A Precipitating Role for Truncated α-Synuclein and the Proteasome in α-Synuclein Aggregation

Parkinson disease and other α-synucleinopathies are characterized by the deposition of intraneuronal α-synuclein (αSyn) inclusions. A significant fraction (about 15%) of αSyn in these pathological structures are truncated forms that have a much higher propensity than the full-length αSyn to form aggregates in vitro. However, little is known about the role of truncated αSyn species in pathogenesis or the means by which they are generated. Here, we have provided an in vitro mechanistic study demonstrating that truncated αSyns induce rapid aggregation of full-length protein at substoichiometric ratios. Co-overexpression of truncated αSyn with full-length protein increases cell vulnerability to oxidative stress in dopaminergic SH-SY5Y cells. These results suggest a precipitating role for truncated αSyn in the pathogenesis of diseases involving αSyn aggregation. In this regard, the A53T mutation found in some cases of familial Parkinson disease exacerbates the accumulation of insoluble αSyns that correlates with the onset of pathology in transgenic mice expressing human αSyn-A53T mutant. The caspase-like activity of the 20 S proteasome produces truncated fragments similar to those found in patients and animal models from degradation of unstructured αSyn. We propose a model in which incomplete degradation of αSyn, especially under overloaded proteasome capacity, produces highly amyloidogenic fragments that rapidly induce the aggregation of full-length protein. These aggregates in turn reduce proteasome activity, leading to further accumulation of fragmented and full-length αSyns, creating a vicious cycle of cytotoxicity. This model has parallels in other neurodegenerative diseases, such as Huntington disease, where coaggregation of poly(Q) fragments with full-length protein has been observed.

Formation of intraneuronal ␣-synuclein (␣Syn) 1 aggregates is a characteristic of several neurodegenerative diseases, in-cluding Parkinson disease (PD), termed ␣-synucleinopathies (1). For example, ␣Syn is the major component of Lewy bodies (LB), a pathologic marker for PD (2). A direct genetic connection between ␣Syn and PD has been established by evidence that three missense mutations (A30P (3), E46K (4), and A53T (5)) in ␣Syn underlie rare early onset hereditary PD. As the majority of the PD cases are sporadic, elevated oxidative and metabolic stresses are thought to contribute to pathogenesis (6). Impaired mitochondrial complex I elevates concentration of reactive oxygen species, which can accelerate ␣Syn aggregation (7). An impaired ubiquitin-proteasome degradation system may also play a role in initiating PD pathogenesis (6). Increased misfolding of proteins due to oxidative damage could overwhelm and inhibit the proteasome degradation machinery. Inefficient degradation may thus simply cause the accumulation of ␣Syn and, by elevating ␣Syn concentration, increase the rate of ␣Syn aggregation. Such a scenario is supported by direct genetic evidence from a familial PD with triplicate ␣Syn genes (8), animal models (9,10), and a yeast model (11) in which higher concentrations of ␣Syn are sufficient to induce pathogenesis.
Analysis of Lewy bodies from patients revealed that they contain multiple proteins, with ␣Syn being predominant. Different forms of modified ␣Syn have been identified in these pathologic samples, including phosphorylated, nitrated, and mono-, di-, or tri-ubiquitinated ␣Syn (12)(13)(14). Notably, about 15% of the ␣Syn in these aggregates is truncated (15,16). The tendency of truncated ␣Syn species to rapidly aggregate (17,18) suggests that they may have played a role in inducing Lewy body formation. Here, we have examined the precipitating role for truncated ␣Syn in aggregation and cytotoxicity and have investigated the means by which they are produced. The data suggest a precipitating role of truncated ␣Syns in PD pathogenesis and demonstrate that the proteasome can produce truncated species via the degradation of ␣Syn not bound to membranes.

EXPERIMENTAL PROCEDURES
Antibodies-SNL-4 and SNL-1 are rabbit antibodies raised against synthetic peptides corresponding to amino acids 2-12 and 104 -119 in ␣Syn, respectively. Syn204 and Syn211 are mouse monoclonal antibodies specific for human ␣Syn (19). The ␣Syn mouse monoclonal antibody from BD Transduction Laboratories recognizes an epitope between 83 and 100 amino acids determined using C-terminal-truncated ␣Syn proteins. Mouse anti-␤-actin monoclonal antibody was from Chemicon. Mouse anti-c-Myc monoclonal antibody was from Santa Cruz Biotechnology.

