Propagation and spread of pathogenic protein assemblies in neurodegenerative diseases

Abstract

Many neurodegenerative diseases such as Alzheimer’s disease, Parkinson’s disease, and amyotrophic lateral sclerosis are characterized by the progressive appearance of abnormal proteinaceous assemblies in the nervous system. Studies in experimental systems indicate that the assemblies originate from the prion-like seeded aggregation of specific misfolded proteins that proliferate and amass to form the intracellular and/or extracellular lesions that are typical of each disorder. The host in which the proteopathic seeds arise provides the biochemical and physiological environment that either supports or restricts their emergence, proliferation, self-assembly and spread. Multiple mechanisms influence the spatiotemporal spread of seeds and the nature of the resulting lesions, one of which is the cellular uptake, release, and transport of seeds along neural pathways and networks. The characteristics of cells and regions in the affected network govern their vulnerability and thereby influence the neuropathological and clinical attributes of the disease. The propagation of pathogenic protein assemblies within the nervous system thus is determined by the interaction of the proteopathic agent and the host milieu.

In 1889, Stephen Paget coined the phrase ‘seed and soil’ to describe how the metastasis of cancer cells is governed by the nature of the cells (the seed) and the site of secondary growth (the soil) ¹. The basic concept remains valid today, and furnishes useful insights into the selective spread of metastatic cancer to other organs. A wealth of research in recent years reveals that many of the most common age-associated neurodegenerative diseases – Alzheimer’s disease (AD), Parkinson’s disease (PD) and amyotrophic lateral sclerosis (ALS) among them – result from the transformation and accumulation of specific proteins within the nervous system. Although mechanistically different, this disease process can be likened conceptually to metastatic cancer, except that the disease agents that proliferate in these degenerative brain disorders are transformed proteins rather than transformed cells. As in the case of cancer cells, the dissemination of abnormal proteins and the nature of the resulting disease depend on both the proteinaceous agent – the seed – and the host milieu – the soil. The archetypical proteopathic seed is the prion, which was first identified as an unconventional infectious agent in a family of uniformly fatal brain diseases of humans and nonhuman species ².

Prions: paradigmatic proteinaceous disease agents

Prions were initially defined as ‘proteinaceous infectious particles’ ², though the definition has since broadened to ‘proteins that acquire alternative conformations that become self-propagating’ ³. The prion diseases include Creutzfeldt-Jakob disease, Kuru, Gerstmann-Sträussler-Scheinker syndrome, fatal insomnias, and variably protease-sensitive prionopathy in humans ^{4, 5}, as well as scrapie in sheep and goats, bovine spongiform encephalopathy in cattle, chronic wasting disease in cervids, and several other nonhuman prionoses ⁶.

Prion diseases are unorthodox in that they can be infectious, genetic, or idiopathic (sporadic) in origin ⁷. They arise when normal cellular prion protein molecules (PrP-Cellular, or PrP^C) misfold, self-assemble, and spread through the nervous system ³. Once established in the living organism, the aberrant proteins propagate by the corruptive templating of like molecules, which are continually produced in the natural course of cellular metabolism. The misfolded, pathogenic PrP molecules (conventionally referred to as PrP-Scrapie, or PrP^Sc) bind to one another and, in a crystallization-like process ⁸, the assemblies grow, fragment, and proliferate, eventually occupying many regions of the nervous system. In some (but not all) cases, PrP^Sc polymerizes into distinctive fibrils that amass to form amyloid (fibrillar, proteinaceous, congophilic deposits that birefringe in cross-polarized light), although small, oligomeric assemblies of PrP^Sc, which are not in the canonical (fibrillar) amyloid state, can be particularly pathogenic ^{3, 8–10}. Importantly, characteristics of both the seeds and the host influence the infectivity of PrP^Sc as well as the nature of the ensuing disease ^{7, 10}.

PrP^Sc in the broad sense thus comprises a range of pathogenic structures that are referred to simply as prions. However, with the expansion of the prion concept to include other self-propagating protein assemblies ^{8, 11, 12}, and to minimize concern that non-PrP cerebral proteopathies might be similarly infectious, we here refer to prototypical (PrP) prions as PrP-prions.

