The roles of connectivity and neuronal phenotype in determining the pattern of α-synuclein pathology in Parkinson's disease

Parkinson's disease (PD) is the most common neurodegenerative movement disorder, and motor dysfunction has been attributed to loss of dopaminergic neurons. However, motor dysfunction is only one of many symptoms experienced by patients. A neuropathological hallmark of PD is intraneuronal protein aggregates called Lewy pathology (LP). Neuropathological staging studies have shown that dopaminergic neurons are only one of the many cell types prone to manifest LP. Progressive appearance of LP in multiple brain regions, as well as peripheral nerves, has led to the popular hypothesis that LP and misfolded forms of one of its major components – α-synuclein (aSYN) – can spread through synaptically connected circuits. However, not all brain regions or neurons within connected circuits develop LP, suggesting that cell autonomous factors modulate the development of pathology. Here, we review studies about how LP develops and progressively engages additional brain regions. We focus on how connectivity constrains progression and discuss cell autonomous factors that drive pathology development. We propose a mixed model of cell autonomous factors and trans-synaptic spread as mediators of pathology progression and put forward this model as a framework for future experiments exploring PD pathophysiology.


Introduction
is characterized by the gradual appearance of α-synuclein (aSYN)laden intraneuronal inclusions, referred to as Lewy pathology (LP) (Goedert et al., 2013), and neurodegeneration in certain brain regions. For the last two decades, one of the dominant theories about the origin of this pathology is that LP spreads through synaptic coupled networks from the periphery, inducing neurodegeneration and the symptoms of PD (Brundin and Melki, 2017). Braak and colleagues initially advanced this hypothesis based upon post-mortem examination of the brains of premanifest and symptomatic PD patients using immunocytochemical localization of aSYN aggregates (Braak et al., 2003). Since then, there has been an explosion of experimental studies supporting the proposition that aSYN pathology can spread from one neuron to another in the brain. Much of this work has been driven by the discovery by Lee and colleagues that the pathology induced by injection of fibrillar forms of aSYN into the rodent brain readily spreads from the site of inoculation (Luk et al., 2012). Subsequent work by Recasens et al. found a similar phenomenon in the primate brain following inoculation with material derived from human LP, solidifying the connection to PD (Recasens et al., 2014). Because spreading of aSYN pathology depends upon the recruitment of endogenous aSYN to misfolded aSYN templates, the term 'prionlike' has commonly been used to describe this phenomenon (Brundin and Melki, 2017).
Although attractive in its simplicity, there are many unresolved questions about how misfolded forms of aSYN spread and the relevance of experimental studies to the human condition. The goal of this review is to rigorously explore some of these questions, critically discuss the existing literature and to identify key gaps in our understanding. There are four sections to the review. First, we discuss the evidence from clinical autopsy studies suggesting that aSYN pathology is triggered in restricted areas of the nervous system and then spreads along the synaptic connectome. Second, we discuss observations in model systems consistent with the proposition that aSYN pathology can propagate transsynaptically in a prion-like fashion, and the relationship between these studies and human autopsy data. Third, we discuss potential mechanisms mediating seeding and propagation of aSYN pathology in the brain, as well as neurodegeneration, highlighting the potential role of cell autonomous factors. Last, we consider the possibility that the LP characteristic of PD reflects a combination of trans-synaptic spread of aSYN aggregates and cell autonomous factors. 1998b; Spillantini et al., 1997). These studies have shown that fibrillar aSYN is a significant protein component of LP and possible seed for PD progression. Several lines of evidence now suggest that LP in PD is seeded either in the periphery or in the brain and then spreads. One line of evidence comes from the progression in symptoms associated with PD. For example, James Parkinson noted that constipation was an early and consistent complaint of people affected by "paralysis agitans", implicating gastrointestinal or brainstem pathology in prodromal PD (Parkinson, 2002). More recently, longitudinal studies of PD patients suggest that deficits in olfactory sensation and gastrointestinal function may appear decades before a clinical diagnosis of PD (which requires manifest motor disability) (Fereshtehnejad et al., 2019;Postuma and Berg, 2019;Schaeffer et al., 2020). Although these symptoms have a relatively low specificity for PD, it is suggestive of a prolonged prodromal period in which LP is accumulating. Even within the basal ganglia structures, whose dysfunction underlies the cardinal motor symptoms of PD, pathology may be accumulating well before motor symptoms. At the time of diagnosis, there already has been a dramatic loss of proteins associated with dopaminergic transmission in the striatum and significant loss of phenotypic markers in the substantia nigra itself (Cheng et al., 2010;Fearnley and Lees, 1991;Greffard et al., 2006;Kordower et al., 2013). It is highly likely that these changes started decades before and that the basal ganglia circuitry has compensated for declining dopaminergic dysfunction until a 'tipping point' is reached (Bezard et al., 1999;Bezard et al., 2001;Nandhagopal et al., 2011). Even after diagnosis, there is often a progressive decline in cognitive function, and a high percentage of PD patients progress to dementia (Aarsland et al., 2003).