Fractionation and Characterization of Human and Mouse
Brain Tissues-Sample sources were: cingulate cortex from three controls (ages 73, 76, and 83 years old) and three dementia with Lewy bodies (DLB) patients (ages 77, 79, and 89 years old); substantia nigra from two Alzheimer disease (AD) patients (ages 85 and 86) as controls, and one PD patient (age 75); and cortex and brain stem from different ages of transgenic mice expressing human ␣Syn or human ␣Syn A53T mutant. Samples were sequentially fractionated as follows: ground tissue was homogenized on ice in 10 ml of buffer A (20 mM Tris-HCl, pH 7.6 at 4°C, 750 mM NaCl, 5 mM EDTA with complete protease inhibitor mixture (Roche Diagnostics)) for each gram of tissue. The homogenates were centrifuged at 100,000 ϫ g for 30 min at 4°C. The supernatants were the high salt (HS)-soluble fraction. The HS pellets were rinsed twice with buffer A and then sonicated within 5 ml of buffer B (20 mM Tris-HCl, pH 7.6 at 4°C, 4% SDS with complete protease inhibitor mixture) for each gram of tissue. Samples were centrifuged at 100,000 ϫ g for 30 min at 25°C. The supernatants were the SDS-soluble fraction. The SDS pellets were rinsed twice in buffer B at room temperature and extracted by sonication in 2.5 ml of buffer C (20 mM Tris-HCl, pH 7.6 at 4°C, 8 M urea, and 4% SDS) for each gram of tissue. After 30 min of centrifugation at 100,000 ϫ g at 25°C, the supernatants were the urea-soluble fraction. 3 l of HS-soluble, 6 l of SDSsoluble, and 12 l of urea-soluble fractions were diluted to 40 l with 1ϫ SDS sample buffer, separated by electrophoresis in 12.5% SDS-polyacrylamide gels, and transferred to nitrocellulose (Millipore). A panel of ␣Syn antibodies was used to identify both full-length and truncated ␣Syn species.
␣Syn Aggregation Assay-␣Syn proteins were expressed in Escherichia coli BL21 (DE3). Cells were lysed in buffer D (20 mM Tris-HCl, pH 7.6 at 4°C, 20 mM NaCl) with complete protease inhibitor mixture. After centrifugation, the supernatant was boiled at 100°C for 10 min. Aggregated bacterial proteins were removed by centrifugation, and ␣Syn in the supernatant was purified on a hydrophobic interaction (butyl-Sepharose) column and an anion exchange (DEAE) column. Protein purity in relevant fractions was Ͼ95% as determined by Coomassie blue-stained SDS-PAGE. For aggregation assay, proteins were equilibrated in buffer E (20 mM phosphate, pH 7.4) by extensive dialysis. ␣Syn aggregation was detected by fluorescence enhancement of thioflavin T (TfT) upon binding. TfT is a dye with specificity toward amyloid fibrils. Aggregation reactions were set up in triplicate in 96-well Costar white plates. Each well had one 1/8-inch Teflon bead (McMaster-Carr) and 150 l of protein in buffer E supplemented with 0.02% NaN 3 and 20 M TfT. Plates were sealed with an ABI PRISM optical adhesive cover (Applied Biosystems) and continuously shaken at 37°C except for 1 min each 10 min as required for data acquisition. TfT fluorescence changes were monitored at 450 nm (excitation)/480 nm (emission) on a Gemini plate reader (Molecular Devices).
Formation of Hybrid Protofibrils between Full-length and Truncated ␣Syns-Protofibrils were prepared according to published methods (20). Briefly, proteins were equilibrated in 20 mM NH 4 HCO 3 and then lyophilized and redissolved in buffer E at ϳ500 -700 M final protein concentration. Proteins were incubated on ice for 1 h and then centrifuged at 16,000 ϫ g for 5 min to pellet aggregates. The protofibrillar population in the supernatant was then isolated on a Superdex 200 size exclusion column (Amersham Biosciences). To assess whether hybrid protofibrils between full-length and truncated ␣Syns formed, equal molar amounts of His 6 -tagged full-length ␣Syn were mixed with non-tagged ␣Syn110 or ␣Syn120 prior to lyophilization. Purified protofibrils were then mixed with 10 l of Ni-NTA-agarose (Qia-gen) and rocked at 4°C for 1 h. Beads were washed three times with buffer E plus 0.5 M NaCl and resuspended in 60 l of 1ϫ SDS sample buffer with 5 M urea. Western blotting analysis was used to detect whether the beads pulled down both tagged fulllength and untagged truncated ␣Syns.
