C-terminal α-synuclein truncations are linked to cysteine cathepsin activity in Parkinson's disease

A pathological feature of Parkinson's disease (PD) is Lewy bodies (LBs) composed of α-synuclein (α-syn) amyloid fibrils. α-Syn is a 140 amino acids–long protein, but truncated α-syn is enriched in LBs. The proteolytic processes that generate these truncations are not well-understood. On the basis of our previous work, we propose that these truncations could originate from lysosomal activity attributable to cysteine cathepsins (Cts). Here, using a transgenic SNCAA53T mouse model, overexpressing the PD-associated α-syn variant A53T, we compared levels of α-syn species in purified brain lysosomes from nonsymptomatic mice with those in age-matched symptomatic mice. In the symptomatic mice, antibody epitope mapping revealed enrichment of C-terminal truncations, resulting from CtsB, CtsL, and asparagine endopeptidase. We did not observe changes in individual cathepsin activities, suggesting that the increased levels of C-terminal α-syn truncations are because of the burden of aggregated α-syn. Using LC-MS and purified α-syn, we identified C-terminal truncations corresponding to amino acids 1–122 and 1–90 from the SNCAA53T lysosomes. Feeding rat dopaminergic N27 cells with exogenous α-syn fibrils confirmed that these fragments originate from incomplete fibril degradation in lysosomes. We mimicked these events in situ by asparagine endopeptidase degradation of α-syn fibrils. Importantly, the resulting C-terminally truncated fibrils acted as superior seeds in stimulating α-syn aggregation compared with that of the full-length fibrils. These results unequivocally show that C-terminal α-syn truncations in LBs are linked to Cts activities, promote amyloid formation, and contribute to PD pathogenesis.

Understanding the degradation processes that generate C-terminal truncations (⌬C) would aid in the elucidation of new ways to circumvent the progression of PD. Mounting evidence supports the involvement of the lysosome and proteasome in ␣-syn degradation (24,25). However, because the lysosome is generally considered to be responsible for removal of aggregation-prone species, we hypothesize that these truncations stem from incomplete proteolytic events in this organelle. In fact, the lysosomal protease, cathepsin D (CtsD) was shown to generate ␣-syn⌬C species (26,27). More recently, the lysosomal cysteine cathepsin asparagine endopeptidase (AEP), found to be elevated in PD brains, was reported to generate an ␣-syn fragment composed of residues 1-103, which enhanced neurotoxicity in a PD mouse model (28). Although our interests are in a lysosomal role in generating ⌬C-terminal ␣-syn truncations, cytosolic proteases such as calpain-I (7-9), caspase-1 (10) and neurosin (11) have also been considered in generating C-terminal truncations.
In this work, we sought to determine which ␣-syn truncations found in LBs are lysosomal in origin. Lysosomes were purified from two disease-related models: brains from transgenic mice overexpressing the PD-associated A53T mutant form of ␣-syn (SNCA A53T ) (34) and cultured N27 rat dopaminergic neuronal cells treated with fibrils formed in vitro by N-terminally acetylated ␣-syn (hereafter, simply abbreviated as ␣-syn). Aged SNCA A53T develops hind leg paresis, weight loss, and difficulty ambulating, leading to progressive neurodegeneration and death, but the actual age of disease onset varies. Here, we compared symptomatic SNCA A53T mice to agematched nonsymptomatic SNCA A53T mice as well as nontransgenic (wt) control mice. Using antibody epitope mapping, we show that C-terminal truncations (corresponding to 12-and 8-kDa bands on Western blots) were enhanced only when the mice became symptomatic. Interestingly, in the second N27 model, two bands of similar weight were also observed after partial digestion of preformed fibrils in lysosomes. The identities of the ␣-syn⌬C species were determined using recombinant ␣-syn and peptide mapping by MS. Selective inhibition studies verified the involvement of CtsB, CtsL, and AEP in generating the 12-kDa band, whereas only CtsB contributed to produce the 8-kDa band. To better understand the mechanism involved, we recapitulated these proteolysis events in situ by AEP degradation of preformed ␣-syn fibrils. Importantly, the AEP-derived 1-122 and 1-103 fibrils stimulated aggregation of soluble full-length ␣-syn. These data unequivocally show that ␣-syn⌬C species in LBs are linked to cysteine cathepsin activities and serve as potent amyloid seeds. Collectively, this work demonstrates a new molecular connection between the lysosome and PD pathology.

