The emerging role of a -synuclein truncation in aggregation and disease

a -Synuclein ( a syn) is an abundant brain neuronal protein that can misfold and polymerize to form toxic fibrils coalescing into pathologic inclusions in neurodegenerative diseases, including Parkinson ’ s disease, Lewy body dementia, and multiple system atrophy. These fibrils may induce further a syn misfolding and propagation of pathologic fibrils in a prion-like process. It is unclear why a syn initially misfolds, but a growing body of lite-rature suggests a critical role of partial proteolytic processing resulting in various truncations of the highly charged and flexi-ble carboxyl-terminal region. This review aims to 1) summarize recent evidence that disease-specific proteolytic truncations of a syn occur in Parkinson ’ s disease, Lewy body dementia, and multiple system atrophy and animal disease models; 2) provide mechanistic insights on how truncation of the amino and carboxyl regions of a syn may modulate the propensity of a syn to pathologically misfold; 3) compare experiments evaluating the prion-like properties of truncated forms of a syn in various models with implications for disease progression; 4) assess uniquely toxic properties imparted to a syn upon truncation; and 5) discuss pathways through which truncated a syn forms and therapies targeted to interrupt them. Cumulatively, it is evident that truncation of a syn, particularly carboxyl truncation that can be dysfunction, functional insufficiency of monomeric a syn

Edited by Paul E. Fraser a-Synuclein (asyn) is an abundant brain neuronal protein that can misfold and polymerize to form toxic fibrils coalescing into pathologic inclusions in neurodegenerative diseases, including Parkinson's disease, Lewy body dementia, and multiple system atrophy. These fibrils may induce further asyn misfolding and propagation of pathologic fibrils in a prion-like process. It is unclear why asyn initially misfolds, but a growing body of literature suggests a critical role of partial proteolytic processing resulting in various truncations of the highly charged and flexible carboxyl-terminal region. This review aims to 1) summarize recent evidence that disease-specific proteolytic truncations of asyn occur in Parkinson's disease, Lewy body dementia, and multiple system atrophy and animal disease models; 2) provide mechanistic insights on how truncation of the amino and carboxyl regions of asyn may modulate the propensity of asyn to pathologically misfold; 3) compare experiments evaluating the prion-like properties of truncated forms of asyn in various models with implications for disease progression; 4) assess uniquely toxic properties imparted to asyn upon truncation; and 5) discuss pathways through which truncated asyn forms and therapies targeted to interrupt them. Cumulatively, it is evident that truncation of asyn, particularly carboxyl truncation that can be augmented by dysfunctional proteostasis, dramatically potentiates the propensity of asyn to pathologically misfold into uniquely toxic fibrils with modulated prion-like seeding activity. Therapeutic strategies and experimental paradigms should operate under the assumption that truncation of asyn is likely occurring in both initial and progressive disease stages, and preventing truncation may be an effective preventative strategy against pathologic inclusion formation.
Neurodegenerative diseases, including Alzheimer's disease (AD), Parkinson's disease (PD), and Lewy body dementia (LBD), are collectively the leading cause of dementia worldwide with devastating human and economic costs (1,2). There currently exist no disease modifying therapies that slow or prevent the onset of dementia in these diseases (2), and further research is needed to understand underlying pathophysiologic processes with the goal of identifying targets for therapeutic strategies. Common to most neurodegenerative diseases is the presence of amyloidogenic aggregates comprised predominantly of misfolded proteins of neuronal origins (3,4); these diseases are typically classified based on the clinical presentations and identity of the misfolded proteins (3).
The consequences of misfolded oligomeric and fibrillar forms of asyn in preclinical models have been extensively studied and reviewed elsewhere (11,19,20), and current evidence suggests that disease progression may be difficult to thwart once these forms of asyn are widespread due to prion-like recruitment of endogenous asyn into further pathologic forms that can spread within and between neuronal and glial cells in a vicious cycle (20,21). A more attractive therapeutic target may be the dysfunctional processes resulting in initial asyn misfolding and fibril formation through mechanisms such as aberrant asyn PTMs if these occur early or are essential for disease progression. Some PTMs have been shown in vitro to alter the propensity of asyn to form pathologic fibrils similar to those isolated from disease inclusions (18,22), and corollary in vivo processes likely responsible for the appearance of these PTMs such as oxidative stress and impaired protein clearance pathways are demonstrable in animal models of synucleinopathy and tissue from diseased patients (23)(24)(25). Indeed, PTMs of asyn are abundant in disease, with 90% or more of asyn being phosphorylated at Ser-129 and 15-20% being C-terminally truncated within pathologic inclusion extracts (16,17,(26)(27)(28). C-terminal truncation of asyn may be particularly detrimental among PTMs, as it has been consistently demonstrated in vitro that C-truncated asyn self-assembles into fibrils far more readily than full-length (FL) asyn or even familial disease causing missense mutant forms of asyn (17,(29)(30)(31)(32)(33)(34)(35)(36). Proteolytic formation of truncated asyn (32,(37)(38)(39)(40)(41)(42), possibly promoted by impairment of proteostasis as in aging (43), may initiate inclusion pathology formation and progression. This review aims to summarize the evidence for the accumulation of diverse types of truncated asyn in disease, survey the physiological processes involved in their formation, and discuss the functional implications in the context of pathologic fibril formation, toxicity, and prion-like disease progression.
asyn background asyn has structurally distinct regions important to function and disease asyn was first isolated from synaptic vesicles in the Torpedo fish and was termed "synuclein" when a homologous protein was detected in rat neuronal synaptic vesicles and nuclear envelopes (44). The majority of asyn is in a cytoplasmic, soluble form as an intrinsically disordered protein (6); in association with synaptic vesicles, the repeat-rich N terminus (residues 1-60) and "nonamyloid component" (NAC) domain (residues 61-95) adopt a helical structure necessary for the regulatory role of asyn in vesicle trafficking that is thought to be a main function of asyn in the neuron (45,46). However, asyn also functions in synaptic maintenance and SNARE protein assembly, as evidenced by its ability to mitigate phenotypes in CSPa null mice (47), and asyn can function as a chaperone, which is made possible by labile intermolecular interactions inherent to unstructured proteins (48,49). These biological functions of asyn are attributed to its three overarching structural domains: the amphipathic N terminus, the hydrophobic NAC, and the acidic C terminus (residues . The N terminus harbors most of the seven imperfect KTKEGV 11-mer repeats important in helix formation in association with phospholipid surfaces (45,46). The NAC region is hydrophobic and acts as a lipid "sensor," mediating specificity for synaptic vesicle binding (5,45). Last, the C terminus remains unstructured in nearly all conformations due to its highly charged 15 acidic residues; the charged state and ever-changing structure of this region permits most of the chaperone activity of asyn along with its ability to bind metals, polyamines, and positively charged proteins such as tau (5,45,48,50,51).
Due largely to the hydrophobic nature of the NAC region (52), a confluence of pathologic processes can result in the disease-associated misfolding of asyn into dense fibrils that are highly insoluble in detergents (8). Using proteolytic digestion of fibrils, antibody accessibility, and structural techniques such as cryo-EM, it was determined that the typical asyn amyloid fibril core is composed roughly of residues ;31-102 stacked as in-register b-sheets (31,51,(53)(54)(55)(56)(57)(58)(59). The extreme N terminus and most of the C terminus are not within the amyloid core, but the N terminus is still relatively structured compared with the fully disordered C terminus (53,55,60). The C terminus may also govern higher-order assembly of fibrils, as its highly negative charge governs lateral packing of fibrils and possibly association of protofilaments into more mature amyloid species (33,56,60).
asyn-laden inclusions are hallmarks in multiple neurodegenerative diseases asyn was first associated to neurodegenerative diseases when the NAC region of the protein was detected in association with amyloid plaques in AD (61,62). A more direct involvement in neurodegenerative diseases was supported by the seminal discovery that the A53T missense mutation in the SNCA gene encoding asyn was responsible for autosomal dominant PD in the Contursi kindred (63). Immediately thereafter, asyn was identified as the major component of Lewy bodies (LBs) and Lewy neurites (LNs) that have been pathologic hallmarks in PD and LBD for 100 years prior (64,65).
asyn pathological inclusions are known to be present in a number of heterogeneous neurodegenerative diseases, including in about ;50% of AD patients (65)(66)(67). Symptomatically and pathologically, various diseases afflicted by asyn inclusions can be very different. PD and LBD are pathologically akin with LB and LN pathology (Lewy-related pathology; LRP) prevalent in the degenerating dopaminergic neurons of the substantia nigra pars compacta (SNpc) that underlies the parkinsonism movement disorder (66,68). In stark contrast to the aforementioned diseases where asyn aggregates are mainly found in the neuronal cytoplasm as LBs and LNs, MSA is a movement disorder that may present with cerebellar signs or parkinsonism and is characterized by asyn-containing glial cytoplasmic inclusions (GCIs) within oligodendrocytes (7,66,69). GCIs are predominantly formed in white matter tracts, but morphologically unique nuclear neuronal asyn inclusions and neuronal cytoplasmic inclusions can also be found in various brain regions in MSA.