In Vitro Cell Death Assay for Cytotoxicity of Protofibrils-SH-SY5Y cells (American Type Culture Collection) were cultured in Dulbecco's modified Eagle's medium (Invitrogen) with 10% fetal bovine serum and 1% penicillin/streptomycin. Cells were maintained in 5% CO 2 at 37°C. Cells were plated at a density of about 10,000 cells/well on a 96-well culture plate (Greiner Bio-one) in 100 l of fresh medium. After overnight incubation, the medium was exchanged with 100 l of Dulbecco's modified Eagle's medium/10% fetal bovine serum without phenol red and supplied with 50 nM/well purified protofibrils. After 4 h of incubation, cell death was assayed using a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) reduction kit (Sigma) according to the manufacturer's instructions.
In Vitro 20 S Proteasomal Degradation of ␣Syn-20 S proteasome was purified from bovine red blood cells (21). Proteasomal degradation was carried out in buffer F (20 mM Tris-HCl, pH 7.1 at 37°C, 20 mM NaCl, 1 mM EDTA). 200-l reactions were set up at 37°C containing 20 nM 20 S proteasome and either 0.4 or 4 M ␣Syn. At each time point, 40-or 4-l samples were withdrawn, and reactions were stopped by the addition of SDS sample buffer. Degradation was assayed on 12.5% SDS-polyacrylamide gel by direct Coomassie staining or Western blotting analysis. Site-specific inhibitors, as stated in the figure legends, were used to identify which activity is responsible for the production of truncated ␣Syn. To identify the fragments, a 500-l reaction containing 20 nM 20 S proteasome and 100 M ␣Syn-His 6 in buffer F was incubated for 30 min at 37°C. 20 S proteasome was removed with a YM-100 (Millipore) spin column, and the flowthrough containing the remaining ␣Syn-His and degradation fragments was incubated with 25 l of Ni-NTA-agarose and rocked at 4°C for 1 h. Beads were isolated by centrifugation with the supernatant containing non-His-bound fraction. Beads were washed three times with buffer F plus 0.5 M NaCl, and the His-bound fraction was eluted with 0.2 M imidazole. Both frac-tions were subjected to electrospray mass spectrometry analysis (UT Southwestern mass spectrometry core).

Two C-terminal-truncated ␣Syns and One Central NAC Fragment Coaggregate with Full-length ␣Syn in Patients and
Transgenic Mice Expressing Human ␣Syn-A53T-We first identified truncated ␣Syn species in both PD and dementia with Lewy bodies (DLB) patients using a spectrum of ␣Syn antibodies. Brain samples from PD and DLB patients and controls (age-matched normal brains or AD patients) were sequentially extracted with a HS buffer, a 4% SDS buffer, and an 8 M urea buffer to obtain soluble ␣Syn, SDS-soluble aggregates, and urea-soluble/SDS-insoluble aggregates, respectively. Two C-terminal-truncated forms of ␣Syn, H1 and H2, were detected in the HS-soluble fractions of controls, PD, and DLB patient brains ( Fig. 1, A, B, and D). A small amount of H1 was also observed in the SDS-soluble fractions of the control samples. However, only PD and DLB patient brains contained appreciable amounts of ␣Syn truncations, as well as higher molecular weight species, in both the SDS-and urea-soluble fractions. These fractions also were the only ones to contain a third fragment (H3) that is both N-and C-terminal truncated and also contains the highly amyloidogenic NAC region of ␣Syn ( Fig. 1, A, B, and D), possibly explaining why this hydrophobic fragment was detected only in aggregates.