␣-Syn⌬C species are enriched in lysosomes isolated from symptomatic SNCA A53T mice
Mice overexpressing human SNCA A53T were euthanized when symptoms developed, typically at ϳ16 months of age (34). Samples from age-matched nonsymptomatic (ϳSNCA A53T ) as well as nontransgenic (wt) mice were compared with those from the symptomatic mice. Lysosomes were isolated and enriched from mouse brain samples by density gradient centrifugation and lysosomal extracts were prepared by repeated freeze-thaw cycles (Fig. 2a). Immediately after generating lysosomal extracts, we probed for the presence of endogenous ␣-syn in lysosomes to capture levels of undigested ␣-syn. First, the presence of Ser-129 -phosphorylated ␣-syn, a hallmark of LBs (35), was evaluated, and lysosomes from symptomatic SNCA A53T mice were found to have dramatically more Ser-129 -phosphorylated ␣-syn than nonsymptomatic ϳSNCA A53T littermates (Fig.  2b), indicating that ␣-syn is hyperphosphorylated only in a disease-state (35). Interestingly, lysosomes from wt mice showed no immunoreactivity toward the Ser-129 antibody, even though the epitope region is conserved between murine and human sequences (Fig. S1).
The antibody that recognizes epitope 118 -123 of ␣-syn ( Fig.  2f) showed an additional band at 12-kDa (denoted by an asterisk) in lysosomal samples from symptomatic SNCA A53T mice. Because this band was not present using all other C-terminal antibodies (Fig. 2, d, e, g, and h), it can be inferred that identity of this C-terminal truncation results from a cleavage site around residues 118 -123. Analysis with an antibody that recognizes an epitope at residues 91-99 of human and murine ␣-syn (Fig. 2i) also consistently showed this 12-kDa band in lysosomes from the symptomatic mice.
Antibodies specific for the N-terminal region (residues 1-64) of ␣-syn revealed an 8-kDa band (Fig. 2, denoted by two asterisks) in addition to the 12-kDa band. Specifically probing with antibodies that recognize epitope regions 15-64 (Fig. 2j), 1-50 (Fig. 2k), and 11-26 (Fig. 2l) demonstrated that truncations were more abundant in lysosomes from SNCA A53T mice. Because this band was not observed by the antibody that recognizes epitope 91-99, the protein is likely cleaved N-terminal to residue 99 and likely contains the majority of the N terminus. The N-terminal sequence of ␣-syn is highly conserved between humans and mice except for two amino acid differences at positions 53 and 87 (Fig. S1). The lack of truncated forms in wt lysosomes suggests that the degradation pattern of murine

C-terminal ␣-syn truncations
␣-syn differs from the human form. Clearly, these data show not only an increase in full-length ␣-syn but an enrichment in ␣-syn⌬C species in lysosomes from symptomatic SNCA A53T mice.

Evaluation of cathepsin activities in lysosomes isolated from SNCA A53T mice
To assess specific lysosomal protease activity, fluorogenic substrates specific for CtsB, CtsL, CtsD, and AEP activity were used. Two biological replicates of lysosomal extracts (1 g) from wt, ϳ SNCA A53T , and SNCA A53T mice were incubated in the presence of fluorogenic substrates (Ac-RR-AMC for CtsB, Ac-FR-AMC for CtsL, MCA-GKPILEFRKL(Dnp)-D-R-NH 2 for CtsD, and AENK-AMC for AEP) at pH 5 ( Fig. 3). Analysis of individual activities as a function of time showed no significant changes in activity when mice became symptomatic. One exception, seen only in one dataset (Fig. S2) yet not in the other (Fig. 3), was an elevation of CtsB activity in lysosomes from ϳSNCA A53T and SNCA A53T mice. Additional biological replicates are needed to determine the significance of this observation. Nevertheless, the data indicate that it is not the loss of protease activity in lysosomes from symptomatic mice but rather the overburden of ␣-syn levels that is responsible for the incomplete degradation and the appearance of ␣-syn⌬C.