In the stereotyped pathologic progression of PD, asyn inclusions are typically observed early in the peripheral nervous system (particularly in the gastrointestinal tract) and caudal brainstem coinciding with minor autonomic symptoms, followed by the onset of parkinsonism with aggregate formation and neurodegeneration in the SNpc, with telencephalic pathology and cortical symptoms such as dementia and psychosis occurring in late stage (65,66). In LBD, progression of disease is more varied than PD in that LRP may appear in limbic and cortical regions earlier than in PD, resulting in more rapid onset of dementia that can even precede motor symptoms or SNpc pathology (66,67). MSA has its own unique progression patterns, further demonstrating the variance in disease states induced by misfolding of the same protein (7,66,69). Additionally, ;40-60% of AD cases present LRP, often restricted to limbic regions, that bears similarities to that of PD and LBD, albeit with some differences (66,67). There exist many other less common neurodegenerative diseases exhibiting diverse asyn aggregates either as primary or secondary lesions with their own unique symptoms and progression patterns (65). In addition to the diversity in progression and morphology of asyn inclusions between diseases and even across brain regions in the same disease, there is diversity in the cell types affected, as astrocytic and oligodendroglial asyn pathologies are present in addition to neuronal asyn aggregates in varying abundance, depending on the disease (21,66,67).
Variations among the synucleinopathies in terms of pathology, symptomatology, and rate of progression suggest that the form of pathologic asyn present may differ between these diseases at the molecular level similar to the concept of strains in prion disease (20,66,70). Additionally, the pathologic and symptomatic progression of PD in particular seem to be linked to the "spread" of pathologic asyn aggregates through interconnected neurons in the autonomic nervous system, which again bears similarities to prion disease (71,72). Indeed, much of the recent research on the molecular mechanisms of synucleinopathies is now based on the assumption that misfolded asyn harbors prion-like activity.
Molecular mechanisms of asyn aggregation, prion-like spread, and toxicity are key to understanding synucleinopathies Aggregates comprised of misfolded asyn are directly implicated in mechanisms of cytotoxicity and disease progression. Evidence for the toxic role of asyn is supported to date by the identification of seven missense mutations in the SNCA gene resulting in autosomal dominant synucleinopathies with associated symptoms (73)(74)(75). These mutant forms of asyn are thought to cause disease by either accelerating polymerization into oligomers/fibrils, altering the physiologic function of asyn, or a mixture of these mechanisms (74,76). Additionally, increased gene dosage through duplication or triplication of the SNCA gene similarly results in familial synucleinopathy (73,74). In animal models, overexpression of asyn and usage of aggregation-prone familial mutant forms such as A53T asyn results in robust Lewy-like neuronal asyn inclusions, neuronal death, and motor symptoms (77,78). Increasing the tendency of asyn to polymerize through mutation or increased gene dosage leads to disease in these familial cases, and in sporadic synucleinopathies, aggregation-promoting PTMs may play a similar role (22,67).
The assembly of asyn monomers into oligomeric species and eventually b-sheet-rich fibrils is important to the disease process; the formation and continued presence of asyn oligomers and fibrils in inclusion-bearing cells has been directly implicated in neuronal and glial toxicity through a number of pathways, including impairment of axonal transport, blockage of lysosomal autophagy and other proteolytic pathways, mitochondrial toxicity and oxidative stress, synaptic dysfunction, and functional insufficiency of monomeric asyn from its typical presynaptic location (reviewed in Refs. 19, 79, and 80). In addition to the direct toxic mechanisms of intracellular aggregated asyn, neuroinflammation is induced by glial detection of the asyn aggregates and cytotoxicity, which may result in further damage (reviewed in Refs. 21, 81, and 82). Although there is debate as to which form of misfolded asyn (oligomers, protofibrils, fibrils, etc.) is the most toxic, it is likely that all of these species are present in diseased cells (80).
In addition to their purported toxicity, asyn aggregates demonstrate prion-like induction of further asyn pathology and progression of disease (83). It was suggested that asyn could potentially act similar to a prion when it was found that fetal stem cell grafts placed into the brains of PD patients developed asyn inclusions within the grafts, demonstrating that a "spread" of pathology had occurred (84,85). Stemming from the inregister b-sheet structure of asyn fibrils (58) and the intrinsic lability of monomeric asyn in adopting diverse conformations necessary for its physiologic function, it became apparent from early in vitro observations that the addition of preformed asyn fibril "seeds" to a solution of asyn monomers could induce fibril elongation as monomers assume a pathologic b-sheet conformation induced from the exposed "templates" at the ends of the fibrils in a process termed "conformational templating" (13,83,(86)(87)(88). This seeded templating process bears similarities to the protein misfolding cyclic amplification technique common to the prion field, and further prion-like properties of asyn were quickly explored (89). The introduction of preformed asyn fibrils (PFFs) to cultured cells expressing monomeric asyn induced a polymerization phenomenon similar to that seen in vitro (90)(91)(92)(93)(94). In animal models, intracerebral or even peripheral injection of PFFs or human synucleinopathy brain lysate into rodents can robustly trigger the formation of LB-like pathology and motor symptoms, further corroborating the importance of prion-like conformational templating in synucleinopathies (20,(95)(96)(97)(98). The prion-like theory of asyn conformational templating may explain propagation of pathology in synucleinopathies, where for PD in particular it has been proposed that spread of fibrils through anatomic connections sufficiently predicts which regions of the brain will be stereotypically affected (the Braak staging schema) and subsequently when various symptoms will arise based on regions affected (71,72,99,100). Indeed, experimentation in various preclinical models has demonstrated that even nanomolar quantities of asyn fibrils are able to spread between neurons and even glial cells through multiple extracellular and intercellular uptake mechanisms (reviewed in Refs. 20 and 21), suggesting that the appearance of misfolded asyn aggregates may be sufficient to kick start a vicious cycle of further asyn aggregation and prionlike spread.
Pathologic asyn adopts different strains that may underlie heterogeneity in disease Like prions, misfolded asyn within fibrils can also adopt varied conformations, which bear similarities to "strains" in prion disease whereby the unique conformations of misfolded prion protein can be propagated, resulting in characteristic pathologies and symptoms particular to the initiating strain (reviewed in Ref. 70). Strains may explain why aggregation of the same protein, asyn, is able to cause a spectrum of diseases, each with overlapping but separate pathologies, symptoms, regional and cellular distribution of inclusions, and rates of progression (66).
Monomers of recombinant asyn can be induced to form into PFFs similar to fibrils found in LBs when incubated with agitation for several days (31); it was noted that upon microscopic examination, extensive heterogeneity exists in the ultrastructural appearance of these PFFs, with some displaying helical periodicity of paired protofibrils, which others lack, hinting at the existence of differing strains of asyn fibrils (31,54,57). Further experimentation in vitro demonstrated that modifying the buffer parameters used (pH, ionic strength, etc.) for PFF production results in PFFs with unique structural and biochemical characteristics stemming from the modified conformations of the misfolded asyn monomers comprising the aggregates (58,70,(101)(102)(103)(104)(105). In cellular and animal models, these variant PFFs can be functionally different in terms of their potency in seeding asyn pathology, rate of disease progression, the cell types they affect, and even the morphology of inclusions that result in mice injected with the variants PFFs, which is strongly reminiscent of differing diseases resulting from prion strains (14,70,103,105).
Extending the strain concept to human disease, asyn fibrils from MSA brain lysate have been compared with lysate from LBD and PD brains to investigate whether there are unique structural and functional differences in fibrils between these diseases (reviewed in Ref. 70). Structurally, there are purported differences in the biochemical and ultrastructural characteristics between asyn fibrils derived from MSA and LBD/PD brains; MSA fibrils appear to have more "twisted" fibril variants characterized by undulating fibril width along their length, which may not be as prevalent in LBD/PD asyn fibrils (10,31). In addition, MSA asyn fibrils can be more stable to protease digestion (106) and detergent extraction (107) and less stable in denaturation with specific solvents (105) and interact with protein aggregate binding dyes differently from LBD/PD asyn fibrils (108), which cumulatively demonstrates significant biochemical, and possibly conformational, differences in the fibrils between these diseases. Additionally, asyn inclusions between MSA and LBD/PD are not only different in their morphologies and cellular locations, but also in detection, as evidenced by monoclonal antibodies that selectively label asyn in MSA inclusions over LBD/PD (106) and the Gallyas-Braak silver staining technique that detects MSA inclusions and not typically those of LBD/PD (109).