Truncated ␣Syn species were also observed in transgenic (Tg) mice expressing human ␣Syn (22,23). The human ␣Syn-A53T mutant Tg mice develop an age-dependent pathological phenotype (complex motor neuron impairment leading to paralysis and death) associated with the accumulation of pathological ␣Syn inclusions (22,23). To examine whether truncated ␣Syns were also associated with pathology, we analyzed brain stem samples from ␣Syn-A53T Tg mice not yet exhibiting pathology (4-month old, normal behavior) and exhibiting pathol-ogy (11-month old, displaying quadriparesis, stiff back, and impaired movement). HS-soluble ␣Syn truncations were identified in both 4-and 11-month-old mice, but animals showing pathology had a greater amount of ␣Syn present in SDS-soluble, aggregated forms ( Fig. 1C, M1, M2, and M3), indicating that truncated species accumulate in pathological inclusions. One notable difference between human patients and ␣Syn-A53T Tg mice is that the Tg mice do not produce urea-soluble aggregates. ␣Syn epitope mapping ( Fig. 1D) showed that the truncated ␣Syns produced by human ␣Syn-A53T Tg mice are very similar to those found in human PD and DLB patients. Thus, both patient samples and pathological Tg mouse brains contain at least three truncated ␣Syn forms: two C-terminaltruncated forms (approximately ending between amino acid residues 102-125 and 83-110, respectively) and one central fragment that contains the NAC domain. Similar fragments were also detected by the same spectrum of ␣Syn antibodies in Neuro 2a cells when full-length human ␣Syn was overexpressed (data not shown). A study that appeared during the review of this report observed a nearly identical pattern of fragments in patient and animal model samples (24).
C-terminal-truncated ␣Syns Induce Rapid Aggregation of Full-length Protein at Substoichiometric Ratios-To explore the possible mechanism underlying the coaggregation of truncated ␣Syn with full-length protein as observed in patients and Tg mice, the effect of two C-terminal-truncated fragments, ␣Syn1-110 (␣Syn110) and ␣Syn1-120 (␣Syn120), on the rate of fibrillization of full-length protein was determined. In vitro, recombinant ␣Syn110 and ␣Syn120 adopt a random coil conformation indistinguishable from the full-length protein as detected by circular dichroism spectroscopy (data not shown).
␣Syn aggregation is a nucleation-dependent process (25) involving at least three steps: monomer 3 oligomer (seed) 3 protofibril 3 fibril. TfT binding, a fluorescent dye assay that ; ␣Syn in cingulate cortex of age-matched control (AD patient or healthy control) and DLB patient (B); and ␣Syn in the brain stem of non-pathological (NP, 4-month-old) and pathological (P, 11-month-old) transgenic mice expressing human A53T mutant ␣Syn (C). Brain samples were separated into high salt-soluble (hs), SDS-soluble (s), and urea-soluble (u) fractions, respectively, by sequential extraction with a high salt buffer (buffer A), a 4% SDS buffer (buffer B), and a 8 M urea buffer (buffer C). D, two C-terminal-truncated fragments and one internal fragment of ␣Syn are present in the aggregates in human patients and ␣Syn Tg mouse brains. Truncated forms of ␣Syn in panels A, B, and C were identified using a spectrum of ␣Syn-specific antibodies recognizing different epitopes. specifically reports amyloid fibril formation, showed that both model C-terminal-truncated ␣Syns aggregate much more rapidly than the full-length protein, as previously described (17,18). Significantly, substoichiometric amounts (5%) of truncated ␣Syn110 and ␣Syn120 also dramatically accelerate full-length ␣Syn aggregation ( Fig. 2A), suggesting a possible scenario in which more aggregation-prone C-terminal-truncated ␣Syns form oligomers and then these oligomers nucleate the aggregation of full-length ␣Syn by forming hybrid protofibrils, which then develop into fibrils (Fig. 2B).
To test this seeding hypothesis, we first examined whether hybrid protofibrils can be formed between the full-length and truncated ␣Syn species. Soluble protofibrils prepared from equimolar mixtures of His 6 -tagged full-length ␣Syn (␣Syn-His) and non-His-tagged ␣Syn110 or ␣Syn120 were purified by gel filtration chromatography. Protofibrils eluted at a position consistent with a 2,000-kDa molecular mass (Fig. 2C). Ni-NTAagarose pulldowns of purified protofibrils detected both ␣Syn-His and untagged ␣Syn110 or ␣Syn120 (Fig. 2D). Nonspecific binding could be excluded because no protein was pulled down in protofibrils prepared from a mixture of non-tagged fulllength ␣Syn and non-tagged ␣Syn110 or ␣Syn120 (Fig. 2D). These results demonstrate that formation of hybrid protofibrils does occur in vitro. Furthermore, TfT binding assays demonstrate that purified protofibril mixtures can effectively nucleate full-length ␣Syn aggregation (Fig. 2E). These data support a model by which C-terminal-truncated ␣Syns induce protein aggregation by forming hybrid protofibrils with full-length ␣Syn, which subsequently accelerates aggregation of the remaining proteins (Fig. 2B).