Identification of ␣-syn⌬C species
By monitoring degradation of the remaining endogenous ␣-syn in lysosomes from SNCA A53T mice as a function of time, it was clear that the 12-and 8-kDa bands originated from fulllength ␣-syn (Fig. 4a). Enrichment of these truncations after a 20-h incubation at pH 5 affirmed that these lysosomal extracts were active and continued to degrade endogenous ␣-syn over time. Because levels of endogenous ␣-syn could not be detected by MS under these experimental conditions, we turned to the use of preformed ␣-syn fibrils (Fig. S3) to identify these truncations. Fibrillar material was chosen because 1) Ser-129 -phosphorylated ␣-syn, a well-accepted posttranslational modifica-tion for aggregated ␣-syn (35), is highly enriched in SNCA A53T lysosomes ( Fig. 2b) and 2) prior work on ␣-syn f degradation by CtsL established that both the N-(residues 1 to 9) and C-terminal region (residues 101 to 140) are the most protease-sensitive regions of the fibril structure (29).
To figure out the identity of the 8-kDa band, the degradation pattern of soluble ␣-syn was evaluated. Upon incubation of ␣-syn with lysosomal extracts, one main species with molecular mass ϳ8 kDa was detected by SDS-PAGE and Coomassie Blue staining (Fig. S4a). Corresponding LC-MS analysis suggested this mass is likely peptide fragment 1-90 (Fig. S4b). Aside from   Table S1. d, schematic representation of the primary amino acid sequence of ␣-syn with identifiable cleavage sites generated from degradation of soluble ␣-syn (black) and ␣-syn f (green) by lysosomes from SNCA A53T mice. Cutting that occurs in both soluble and fibrillar ␣-syn is indicated by #. Cathepsin(s) responsible for each cleavage site is as indicated based on this and prior work (27,29). MS analyses are reported in Tables S1 and S2. Amyloid core (residues ϳ37-99) determined by cryo-EM (68 -70) is highlighted in cyan.  Table S2). Notably, 5-140, 65-140, and 1-103 truncations found in LBs were observed (Fig. 1a).

C-terminal ␣-syn truncations
Cysteine cathepsins are responsible for ␣-syn⌬C generation Next, we asked which protease(s) are responsible for generating ␣-syn⌬C species by using the following selective lysosomal protease inhibitors: pepstatin A (PePA) for CtsD, CA-074 for CtsB, z-FY-DMK for CtsL, and z-AAN-AMC for AEP. Limited proteolysis experiments of ␣-syn f with SNCA A53T lysosomal extracts preincubated with individual inhibitors showed that none of the inhibitors significantly affected formation of the 12-kDa band (Fig. S5). LC-MS analysis (Table S3) showed that when selectively inhibiting either CtsB or CtsL, a loss in activity for one could be rescued by the other, because of a shared substrate site, cutting ␣-syn f at Asn-122/Glu-123 to generate the peptide fragment 1-122. The other two C-terminal truncations at residues 114 and 124 were attributed to CtsB and CtsD, respectively (Fig. 4d). N-terminal cleavage sites were also assigned (27): Phe-4/Met-5 to CtsD, Met-5/Lys-6 and Ser-9/Lys-10 to CtsL, and Gly-14/Val-15 to CtsB activities, respectively.
For the 8-kDa band, selective inhibition experiments confirmed a complete absence of fragment 1-90 with the CtsB inhibitor CA-074, whereas PePA (CtsD inhibitor) significantly reduced this truncation (Fig. S4a). Although only Phe-4/Met-5 and Phe-94/Val-95 were inhibited with PePA, all other activities were inhibited by CA-074 as characterized by LC-MS (Table S4). Cleavage at site Ala-90/Ala-91 is unique to CtsB activity (27); thus, its reduction in the presence of a CtsD inhibitor requires a different interpretation to reconcile this observation. One scenario implies a more indirect role for CtsD activity, where CtsD directly activates CtsB (36, 37), which is then responsible for the generation of fragment 1-90. Alternatively, CtsD can first cleave ␣-syn at residue Phe-94 to generate the fragment 1-94, which could be a preferred substrate over full-length ␣-syn for CtsB activity.