Another line of evidence that is consistent with the spreading hypothesis comes from autopsy data. Identification of aSYN as a major protein component of LP (Spillantini et al., 1998a;Spillantini et al., 1997) led to a detailed immunocytochemical analysis of its distribution in post-mortem brains. Based upon this kind of analysis of 'pre-symptomatic' and PD brains at varying times after diagnosis, Braak and colleagues (Braak et al., 2003) advanced the hypothesis that in the prodromal stages of PD, LP is found in the dorsal motor nucleus of the vagal nerve (DMV), locus coeruleus (LC) and olfactory system structures. At the time of diagnosis (stage 3), the distribution of LP expands in the brainstem, the substantia nigra pars compacta (SNpc), the diencephalon and the basal forebrain. In later stages (4-6), LP worsens in those locations where it is seen previously and expands to include regions of the telencephalon. This broadening of LP and neurodegeneration is correlated with worsening of motor symptoms, and cognitive decline (Braak et al., 2006b;Irwin et al., 2017;Irwin et al., 2012).
Braak and others followed up this initial staging with investigations of the peripheral nervous system and found that LP can be found in the gastrointestinal tract in regions innervated by the DMV through the vagal nerve (Braak et al., 2006a). This finding is consistent with a peripheral seeding of LP that spreads to the brain. But it is important to note that only about half of PD cases follow the Braak staging (Beach et al., 2009). The alternative patterning of LP is consistent with the seeding of LP in the brain itself, which then spreads (Borghammer, 2021;Schaeffer et al., 2020;Borghammer, 2021).
Recently, imaging studies have provided tentative support for Braak staging using brain atrophy as a proxy (Dagher and Zeighami, 2018;Laansma et al., 2021;Pandya et al., 2019). In both cross-sectional and longitudinal studies, brain atrophy in PD patients is progressive and correlated with connectivity (Dagher and Zeighami, 2018;Pandya et al., 2019). Modeling these data have allowed inferences to be drawn about potential sites of disease origin (Dagher and Zeighami, 2018), and may allow a personalized approach to disease progression (Brown et al., 2019). Specifically, if the brain region of disease origin is known and the parameters controlling disease progression are well-defined, then physicians could provide a more personalized disease prognosis for patients and possibly identify populations who would be most likely to benefit from therapies targeted to specific brain regions or cell types. While the relationship between regional atrophy and underlying disease mechanisms remains to be determined, having an imaging ligand for LBs would at least allow an assessment of its potential pathogenic role.
Another set of observations that are consistent with the transmissibility of LP come from post-mortem studies of the brains of patients who previously received grafts of fetal midbrain dopamine neurons. These revealed the presence of LP-like aggregates in young dopaminergic neurons inside the transplants (Kordower et al., 2008a;Li et al., 2008). Specifically, several cases have now been documented showing that 2-12% of grafted dopaminergic neurons exhibit aSYN aggregates between 10 and 24 years after transplant surgery (Brundin and Kordower, 2012;Kordower et al., 2008a;Kordower et al., 2008b;Kurowska et al., 2011;Li et al., 2008;Li et al., 2016). Notably, such aggregates were not observed in patients who died 2-5 years after surgery (Kordower et al., 1995;Kordower et al., 1996;Mendez et al., 2005;Olanow et al., 2003), which suggests that a lag-period was required for the LP to develop in the young, transplanted neurons. The cytosolic levels of aSYN, presumably in a soluble form, were also increased in 40% and 80% of the grafted neurons at 12-and 16-years post-grafting, respectively (Li et al., 2008). These increased levels of cytosolic aSYN resemble those seen in normal aging brains (Chu and Kordower, 2007) and suggest that the grafted neurons might be more susceptible to seeding of aggregates by internalization of host-brain derived aSYN assemblies. However, there are caveats to this interpretation. One is that, as described below, dopaminergic neurons create a cellular environment that is conducive to the formation of aSYN aggregates, particularly in stressful situations, as those that might be found in a graft.