Coexistence of Full-length and Truncated ␣Syns Increases
Cellular Vulnerability to Oxidative Stress-The pathology of ␣-synucleinopathies is accompanied by the formation of ␣Syn inclusions, but whether these aggregates are the species that cause neuronal cell death is as yet unclear. Recent evidence suggests that the aggregation intermediates (soluble oligomers) may be the major cytotoxic species causing neuronal cell death (26,27) and that mature fibrils may have a neuronal protective effect (28). We used a well established in vitro cell death assay to test whether purified ␣Syn protofibril mixtures are cytotoxic. We added 50 nM protofibrils or 5 M monomeric ␣Syn to the cell culture medium. After 4 h, about 30% of human neuroblastoma SH-SY5Y cells treated with protofibrils died, as assayed by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide reduction. In contrast, no cell death was observed after treatment with either 5 M monomeric ␣Syn alone or a mixture of monomeric truncated and full-length ␣Syn (Fig. 3A). The cellular toxicity of purified protofibril mixtures suggests that C-terminal-truncated ␣Syns may promote cell death by accelerating protofibril formation and accumulation when promoting full-length protein aggregation.
To further examine the role of truncated ␣Syns in development of cellular toxicity, truncated ␣Syns were co-overexpressed with full-length protein in SH-SY5Y cells. Overexpression of full-length or C-terminal-truncated ␣Syns increased neuronal cell vulnerability to oxidative stress (29,30). When truncated ␣Syns were co-overexpressed with fulllength protein at a ratio of about 1:10, cellular vulnerability to oxidative stress was significantly increased as compared with the presence of equivalent amounts of full-length pro- tein alone (Fig. 3B). These results suggest that substoichiometric amounts of C-terminal-truncated ␣Syns exacerbate ␣Syn-dependent cytotoxicity.
The A53T Mutation Exacerbates the Effect of Truncated ␣Syns on the Aggregation of Full-length Protein-The A53T mutation in ␣Syn causes an early onset familial PD (5). In vitro, ␣Syn-A53T aggregates more rapidly than the wild type ␣Syn (31). The cross-seeding mechanism (Fig. 2B) predicts that ␣Syn-A53T truncations may also promote the aggregation of full-length ␣Syn-A53T. Consistent with this hypothesis, C-terminal-truncated ␣Syn110-A53T and ␣Syn120-A53T mutants aggregate much more rapidly than full-length ␣Syn-A53T (data not shown). Significantly, C-terminal-truncated A53T mutants promote aggregation of full-length ␣Syn-A53T much more rapidly (a much shorter lag phase) than parallel combinations without the mutation (Fig. 4, A and B). This dramatic acceleration of seeding and aggregation would be predicted to contribute to the early appearance of ␣Syn inclusions in patients with the A53T mutation and in ␣Syn-A53T Tg mice, thereby causing pathology.
To evaluate this hypothesis, the age-dependent accumulation of truncated ␣Syn species was assessed in Tg mice expressing human ␣Syn-A53T mutant (line M83, which develops an age-dependent severe and complex motor disorder leading to paralysis and death). These studies were conducted in parallel with Tg mice expressing equivalent levels of wild type human ␣Syn (line M7, which does not develop pathological inclusions or any abnormal phenotype) (23). Immunoblot analyses were performed to evaluate both full-length and truncated ␣Syns in their aggregated (SDS-soluble) and HS-soluble forms of agematched M83 and M7 Tg mice (Fig. 4, C and D). Samples from 11-month-old M83 Tg mice exhibiting pathology were also included. The early onset and enhanced extent of deposition of both truncated and full-length ␣Syn aggregates (SDS-soluble) in the M83 mice parallel the pathology (Fig. 4C). The amount of soluble full-length ␣Syn extracted by HS buffer remains constant between 1 and 8 months of age for both A53T mutant and wild type ␣Syn Tg mice (Fig. 4D); however, the production and accumulation of truncated ␣Syns in both M83 and M7 lines are age dependent. The M83 line reaches a steady state of about 20% of total ␣Syn being soluble truncations at month 8, corre-lating well with the average age of disease onset (7 months) (23). Line M7 expressing wild type ␣Syn accumulates the same level of truncations even earlier (2 months), but no pathology phenotype was observed. These results are consistent with the reduced ability of the wild type ␣Syn fragments to form aggregates in vitro compared with ␣Syn-A53T fragments as seen in Fig. 4, A and B.