Progression of neurodegenerative diseases and the prion paradigm

Experimental evidence now supports the concept that certain proteins involved in multiple neurodegenerative diseases acquire their pathogenicity by a prion-like mechanism (Figure 1). Some of these proteins (and the lesions they form) include amyloid-β (Aβ) (amyloid plaques and cerebral amyloid angiopathy [CAA] in AD), tau (neuronal and/or glial tauopathies in AD, chronic traumatic encephalopathy and other neurodegenerative disorders), and α-synuclein (Lewy bodies and Lewy neurites in PD, Lewy body dementia, and glial cytoplasmic inclusions in multiple system atrophy) ^{3, 8, 11, 12}. In addition, evidence is growing that huntingtin (inclusion bodies in Huntington’s disease) and several proteins associated with ALS-frontotemporal dementia spectrum disorders, including superoxide dismutase 1 (SOD1) and TAR DNA-binding protein-43 (TDP-43) also acquire pathogenicity by a prion-like molecular process.

Figure 1.

Factors governing the genesis, replication, and spread of proteopathic seeds. Disease-specific seeds (orange) are generated when certain normally produced proteins (green) misfold, in which state they structurally corrupt like proteins and self-assemble into multimers. Whether seeds are produced throughout life and usually are actively removed or whether this is a rare event that inevitably marks the beginning of disease remains to be determined. The seeds move from one location to another by any of several potential mechanisms; in some instances, the affected site can be extracellular. All of these phenomena may contribute to selective local vulnerability; they can vary in different cell types and regions of the nervous system, where such factors as the presence of auxiliary agents for replication, transport and uptake mechanisms differ (dark blue, a cell in which different agents restrict further propagation). In addition, the expression level or isoform of the cognate proteins may differ among cells and compartments, thereby further supporting or restricting the spread of the seeds and/or defining the strain of seed that is propagated (see Figs. 2 and 3).

Although the formation and amplification of proteopathic seeds is fundamental to these disorders, the disease agents must also translocate among cells and to different regions (Figure 1). To this end, cross-sectional histopathologic analyses have consistently indicated that proteinaceous lesions do not appear and spread randomly; rather, they develop in disease-specific spatiotemporal patterns ^13–16. More recently, in vivo imaging investigations have begun to confirm postmortem histopathological findings implicating the neural connectome as an important mediator of the region-to-region progression of proteopathic lesions ^17,18. Ultimately, the development of pathology in recipient compartments is governed by host factors that render the local environment receptive or resistant to the propagation of the abnormal proteins (Figure 1).

Postmortem histopathologic analyses and in vivo imaging modalities rely on the presence of distinctive deposits such as extracellular amyloid or intracellular inclusions to map the lesions in the human brain, but such studies only indirectly disclose the dynamic process by which the lesions spread. Moreover, it is important to recognize that these obvious lesions may not fully represent the distribution of the proteopathic seeds and their associated pathologic sequelae ¹⁰. Like PrP^Sc, many disease-related proteins have an enhanced tendency to form amyloid, but they may also comprise small, self-propagating oligomeric assemblies that can disrupt the function of cells and tissues, but which can be difficult to analyze unambiguously in biological samples. Hence, the amyloid state is indicative of a proteopathic process, but it is not always required for the manifestation of disease.

Experimental models have allowed researchers to methodically investigate the trafficking of seeds and the spatiotemporal emergence of anomalous proteinaceous lesions along neural pathways. In an in vivo exogenous seeding paradigm, both the characteristics of seeds and the site at which seeding originates can be carefully defined.

Propagation of proteopathic lesions

The prion-like propagation of proteopathic assemblies in neurodegenerative diseases other than PrP-prion disease was first established by the demonstration that Aβ seeds in brain extracts are necessary and sufficient to induce the aggregation of Aβ in transgenic mouse models (reviewed in ¹¹). The prion concept has since expanded to include many of the aberrant proteins that characterize human neurodegenerative diseases (Figure 2; Table 1).

Figure 2.