Despite the evidence that LP spreads in the brain through the course of PD, it appears that some regions are more vulnerable to disease than others. Thus, the pattern of LP is restricted in distribution and does not appear in many regions that are synaptically coupled to those that manifest LP early in the disease (Surmeier et al., 2017). Within nuclei manifesting LP, typically only a small percentage (5-15%) of neurons manifest LP even at end stage (Braak and Del Tredici, 2009;Dugger and Dickson, 2010). It is unclear whether this is due to LP formation being stochastic, to these cells being particularly vulnerable to development of LP, or to whether other cells that developed LP earlier have already died and are therefore not available for analysis (Greffard et al., 2010). The uneven distribution of LP between brain regions and between neurons within a given anatomical structure suggests that there might exist cell-and region-specific factors that modulate susceptibility to degeneration or that individual neurons might be more resilient to LP than their immediate neighbors.

Evidence for propagation of aSYN pathology in model systems
While it is currently not possible to longitudinally track the evolution of aSYN pathology in PD patients, cell and animal models have shown that misfolded conformers of aSYN can be released and internalized by neurons, inducing the misfolding of endogenous aSYN. This process is commonly referred to "prion-like" because unfolded endogenous proteins are induced to form a misfolded aggregate by an exogenous template.
Induction of LP-like aggregates in neuron cultures has enabled the assessment of several cell autonomous mediators of aSYN pathology. For example, neurons lacking endogenous aSYN will not develop inclusions following treatment with PFFs (Henderson et al., 2017;Volpicelli-Daley et al., 2011). Therefore, aSYN expression level is a cell autonomous factor mediating LP vulnerability. Several other cell autonomous factors modulate aSYN pathology. For example, cell surface proteins can bind misfolded aSYN and promote internalization (Aulic et al., 2017;Emmenegger et al., 2021;Mao et al., 2016). Reduced lysosomal integrity due to glucocerebrosidase inhibition or pH disruption can elevate susceptibility to PFFs (Henderson et al., 2020;Karpowicz Jr. et al., 2017). Protein kinases involved in axonal trafficking also may influence how well a cell handles toxic forms of aSYN (Henderson et al., 2017). Neuronal spiking enhances the release of aSYN and the internalization of misfolded aSYN (Ueda et al., 2021;Wu et al., 2020), potentially increasing vulnerability.
Animal models also have provided important insights into the mechanisms underlying LP formation and spreading. PFFs can be stereotaxically injected into the brains of rodents or primates, inducing LP-like aggregates (Luk et al., 2012;Masuda-Suzukake et al., 2013). The most studied model is one in which PFFs are injected into the striatum where the highly arborized axons of vulnerable dopaminergic neurons terminate. In this model, aggregates of detergent-insoluble, phosphorylated aSYN and other LP components (Luk et al., 2012;Masuda-Suzukake et al., 2013) form earliest in the SNpc dopaminergic neurons and other regions of the brain that project to the striatum (e.g., cerebral cortex). In SNpc dopaminergic neurons, formation of aggregates is coupled to cell death. The ability to induce LP-like pathology in a wide range of animal models has been a major advantage of the PFF model.
For example, PFF injection in wild-type animals has demonstrated that misfolded forms of aSYN can propagate in the normal brain, establishing relevance of the phenomenon to PD.
The spatiotemporal control of aSYN pathology afforded by localized stereotaxic injection of PFFs into animal brains has enabled the generation of detailed regional maps of the resulting LP-like pathology and how they change as a function of time following the initial seeding event. Several recent studies have combined brain-wide pathology assessment in PFF-treated mice to assess the determinants of spreading (Henderson et al., 2019b;Henrich et al., 2020;Mezias et al., 2019). For example, one study employing intrastriatal PFF injection used a model based on linear diffusion to show that brain-wide pathology followed largely retrograde neuroanatomical connections (Henderson et al., 2019b). Differences between estimated spread through connectivity and regional pathology were partially accounted for by variation in the expression level of aSYN.