The 20 S Proteasome Produces C-terminal-truncated ␣Syn Fragments from the Degradation of Unstructured, but Not Vesicle-bound, ␣Syn-The possible role of ␣Syn truncations in pathogenesis necessitates the discovery of the mechanism by which they are produced. The model predicts that inhibition of this activity could retard the formation of intracellular ␣Syn aggregates. The 20 S proteasome is capable of degrading natively unfolded, non-ubiquitinated ␣Syn in vitro (32,33) except in the case of the minor fraction of O-glycosylated ␣Syn that is a substrate for the E3 ubiquitin ligase, Parkin (34).
We performed in vitro proteasomal degradation assays to examine whether the proteasome produces fragments of ␣Syn as it degrades the full-length protein. Purified 20 S proteasome degraded full-length ␣Syn to produce C-terminal-truncated fragments as detected by an ␣Syn N terminus antibody, with fragment accumulation more substantial at higher ratios of ␣Syn to 20 S proteasome (Fig. 5A). In cells, ␣Syn may bind on the surface of vesicles and form an ␣-helical conformation (35)(36)(37). As the 20 S proteasome is capable of degrading unfolded or damaged proteins, but not stable proteins (38), we hypothesized that the vesicle-bound ␣Syn would not be a good substrate for the 20 S proteasome. Consistent with the hypothesis, vesicle-bound ␣Syn is predominantly ␣-helical as confirmed by CD spectroscopic analysis (data not shown) and is not degraded by the 20 S proteasome (Fig. 5B). The vesicles themselves did not inhibit activity of the proteasome against a control flurogenic substrate (data not shown). This result suggests that only a fraction of intracellular ␣Syn, which is not bound to vesicles, is a substrate for the 20 S proteasome and that loss of vesicle binding because of mutation (39) or oxidative damage (7,40) may result in enhanced 20 proteasome access to ␣Syn and the production of truncated fragments.
The Caspase-like Activity of the Proteasome Is Responsible for Fragment Production-We used inhibitors specific for the in- dividual catalytic sites of the proteasome to identify which activity was responsible for fragment production. YU-102, an inhibitor of the caspase-like activity (41), retards fragment production as determined by Western blotting (Fig. 6A). Notably, a larger fragment accumulated at a 1:5000 ratio of protea-some to ␣Syn (Fig. 6A), and inhibition of trypsin-like activity increased fragment production (Fig. 6A). To identify the preferred proteasomal cleavage sites, degradation products of ␣Syn-His 6 were separated into Ni-NTA-bound and Ni-NTA flow-through fractions. Mass spectrometric analysis of the NTA flow-through fraction identified three predominant fragments: ␣Syn (1-119), ␣Syn (1-110), and ␣Syn (1-83) (Fig. 6B). Interestingly, corresponding fragments ␣Syn (120 -149), ␣Syn (111-149), and ␣Syn (84 -149) are present in the NTA-bound fraction, suggesting that the endoproteolytic activity of the proteasome identified previously (33) may be responsible for production of the fragments. The occurrence of all three scissile bonds after an acidic residue (Asp-119, Glu-110, and Glu-83) is consistent with the results of the inhibition assay, which implicated the caspase-like activity of the proteasome in fragment production (Fig. 6A). It is also notable that the larger fragments observed in vitro (e.g. 1-119, 1-110) are very similar to those observed in PD patients (H1, H2) and in animal models (M1, M2) ( Fig. 1) (24).