Experimentally, MSA asyn fibrils appear to be up to 1000 times more potent in their prion-like activity compared with LBD/PD asyn fibrils or any PFF variant as measured by the time to inclusion formation and disease onset, amount of pathology, and amount of insoluble asyn formed in animal and cellular models injected with various disease lysates and PFFs (106,(110)(111)(112)(113). In addition to the increased potency, MSA asyn fibrils can impart their unique structural conformation in induced pathology measured in cell culture assays and passaging studies in mice (105,106,114). It is increasingly evident that different strains of asyn fibrils likely contribute to the different synucleinopathies or even rates of progression within the same synucleinopathy; however, it is still not understood why specifically these different strains arise.
Mechanisms by which the differing strains of misfolded asyn and subsequent fibrils could arise include modification of asyn through PTMs or altered subcellular environment promoting certain conformations. It is likely that PTMs could induce unique asyn strains, as even single residue mutations in asyn are able to structurally alter resulting asyn fibrils and functionally impact prion-like seeding and biochemical properties in cellular and animal models (115-118). Likewise, there exists a "species barrier" between mouse and human asyn, where due to their 7-amino acid difference, fibrils comprised of the asyn of one species do not seed monomers of the other species as efficiently due to presumed structural incompatibility (119)(120)(121)(122). The majority of misfolded pathologic asyn harbors PTMs such as phosphorylation at Ser-129 (pSer-129), and recent work is establishing the role of these PTMs in modifying prion-like seeding of asyn fibrils. For pSer-129 in particular, two papers have observed that asyn inclusions in MSA appear to be less reactive than LBD for common antibodies targeted against pSer-129, and biochemical analysis suggests less pSer-129 in the insoluble asyn fibrils extracted from MSA brains, suggesting that differing PTMs may indeed result in different fibril strains (106,123).
Experimentation using animal models to explore the effect of PTMs on aggregation is extensive, but far fewer studies have been conducted to determine how the presence of PTMs within asyn fibrils impacts prion-like seeding (18). Compared with a single point mutation or phosphorylation site, removal of multiple residues through truncation of asyn may have a large impact on prion-like seeding activity and strain-like alterations in structural and biochemical properties, which will be discussed herein to determine how the presence of this PTM in disease could impact progression.
Evidence for truncated asyn in health and disease

Detection in human neurodegenerative diseases
Although truncation of asyn may theoretically increase the propensity of the protein to aggregate into pathologic forms, it is important to first confirm that truncated asyn is actually present in human disease and, if so, which specific ones. Note that post-mortem interval does not affect the observation of truncated asyn, showing that truncation is not an artifact of post-mortem decay (16,26,124). Studies aiming to detect truncated forms of asyn typically rely on immunoblotting, immunohistochemistry (IHC) with truncation-specific antibodies, or MS. Immunoblotting in the form of Western blotting (WB) is the most common method utilized; however, this analysis is not always straightforward, as human asyn has an aberrant apparent mobility of ;19 kDa on SDS-PAGE compared with the predicted 14.4 kDa (6); thus, some truncations can result in unexpected shifts in mobility. In addition, the C-terminal truncation of asyn removes acidic residues, resulting in an increased protein pI, and thus typical WB procedures may not detect the full extent of C-truncated forms of asyn, as the pH levels of common blotting buffers are not ideal for the electrophoretic transfer onto membranes (22,26,125). The IHC approach that relies on antibodies specific for modified forms of the protein presents its own challenges, as it is difficult to produce antibodies specific for a truncated form of a protein without any cross-reactivity for the full-length form. MS avoids many of the shortcomings inherent to other techniques but is also the least common approach for detecting truncated asyn. Despite these challenges, robust evidence exists that truncation of asyn is occurring in a way that is largely disease-specific, and the estimated or exact truncated forms of asyn identified are summarized in Table 1 (16, 17, 26-28, 62, 67, 124, 126-135). Overall, studies relating to truncated asyn forms noted their presence on WB from LBD, PD, or MSA brain lysate; however, a few were particularly comprehensive in detecting their presence and attempting to identify exact truncation sites through the generation and IHC application of truncation-specific antibodies (16,(128)(129)(130)(131)(132)(133) and MS (16,27,132,134) to study diseased brain lysate.
In almost all studies utilizing WB of LBD and PD lysate, a common trend is observed where in the insoluble fraction containing aggregated asyn there is a major ;17-19 kDa band representing FL asyn; a ;12-15 kDa band only reactive to Nterminal asyn antibodies, suggesting C-truncated forms of asyn; and 1-2 additional minor bands of ;9-10 and ;6-8 kDa, which may be both N-and C-truncated asyn (Table 1). Each truncation band seems to contain a mix of 2-3 truncated asyn species that can be resolved using MS or specific antibodies. Confirmed C-truncated forms of asyn found in detergent- Table 1 Summary of observational studies of truncated asyn in human synucleinopathies The nomenclature for truncation when only approximate is indicated by a slash; 1-115/120 indicates C-terminal truncation occurred somewhere between residues 115 and 120.

Reference
Design Identified truncations Remarks 126 WB of MSA brains ;12-13 kDa Strong 12 kDa band in MSA and not controls. 62 WB of LBD, PD brains 12 kDa 6 kDa Both bands detected only with NAC antibody. 6 kDa more disease-specific. 28 WB MS, WB, and IHC with truncationspecific antibody at residue 103 for PD and LBD brains 1-103 C-truncated asyn 1-103 asyn present in both soluble and insoluble fractions of PD and LBD but not controls. Inclusions labeled in PD and LBD but not controls with the 103-specific antibody. Ratio of 1-103 asyn to FL asyn increased in SNpc relative to cortex in PD. 133 WB of PD brain lysate and IHC with antibody specific for pSer-129 only if also C-truncated  (Table 1). In particular, 1-119 and 1-122 appear to be some of the most common forms of truncated asyn and comprise the major portion of the ;12-15 kDa band; these have each been detected with multiple specific antibodies and MS experiments where their relative abundances were as high as 20-25% that of FL asyn in the insoluble brain fraction ( Table  1). The forms of asyn truncated at only the extreme N or C terminus, such as 1-135 and 5-140, have also been determined to be highly abundant in two separate MS experiments (16,27); however, they may not be distinguishable from FL asyn on a typical WB due to size similarities and both becoming pSer-129-positive (16) ( Table 1). A caveat to determining the most common truncated forms of asyn is that each study often examined tissue from only one region, and the few studies that compared truncation in differing regions often demonstrated variation in the amount of truncated asyn from one region to the next (67,132,134). It is difficult to arrive at even a rough number for the percentage of asyn that is truncated in disease compared with controls due to the variety of truncated forms, regional variation, and detection issues; however, based only on the 2-3 major truncation bands visible on Western blots from multiple studies, it seems that ;15-30% of insoluble asyn in PD/LBD is truncated (Table 1). Notably, almost all studies examining truncation have focused on LBD and PD and not MSA, although at least two studies have observed a ;12-15 kDa band in MSA and 119 C-truncated asyn in GCIs with a specific antibody (16,126).
In the detergent-insoluble fraction of LBD/PD lysate (containing aggregated forms of asyn), multiple studies have noted an increase in the truncated asyn/FL asyn ratio compared with the soluble fraction and control fractions, suggesting that the presence of certain truncations may be enriched in pathologic inclusions with disease implications (16, 17, 26, 28, 67, 127-129, 131, 132, 134) (Table 1). IHC of pathologic inclusions from synucleinopathy samples using truncation-specific antibodies corroborates the suggested importance of these species. Antibodies specific for asyn truncated at residues 9, 103, 110, 119, and 122 have been used to detect truncated asyn within LBs, LNs, and GCIs (16,(128)(129)(130)(131)(132)(133), where three studies have stated that the truncated forms of asyn are more centrally located in these inclusions compared with FL asyn, suggesting a possible early role for truncated asyn in the generative processes of these inclusions (124,128,130). Regional variation in the amount of truncated asyn may be important in understanding vulnerability of various loci to early asyn aggregation if truncated asyn is linked to initial inclusion formation. Indeed, PD is thought to have early pathologic asyn formation in the enteric nervous system (98), and it was observed using the appendix from PD patients that levels of truncated asyn, particularly C-truncated at residue 125, were greatly elevated compared with the SNpc from the same patients and the appendices of controls (134). Another study found elevated levels of 103 C-truncated asyn in the colon of PD patients (135). Like-wise, findings from our laboratory showed that C-terminal truncation may be highly prevalent in the medial temporal lobe in LBD, which is afflicted early in disease, compared with the cingulate cortex (67). Regional variation of truncated forms of asyn may also have implications for toxicity; for the 1-103 Ctruncated asyn, a specific antibody was able to detect higher amounts of 103 C-truncated asyn in the SNpc, where extensive cell death is present, compared with the cortex in PD cases (132). Another study using a truncation-specific antibody in PD brains found that a doubly N-and C-truncated form of asyn was prone to form punctate aggregates in mitochondria that were associated with mitochondrial dysfunction (133). The consequences of truncated asyn in pathophysiologic processes are more easily studied in preclinical models, which will be further discussed; however, upon examination of post-mortem samples from diseased human tissue, it is evident that truncated forms of asyn are involved in the formation of toxic asyn aggregates.