Compatibility of seed and cognate host protein regulates the propagation of proteopathic seeds at the organismic level. A,B, Proteopathic seeds isolated from a human brain are conformationally heterogeneous (colored dots). In a murine host brain milieu that is permissive for amplification of the dominant conformation (orange), lesions with a molecular structure similar to that in the donor brain will be preferentially propagated following introduction of the exogenous seeds. A, To facilitate seeding, the host mice are often transgenic (Tg), and are engineered to express the human protein that forms specific lesions in the human brain. Seeding efficiency is augmented by high (transgenic) expression of the cognate protein in the host (orange mouse). B, Wild-type (WT) mice usually are more restrictive in propagating the human conformation (for instance, owing to different amino acid sequences in the proteins), but in some cases they may permit the propagation of a subconformation (gray dots). The more abundant the exogenous seeds and the closer their structural characteristics to the host protein, the more likely and efficient their propagation in the host. Hence, transmission of proteopathic lesions from a human donor to a WT mouse (gray) typically requires longer incubation times and sometimes may never occur during the lifetime of the mouse (see also Table 1).

Table 1.

Framework for the transmission of proteopathic seeds to mice or humans either by intracerebral (i.c.) or peripheral (i.p.) inoculation. The year when the first transmission of brain- derived, synthetic, or recombinant seeds was reported is indicated; darker colors denote earlier proof of transmission. Brain symbols, brain-derived seeds; fibril schematic (brown), synthetic fibrils. *Conclusive evidence that synucleinopathy in human tissue grafts is induced by α -Syn seeds from the host is lacking, hence, supporting evidence comes from animal studies. +Huntingtin inclusions have been shown in WT grafts of Huntington’s disease patients¹⁵¹ or in mutant grafts in wild-type mice¹⁵². However, a prion-like proteopathic process in vivo, i.e., the sustained, seeded misfolding and accumulation of huntingtin initiated by minute quantities of inanimate seeds, has not yet been shown.

Propagation of Aβ-proteopathy

As in the case of PrP-prionopathies, Aβ deposition can be instigated in the brain by the introduction of minute amounts of brain-derived Aβ seeds into suitable hosts ^19–26. Aβ seeds delivered to one brain area induce protein aggregation that spreads to interconnected regions, reminiscent of the neuronal transport and trans-synaptic spread of PrP-prions ²⁷. For example, injection of Aβ-rich brain extract into the hippocampus induces local Aβ deposition, as expected. Once initiated, Aβ deposition then propagates non-randomly to axonally linked parts of the brain ^{28, 29}. The involvement of neurons in the systematic emergence of lesions is supported by in vitro studies showing that Aβ aggregates are conveyed by axonal transport ^{30, 31}, and by the occurrence of seeding-active intracellular Aβ assemblies ³². These studies do not rule out spread by passive diffusion (via the cerebrospinal fluid or interstitial fluid), but the early appearance of deposits at non-contiguous sites, their anatomically orderly proliferation, and the ability of neurons to transport Aβ in vitro argue for an important role of neurons themselves in the spread of disease within the brain.

Aβ seeds also have been shown to traffic from the periphery to the brain ^33–35. Aβ deposits seeded from the periphery (by delivery of seeds into the peritoneal cavity or directly into the circulatory system) are predominantly associated with cerebral blood vessels (Aβ-CAA), suggesting a vascular route of transport and the possible participation of innate immune cells that endocytose and translocate Aβ seeds, although the mode of neuroinvasion remains uncertain ^33–35.

The trafficking of Aβ seeds from periphery to brain has been proposed to underlie the intracerebral deposition of Aβ in people many years after having been treated as children with human-derived growth hormone ³⁶. Some of the recipients later succumbed to CJD, apparently because the growth hormone was extracted from large batches of cadaveric pituitary glands that included glands from donors who had died while incubating PrP-prion disease ^37–39. In a post-mortem analysis of 8 of these iatrogenic CJD cases (ranging from 36–51 years of age), four of the subjects also had extensive Aβ plaques and Aβ-CAA in the brain, and two others had sparse Aβ deposits ³⁶. Aβ deposition also has been reported in the brains of cadaveric growth hormone recipients who died of causes other than CJD ^{38, 40}, and in CJD patients who had received dura mater transplants contaminated with PrP-prions ^{38, 41–43}.