To examine whether the fragments produced from the 20 S proteasomal degradation are capable of inducing full-length protein aggregation, mixtures of ␣Syn from proteasomal degradation were assessed for fragment production by Western blotting (Fig. 6C, inset) and rate of fibril formation. TfT binding demonstrated that fragments produced by the caspase-like activity of the 20 S proteasome induced earlier initiation of the aggregation (Fig. 6C). DISCUSSION Although truncated ␣Syns are one of the major modified ␣Syn species in Lewy bodies of PD, little is known about their role in pathogenesis or the mechanism of their production (42). We first identified one central NAC-containing and two C- As controls, aggregation of 5 M ␣Syn120/110-A53T mutant (curve 5), ␣Syn120/110 (curve 6), ␣Syn-A53T (curve 7), and ␣Syn (curve 8) in buffer E (20 mM phosphate, pH 7.4) with 0.02% NaN 3 and 20 M thioflavin T are also shown. C, human ␣Syn-A53T mutant Tg mice deposit both full-length and truncated ␣Syn aggregates earlier and to a greater extent than Tg mice expressing wild type ␣Syn. Immunoblot analysis of the SDS-soluble extracts from 1-to 8-month-old A53T mutant and wild type Tg mice. 11-month-old pathological A53T mutant Tg mice were also evaluated. D, accumulation of soluble truncated ␣Syns in both wild type and A53T mutant Tg mice is age dependent. Immunoblot analysis of HS-soluble extracts from 1-to 8-month-old A53T mutant and wild type Tg mice. 11-month-old A53T mutant Tg mice exhibiting pathology are also included. Bands a and b mark the full-length and the predominant truncated form of ␣Syn, respectively. terminal-truncated ␣Syn fragments in the Lewy bodies of PD and DLB patients and in inclusions of human ␣Syn-A53T Tg mice exhibiting pathology (Fig. 1). These fragments are analogous to those recently identified by Li et al. (24) in patients, animal models, and cell culture systems. Truncated ␣Syns are not unique to PD/DLB patients and human ␣Syn-A53T Tg diseased mice, as they also are present as soluble forms in non-pathology age-matched controls, AD patients, and nonpathology human ␣Syn-A53T Tg mice (Figs. 1 and 4D). In addition, data from Li et al. suggested that C-terminal-truncated ␣Syn fragments are generated by the normal cellular processing of the full-length ␣Syn (24).
Notably, aggregated ␣Syn truncations are unique to PD/DLB patients, as fragments in age-matched controls are present only in soluble forms (Fig. 1). Because of the greater tendency of the C-terminal-truncated ␣Syns to aggregate as compared with the full-length protein (17,18), we suspected that truncated ␣Syns may play a role in PD pathogenesis and other ␣-synucleinopathies. To explore the mechanism underlying the coaggregation of truncated and full-length ␣Syns in PD/DLB brains, we performed in vitro aggregation assays demonstrating that truncated ␣Syns induce rapid aggregation of the fulllength ␣Syn at very low substoichiometric ratios, possibly by a seeding mechanism in which more aggregation-prone truncated ␣Syns function as seeds to nucleate the full-length protein aggregation as we tested in vitro (Fig. 2). Others (18,24) also observed acceleration of aggregation by similar C-terminal-truncated ␣Syns but did not investigate the mechanism by which the rate was enhanced. Together, these results suggest that truncated ␣Syns may play a precipitating role in inducing the formation of inclusions in PD/DLB brains.
Promoting ␣Syn aggregation with substoichiometric amounts of truncated ␣Syn results in earlier formation and accumulation of aggregation intermediates (protofibrils) and, subsequently, final mature fibrils. Recent data suggest that these or related aggregation intermediates are the cytotoxic species that cause neuronal cell death (26,27), whereas mature fibrils may have a neuronal protective effect (28). A well established in vitro cell death assay (26,27) demonstrated that purified soluble protofibrils, but not monomers, from mixtures of truncated and the full-length ␣Syns cause SH-SY5Y cell death (Fig. 3A). Although the mechanism underlying the cytotoxicity of the protofibrils is not clear, earlier studies showed that ␣Syn protofibrils disrupt synthetic vesicles in vitro (43,44), similar to the mechanism utilized by cellular toxins (45), causing imbalance of cellular ions and fluid, and thus, cell death. Consistent with this mechanism, in these studies the toxic conformers of cytoplasmic proteins were toxic even when administered extracellularly. Because ␣Syn is an intracellular protein, we also examined the consequences of coexistence of small fractions of truncated ␣Syn with the full-length ␣Syn in the cytosol by transiently co-overexpressing ϳ10% of truncated ␣Syn110 or ␣Syn120 with the full-length ␣Syn in SH-SY5Y cells. Cell death assays demonstrated that the coexistence of truncated ␣Syns with full-length ␣Syn increased cell death during an oxidative challenge (Fig. 3B). As elevated oxidative stress is one of the events arising from impaired mitochondria function and perhaps plays a role in PD pathogenesis (6), these results suggest that the induction of rapid ␣Syn aggregation by truncated forms could accelerate neuronal cell death.