Although not confirmed to be due to truncation, unique histologic features detectable only with antibodies raised against central epitopes of asyn and not those against the extreme N or C terminus (presumably lost due to truncation) may provide additional evidence that truncated forms of asyn are involved in disease. Indeed, central asyn epitope antibodies have been reported by our laboratory and others to prominently detect pathologic forms of asyn in astrocytes in LBD that are not often detectable with N-or C-terminal antibodies (21,67,(136)(137)(138). Likewise, central asyn epitope antibodies have been reported to detect high-molecular-weight bands in the insoluble fraction of LBD lysate that are not apparent when using N-or C-terminal antibodies; this may be due to oligomeric or ubiquitinated forms of truncated asyn (67,136,138).
Cumulatively, these studies show that truncated forms of asyn are present in disease-associated aggregates in synucleinopathies and may have implications for the initiation and progression of disease.

Detection in healthy brains and splice variants
In the soluble fraction of aged control brains that do not have a synucleinopathy, there is quantitatively less truncated asyn than in insoluble disease fractions (16, 17, 26, 28, 67, 127-129, 131, 132, 134) (Table 1); the only truncation band present in some studies appears to be the C-truncated 12-15 kDa band, particularly 1-119 asyn, which is typically of lesser intensity than insoluble disease fractions, although a few studies have not found significant differences (16,17,26,27,62,124). At least one article states that more 1-119 asyn is detected in the soluble fraction of diseased brains compared with controls (26), whereas other studies see little to no change (16,17,27,124). Although 1-119 asyn is found to some degree in the soluble fraction of both diseased and control brains, it is only found to form insoluble pathologic inclusions in disease (16,130).
Although not necessarily truncated by a protease, there exist at least three splice variants of asyn that are missing either exon 5 (residues 103-130), exon 3 (residues 41-54), or both, resulting in variants termed asyn 112, asyn 126, or asyn 98 (139). These variants, particularly those missing C-terminal residues through exon 5 removal, may display properties similar to those of Ctruncated asyn, such as increased aggregation, which will be further discussed (reviewed in Ref. 139). There have been multiple observations that the mRNA ratios of these splice variants relative to FL asyn are altered in disease, often in a region-specific manner (reviewed in Ref. 139); however, it is difficult to correlate this with resulting changes in asyn proteoforms due to the low abundance of the variants, and consequently they are not typically observable on a Western blotting or IHC even with a splice variant-specific antibody (26,124). Additionally, the majority of truncated forms of asyn isolated from LB extract are consistent with formation from various proteases (discussed below), suggesting that alterations in splicing machinery are not the major source of asyn variants in disease (41).

Detection in mouse models of disease
In murine models based on overexpression of FL human asyn (often containing the aggregation-prone A53T mutation), truncated asyn is similarly observable when insoluble LB-like aggregates form (140). Two studies found that overexpression of human A53T asyn in mice results in clear truncated asyn bands of 12, 10, and 8 kDa in size that are enriched in the insoluble fraction in regions containing LB-like pathology, which is similar to the patterns described for human disease (26,140). These bands are not found in nontransgenic mice (26,140), as they do not develop insoluble asyn. Also, it has been found that overexpression of human asyn in cultured cells results in more asyn truncation compared with mouse asyn overexpression (26). A similar pattern was observed with a separate A53T asyn mouse model, where three truncation bands were observed in the insoluble fraction of sick mice with two bands (;13 and ;10 kDa) being only C-truncated and one (;7-8 kDa) being both N-and C-truncated; these truncation bands were roughly similar to those seen in the insoluble fraction of LBD lysate in the same study (17). A more recent experiment used MS to determine that truncated asyn in sick A53T human asyn transgenic mice (same mouse line as the previous study) is mainly the 1-122 form found in humans, but also 1-90 that has not previously been detected in human disease; thus, some differences may be present between mouse models and human disease; lesser amounts of other forms of truncated asyn were also detected, including 1-103, 1-124, and 5-140, that are prominent in human LBs (141) ( Table 1). In a mouse model overexpressing A30P asyn, punctate inclusions in sick mice were labeled with truncation-specific antibodies but not in non-sick transgenic mice or age-matched nontransgenic controls, further suggesting that the appearance of truncated asyn is linked to the appearance of pathologic inclusions; other truncations were identified using MS (128). Even in mice overexpressing WT human asyn, accumulation of C-truncated asyn at residue 122 was noted upon Western blotting of symptomatic mice that was not present in nontransgenic mice; 122 C-truncated asyn was also noted to be present in punctate inclusions and dystrophic neurites with similar patterns to LBD samples in the same study (131).
In addition to overexpression models of synucleinopathy, truncation of asyn appears to be present in prion-like models as well. When nontransgenic mice were intracerebrally injected with PFFs, a truncation-specific antibody was used to detect a doubly N-and C-truncated form of asyn linked to mitochondrial toxicity in human samples, which was not present in controls (133). From these reviewed studies, it is apparent that truncated asyn forms in murine models of synucleinopathy in relation to the development of pathologic asyn inclusions and subsequent symptoms.
Pathologic consequences of asyn truncation C-terminal truncation of asyn increases pathologic aggregation The C terminus of asyn harbors multiple protective features to limit pathologic misfolding and aggregation due to several structural factors, and loss of these through truncation promotes fibril formation. First, the C terminus is the most highly charged portion of asyn due to 15 acidic residues that promote a disordered protein structure and maintain protein solubility (142)(143)(144). Events neutralizing the C-terminal charge, such as lowering the pH, binding of cations to residues 119-124, or mutation of the negative residues to neutral or positive ones, promote robust spontaneous aggregation of asyn into oligomers or fibrils, whereas the addition of extra negative residues decreases fibril formation (36,142,143,(145)(146)(147)(148)(149)(150). Furthermore, the fusion of the C terminus of asyn to other proteins is sufficient to solubilize them and protect against heat-induced aggregation, demonstrating its anti-aggregative properties (48,151). Second, the C terminus has autochaperoning abilities, whereby asyn often adopts conformations in which the C terminus contacts the hydrophobic NAC to shield it from pathologic templating interactions; this behavior is mediated by hydrophobic motifs in residues 115-119 and 125-129 (36,(152)(153)(154)(155)(156)(157). When long-range contacts are impaired by the introduction of nanobodies against the C terminus, or the transient interaction with NAC is unfavorable due to high temperature, aggregation is accelerated similar to the impairment of charge (145,147,153,156). In addition, the C terminus is the most proline-rich region in asyn, with 5 proline residues known to be unfavorable in b-sheet formation; site-directed mutagenesis of these prolines to alanines promotes aggregation (142,158). All of these protective features are lost upon C-terminal truncation, and consequently C-terminal truncation of asyn has been repeatedly shown to promote in vitro oligomer and fibril formation far more than FL asyn (17, 22, 26, 29-31, 33-35, 132, 134, 159-164). Comparatively, N-terminal truncation has been shown to either not affect aggregation or mildly decrease it (56,165,166), although one study did observe increased aggregation with the removal of N-terminal repeats (167). The lesser effect of N truncation on aggregation is expected due to the comparatively inconsequential changes in basic biochemical properties, such as pI and hydrophobicity, that are induced when the amphipathic N terminus is truncated (22,165).