Evidence of tauopathy was minimal ^{36, 38, 40, 44} or absent ⁴⁵ in these iatrogenic cases. Abnormal tau is present in pituitaries from Alzheimer patients ⁴⁶, and tau was detected in some batches of cadaver-derived human growth hormone ⁴⁴ (abnormal tau has not been reported in dura mater). Why tauopathy is sparse in the growth hormone recipients is uncertain, considering that tau can seed tauopathy directly in animal models (below).

Although other interpretations cannot be ruled out, the most likely explanation for Aβ-proteopathy in these subjects is that some lots of growth hormone and dura mater were contaminated with Aβ seeds from AD (or incipient AD) donors. In support of this possibility, aggregated Aβ was detected in batches of cadaver-derived growth hormone ⁴⁴, in pituitary glands from AD patients ⁴⁶, and in samples of the dura mater implicated in transmitting CJD ⁴⁵. In addition, Aβ-CAA was abundant in many of these human cases, similar to the increased vascular Aβ deposition in APP-transgenic mice following peripheral administration of Aβ seeds ^33–35 (above). The possibility that Aβ deposition is somehow actuated by prion disease is unlikely in light of the findings that some non-CJD patients developed Aβ-proteopathy ⁴⁰ and that PrP-prions do not induce Aβ deposition in mouse models ⁴⁷. Whether the surviving recipients of tainted biologics are at a higher risk of developing the full clinicopathologic phenotype of AD is not known. Given the long, clinically silent incubation period for AD ⁴⁸, signs of dementia would not be expected for years or even decades following the initiation of Aβ-proteopathy.

Propagation of tauopathy

Tauopathy is associated with over 20 different disorders ⁴⁹, and thus is one of the most common proteopathies of the nervous system. Similar to PrP-prionopathies and Aβ-proteopathy, there is clear evidence that tau can self-assemble and propagate in vivo by a prion-like molecular process. The aggregation of hyperphosphorylated tau is inducible by the intracerebral infusion of small amounts of tau seeds into transgenic mice expressing human tau ^{50, 51}, and, to a lesser degree, into wild-type mice ^52–54 (Figure 2). This exogenously induced form of tauopathy then spreads systematically from the site of injection to axonally linked brain regions ^55–57, indicative of the neuronal endocytosis, templated amplification, transport, and release of tau seeds ^58–60. In addition, like PrP-prionopathies, Aβ-proteopathy, and α-synucleinopathy (below), tauopathy can be induced in the brain by delivery of tau seeds into the peritoneal cavity ⁶¹.

To model the endogenous emergence and spread of cerebral tauopathy, expression of a human tau transgene was restricted principally to projection neurons of the entorhinal cortex in genetically modified mice ^{62, 63}. The mice developed tauopathy first in the entorhinal cortex, and subsequently in axonally-coupled areas ^{62, 63}. Later studies found that the tau transgene is weakly expressed in other brain regions, which could influence the spatiotemporal pattern of lesion progression ⁶⁴. However, in light of the stereotypical localization of tauopathy in interconnected brain regions in AD, chronic traumatic encephalopathy, and FTLD-tau ^{13, 16, 65, 66}, the experiments in mouse models support the view that neuronal trafficking mechanisms contribute to the connectomic distribution of tau seeds within the nervous system.

In AD, genetic and biomarker analyses indicate that tauopathy is downstream of Aβ aggregation ^{48, 67}. Experimentally, aggregated forms of Aβ have been shown to induce tau lesions and to promote the spread of tauopathy in mice ^68–72. How the two proteins interact is incompletely understood, but it may involve the formation of tau seeds within Aβ-induced dystrophic neurites ^{71, 72}, heterotypic tau seeding by Aβ ⁷³, or the stimulation of tau release from neurons by Aβ-mediated neuronal hyperexcitability ^{74, 75}. Whether the presence of Aβ seeds is necessary to continually drive the spread of tau, or whether Aβ assemblies simply trigger the self-sustaining propagation of tauopathy, is an open question with implications for therapeutic strategies targeting AD.

