyloid deposition as well as transfer of amyloidogenic species.

ANIMAL MODELS

Propagation of amyloid proteins, as measured by the presence of amyloid plaques and Lewy bodies and Lewy neurites, follows a stereotypical pattern[6,8] in AD and PD patients. To explain this, it is hypothesized that Aβ and α-syn transfer through neuronal networks and seed the misfolding of endogenous protein in naïve cells. Several experimental paradigms support this hypothesis, particularly, seeding and propagation of pathology in animals treated with exogenous misfolded protein (Table 1). Prion-like seeding of Aβ was demonstrated by injection of AD brain homogenates into APP-overexpressing mice leading to senile plaque formation near the injection site, 5 mo post-injection[97]. Although this provided the first evidence of Aβ seeding in vivo, it did not provide unequivocal evidence that Aβ was responsible for the seeding of pathology. Contamination of homogenate tissue with other seeding agents such as gangliosides and/or other molecules implicated in amyloid aggregation may have accelerated the aggregation kinetics of endogenous Aβ. Meyer and colleagues, demonstrated the necessity for Aβ peptide after immunodepletion of Aβ from the injected extracts, protein denaturation, or Aβ immunization of APP23 mice attenuated seeding activity in the injected hosts[98]. Use of extracts from different AD mouse models, APP23 mice carrying the Swedish mutation and APP-PS1 mice carrying the Swedish and PS-1 L166P mutations, leads to differing seeding patterns in the hippocampus; APP23 extracts promote formation of primarily diffuse and filamentous lesions while APP-PS1 extracts promote formation of compact, punctate lesions[98]. Synthetic or cell-culture derived Aβ failed to demonstrate seeding altogether. The failure of synthetic Aβ aggregates generated in an in vitro environment to seed was attributed to Aβ peptide polymorphism and the possibility of differing strains of Aβ exhibiting different seeding potential[98]. Langer et al[99] characterized these proteopathic species and revealed that both fibrillar proteinase K-resistant and lower order soluble, proteinase K sensitive Aβ species from the same AD brain extract show seeding activity in vivo. Since different forms of Aβ found in AD brain extracts show more robust induction of pathology compared to synthetic Aβ fractions, it was proposed that endogenous Aβ species contain a structural feature capable of templating endogenous Aβ that is not generated in synthetic Aβ oligomers/protofibrils in vitro. Additionally, the soluble and insoluble fractions of brain extracts differ in the distribution pattern and morphology of aggregates formed within the brain; this is hypothesized to be due to differing propagation/transfer of both species in vivo. Stöhr et al[100] was able to show that fibrillized synthetic Aβ₄₀ and (AβS26C)₂, a mutant Aβ₄₀ peptide that forms covalent dimers and aggregates into neurotoxic fibrils, induced amyloid plaques that propagated throughout the brain from a single ipsilateral injection. Albeit, it was noted that synthetic fibrils were less potent than both AD brain lysates and Aβ purified from these extracts. Purified Aβ reduced seeding time by 40% and was attributed to 10-fold higher levels of pure Aβ than those in extracts, which suggested that the effect was concentration-dependent. Nonetheless, purification of Aβ from in vivo homogenates and recapitulation of results generated by injection of total homogenates demonstrated that Aβ alone was sufficient to generate seeding in vivo. More importantly, seeding by synthetic fibrils indicates that cofactors of aggregation found in vivo are not critical for induction of proteopathic conformations.

Table 1 Seeding of Alzheimer’s disease and Parkinson’s disease-like pathology in mice.

	AD	PD
Misfolded peptide/protein	Aβ peptide	α-syn protein
	τ protein
Pathological deposits	Intracellular (neurofibrillary tangles)	Intracellular (lewy bodies/lewy neurites)
	Extracellular (amyloid plaques)
Exogenous seeding of brain homogenates
Transgenic mouse models	Human AD brain extracts into APP23[98,103,105], P21[102] and HuAPPwt[101]	No reports
	Rodent APP brain extracts into APP23[98,100,103], APP-PS1[98]	Rodent α-syn brain extracts into M83[75,106]
Non-transgenic mice	Human AD brain extracts into Fischer 344 rats-no seeding[102]; into mice-no seeding[97]	Human Dementia with Lewy Bodies brain extracts into C57BL/6 J mice[76]
	Rodent APP brain extracts into C57BL/6 J-no seeding[98]
Exogenous seeding of synthetic peptides/protein
Transgenic mouse models	Aβ40 and AβS26C into APP23[100]	Mouse α-Syn Fibrils into M83[75]
	Aβ40/42 into APP23-no seeding[98]	Human α-Syn Fibrils into M83 and M47[108,109]
Non-transgenic mice	No reports	Mouse α-Syn Fibrils into C57BL/6 J[74,76,107,108]
		Human α-Syn Fibrils into C57BL/6 J[76,107,108]

Aβ: Amyloid-beta; τ: Tau; α-Syn: Alpha-synuclein; AD: Alzheimer’s disease; APP: Amyloid precursor protein; PD: Parkinson’s disease.