To determine if the seeding site affected spreading, another recent study injected PFFs into the pedunculopontine nucleus (PPN), a region that exhibits LP early in the course of PD (Henrich et al., 2020). The authors found that S129 aSYN immunoreactivity following small PFF injections was almost exclusively in cholinergic neurons and difficult to detect in neighboring glutamatergic and GABAergic projection neurons, as is the case in human PD. Using monosynaptic rabies virus mapping approaches to quantitatively map inputs specifically to cholinergic projection neurons, they found that all neurons innervating these 'starter' neurons manifested aSYN pathology (pS129 aSYN immunoreactivity) within 6 weeks of the injection. However, in many of these regions, aSYN pathology peaked and then declined or completely cleared within 12 wks, suggesting pathology was transient. Importantly, there was no significant correlation with the nominal strength of the projection and semi-quantitative scoring of PFF-induced pathology. That said, the monosynaptic rabies virus approach does not provide a measure of the functional strength of synapses or their activity, both of which might be factors in propagation (Ueda et al., 2021).
Other studies also have suggested that aSYN pathology might decline following longer survival times (typically many months) after the initial PFF injection, but it is not yet clear whether this is due to the LP-like pathology being cleared by the neurons or if the affected neurons die, leading to fewer aSYN aggregates being visible at the time of sacrifice (Rey et al., 2017). Mice with defective mitochondrial function, due to the absence of one allele of the Engrailed1 transcription factor, also display exacerbated aSYN pathology following intrastriatal PFF injection (Chatterjee et al., 2019) (see below). Another important, but relatively unexplored, determinant of pathological aSYN spreading is inflammation. With stereotaxic injection of foreign material into the brain there is invariably damage and inflammation, which may contribute to spreading and neurodegeneration (Duffy et al., 2018;Rey et al., 2016a;Zheng and Zhang, 2021). A closely related question is the dose-response relationship of aSYN pathology and spreading. One interesting study that has addressed this question found that this relationship was non-linear, and dependent upon the brain region as well as the species of the host (Abdelmotilib et al., 2017). Finally, the conformation of aSYN fibrils should be considered. Several studies have found remarkable differences in the pathogenicity of aSYN derived from multiple system atrophy and PD brains (Peng et al., 2018;Prusiner et al., 2015;Recasens et al., 2014;Tanudjojo et al., 2021).
Furthermore, experiments using different strains of recombinant aSYN can seed aSYN pathology differentially in mouse brain . While recombinant aSYN PFFs induce Lewy-like pathology in mice, differences between mouse models and human disease may be attributable to differences in fibril conformation (Schweighauser et al., 2020).

Evidence for a cell autonomous contribution to aSYN pathology
As noted above, the distribution of LP in PD, particularly in its early and middle stages is not random but is restricted to a set of nuclei. Notably, nuclei adjacent to those that manifest LP are typically not affected in PD. For example, the largest synaptic input to LC -a nucleus with early and robust LP -comes from the cerebellum; but the cerebellum is devoid of LP (Braak et al., 2003;Schwarz et al., 2015). The striatum, globus pallidus and subthalamic nucleus robustly innervate vulnerable SNpc dopaminergic neurons but manifest few Lewy bodies themselves (Beach et al., 2009;Braak et al., 2003;Halliday et al., 2011). LP in the SNpc does not spread to the substantia nigra pars reticulata (SNr), a few hundred microns away and even though dopamine-releasing dendrites of the most vulnerable ventral tier dopaminergic neurons heavily innervate the SNr itself (Beach et al., 2009;Braak et al., 2003). Even within nuclei that manifest LP, there is heterogeneity. For example, in the PPN, LP is almost exclusively in cholinergic neurons and does not appear in interdigitated GABAergic and glutamatergic neurons (Giguere et al., 2018;Pienaar et al., 2013). Even within the SNpc dopaminergic population, susceptibility to LP formation is not equally distributed, but is much higher in the ventral tier than in dorsal and medial tiers (Dickson, 2012;Wakabayashi et al., 2007).