Propagation of α-synucleinopathy

α-Synuclein misfolds and self-aggregates into characteristic inclusions known as Lewy bodies and Lewy neurites in α-synucleinopathies such as PD and dementia with Lewy bodies, and into glial cytoplasmic inclusions in multiple system atrophy ⁷⁶. Interest in the seeding capacity of abnormal α-synuclein was piqued with reports that Lewy bodies materialize in fetal brain cells that had been transplanted intracerebrally into PD patients in an attempt to alleviate the behavioral manifestations of the disease ^{77, 78}. Examination of the brains of subjects who died years later disclosed that some of the transplanted cells had developed Lewy-pathology, suggesting (but not proving) that α-synuclein seeds in the host brain induced the misfolding and aggregation of the protein in the transplanted cells.

Experimental studies in animals subsequently provided further support for the self-propagation of α-synuclein seeds (Figure 2; Table 1). Exogenous introduction of brain-derived or synthetic α-synuclein seeds instigates progressive neurodegenerative disorders in animals that recapitulate some characteristics of human PD ^79–83 or multiple system atrophy ^{84, 85}. In addition, α-synuclein pathology seeded in one region of the brain propagates along anatomically connected structures ^{80, 81, 86, 87}, suggestive of selective neuronal transport of the seeding agent.

Similar to other cerebral proteopathies, α-synucleinopathy is inducible in the brains of experimental animals by the peripheral infusion of α-synuclein seeds ^88–91. This finding has re-invigorated consideration of the mechanisms underlying one of James Parkinson’s (1817) seminal observations of the “shaking palsy” ⁹² (now known as PD). Parkinson noted that constipation is a frequent symptom of the disease, and he even considered how a gastrointestinal disorder and a brain disorder could be related ⁹². Since then, immunoreactive α-synuclein inclusions in the autonomic nervous system have been described, whence α-synuclein seeds are hypothesized to travel to the brain via neuronal connections ⁹³.

Propagation of other neurodegeneration-associated protein assemblies

In addition to PrP, Aβ, tau, and α-synuclein, some proteins associated the ALS-frontotemporal dementia spectrum exhibit self-propagating properties and spread in mouse models, including SOD1 ⁹⁴ and TDP-43 ⁹⁵ (Figure 2; Table 1). Although in vitro studies indicate molecular prion-like processes for disease-associated proteins such as FUS and polyglutamine-containing proteins (above), as well as dipeptide repeat proteins ⁹⁶, definitive evidence for a bona fide prion-like mechanism in vivo remains to be demonstrated in these instances.

Heterogeneity of proteopathic seeds

Strains and clouds

In PrP-prion diseases, conformation-sensitive assays and molecular probes indicate the existence of structurally heterogeneous assemblies of PrP^Sc within the brain. Such variants are referred to as PrP-prion strains, and their heterogeneity constitutes what are known as conformational clouds ^{10, 25}, i.e., a group or ‘cloud’ of related conformations within the same brain. PrP-prion strains can change and undergo differential amplification under selection pressure ^{10, 97, 98}. Strains and clouds have been linked to the species (transmission) barrier and to the variable phenotypic expression of PrP-prion disease ⁹⁹. The occurrence of strains and clouds of distinct conformations is a predicted feature of all amyloidogenic proteins ⁹. Indeed, misfolded Aβ, tau, and α-synuclein share with PrP^Sc the properties of conformational strains and clouds, a phenomenon that can influence both the propagation and characteristics of the respective proteinaceous lesions (Figure 2, 3).

Figure 3.

Compatibility of seed and cognate cellular protein governs propagation at the cellular or compartmental level. For seeded propagation and spread, proteopathic seeds must translocate from cell-to-cell or compartment-to-compartment, and they must be replicated at each successive location (see also Fig. 1). Both steps are dependent on the host, and can vary in different cell types such as neurons and glia (upper and lower cells, respectively, in panels A–C), in which such factors as protein expression, isoforms, and auxiliary molecules influence cell tropism. As a result, some cells resist seeding, and others may select for particular proteopathic conformations (different colored dots). A, Neurons and glia both select for the same strains. B, Neurons and glia select for different strains; here the glia generate secondary seeds that differ from the initial seed. C, The glia are incapable of replicating any pathogenic form of the protein.