Motivated by our limited lysosomal degradation results
showing that the 12-kDa ␣-syn⌬C species is derived from fibrillar ␣-syn, we sought to recapitulate this observation in cultured neuronal cells. Here, we asked whether ␣-syn truncations are generated after fibrils are endocytosed and trafficked to lysosomes in live cells (38 -40). We chose immortalized N27 rat cells because they are dopaminergic (41). Cells were treated with both soluble and fibrillar ␣-syn at low and high protein concentrations (100 nM and 1 M). To determine an appropriate treatment time frame, we used live cell imaging to evaluate cell viability. Over the course of 60-h treatment, only the cells treated with fibrils showed any noticeable effects. Fewer cells were observed, with many losing their dendritic morphology, and eventually disintegrating over time (Fig. S6a).
To verify uptake, Western blot analysis was performed on cells collected 48 h post treatment. Using a N-terminal epitope antibody, no bands were detected from cells treated with soluble ␣-syn, but both the full-length and a 12-kDa band were observed after treating with fibrils at the low and high concentrations (Fig. S6b). A series of higher molecular mass species were also present, indicating an aggregated state of ␣-syn. After treatment with soluble ␣-syn, little immunoreactivity was observed, likely because of its rapid degradation in N27 cells.
To observe ␣-syn association with lysosomes in treated N27 cells, immunofluorescence experiments using ␣-syn (epitope 1-50) and CtsL antibodies were performed. After treating N27 cells for 48 h with fibrillar (Fig. 5a, top; see Fig. S7 for additional fields of view) and soluble ␣-syn (Fig. 5a, bottom), immunoreactivity and colocalization were only observed in fibril-treated cells. Images for untreated cells are shown in Fig. S7. After treating with ␣-syn f (1 M), ␣-syn-positive puncta are clearly  Table S5. Asterisk denotes assignment of 12-kDa band observed in panel d.