What cell-type specific factors might control release and uptake of misfolded aSYN from the extracellular space (Fares et al., 2021)? One factor contributing to uptake that has already been mentioned is neuronal activity. Regenerative spike activity promotes PFF internalization from the extracellular space through a form of micropinocytosis (Ueda et al., 2021;Wu et al., 2020). How activity regulates this process is unclear, although Ca 2+ entry is likely to be a factor. Ca 2+ dependent release of exosomes containing aSYN also has been reported (Emmanouilidou et al., 2010). That said, there are many very active neurons in the brain (e.g., SNr neurons, cerebellar Purkinje neurons) that never manifest LP; whether they take up PFFs remains to be determined. Another factor contributing to uptake are putative surface receptors like LAG3, neurexin 1b, Aβ precursor-like protein 1, prion protein, or the α3-subunit of the Na + /K + -ATPase (Aulic et al., 2017;Mao et al., 2016;Shrivastava et al., 2015) and heparan sulfate proteoglycan mediated micropinocytosis (Holmes et al., 2013). However, it is unclear to what extent neuronal internalization can be attributed to receptor mediated endocytosis; for example, other groups have found that LAG3 is not expressed in neurons and therefore does not modulate propagation of aSYN pathology between neurons (Emmenegger et al., 2021). Transfer of aSYN pathology through tunnelling nanotubes, which may be created between specific cell types, also has been suggested (Abounit et al., 2016), and was recently shown to mediate transfer of aSYN fibrils between microglia (Scheiblich et al., 2021).
What other cell type-specific factors might contribute to the growth and persistence of intracellular aggregates? For misfolded aSYN that is taken up into the endosomal system, transit to the cytosol (where LP is found) is not well understood (Fares et al., 2021), despite evidence that endosomes/lysosomes may rupture (Jiang et al., 2017). Once released into the cytosol, what drives aggregation? At least in part, an answer might come from a consideration of what promotes fibril formation in the test tube. Two factors that have relevance are the concentration of aSYN, and divalent ions (like Ca 2+ ) that charge shield the carboxyl terminal region of aSYN (Nath et al., 2011;Rcom-H'cheo-Gauthier et al., 2016). As noted above, the expression level of aSYN is correlated with PFF-induced spreading of pathology in rodent brain (Henderson et al., 2019a), primary neurons (Courte et al., 2020), and in PD (Erskine et al., 2018). Indeed, a key trait of the neurons that are most vulnerable to death in PD -dopaminergic SNpc neurons, noradrenergic LC neurons, serotonergic raphe neurons, cholinergic PPN and cholinergic basal forebrain neurons is that they have a massive axonal arbor with long and highly branched axons invested with transmitter release sites. In the human SNpc, the number of release sites is estimated to 1-2 million for one individual axon (Bolam and Pissadaki, 2012;Diederich et al., 2019). With this massive axon must come elevated expression of the presynaptic protein aSYN. Consistent with this idea, decreasing the axonal size of the SNpc dopaminergic neurons in culture significantly reduces their vulnerability to PD-linked stressors (Pacelli et al., 2015). Relatively low aSYN expression levels in regions like the cerebellum, SNr, and GPe could very well be a major factor in their resistance to LP (Courte et al., 2020;Taguchi et al., 2019).
In addition to having elevated expression of aSYN, many of the most vulnerable neurons in PD have high basal levels of cytosolic Ca 2+ , a divalent cation which may promote aggregation (Rcom-H'cheo-Gauthier et al., 2016). At-risk neurons in the SNpc, LC, basal forebrain, PPN, and DMV are slow autonomous pacemakers with broad action potentials (which promote Ca 2+ entry), low intrinsic Ca 2+ buffering and large activity-dependent fluctuations in cytosolic Ca 2+ concentration (Dryanovski et al., 2013;Goldberg et al., 2012;Guzman et al., 2010;Sanchez-Padilla et al., 2014). In SNpc dopaminergic neurons and LC noradrenergic neurons, where this has been studied in more depth, the cytosolic Ca 2+ oscillation is triggered by plasma membrane voltage-dependent Ca 2+ channels with a Ca v 1.3 pore-forming subunit. As in other excitable cells, these L-type channels are juxtaposed to ryanodine receptors (RYRs) in the endoplasmic reticulum (ER) and induce Ca 2+ -induced Ca 2+ release (CICR) (Chan et al., 2007;Diaz-Vegas et al., 2018;Guzman et al., 2009;Viola et al., 2009). CICR promotes Ca 2+ loading of mitochondria at specialized junctions (mitochondria-associated membranes). This loading stimulates oxidative phosphorylation (ATP production) by activating the malate-aspartate shuttle and disinhibiting matrix tricarboxylic acid cycle enzymes. As in muscle (Diaz-Vegas et al., 2018;Viola et al., 2009), this pathway serves as a feed-forward control mechanism for mitochondrial ATP production that anticipates demand, rather than waiting for ATP levels to fall and disinhibit mitochondrial complex V (Zampese and Surmeier, 2020). This should help to ensure that in times of stress or continuous demand, these critical neurons, which modulate large swaths of motor and cognitive circuitry do not shut down (Surmeier et al., 2017).