In studying how C-terminal truncation affects aggregation in vitro, most studies first generate truncated asyn through purification of recombinant truncated asyn or through incubation with common cellular proteases. Experimentation to measure aggregation rates usually relies on incubating truncated monomers of asyn at physiologic temperatures with shaking and then measuring the rate of amyloid-binding dye fluorescence or biochemical insolubility over time (17, 22, 26, 29-31, 33-35, 132, 134, 159-164). A few studies examined aggregation of truncated forms of asyn confirmed to exist in LBs (22,34,132,142,161,163) or formed from common proteases (132,134,160,162,164,168), but most experiments utilized arbitrary C truncations. In general, the in vitro studies cumulatively show that the exact site of truncation is not as important as the amount of the C terminus that is lost when assessing aggregation propensity. Removal of only a couple residues, such as with 1-133 and 1-135 asyn, does not increase the aggregation rate or extent much compared with larger truncation, such as with 1-115 asyn (22,29,31,34,142); our laboratory noted that aggregation propensity as measured by both the rate and total extent of aggregation increases rapidly once more than ;10 Cterminal residues are removed (22). Removal of C-terminal asyn residues somewhat linearly increases aggregation rate and extent up to residues ;85-90, where the NAC domain begins; as NAC residues form the core of the amyloid fibrils, further truncation decreases aggregation (29,32,36,52). C-truncated asyn may be pathologically relevant not only due to the robust aggregation propensity, but also in synergistically accelerating the aggregation of FL asyn. Experiments conducted in vitro have reported an enhancement of FL asyn aggregation when co-fibrillized with C-truncated asyn monomers (17,22,26,29,132,160). Indeed, many of these experiments used substoichiometric quantities of truncated asyn and demonstrated that their presence can induce FL asyn to aggregate quickly and to a larger extent than would be observed without the truncated asyn present (22,26,29). Moreover, it was demonstrated that this phenomenon can lead to FL asyn aggregation at substantially lower concentrations, where it would otherwise not occur, close to the physiologic 3 mM cellular concentration of asyn (13). The extent to which C-truncated asyn induced FL asyn to aggregate depended on the extent of truncation, where more heavily truncated 1-115 asyn induced the most FL asyn aggregation compared with 1-135 asyn (22). The mechanism for this synergistic aggregation appears to be co-polymerized C-truncated/FL asyn fibril formation, as both FL and C-truncated forms of asyn became insoluble in equal proportions, and both forms are found co-assembled into the same fibrils, as shown with immuno-EM labeling (22).
It is now well-established that C-terminal truncation of asyn, in line with predicted biochemical changes, greatly increases the tendency of asyn to misfold into pathologic fibrils. The appearance of even small amounts of C-truncated asyn may catalyze the misfolding of FL asyn into fibrils that may initiate the prion-like cycle, as will be discussed.

Truncation of asyn increases aggregation in cultured cells and murine models
C truncation of asyn increases aggregation in vivo as well; in addition to the structural aspects, removal of the C terminus may further enhance aggregation in a cellular milieu through impaired asyn degradation due to loss of consensus motifs for ubiquitin ligases and chaperone mediators of autophagy (169)(170)(171) and through loss of normally protective C-terminal interactions, such as binding of oxidized dopamine (172,173). Although not as numerous as the in vitro studies, experiments have been conducted demonstrating that C truncation of asyn promotes aggregation into pathologic inclusions in cultured cells and animal models (22,132,159,(174)(175)(176)(177)(178)(179)(180)(181)(182)(183). These same studies often demonstrated toxic consequences from C truncation of asyn that will be discussed further.
In cultured cells, C-truncated asyn can spontaneously aggregate more readily than FL asyn and is more amenable to misfolding induced by PFFs. In stably transfected astrocytoma cells, it was demonstrated that 1-111 asyn formed more inclusions than FL asyn-containing cells and additionally, the 1-111 asyn cells formed higher-molecular weight species detectable on WB that may be oligomers not seen with FL asyn (177,178). Likewise, overexpressions of 1-120, 1-108, and 1-103 asyn have been demonstrated to form more and/or larger inclusions than FL asyn in cultured cells, including with unique inclusion morphology, as 1-120 asyn formed larger-volume inclusions than FL asyn in primary neurons (132,175,183). In a prion-like induction cell model, our group studied eight Ctruncated forms of both human and mouse asyn overexpressed in HEK293T cells and demonstrated that with increasing extent of C truncation, there is more aggregation compared with FL asyn when exposed to the same PFFs as measured by amount of insoluble asyn formed and number of inclusions observed (22,159).
Overexpression through viral transduction or transgenesis of certain C-truncated forms of asyn in Drosophila and in murine models have demonstrated that these truncations tend to aggregate into pathologic inclusions more than the FL protein, with additional toxic consequences (132,174,(179)(180)(181)(182)(184)(185)(186). Transgenesis with 1-120 asyn has been the most widely used mouse model for truncated asyn, as three different lines using different promoters (CamKII-a or rat TH) have demonstrated pathologic inclusion formation in disease-relevant regions, such as the SNpc, striatum, cortex, and olfactory bulb (174,179,181). In Drosophila overexpressing either FL or 1-120 asyn, it was noted that 1-120 asyn resulted in more inclusions and proteinase K-resistant asyn (180). Adeno-associated virus transduction of either 1-103 or FL asyn into the SNpc of nontransgenic mice demonstrated that the 1-103 asyn resulted in more inclusions and dopaminergic cell death (132). Additionally, 1-130, 1-119, or 1-93 asyn-overexpressing animals were created that did not show evidence of inclusion formation but did all have various pathologies and dopaminergic deficits (184)(185)(186).
The tendency of C-truncated asyn to readily aggregate into pathologic inclusions has been studied in vivo, but additionally the synergistic aggregation with FL asyn observed in vitro also occurs in cellular and animal models. Our group demonstrated in HEK293T cells that co-expression of FL asyn with increasingly C-truncated asyn leads to a greater proportion of aggregated, insoluble FL asyn with the presence of more extensive C truncation (22). Likewise, two animal models demonstrated increased asyn inclusion formation when both FL asyn and one of 1-110 or 1-120 asyn were present, which was more than additive (181,182). These results further demonstrate that asyn C truncations may assist in initiating FL asyn fibrillization. Cumulatively, the in vitro, cell culture, and animal experiments reveal that C-truncated asyn is primed to aggregate, which may be crucial in initial disease pathogenesis.

Fibrils containing truncated asyn have distinct structures and biochemical properties
Both N and C truncation of monomeric or fibrillar asyn may result in eventual pathologic fibrils with unique structural and biochemical features, as measured through a battery of assays, including CD, EM, protease digestion and accessibility, and others that will be discussed. Differences in these properties may be suggestive of differing conformations and strains, with resulting alterations in prion-like properties.
As expected from removing a highly acidic region, C truncation of asyn alters biochemical properties, including increased pI and hydrophobicity (22); for N truncation, the trend is reversed, and increased hydrophilicity and net electrostatic charge is observed (165). CD analysis of C-truncated (29,31,36,187,188) and N-truncated (167) forms of asyn mostly demonstrate the typical random coil spectra of FL asyn (6). One study did note that incubation of monomeric FL asyn with a protease implicated in ;10-122 asyn formation, calpain I, resulted in increased b-sheet structure (128). When placed in solvent that promotes structure formation, FL asyn became partially a-helical, which is physiologic (167); however, N-truncated forms (167) and C-truncated forms (187) became more structured with altered spectra (167,187,188), which has been interpreted as the possible formation of misfolded intermediates that may be oligomer-prone (147,188).
When fibrils were produced from monomers of truncated asyn in vitro, immediate structural differences were noted compared with FL asyn fibrils. For C truncation, CD analysis and similar techniques indicate a higher percentage of b-sheet structure than FL asyn fibrils in most studies (29,31,56) (although one study detected a decreased percentage (160)), and amyloid-binding dyes may have altered affinity for C-truncated fibrils, indicating an alternative fibril structure (29,60,160,163,189). The nature of b-sheets within the fibril core of C-truncated asyn fibrils may be fundamentally different from that of FL asyn, as one study found evidence that b-sheets are highly "twisted" in C-truncated fibrils (33). Biochemically, Ctruncated fibrils have demonstrated altered protease digestion patterns compared with FL asyn fibrils, which again could imply variable epitope exposure through unique fibril conformations (22,161,189).
When studied with EM and atomic force microscopy, most studies find that C-truncated fibrils are unbranched and typical amyloid in appearance with some differences from FL asyn fibrils (22, 29-31, 33, 35, 36, 56, 60, 132, 160, 161, 189). Fibril width is decreased after C truncation of ;20 residues, which is consistent with the theory that the unstructured C terminus "hangs" off the fibril core (22,29,33,56,60). These fibrils also tend to be shorter longitudinally, likely due to the increased aggregation rate (22,31,35,189), and have denser lateral packing between fibrils, which may be due to less C-terminal charge (22, 31, 35, 36, 60, 189). Other morphologic properties of C-truncated fibrils observed with EM and other techniques include the formation of "ribbon-like" fibrils (31) and increased "twisting" of paired protofilaments within the fibril compared with FL asyn (33). The increased twisting in particular may be a key difference, as recent characterization of C-truncated and FL asyn fibrils through the use of cryo-EM shows that the protofilaments in the truncated fibrils do have a higher helical periodicity compared with FL asyn, and the increased periodicity is more extreme with more extensive C truncation (58,161,190,191). This increase in twisting upon C truncation is attributed to less steric hindrance between asyn subunits in the fibril core as the unstructured C termini are removed, leading to tighter packing and increased twisting, resulting in a fundamentally different fibril structure (33,161).