C-terminal ␣-syn truncations
visible, and most also co-staining with the lysosomal marker CtsL. Monomer treatment showed dispersed immunoreactivity of ␣-syn and no colocalization with CtsL.
To unequivocally prove that C-terminal truncations are directly derived from lysosomal degradation of ␣-syn f , lysosomes were purified from N27 cells pretreated with exogenous fibrils. A total of 31 T75 flasks containing confluent (ϳ70%) N27 cells grown for 48 h in the presence of 1 M fibrils were used to obtain enough lysosomal extract for immunoblotting. Using an N-terminal ␣-syn antibody (epitope 1-50), full-length ␣-syn and both 12-and 8-kDa species were observed (Fig. 5b). Higher molecular mass bands were also present, indicative of aggregated ␣-syn. Using an antibody that recognizes an epitope at residues 91-99, only the 12-kDa band could be observed, indicating that the 8-kDa band had lost this epitope region. To ensure that the lysosomal extracts were still active, Western blot analysis was performed after 20 h (Fig. 5c). It was evident that proteolysis was still occurring, as the full-length protein was greatly diminished over time, whereas the 12-kDa band, along with some higher molecular mass species, remained, indicating that ␣-syn⌬C was more resistant.
Unfortunately, direct MS analysis on the limited amount of lysosomal material from N27 cells was not possible, so we again turned to the use of exogenous ␣-syn. After adding exogenous ␣-syn f (15 M) to lysosomal extracts (2 and 5 g) from N27 cells, the 12-kDa band, visualized by SDS-PAGE (Fig. 5d), was shown by LC-MS to be ␣-syn⌬C fragments with or without N-terminal truncations (Fig. 5e). At the C-terminal end, the major cleavage site is at Asn-122/Gly-123 along with minor sites at Gln-109/Gly-110 and Gly-114/Asp-115 (Table S5). Cut sites in the N terminus were seen at Phe-4/Met-5, Ser-9/Lys-10, Gly-14/Val-15, and Ala-17/Ala-18, corresponding to fragments 5-122, 10 -122, 15-122, and 18 -122. These peptides were also observed using fibrils formed in vitro from the PD-associated A53T ␣-syn mutant ( Fig. S8; Table S6), indicating that this point mutation does not change fibril degradation at the N and C terminus. These data complement results from lysosomes from SNCA A53T mice showing that the 12-kDa species is derived from C-terminal cleavage at Asn-122/Glu-123 and can be attributed to cysteine cathepsin digestion. The fragment 1-114 was also seen from lysosomal activity in SNCA A53T mice and is because of CtsB activity; however, the 1-109 fragment appeared to be specific to N27 cells and its origin could not be identified. The N-terminal cut at Gly-14/Val-15 is the consequence of CtsB whereas Ser-9/Lys-10 and Ala-17/Ala-18 result from CtsL activity (27,29). No assignment of the 8-kDa fragment could be made because of its low abundance. Collectively, these data strongly support that both N-and C-terminal truncations of ␣-syn originate from incomplete lysosomal degradation and establish that in addition to the 12-kDa species, the 8-kDa band is derived from a fibrillar state of ␣-syn.

␣-Syn⌬C generated in situ accelerates ␣-syn amyloid formation
To assess the aggregation potential of ␣-syn truncations after incomplete lysosomal digestion, we sought to recapitulate these events using purified cathepsin to investigate its impact on fulllength ␣-syn. Specifically, we asked whether the C-terminal truncations formed via limited proteolysis could act as seeds and propagate soluble ␣-syn aggregation at physiological pH. The cysteine cathepsin AEP was chosen, because this protease has limited substrate specificity on ␣-syn, cleaving mainly at three positions: Asn-65/Val-66, Asn-103/Glu-104, and Asn-122/Glu-123. Because residues Asn-65 and Val-66 are in the amyloid core, they should be less accessible in the fibrillar state (Fig. 1b). Upon fibril digestion, truncations 1-103 and 1-122 should be more abundant. Both of these C-terminal truncations are seen in Lewy bodies (Fig. 1) and have been identified from lysosomal extracts in this work (Figs. 4c and 5e).
As anticipated, incubation of fibrillar ␣-syn (100 M) with AEP (200 nM) at pH 5 for 20 h at 37°C showed that 1-122 and 1-103 were the dominant species, with a minor population of peptide fragments from cleavage at Asn-65/Val-66 ( Fig. 6a; Table S7). Attempts to completely remove the C terminus from full-length ␣-syn f by increasing the AEP was unsuccessful. TEM images of AEP-treated fibrils (␣-syn AEP ) showed multiprotofibril structures (ϳ5 nm each) that were laterally associated to form fibrils that had a high degree of twisting (Fig. 6a,  bottom). By comparison, greater lateral association of protofibrils was observed for untreated fibrils at pH 5, where some fibrils (Ͼ40 nm in diameter) were composed of multiple protofibrils (Fig. 6a, top).
Next, preformed ␣-syn f and ␣-syn AEP (5, 10, and 20%) seeds were incubated with 40 M ␣-syn at pH 7.4 with continuous shaking at 37°C (0 -40 h) and monitored by thioflavin T (ThT), an extrinsic fluorophore that increases in intensity upon binding to amyloid fibrils (42). In the absence of seeds, aggregation of soluble ␣-syn proceeded with a lag phase (ϳ10 h) followed by a shallow elongation phase (Fig. 6b, red). Addition of preformed ␣-syn f seeds had little effect on the observed kinetics, but rather produced higher final ThT signals (Fig. 6b, top). SDS-PAGE analysis of the pelleted samples taken after 40 h showed comparable protein levels, suggesting the ThT difference is not because of increased fibrillar material (Fig. 6c, top). In the presence of ␣-syn AEP seeds (Fig. 6b, bottom), ␣-syn aggregation is noticeably faster. The lag phase is hastened as a function of seed concentration, and higher ThT fluorescence signal was reached at the end of the reaction. Again, SDS-PAGE analysis of ultracentrifuged protein samples taken post aggregation suggested that ThT signal changes are not indicative of fibril concentration. Our data clearly show that the truncated fibrils act as superior seeds in promoting ␣-syn amyloid formation compared with that of the full-length fibrils.
Representative TEM images taken post seeding with ␣-syn f and ␣-syn AEP displayed paired protofibrils (5 nm each) laterally associating and forming varying twists (Fig. 6d). Because of the heterogenous nature of these fibrils, which also contain some nontwisting fibrils, it is difficult to define morphological differences between seeding conditions. Although the TEM images were inconclusive, the striking differences in ThT signals between ␣-syn AEP -seeded versus self-seeded ␣-syn reactions suggested different fibril polymorphs. To further address this, a second round of seeding was performed. Using the first generation of ␣-syn AEP seeds (10%), the lag phase was abrogated, and a higher ThT signal was observed compared with the selfseeded reaction (Fig. S9), supporting the propagation of a dif-C-terminal ␣-syn truncations ferent polymorph in the ␣-syn AEP -seeded fibrils. These results highlight the potency of the original C-terminal truncated seeds. Although TEM images did not discern obvious morphological differences after a second round of seeding, it is clear that the seeded samples are more homogenous than the nonseeded controls (Fig. S10).