This feed-forward control comes at a cost. The electron flux necessary for oxidative phosphorylation generates reactive oxygen species, particularly at times when the stimulation precedes demand and mitochondrial membrane potential is high (Brookes et al., 2004). Indeed, basal mitochondrial oxidant stress in SNpc, LC, DMV and PPN neurons is high, particularly in dendrites and axons (Goldberg et al., 2012;Guzman et al., 2010;Sanchez-Padilla et al., 2014). Reactive oxygen and nitrogen species generation are associated with mitochondrial DNA deletions (Bender et al., 2006;Dolle et al., 2016;Sanders et al., 2014) and the loss of mitochondrial complex I function (Reeve et al., 2014;Schapira et al., 1990). This stress may be exacerbated by iron accumulation and the Fenton reaction in some at-risk neurons (Berg and Hochstrasser, 2006). Indeed, basal mitochondrial stress in SNpc dopaminergic neurons is sufficient to increase turnover of mitochondrial proteins (Guzman et al., 2018). In addition to damaging mitochondrial proteins, mitochondrial oxidant stress compromises lysosomal function (Burbulla et al., 2017) and may contribute to disruption of chaperone-mediated autophagy (Martinez-Vicente and Cuervo, 2007), both of which are linked to degradation of aSYN. Moreover, the Ca 2+ signaling in vulnerable neurons leads to pacemaking-associated oscillations in dendritic Ca 2+ concentration that may reach the micromolar level necessary to activate the calcineurin and protease calpain, both of which promote aggregation (Dehay et al., 2010;Diepenbroek et al., 2014;Dufty et al., 2007).
Another factor that may contribute to the vulnerability of dopaminergic neurons is the reliance upon dopamine (DA) as a transmitter. The combination of DA, Ca 2+ and aSYN promotes the degeneration of murine dopaminergic neurons in vitro (Mosharov et al., 2009). DA modification of aSYN also blocks chaperone mediated autophagy (Martinez-Vicente et al., 2008). In human iPSC-derived dopaminergic neurons, mitochondrial oxidant stress drives oxidation of DA, which promotes lysosomal dysfunction and aSYN aggregation (Burbulla et al., 2017). Interestingly, in human neurons, cytosolic DA concentrations were higher than those in rodent-derived neurons, creating a condition that could underlie increased mitochondrial oxidant stress and neuromelanin accumulation (Burbulla et al., 2017;Graves et al., 2020). Although DA metabolism to neuromelanin has long been thought to protect dopaminergic neurons by sequestering reactive DA metabolites (Zucca et al., 2017), recent work has shown that too much neuromelanin can be toxic, probably as a consequence of crowding-induced inhibition of proteostasis (Carballo-Carbajal et al., 2019).
Thus, there may be a 'perfect storm' in vulnerable neurons with aging: 1) high levels of aSYN, Ca 2+ , reactive oxygen species (and DA in SNc neurons) which create a cellular environment that promotes aSYN aggregation and 2) flagging mechanisms for aggregate clearance. This possibility is supported by what we know about familial forms of PD (Kalia et al., 2015;Pickrell and Youle, 2015;Poulopoulos et al., 2012). Moreover, as mentioned above, partial deletion of Engrailed 1 in mice leads to exacerbated aSYN pathology (Chatterjee et al., 2019).