Multiple experimental approaches reveal a diversity of Aβ aggregates^{, 24, 25, 100–103} and tau aggregates ^{51, 104} in the human brain. Using solid-state nuclear magnetic resonance on AD-seeded, synthetic Aβ fibrils ¹⁰⁰, or conformation-sensitive assays of brain samples ¹⁰³, variant molecular structures have been detected that correspond to either typical AD or a rapidly progressing form of the disorder. In addition, conformation-sensitive amyloid-binding dyes confirm the variability of Aβ aggregates both within (clouds) and among AD brains ^{25, 101}. These studies further show that the molecular attributes of aggregated Aβ deposits differ among familial and idiopathic AD patients ^{25, 101}. High-resolution cryo-electron microscopy is becoming an important tool for defining the molecular architecture of aberrant Aβ ¹⁰⁵ and other proteins within the brain. Recently, the variant structures of two different types of tau filament (paired helical and straight filaments) in an AD case were revealed by cryo-electron microscopy ¹⁰⁶, and tau strains have been isolated from AD brains and propagated in clonal cells ⁵¹, although it remains to be determined whether the structures of the propagated strains have the same conformations as the those in the brain.

Several experiments indicate that it may be possible to replicate in animal models the strain-like features of aggregated proteins from human diseased brains. The intracerebral injection of brain extracts containing aggregated Aβ from different mice ¹⁰⁷ or etiologically-different AD cases ^{24, 25} into susceptible host mice induces cerebral Aβ deposits with molecular traits that partially recapitulate those in the donor brains. Similarly, tau seeds extracted from the brains of humans who died of different tauopathies (AD, progressive supranuclear palsy (PSP), frontotemporal lobar degeneration-tau (FTLD-tau), or corticobasal degeneration (CBD)) induce tau inclusions in mice that are remarkably similar to the corresponding human lesions (including astroglial and oligodendroglial inclusions for PSP-tau and CBD-tau seeds, and neuronal inclusions for AD-tau seeds ^{52, 104, 108}).

Aggregates of α-synuclein from PD brains exhibit differential proteinase-K cleavage patterns, indicative of variant molecular conformations of α-synuclein ¹⁰⁹. Furthermore, brain extracts from patients with multiple system atrophy (MSA) or PD have been found to induce different phenotypes upon seeded transmission of α-synucleinopathy to mice ⁸⁵. Indeed, oligodendrocytes are specifically able to convert α-synuclein into the MSA strain, which shows a much higher seeding capacity compared to neuronal α-synuclein seeds ⁸⁴. Finally, recombinant α-synuclein fibrils can cross-seed tau fibrillization, and the efficacy of this cross-seeding is governed by strain-like variations in the α-synuclein seeds ¹⁰⁹, although the in vivo relevance of synthetic α-synuclein strains remains uncertain (e.g. ¹¹⁰; see also below).

Durability and activity

Although the detailed molecular conformation of PrP^Sc and its variants is still tentative ¹¹¹, certain structural and functional properties common to PrP^Sc and other proteopathic seeds contribute to their shared pathobiology. The enhanced ability to form amyloid renders some proteopathic seeds resistant to physicochemical degradation by harsh treatments such as heat, formaldehyde, or exposure to proteases. Similar to PrP^Sc, resistance to inactivation by formaldehyde has been shown for Aβ seeds ¹¹² tau seeds ¹¹³ and α-synuclein seeds ^{114, 115}. In addition, a subset of Aβ seeds, like PrP^Sc, are resistant to degradation by heat ^{20, 116} and proteinase-K ²¹. Notably, some Aβ seeds ¹¹⁷ and PrP-prions ¹¹⁸ can persist in the living brain for months following exogenous infusion.