Comparatively, fibrils comprised of increasingly N-truncated asyn had less b-strand structure than FL asyn fibrils in one study (165) but more in two others (56,167) and also altered amyloid dye binding (189). Upon EM and atomic force microscopy examination, the N-truncated asyn formed typical amyloid fibrils (56,167,189); however, N truncation also decreases the fibril width, as part of the N terminus is not incorporated into the fibril core (56,167), and in contrast to C-truncated fibrils are longitudinally longer (189). Additionally, protease digestion patterns suggested that N truncation harbors a different fragmentation pattern compared with FL or C-truncated asyn (189). These results suggest that unique fibril structures may be present for N-truncated asyn; however, far fewer experiments have been conducted relative to C truncation, and the results are less clear and require further study.
Although experiments utilizing C-truncated asyn fibrils produced from recombinant C-truncated asyn are useful, it has been suggested that more physiologically relevant information may be gathered from studying fibril structures incubated with proteases, as this likely results in a mix of N-and C-truncated and FL asyn in the same fibril. Many of the same structural alterations seen with recombinant C-truncated fibrils are observed on FL asyn fibrils exposed to common proteases, including a fibril width reduction (32,39,56) and an increase in fibril twisting (141). One study noted that a unique single-protofilament fibril structure observed only with recombinant doubly N-and C-truncated asyn was able to be generated by protease digestion of a FL asyn fibril, suggesting that truncated fibrils produced through use of recombinant protein may be similar to those produced through proteolysis (56). It seems paradoxical that cleavage of the extreme N and C terminus of asyn would fundamentally alter the fibril structure, as they are thought to be largely unstructured in fibrils (58); however, the cumulative evidence suggests that C truncation in particular can induce important structural changes with consequences for prion-like seeding, as will be discussed.

Fibrils containing truncated asyn may alter prion-like seeding of pathology in vitro
The effect of asyn truncation is important to understand in relation to prion-like seeding, as asyn fibrils in healthy neurons are quickly trafficked for lysosomal processing, where extensive truncation of exposed N-and C-terminal regions can readily occur (20,41,141,192). Additionally, C-truncated asyn may have outsized importance in initial disease regions as discussed previously, and reducing prion-like activity at this stage may be key to disease prevention. The presence of truncated asyn in fibrils can alter fibril structure and thus theoretically affect prion-like seeding in a strain-like manner. Unlike aggregation properties, the exact truncation site may be important in prionlike seeding analogous to the differing templating properties of Ab 40 and Ab 42 (193) or the species barrier in prion phenomena (120), where even a slight difference in the amino acid sequence can have drastic consequences for pathologic propagation.
The ability of truncated asyn-containing fibrils to induce prion-like conformational templating has been studied in multiple experimental paradigms by producing PFFs containing truncated asyn (using either recombinant protein or incubation with proteases) and then measuring the formation of asyn fibrils when added to an in vitro solution of physiologic FL asyn, cultured cells expressing asyn, or mouse models harboring endogenous asyn. Although seemingly straightforward, in vitro studies have produced mixed results, with some observing increased templating ability (132,141,161,189) of FL asyn by truncated fibrils and others observing a decrease (32, 33, 163) compared with FL asyn fibrils. Three of the studies that observed increased templating activity with truncated fibrils used either the fibrils comprised of 1-103 asyn or fibrils digested by the protease implicated in formation of 1-103 asyn, AEP (132,141,161). Furthermore, it was demonstrated that the 1-103 asyn fibrils or AEP-cleaved fibrils can propagate their unique twisted fibril morphology onto FL asyn, demonstrating what could be construed as true prion-like strain behavior (141,161). The only other truncated asyn fibrils to show increased seeding activity were comprised of 1-120 asyn, where FL asyn was rapidly assembled into fibrils in the presence of the 1-120 asyn fibrils; furthermore, the resulting templated fibrils again had a unique twisted morphology imparted onto them that was not seen when seeded with FL asyn fibrils (189). All other examined truncated asyn fibrils displayed reduced seeding activity of FL asyn compared with FL asyn seeds, including fibrils comprised of 1-130 asyn, 1-110 asyn, 1-108 asyn, 1-119 asyn, 10-140 asyn, 30-140 asyn, and calpain Icleaved asyn fibrils that were characterized to contain 1-114 and 1-122 asyn (32, 33, 163); 21-140 asyn fibrils had similar seeding activity to FL fibrils (189), and a separate study using 1-103 asyn fibrils observed decreased seeding in contrast to the other 1-103 asyn studies (163). These results show little trend regarding extent or exact site of truncation, suggesting that prion-like activity of fibrils comprised of each form of truncated asyn may differ.
The heterogeneous outcomes of the in vitro seeding studies with truncated fibrils may be somewhat understood in the context of homotypic seeding. Homotypic seeding is the theory that there must exist structural compatibility between the seeding fibril and templated monomeric protein; this is usually applied to understanding the species barrier in prion propagation, where even one amino acid difference can greatly impair the addition of monomers to the seed fibril (120). One study noted improved seeding capacity of the truncated fibril when added to monomeric asyn harboring the same truncation (33).
Indeed, although 1-108 asyn fibrils seeded FL asyn poorly, they promptly templated 1-108 asyn monomers to elongate the seed fibril (33). This may be a simplistic explanation, as it has also been observed that FL asyn fibrils may seed C-truncated asyn monomers equally or superiorly to FL asyn monomers (22,33,159,163), as the increased aggregation propensity of C-truncated asyn must also be taken into account, but the general idea of compatibility between seeding fibril and templated monomer is likely occurring to a degree.
It has been proposed that the highly twisted structure of Ctruncated asyn fibrils may impair the addition of FL asyn monomers to elongate the fibril, as there is increased steric hindrance for the added FL asyn monomer still containing a C terminus; the reverse, where C-truncated asyn elongates a FL asyn fibril, would not be a problem (22,33). It seems likely that the incompatibility between truncated asyn fibrils and FL asyn monomers explains the findings for most experiments; however, for certain truncated fibrils, such as those containing 1-103 asyn and 1-120 asyn, there may be true prion-like strain propagation to the templated FL asyn. Overall, no theory adequately explains the current in vitro findings for seeding with truncated asyn fibrils, and prion-like activity of truncated fibrils must be evaluated on a case-by-case basis, depending on the truncation present.

Fibrils containing truncated asyn have strain-like variation in seeding capacity in cultured cells and murine models
Extending the study of prion-like seeding capacity of truncated asyn fibrils to cellular and animal models is complicated due to additional extrinsic factors now involved, such as modulation of cellular uptake and spread of fibrils due to truncation (194,195), trafficking to subcellular compartments (196), and structural changes to the truncated fibril due to additional PTMs; most of these variables have not been studied in relation to truncated asyn fibrils or even typical asyn seeding experiments. Nonetheless, cellular (22, 38, 91-93, 103, 120, 159, 189, 197, 198) and animal (135,159,189,199) experiments using fibrils containing truncated asyn have shown some similar findings to in vitro studies.
In cellular and animal models, there are again heterogeneous results, depending on the exact truncation, where most studies observe decreased seeding of FL asyn when fibrils are comprised entirely of truncated asyn (22,103,120,159,189), some see increased seeding of FL asyn (38,160,189), and others observe equal seeding capacity (91-93, 97, 159, 189, 197-199) compared with FL asyn fibrils. In most of these studies, analyzing the seeding behavior of truncated asyn fibrils was not the focus, and likewise mostly qualitative results are available. Fibrils comprised of 21-140 asyn (91,93,97,189,198,199), 58-140 asyn (92), and 1-89 asyn (92) have been observed to induce FL asyn aggregation and inclusion formation equally to FL asyn fibrils, whereas 1-99 asyn fibrils showed a decrease (120). Despite the in vitro findings that 1-120 and 1-103 asyn fibrils were excellent seeds of FL asyn, in cellular models, 1-120 asyn was either equal to or less than FL asyn fibrils in seeding (91,92,103,189,198) and likewise for 1-103 asyn (135).