More recently, induction of Aβ aggregation in mouse[101] and rat[102] models that express human APP, but never develop AD pathology, was shown after injection of brain extracts from advanced AD patients. Despite never developing pathology during normal aging, the rat model (APP21-transgenic line) expresses a human APP gene carrying the Swedish and Indiana mutations and represents a model of AD-susceptibility. HuAPPwt mice, express the wildtype human APP gene, demonstrated that susceptibility to templating was not dependent on the presence of pathological mutations in Aβ. These results also suggest that murine APP retains resistance to templating, strengthening the importance of identifying conformational variants of amyloid (strains) for future propagation studies. Evidence for the existence of conformational variants of misfolded Aβ was demonstrated when brain extracts from AD patients carrying different mutations (Arctic vs Swedish) or with sporadic AD displayed different patterns of pathology in inoculated APP23 mice[103]. Extracts from all AD cases led to similar deposition of parenchymal Aβ plaques but differences were detected in cerebrovascular deposition. Patterns of Aβ deposition that surround blood vessels and spread into the parenchyma in mice inoculated with Arctic mutation-containing extracts; contrastingly, extracts from sporadic and Swedish mutation-carrying cases displayed a thin, compact layer of Aβ surrounding vessels, with the exception of a sporadic case displaying a combination of phenotypes[103]. These differences were maintained when homogenates from mice that had developed pathology post-inoculation were injected into younger mice, demonstrating serial transmission of seeding that was dependent on the properties of the seeds[103]. Interestingly, transgenic mice overexpressing the Arctic and Swedish mutations show the spreading cerebral amyloid angiopathy phenotype as compared to mice carrying only the Swedish mutation, which show thin, compact layers of Aβ perivascularly[104]. This suggests that the Arctic mutation promotes a dominant phenotype of cerebral amyloid angiopathy. Fritschi et al[105] further supported the existence of Aβ variants when it was shown that Aβ seeds derived from soluble AD brain extracts, but not Aβ CSF fractions, led to Aβ plaque deposition in APP transgenic mice. Despite Aβ levels being significantly higher in the CSF, they failed to seed pathology and this was attributed to their processing; CSF Aβ was not N-terminally truncated like Aβ peptides found in the soluble brain extract[105].