A related question is whether LP is an essential precondition for the development of PD symptomatology, or parkinsonism in broader terms? Duplications or triplications of the SNCA gene (PARK4) as well as point mutations (PARK1), like A53T, A30P, or E46K, are generally associated with early onset cPD and widespread LP (Poulopoulos et al., 2012). However, in PD cases driven by loss-of-function mutations in PINK1, PARK2, or PARK7, LP is sparse or absent (Doherty et al., 2013;Poulopoulos et al., 2012). Patients carrying a mutation in LRRK2 (encoding leucine-rich repeat serine/threonine-protein kinase 2), the most common autosomal-dominant cause of PD, have fairly heterogenous neuropathology, and LP is absent in a significant number of them (Hasegawa et al., 2009;Henderson et al., 2019c;Kalia et al., 2015;Poulopoulos et al., 2012;Ross et al., 2006;Wszolek et al., 2004). Dopaminergic neurons are also susceptible to age-related loss in the absence of disease (Collier et al., 2011;Giguere et al., 2018). These findings show that LP is not necessary for the development of parkinsonism or motor abnormalities.
While in this review we focus on neuronal factors that might play a role in dictating the propagation of LB pathology between brain regions, emerging findings suggest that glial cells also exhibit differences between brain regions (Boisvert et al., 2018;Clarke et al., 2018;Grabert et al., 2016;Soreq et al., 2017). Because both microglia and astrocytes have been suggested to influence both toxicity and cell-to-cell transfer of pathogenic aSYN conformers (Liddelow et al., 2017;Scheiblich et al., 2021;Yun et al., 2018), they too might influence how LP spreads between brain structures.

A mixed model of the spreading of aSYN pathology and future directions
Taken together, there is unequivocal evidence that fibrillar forms of aSYN can -once injected in the brain, gut, or muscle -spread along the connectome and seed LP-like pathology in mice, rats, and non-human primates. But the pattern of propagated pathology within these models does not simply reflect the synaptic connectome of neurons at the seeding site (Henderson et al., 2019a;Henrich et al., 2020). Rather, the distribution (and persistence) of aSYN pathology appears to be modulated by cell-autonomous and regional factors.
If we consider a simple model of aSYN spreading (Fig. 1a), there are several inferences that can be drawn. First, neuronal connectivity constrains and helps predict the rate at which aSYN pathology spreads. That is, the further away you are in the network, the longer it will take to possibly reach you (Fig. 1b). If connectivity was the only driver of pathology spread, pathology would be expected to reach nearly every cell in the brain eventually. This does not happen. Hence, cell autonomous or regional factors must impact either the development of aSYN pathology or its persistence (Fig 1c).
However, it is difficult to imagine how cell-autonomous factors alone could produce a pattern of pathology consistent with that observed in PD. First, neurons with certain vulnerability factors (high levels of aSYN, basal oxidant stress and elevated cytosolic Ca 2+ levels, limited 'spare' capacity to degrade misfolding proteins because of the proteostatic demands placed on them by their axon and high rate of mitochondrial turnover) would all be predicted to manifest pathology. Indeed, with normal healthy aging, SNpc dopaminergic neurons are progressively lost (Giguere et al., 2018). That said, most people do not develop PD in their lifetimes, suggesting that other factors must accelerate the naturally occurring loss enough to reach the threshold for clinical manifestation. It is also important to note that some neurons that do not appear to have the traits of the most vulnerable neurons, do eventually manifest LP (e. g., cortical pyramidal neurons). While the impact of aging on these neurons may result in at least a subset of the factors linked to vulnerability (e.g., metabolic stress and elevated Ca 2+ levels Rcom-H'cheo-Gauthier et al., 2016;Collier et al., 2011), a more parsimonious hypothesis is that LP in these neurons reflects spreading from another seeding site.
Thus, a 'mixed' model of LP that involves both propagation and cell autonomous factors has great appeal. This model posits that: 1) neuroanatomic connectivity constrains the spatiotemporal parameters of spreading, but 2) the persistence, magnitude and cellular consequences of the pathology are determined by local factors (Fig. 1d). This model does not exclude the possibility that aSYN pathology arises independently of propagated pathology, resulting in spread from those central sites, rather than the periphery. Recent work examining the longitudinal distribution of tau pathology in Alzheimer's disease also is consistent with a mixed model involving both trans-synaptic spreading and local factors (Meisl et al., 2021).