Examining only studies that explicitly set out to study the seeding behavior of truncated asyn in cellular or animal models provides more outcome measures and discernible trends (22,38,159,160,189). The only four truncated forms of asyn to demonstrate increased seeding of FL asyn as measured by inclusion counts or insoluble asyn formation were fibrils comprised of 31-140 and 11-140 asyn in cells and mice (189), 1-121 asyn in cells (160), and FL asyn fibrils cleaved by cathepsin B in cells (38). In one comprehensive study, it was found that removal of 10, 20, or 30 C-terminal residues or 20 N-terminal residues resulted in equal or slightly less seeding of FL asyn pathology in cultured cells and nontransgenic mice, whereas removal of 10 or 30 N-terminal residues increased seeding in the same models; this was in contrast to the in vitro results that predicted lesser seeding for all except 1-120 asyn (189). In our laboratory we have observed a trend utilizing eight different C truncations (residues 1-135) in both mouse and human asyn and observed that in general, increasing C truncation leads to lessened seeding of FL asyn in both cellular and animal models (22,159). Overall, whereas removal of the C terminus appears to impair cross-seeding of FL asyn to varying degrees, depending on the exact truncation, there appear to be a few truncations that similarly to in vitro results may demonstrate true prion-like strain propagation and overcome structural incompatibility, and particularly N-terminal truncation may lead to a potent increase in seeding capacity for FL asyn. Additionally, many fibrils comprised of truncated asyn demonstrating potent seeding in vitro failed to produce similar findings in cellular and animal models, suggesting that prion-like properties of these fibrils should be chiefly studied in the more complex cellular environment.
Last, whereas evaluation of the prion-like properties of fibrils fully comprised of truncated asyn may be of benefit, physiologically it is unlikely that a fibril would become entirely truncated. Our laboratory has demonstrated that fibrils containing both C-truncated and FL asyn in a 1:1 ratio seed similarly to, or slightly less than, FL asyn fibrils, suggesting that the extreme structural alterations of fibrils comprised of C-truncated asyn are attenuated in this configuration (22,159). Additionally, fibrils comprised of truncated asyn may interact with other neurodegenerative proteins as well, as a seed containing both 1-103 truncated asyn and truncated tau was superiorly able to induce asyn pathology when injected into mice compared with 1-103 or FL asyn fibrils, which were themselves about equal in prion-like potency (135).
Overall, the strain-like properties of asyn fibrils containing multiple N and C truncations along with other PTMs as would be expected to occur in disease are not currently known and warrant further investigation based on the discussed findings where alterations in the terminal regions can greatly alter prion-like seeding. Most investigative models of synucleinopathy do not appreciate the impact of PTMs on seeding; however, the results discussed here demonstrate that truncation in particular likely contributes to strain-like variation in prionlike properties underlying variance in disease properties.

Truncation of asyn may increase cellular toxicity through direct and indirect mechanisms
Truncation of asyn may worsen certain modalities of toxicity in both monomeric and polymerized forms, including direct toxicity and indirect mechanisms (loss of function due to truncation or potentiation as an inflammatory agent (200)). Particularly, multiple studies have observed either an increase in oxidative stress or inability to handle oxidative stress in cells containing C-truncated, but also N-truncated, asyn (17,133,177,201). Additionally, animal models overexpressing C-truncated asyn in dopaminergic cells have invariably found dysfunction to occur that may be secondary to increased oxidative stress (132, 174, 179-182, 184-186, 202).
Dopaminergic cells of the SNpc are particularly prone to oxidative stress due in part to baseline high bioenergetic demands and production of reactive dopamine quinones, and even slight perturbances in reactive oxygen species production through mechanisms such as mitochondrial dysfunction may induce toxicity (203). When overexpressed, C-truncated forms of asyn appear to be particularly potent in compromising various cell lines in their ability to handle oxidative stress, and the effect is more than that exerted by FL asyn (17,160,177,201). Cell viability was generally decreased when C-truncated forms were expressed compared with FL asyn, demonstrating the toxicity of these species (17,160,177,201). Mechanistically, C-truncated asyn may cause oxidative stress through mitochondrial dysfunction, as a doubly N-and C-truncated protofibrillary form of asyn has been shown to accumulate in the mitochondria of diseased cells with deleterious consequences for the hosting mitochondria (133); similar findings were observed with 1-93 truncated asyn (186). These toxic effects may primarily be mediated by C truncation of asyn even though N truncation can also be present, as N truncation alone has not been demonstrated to be more toxic than FL asyn (166).
Similar to cell culture findings, overexpression of C-truncated asyn in animal models is demonstrably toxic, often more so than FL asyn when comparison is performed. Animal models have displayed motor symptoms, filamentous neuronal inclusion formations, and dopaminergic dysfunction in particular, which may be attributed to the cellular vulnerability to oxidative stress (132, 174, 179-182, 184-186, 202). Indeed, overexpression of multiple forms of C-truncated asyn, including 1-93, 1-110, 1-119, 1-120, and 1-130, leads to impaired dopaminergic cellular function evidenced by reduced striatal dopamine (179,181,184,185,202), nigral TH cell death (132,182,184,186), and resultant behavioral deficits in tests of motor and cognitive ability (132,182,184,186).
Although not well-studied, C-truncated asyn is a potent agonist of TLR4 receptors on immune cells, as one study found increased release of inflammatory cytokines, such as IL-6 and TNFa, with 1-111 asyn compared with FL asyn (200). Inflammasome-related enzyme caspase-1 is also known to produce Ctruncated 1-121 asyn, and toxicity from 1-121 asyn overexpression can itself trigger further caspase-1 activity in a vicious cycle (160,162). We have previously reviewed (21) the role that potentially truncated asyn may play in resident neuroinflammatory astrocytes in the brain in synucleinopathies, and asyn PTMs such as truncation add a new domain of inflammatory mechanisms in synucleinopathies needing further study. Ctruncated asyn is mainly studied due to its remarkable ability to aggregate into pathologic fibrils, but removal of C-terminal residues can also produce cellular dysfunction through mitochondrial and oxidative stress, adding another layer of complexity to the pathologic role of truncated asyn in disease.

Physiologic and pathologic production of truncated asyn
Endogenous asyn is chiefly degraded through autophagic pathways (169,204), whereas pathologic asyn fibrils are trafficked to lysosomes for sequestration and elimination by cathepsins (20,39,41,141,192). In both instances, proteases capable of producing truncations relevant to disease act on asyn, and these proteases are described here and displayed in Fig. 1 along with C-terminal cleavage sites.
The 20S proteasome may have some activity in degrading cytosolic, soluble asyn and appears to mainly C-truncate soluble asyn to the common 1-119 form, with 1-110 and 1-83 also being produced (17,205). Both 1-119 and 1-110 have been detected in LBs (Table 1); however, the proteasome is not the major producer of 1-119 asyn, as proteasome blockade does not affect the appearance of the major ;12-kDa C truncation (124) observed in disease lysate (Table 1). Although not considered the major pathway for asyn clearance, proteasomal dysfunction may have some relevance in synucleinopathies, as monomers of asyn incubated with 20S proteasome demonstrate increased aggregation (17), proteosomal subunits are detected in LBs (205), and proteosomal activity is diminished in the SNpc in PD (216).
The cytosolic proteases calpain I and caspase I may be an important link between cellular stress and asyn aggregation as they both predominantly C-truncate asyn in response to elevated intracellular calcium (calpain I) or inflammasome activation (caspase I). Calpain I has been identified to cleave in both the extreme N terminus and C terminus of fibrillar asyn and also the NAC region of monomeric asyn (40); the main truncation stemming from this protease is cleavage after residue 122, although cathepsins may be the main source of the 122 truncation (37,39). Monomers cleaved by calpain I were observed to readily aggregate in one study (32) but were fully degraded in another (128), which may be due to differences in experimental conditions. Calpain I is co-localized in LBs (32) with cleaved 1-122 asyn, and its activity is increased in the SNpc of PD patients (217), suggesting a role in disease; additionally, two studies utilizing calpain inhibitors were able to alleviate pathologic findings in mouse models of synucleinopathy (218,219).
Caspase I is a cytoplasmic protease that is active upon inflammasome assembly, which can be triggered by various deleterious stimuli, such as lipopolysaccharide, cytokines, or other cellular stressors (162). Uniquely, caspase I only appears to Figure 1. Summary of identified C truncation sites of asyn in human disease tissue and proteases known to cleave at the indicated asyn site. All asyn C truncation sites confirmed with either MS or C truncation-specific antibodies (Table 1)  cleave at one site in asyn, after residue 121. Experiments with 1-121 asyn have demonstrated rapid assembly into fibrils and potent prion-like seeding with associated toxicity, leading to further inflammasome and capsase I activation, which may cause a pathologic positive feedback cycle in disease (160,162). Caspase I has been found to co-localize in LBs (162), and administration of a caspase I inhibitor has shown benefit in preserving TH neurons and decreasing asyn aggregation in a mouse model of MSA (220).