Given the conformational variability of proteopathic seeds, it is not surprising that seed durability and bioactivity also vary, with some seeds being relatively fragile but exceptionally seeding-active. Oligomeric forms of PrP^Sc have a higher specific seeding activity than do larger multimers ^{119, 120}. Similarly, potent seeding activity has been found for smaller aggregates of Aβ ^{21, 23}, and tau ¹²¹. The smallest unit of infectivity/seeding capacity is not known, although estimates for PrP-prions are approximately 6 – 20 PrP molecules per particle ^{119, 120}, and for tau, 3 - >10 molecules per particle ^{122, 123}.

Assemblies of synthetic or recombinant Aβ, tau, and α-synuclein have been consistently shown to have comparatively weak seeding capacity compared to seeds derived from brain ^{20, 54, 81, 83, 124–126}. For example, sub-attomolar amounts of brain-derived Aβ can induce Aβ deposition following intracerebral infusion into APP-transgenic mice ²³, whereas aggregates of synthetic Aβ require 100 to 1000 times more Aβ and longer incubation times to induce histologically detectable seeded deposition ^{20, 126}. Analysis of molecular structure could yield clues to the differential functionality of proteinaceous seeds.

Generating in vivo-active recombinant PrP seeds in vitro had been a longstanding challenge for the PrP-prion field, but infectivity of recombinant PrP can be capacitated by aggregation in the presence of particular cofactors ¹²⁷. Whether other synthetic seeds might be similarly enabled by cofactors is not yet certain, but the in vivo seeding efficacy of synthetic Aβ seeds is enhanced if the Aβ is aggregated on living tissue slices in culture ¹²⁸. The host milieu thus is a key element in the development of proteopathic seeds; host factors also control the susceptibility to disease as well as the resulting phenotype.

Host factors

As in PrP-prion diseases, the host plays a critical role in determining the formation and pathogenicity of other proteopathic seeds. Whether disease-specific seeds are produced throughout life and usually are actively removed, or whether the generation of seeds is a rare event that inevitably marks the beginning of the disease, remains to be determined. In both scenarios, however, the emergence and persistence of seeds is thought to be promoted by the age-related deterioration of the host proteostasis network ¹²⁹. The host also provides the active and passive mechanisms by which seeds spread through the nervous system. Furthermore, as the source of auxiliary molecules such as chaperones along with the naïve protein molecules that serve as the substrate for templated conversion, the host regulates the self-propagation of seeds. Finally, a salient characteristic of neurodegenerative disorders is the selective vulnerability of different cell types and regions of the nervous system to disease. In virtually all neurodegenerative proteopathies, some cells are highly vulnerable whereas others are not, sometimes within the same local environment ^{130, 131}. The topography of disease reflects in part the extended connectome of the afflicted areas, but also the temporal development of lesions and the intrinsic features of cells and tissues that render them selectively susceptible to disease.

Compatibility of host proteins and seeds

The transmission of PrP-prions and other seeds to new hosts follows a fundamentally similar molecular process. In exogenous seeding models, both the concentration of seeds and the structural compatibility with their proteinaceous substrate govern the subsequent self-propagation of aggregates ¹¹. Thus, experimental transmission of human protein assemblies to mice is facilitated when the murine host overexpresses the corresponding human-sequence protein (Figure 2). An important determinant of seed-host compatibility thus is the amino acid sequence of the protein, which constrains the protein’s folding options, although post-translational modifications and other cellular factors also contribute. These, in turn, influence the complementarity of the molecular surfaces that interact to actuate the seeding cascade ⁹.

In addition, expression levels and isoforms of a protein may differ among cells and compartments, and thus the host can differentially select certain conformations for amplification in specific locations (Figure 3). For instance, the conformational features and cellular location of seeded α-synuclein lesions are dependent on the expression level and type of host α-synuclein ^{91, 132}, and the same is true for the seeded induction of tau ^{52, 104} and Aβ lesions ^{34, 107, 133}. Thus, the compatibility of host proteins and seeds is an important factor that regulates the selective propagation of proteinaceous assemblies in different cell types and brain regions ^{84, 130, 133, 134, 135}. Another way in which the host mediates the non-random spread of proteopathic lesions is by the selective translocation of seeds from cell-to-cell or compartment-to-compartment.