Point mutations (A30P, A53T, and E46K) in ␣Syn cause inheritable early onset PD (3)(4)(5). Human ␣Syn-A53T Tg mice, but not wild type human ␣Syn Tg mice, develop a pathological phenotype similar to human patients (22,23). The greater propensity for aggregation of the ␣Syn-A53T mutant compared with the wild type protein (31) likely contributes to the dramatic differences between the A53T mutant and the wild type Tg mice. Age-dependent coaggregation of truncated ␣Syns with the full-length ␣Syn was found to correlate with the development of pathology in A53T mutant Tg mice (Fig. 4). These fragments are very similar to those deposited in human patients as detected by ␣Syn epitope mapping (Fig. 1) and putative mass spectroscopic identification (24). Furthermore, accumulation of soluble ␣Syn fragments in A53T mice is age dependent whereas full-length protein level remains constant, and only the former correlates with the average disease onset time of this Tg line (7 months) (Fig. 4D). The in vitro aggregation assay demonstrated that the A53T mutant exacerbates the accelerating effect of truncated ␣Syns on rapid aggregation of the full-length protein (Fig. 4, A and B). Together, these results suggest that truncated ␣Syns in human ␣Syn-A53T Tg mice may play a precipitating role in inclusion formation in this Tg line and, thus, accelerate pathogenesis.
Partial proteolysis has a role in several neurodegenerative diseases (42). In AD, the highly amyloidogenic A␤ peptide is generated by the sequential action of ␤-secretase and ␥-secretase on amyloid precursor protein (46), whereas in Huntington disease, toxic truncated N-terminal fragments of the Huntington protein that contain an expanded polyglutamine repeat are deposited in the inclusions of Huntington disease postmortem tissue (47,48). Our data indicate a possible role for truncated ␣Syns in inducing inclusion formation. Moreover, they implicate the critical role of the 20 S proteasome in fragment production. Several proteases and degradation pathways have been shown to regulate the degradation of various forms of ␣Syn. In vitro and in vivo data support the involvement of the proteasomal (32,33,49) and lysosomal pathways (50,51), and in vitro data suggest a role for calpain I in degradation of aggregated protein (52). Because the physiological function of ␣Syn may require its association with lipid vesicles where it assumes an ␣-helical conformation, the metabolism of the vesicle-bound ␣Syn may occur through the lysosomal degradation pathway; such a model is in agreement with the relatively long half-life of this protein of about 20 h in certain cell cultures (50,51). Structural studies show that ␣Syn binding to vesicles is dynamic (53) and dissociation from the vesicle could be promoted by mutation (39) and oxidative stress (7,40). Our data demonstrate that the 20 S proteasome degrades this free, unstructured ␣Syn, but not membrane-bound, ␣-helical ␣Syn. Furthermore, the caspase-like activity of the proteasome generates C-terminal-truncated ␣Syn species similar to those deposited in human patients that promote the aggregation of full-length ␣Syn. This acceleration is abolished by inhibition of fragment production by the caspase-like activity inhibitor YU102. Finally, production of C-terminal-truncated ␣Syn fragments is dramatically affected by the proteasome capacity, as greater amounts of fragments are observed under higher substrate to proteasome ratios as well as by inhibition of the trypsin-like activity of the proteasome. Thus, because aggregated ␣Syn has been shown to inhibit some proteasome activity, this mechanism could result in the accumulation of both truncated and full-length ␣Syn and an even further elevation of the concentration of aggregation-prone truncated ␣Syn that, in turn, could initiate the aggregation of full-length ␣Syn, especially in the crowded cellular environment (54).
In summary, our data suggest a model in which incomplete degradation of ␣Syn produces highly amyloidogenic fragments that rapidly aggregate and seed the aggregation of full-length protein. Subsequently, proteasomal activity could be inhibited by further attempts to dispose of these more proteolysis-resistant fragments and aggregates (55). This inhibition would, in turn, cause further accumulation of both full-length and fragmented ␣Syn species, thus creating a vicious cycle of cytotoxicity. This model makes the paradoxical prediction that early interference in the proteolysis of full-length ␣Syn may actually retard aggregate formation by reducing the production of truncated ␣Syn.