Although a great deal of progress has been made, there are several key questions that remain unanswered. One of these is to what extent PFF seeding reproduces the processes responsible for human LP (Fares et al., 2021). Once taken up, do PFFs produce human-like LP, which not only contains misfolded aSYN, but mitochondrial components, lipids and other molecules? Given that most spreading appears to be retrograde, do PFFs need to be modified by local cells (microglia, astrocytes, neurons) once deposited for there to be significant axon terminal uptake? Do cell-specific mechanisms limit terminal uptake? For example, does activity drive internalization at terminals the way it appears to at cell bodies? If there is retrograde propagation and somatodendritic release of misfolded aSYN, then what are the factors governing release?
It also remains to be determined whether the cell autonomous factors that promote aSYN pathology in humans actually affect PFF-induced pathology in model systems. There are a wide range of pharmacological, chemogenetic, optogenetic and molecular approaches that could be used to nail this question down. These same tools can be used to assess how aSYN pathology affects circuits implicated in motor and non-motor features of PD (Tozzi et al., 2021).
Another major question is whether the mechanisms mediating PFF induced neurodegeneration (which are poorly understood) mimic those found in human PD patients. PFF-induced neurodegeneration can be rapid in models, but neuronal loss appears to be very slow in humans (Huynh et al., 2021;McCann et al., 2016). Thus, it remains uncertain whether LP (or a linked aSYN pathology) actually causes neuronal death or is a disease 'tombstone' (Wakabayashi et al., 2007).
Getting clear answers to these questions (and others) should guide translational efforts to diminish LP and neuron death in humans suffering from PD. At present, most efforts to develop therapies that might diminish or reverse LP in humans are focused on enhancing lysosomal function, reducing inflammation or reducing the spread of misfolded aSYN with immunotherapies targeting aSYN (BIIB054, Prasinezumab; Phase II) (McFarthing et al., 2021). The latter therapy is based on the expectation that aSYN becomes extracellular and that clearing it may slow progression. The notion that blocking the uptake of pathogenic aSYN seeds with small molecules is possible has gained traction for reasons discussed above (Aulic et al., 2017;Emmenegger et al., 2021;Holmes et al., 2013;Mao et al., 2016;Shrivastava et al., 2015). That said, the extent to which these approaches will reverse existing pathology and restore function is unclear. Although several previous clinical trials have targeted cell autonomous mechanisms that might play a role in LP spreading, like Ca 2+ entry through Cav1 channels, they have failed to yield significant results (McFarthing et al., 2021). A major issue in these trials has been whether there was adequate target engagement and biological efficacy of the therapeutic intervention. Recent re-examination of the STEADY-PD II and III trials suggest that there was a slowing of disease progression in those patients where there was more robust target engagement by isradipine (either because they were given an extended-release format of the drug or because they cleared it more slowly) Venuto et al., 2021). Nevertheless, designing better strategies for manipulating cell autonomous factors that might contribute to pathogenesis in PD remains a challenge.
In summary, research in the last decade has dramatically improved our understanding of the potential mechanisms underlying the emergence and propagation of aSYN pathology in PD. The available data point to a mixed model in which both trans-synaptic propagation of aSYN pathology and cell autonomous mechanisms govern the pattern, magnitude and persistence of LP. Many questions remain unanswered regarding this mixed model. Filling these gaps in knowledge should facilitate the development of novel therapeutic strategies. We propose that strategies that target both propagation and cell autonomous determinants of aSYN pathology should prove more effective than targeting only one or the other. A mixed model of the spreading of aSYN pathology (a) We consider a very basic neuron network in which each neuron has an intrinsic risk to develop pathology based on cell autonomous factors discussed in the review. (b) Spread through connectivity (in this case in a retrograde direction) is useful to describe the pattern of pathology progression, such that if pathology started at neuron 1, we can estimate which neurons should subsequently exhibit pathology (1>2>3>4). However, if pathology were to spread along all anatomical connections, it would affect all connected regions, including many regions that never exhibit pathology. (c) Cell autonomous factors are useful in describing why some neurons are vulnerable within the network. However, if cell autonomous factors alone drove disease, it would seem likely that all vulnerable neurons would get pathology around the same time (1,1,1,2,2,2) in accordance with their cell autonomous risk. (d) Current evidence points to a mixed model where connectivity provides a conduit for pathology progression through the brain, but cell autonomous factors modulate which regions along that conduit exhibit pathology and which do not. This model posits that neurons that do not express cell autonomous vulnerability factors or are not connected to neurons with pathology will not develop pathology or will do so at a much slower rate. Graphic created with BioRender.com.