Host mechanisms of seed spread

Several active and passive mechanisms may promote the dissemination of seeds (Figure 1). Of these, active axonal transport along defined neural pathways appears to play a major role. Axonal transport mechanisms enable the general translocation of diverse materials such as macromolecules, organelles and viruses to and from neuronal somata. Although fibrillar forms of Aβ, huntingtin and α-synuclein have been demonstrated to travel in both an anterograde and retrograde direction along axons, the rate of transport (at least in vitro) differs for the three proteins ³¹. Variant states of a protein also may influence how they move from place to place, as shown for PrP-prions; in mice expressing PrP with an intact glycosylphosphatidylinositol (GPI) anchor, infectivity traffics mainly along neuronal pathways, whereas in mice expressing PrP lacking the GPI anchor, infectivity is more likely to diffuse through the interstitial fluid ¹³⁶.

Some means of translocation appear to be selective for certain proteopathic assemblies. The protein product of lymphocyte-activation gene 3 (LAG3) has been suggested to be a receptor for the endocytosis and spread of α-synuclein seeds (pre-formed fibrils) in neurons, but the same mechanism appears not to accommodate tau or Aβ seeds ¹³⁷. Unconventional secretion pathways for the cellular release of both tau and α-synuclein have been described ^{138, 139}, which, in one case (mediated by the ubiquitin-specific protease 19), appears to be specific for misfolded α-synuclein but not tau ¹³⁸, although such specificity may vary among cell-types. These findings indicate that differential trafficking of seeds by the host may contribute to the selective vulnerability of different cells and brain areas in neurodegenerative diseases (Figure 3). However, it is also important to note that both the host milieu and the proteopathic seeds are likely to change throughout the long course of neurodegenerative diseases.

Host-seed interactions as disease progresses

The conditions that govern the interaction of host and seeds may evolve with advancing age and disease progression. Studies in transgenic mice show that the specific seeding activity of Aβ changes as Aβ deposition amplifies in the brain, and is highest in the earliest stages of Aβ-amyloidogenesis ¹⁴⁰. The cellular release of Aβ, tau, and α-synuclein, and the spread of the resulting lesions, are promoted by neural activity ^141–143. Neuronal connectivity contributes to the anatomic distribution of proteopathic lesions; as the disease process advances, however, the developing lesions can disrupt the connectome in ways that interfere with normal function and complicate the pathways by which the proteinaceous seeds spread further ^{18, 144}. Finally, a role of microglia in the transport and processing of seeds has been described; hence, the activation and disease state of microglia influence the disease phenotype, probably in complex ways ^145–147. Most importantly, similar to the prionoses, other neurodegenerative proteopathies have a long, quiet phase during which the abnormal proteins proliferate in the nervous system, even before the characteristic lesions can be detected with histological or imaging tools ^{140, 148, 149}. This critical early phase of seed propagation is an important topic for future research.

Perspective: The seed and soil concept in neurodegenerative diseases

Compelling genetic, pathologic and experimental evidence now implicates the prion-like misfolding and corruptive templating of proteins in the pathogenesis of neurodegenerative diseases. In each disorder, specific proteins selectively aggregate in certain parts of the nervous system. The pattern of accumulation reflects the nature of the proteopathic seeds, the various pathways through which they can translocate, and the idiosyncratic features of the affected structures. Conceptually, the proliferation and selective spread of proteopathic seeds is reminiscent of the tissue tropism of malignant cells in metastatic cancer ¹⁵⁰. Another similarity is the heterogeneity of the disease agents, which provides a varied substrate for the Darwinian selection of proteopathic strains or subclonal cancer cells in response to therapy ^{98, 150}. A deeper understanding of the emergence, spread, and selective impact of pathogenic protein assemblies, particularly at early stages of disease, will yield useful insights into the pathobiology of a variety of human afflictions. Just as malignant cells and host factors interact to define the pathogenesis of cancer, the ‘seed and soil’ concept first proposed by Paget ¹ could inform the coherent development of disease-modifying therapies for neurodegenerative disorders involving the seeded aggregation of proteins.