Propagation studies investigating prion-like behavior of α-syn have also demonstrated seeding properties. Mougenot et al[106] injected M83 mice (mouse model of PD expressing α-syn with the A53T mutation) with brain homogenates from older M83 mice that developed intracellular α-syn inclusions and showed that this inoculation accelerated the appearance of synuclein pathology while shortening lifespan. Luk et al[75] showed that young M83 mice inoculated with both lysates of aged M83 mice and synthetic α-syn fibrils developed accelerated pathology, further demonstrating that seeding of pathology was solely dependent on α-syn in vivo. Unilateral injections into the neocortex and striatum led to formation of inclusions containing phosphorylated endogenous α-syn in regions directly and indirectly connected to the injection sites such as the frontal cortex, thalamus, hypothalamus, substantia nigra pars compacta, locus coeruleus, cerebellar nuclei, and the spinal cord. The pathology was bilateral and phosphorylation of α-syn at serine 129 (pSer129) was used as an indicator of pathological fibrillar α-syn. Hyperphosphorylated α-syn-containing inclusions were detected in the SNpc and were accompanied by diminished tyrosine hydroxylase staining. These observations suggested that misfolded wildtype α-syn underwent interneuronal transfer, seeded the formation of intracellular inclusions by recruiting endogenous α-syn, and initiated neuronal degeneration in dopaminergic neurons. It was also demonstrated that induction and propagation of PD-like pathology occurs in wildtype mice[74]. Pathology, as measured by inclusion formation and dopaminergic neurodegeneration, was detected as early as 30 d post-injection and was progressive becoming more severe at 90 d in both transgenic and wildtype mice[74,75]. Masuda-Suzukake et al[76] used human synthetic α-syn fibrils and insoluble fractions of α-syn from patients with Dementia with Lewy Bodies for injection into SNpc of wild type mice. Pathology was detected in regions connected to the injection site, starting at 90 d post-injection and showing maximal severity at 15 mo. Injection of fibrils into the SNpc led to intracellular inclusions appearing in the striatum, amygdala, stria terminalis and dentate gyrus[76]. Masuda et al[76] suggests that the detection of pathology in hippocampal regions, unlike mice injected into the neocortex and striatum, which do not display pathology in the hippocampus, are likely dependent on the injection site. Masuda-Suzukake et al[107] confirmed the importance of the injection site in determining regional transfer of pathology by injecting wildtype mice in three different regions and showing differential spread of pathology. Within one month of inoculation, α-synuclein spread from the substantia nigra to the amygdala and stria terminalis; from the striatum to the amygdala, substantia nigra and throughout the cortex; and from the entorhinal cortex to the dentate gyrus and CA3 regions of the hippocampus, the fimbria and the septal nucleus[107]. Other differences, such as rate of inclusion formation and degeneration of SNpc neurons may be linked to fibril length. Luk et al[75] used fibrils fragmented by sonication whereas Masuda and colleagues initially used full-length fibrils. Masuda-Suzukake et al[107] used fragmented fibrils in later experiments and see pathology as early as 1 mo post-injection. Shorter fragments may increase the surface area available for interaction with the cell leading to accelerated seeding process. Interestingly, in both studies, wildtype mice presented with very similar pathological distribution regardless of whether disease brain lysates or synthetic fibrils were used. This suggests that, in contrast to Aβ, α-syn fibrils displayed limited conformational heterogeneity since in vitro conditions are sufficient to create seeds with equivalent potency to those generated in vivo. Interestingly, a study by Sacino et al[108] questions the propagation of pathology in wildtype mice by injection of wildtype α-syn preformed fibrils. Initially, it was shown that injecting human α-syn preformed-fibrils in M83 mice hippocampi led to widespread inclusion formation throughout the CNS, including the hippocampus, striatum and cortex, three regions that typically do not develop pathology in M83 mice. However, detection of pathological α-syn in white matter tracts was absent conflicting with previous observations[74-76]. Furthermore, injection of both human and mouse α-syn fibrils into non-transgenic mice led to limited inclusion formation that was restricted to the injection sites. Inoculations were bilateral into the hippocampi and cortices of 2 mo old mice and detection of pathology at increasing time points post-injection showed decreasing immunoreactivity of pathological α-syn specific antibodies. The study attributes previously seen widespread staining to cross-reactivity of phospho-specific antibodies targeting phosphorylated α-syn (pSer129) with phosphorylated neurofilament-L (pSer473 NFL)[108]. More recently, Sacino and colleagues injected soluble (Δ71-82) and fibrillar human α-syn into M20 mice (overexpressing Human wildtype α-syn) that do not typically develop Parkinson-like pathology, as measured by presence appearance of Lewy bodies, in their lifespan[109]. By 4 mo post-injection, inoculated mice developed α-syn pathology at the injection site, the hippocampus, as well as the cortex, striatum, midbrain, and brainstem[109]. Induction of Lewy body formation using soluble species that do not fibrillize, implicates mechanisms other than templating in the disturbance of protein homeostasis. Furthermore, results suggest that overexpression of WT α-syn compromises resistance to templating and transmission since similar inoculations into non-transgenic mice failed to show spread to other regions[108]. As such, replication of studies between laboratories will be necessary to distinguish pathological seeding with α-syn.

The variability in results reported by different groups attempting to induce propagation of pathology by exogenously added seeds indicates that the mechanisms of propagation are likely complex. Sacino et al[108] suggests that conformational variations are a major factor explaining some of the inconsistencies seen in literature. Additionally, Sacino et al[108] suggests that seeding/transfer of α-syn possibly occurs as a result of the neurotoxic effects of the injected aggregates as well as the disruption of normal proteostatic mechanisms[108], neuroimmune activation and/or injury response[109]. Guo et al[110] reinforce the need for biochemical identification of conformational variants of amyloidogenic peptides and proteins, such as seen with Aβ[111] and α-syn[112,113]. Future studies should not only thoroughly characterize/profile the species prior to injection, but should extensively analyze changes that occur in vivo as a result of amyloidogenic seeds.

AD and PD exhibit many similarities in mechanisms potentially linked to their etiology. As it stands, strong evidence supports amyloidogenic structures seeding native peptides/proteins to pathologically misfold, facilitating the progression of disease. The neurotoxic effects of various aggregate subpopulations have not been characterized sufficiently to understand whether the mechanisms underlying toxicity contributes to the propagation of pathology during disease progression. As well, oligomers, protofibrils and fibrils have been shown to display overlapping functions, such as membrane disruption, despite differences in structure. Different members of the aggregation pathway may share extensive β-sheet structure, but understanding molecular variations in the tertiary and quaternary structures may help in the understanding of their structure-function relationships and how this relationship progresses during aggregation.