Neurosin, plasmin, and MMPs differ from the other listed proteases in that they are extracellular proteases and thus are unlikely to contribute extensively to intracellular truncated asyn, although they may truncate extracellular asyn. Neurosin can cleave within the NAC and C terminus of monomeric and oligomeric asyn (207)(208)(209) and degrade fibrillar asyn (211) and is mainly thought to be beneficial, as two studies have demonstrated a reduction in asyn-related pathology when neurosin is overexpressed in mice (209,210). Neurosin can be found in LBs and GCIs even though it is active only extracellularly, and reduction in its activity is evident in LBD brains (206,209). Plasmin can truncate monomeric or fibrillar asyn in mainly Nterminal regions, which for extracellular fibrils decreases their ability to translocate into adjacent cells, thus impacting spreading (213). Of the MMPs, it appears that MMP3 mainly truncates asyn in the C-terminal region, and fragments produced from MMP3 cleavage of asyn aggregate more readily into fibrils and protofibrils than FL asyn and produce more toxicity in cultured cells (164,168,186). Incubation of MMP1 with asyn also results in an increase in aggregation of the fragments through truncation (164), whereas other studied MMPs do not exhibit this effect and sufficiently degrade asyn (164,168,212). The ubiquitous presence of these proteases in the extracellular matrix suggests that asyn fibrils may undergo some degree of truncation during intercellular spread, and resulting alterations in prion-like activity may result.
The most common proteases in truncating both monomeric and fibrillar asyn are lysosomal proteases, where most physiologic and pathologic asyn is normally degraded (39,169,204). Of the lysosomal proteases, AEP and cathepsins B, D, and L have been well-characterized in their truncation of asyn. AEP and its major asyn truncation product, 1-103 asyn, have been demonstrated to be uniquely pathologic in their interaction with asyn (132,135,141,161). AEP has been shown to become overactive in excitotoxicity and aging (132), and high activity of this protease is detected in PD in the SNpc and cortex (132) and possibly the colon (135), where asyn pathology may begin. The 1-103 asyn product of AEP is enriched in the brains (132) and colons (135) of PD patients, and this C-truncated asyn has been continuously shown to readily aggregate into toxic fibrils that can template further pathology (132,135,141,161).
The lysosomal cathepsins have been characterized as the main proteases involved in normal breakdown of monomeric and fibrillar asyn (37,39,204). In terms of their ability to degrade asyn, cathepsin D is the least efficient, only being able to C-truncate both monomeric or fibrillar asyn; cathepsin B is able to degrade monomeric but only truncate fibrillar asyn, whereas cathepsin L can fully degrade all forms of asyn (39,141,215). Combined action of these cathepsins results in the appearance of many common truncated forms of asyn identified (Table 1), including C truncation after residues 103 (cathepsin L), 122 (cathepsin B or L), 114 (cathepsin B), and 124 (cathepsin D) (39,141). Cathepsin D has been observed to be dysregulated in mouse models of synucleinopathy, which can be rectified with overexpression of the antioxidant transcriptional activator Nrf2 (221); likewise, in cultured cells it was demonstrated that oxidative stress induced cathepsin D overactivation and the appearance of oligomeric, truncated asyn (214). It has been suggested that oxidative stress can shift the balance of proteolytic activity in favor of cathepsin D over cathepsin B or cathepsin L (222), which would in turn lead to an increase in partially degraded, truncated forms of asyn. Compared with cathepsin D, cathepsin B is not as implicated in the formation of truncated asyn; however, incubation of FL asyn fibrils with cathepsin B has been shown to produce truncated asyn fibrils with increased prion-like seeding that can be prevented with a cathepsin B inhibitor (38).
There is still much work to be done in identifying which proteases are responsible for forming truncated forms of asyn found thus far in human disease (Table 1) and in human cells treated with pathologic asyn (41), as many of these truncations do not have a suggested protease implicated in their formation, which is summarized in Fig. 1 Proteases that are, in theory, particularly prone to partial degradation of asyn into C-truncated forms include the 20S proteasome, calpain I, caspase I, AEP, MMP1 and -3, and cathepsin D and B. Many of these enzymes have been demonstrated to display increased activity in situations of oxidative stress, which in combination with impaired proteostasis is a hallmark of aging, the main risk factor in developing neurodegenerative disease (223). Regional variability in these truncating proteases' activity has been observed (224) and may play a role in initiation and progression of synucleinopathies, particularly in the colon in PD, where the appendix displays extensive proteolysis favoring asyn truncation, which is not seen in healthy controls. Therapies targeting these proteases have proven somewhat effective in preclinical models, and further understanding of detrimental protease activity in synucleinopathies may uncover efficacious treatment strategies aimed at preventing harmful truncation of asyn.

Summary and future directions
The main genetic risk factors for PD are centered around lysosomal activity (225), and dysfunction of autophagy and proteostasis in general is taking center stage in understanding why synucleinopathies begin and how they progress (226). Truncation of asyn has been the focus of this review, and inability of proteostatic mechanisms to fully degrade physiologic or pathologic asyn is the cause of this PTM. Truncated forms of asyn are common occurrences in synucleinopathies where a number of major C truncations are often present, including those cleaved after residues 103,115,119,122,125,133, and 135 (Table 1). Many of these truncations are enriched within the disease-associated insoluble fraction of brain lysate and not commonly found in healthy controls, which is indicative of their role in disease pathogenesis but may also prove useful as a peripheral biomarker. Tissue from the regions involved early in PD and LBD, such as the vermiform appendix and amygdala, are rife with truncated forms of asyn compared with controls (Table 1), and cellular export of these species may signify dysfunctional proteostasis in these vulnerable regions. C truncation of asyn once formed has the capacity to initiate misfolding and assembly of asyn into amyloid fibrils, as loss of the C terminus allows the hydrophobic NAC motif to interact with other asyn proteins to polymerize, which has been demonstrated by a vast repertoire of experimental techniques both in vitro and in vivo. It is difficult to overstate how rapidly the removal of the C terminus of asyn allows pathologic aggregation to occur compared with FL asyn, as removal of even 20 residues results in fibril formation at concentrations that are 4 times less than that required for FL asyn (31). Once formed, these truncated fibrils may lead to prion-like seeding of endogenous FL asyn to kick start a vicious cycle in which new fibrils may spread from one cell to the next and propagate therein (Fig. 2). As opposed to aggregation, where increased C-terminal truncation seems to continually increase aggregation up to a point, prion-like properties were heavily dependent on the exact composition of the truncated fibril in terms of N and C truncations present along with possibly other PTMs or even proteins such as tau. Even after FL asyn fibrils are formed, it is likely that they will come to be truncated as well due to extracellular proteases they are exposed to in intercellular spreading, lysosomal proteases upon endocytic uptake, and cytoplasmic proteases if they manage to escape the lysosome. Indeed, multiple studies have found that FL asyn fibrils added to cells are quickly trafficked to lysosomes and rapidly truncated (41,141), which indicates that these truncated asyn fibrils may be the true prion-like entity that templates endogenous neuronal asyn and results in toxic sequelae.
Truncated asyn may theoretically have a role in nearly every stage of asyn pathologic misfolding and prion-like propagation as we have reviewed, and this should be kept in mind for the design of experiments and therapeutic strategies. As discussed, interventions aiming to preserve proteostatic health and decrease truncated asyn have shown promise (209,210,(218)(219)(220). Likewise, immunotherapeutic targeting of 122 Ctruncated asyn was beneficial in ameliorating pathologic outcomes in a mouse model of PD, as the presence of this truncation may be highly specific for misfolded asyn (227). Ultimately, targeting the factors inducing initial misfolding of asyn, such as truncation and other possible PTMs, or alteration of the cellular milieu, including oxidative or proteostatic stress, in Figure 2. Depiction of pathologic role of asyn truncation in initiation and propagation of asyn aggregation. A, in healthy neuronal cells, lysosomal enzymes are capable of fully degrading both monomeric and fibrillar forms of asyn, which prevents both spontaneous formation of aggregates and prion-like seeding from uptaken extracellular fibrils. B, in unhealthy neuronal cells, due to cumulative insults including impaired lysosomal autophagy and oxidative stress, only partial degradation of monomeric and fibrillar forms of asyn occurs. Accumulation of C-truncated monomeric asyn may kick start initial aggregation and fibril formation, leading to the prion-like seeding cycle of pathology propagation. In continued disease, incomplete degradation of fibrils leads to truncation-containing fibrils with altered prion-like seeding activity. Ultimately, persistence of asyn aggregates is cytotoxic through various mechanisms, including impairment of autophagy and mitochondrial damage. Image initially created with Biorender. areas vulnerable to early pathology will be the most efficacious strategy to prevent prion-like propagation and development of symptomatic disease.
Funding and additional information-This work was supported by National Institutes of Health Grants R01NS089022 and R01NS100876 (to B. I. G.) and the University of Florida Moonshot Initiative. Z. A. S. was supported by National Institutes of Health Grant F30AG063446 (to Z. A. S.). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Conflict of interest-The authors declare that they have no conflicts of interest with the contents of this article.