Discussion
The presence of ␣-syn truncations in LBs is a feature of PD (Fig.  1a), yet the mechanism(s) underlying the formation of these aggregation-prone species is not well-understood. In this work, we could assign Ͼ60% of the cleavage sites to lysosomal proteases (Fig. 1b), supporting a central role for lysosomal function in PD (43,44). For the first time, we show that ␣-syn⌬C species are found in brain lysosomal preparations from a transgenic SNCA A53T PD mouse model. Their abundance was significantly increased in symptomatic versus nonsymptomatic mice. Our results are supported by another transgenic SNCA A53T mouse study, where immunoblot analysis of mouse brainstem and cortical extracts from symptomatic animals also showed similar increases in fulllength and ␣-syn truncations (45).
We suggest that this enrichment of truncations is not because of a loss of lysosomal proteolysis, as individual cathepsin activities were unchanged (Fig. 3; Fig. S2), but rather, because of a change in the balance between cathepsin and ␣-syn levels in the lysosome. Under disease conditions, aggregated ␣-syn is increased in the lysosome. The increased burden leads to incomplete degradation, yielding ␣-syn⌬C species (Fig. 2, f, i-l).
This work revealed that lysosomal fractions in symptomatic SNCA A53T mice had ␣-syn truncations corresponding to 12and 8-kDa in size (Fig. 2, j-l). This was also observed in N27 dopaminergic cells fed with exogenous ␣-syn fibrils and they originated from truncations of fibrils taking place in the lysosome (Fig. 5). From epitope mapping, likely cleavage sites were defined to be between residues 118 to 123 and 91 to 99, respectively. Unlike the 12-kDa species, which we identified to be the  (Table S7). b, scheme of ␣-syn seeding reactions. Aggregation kinetics of ␣-syn (40 M) at pH 7.4 in the absence (red) and presence of 5 (black), 10 (green), and 20% (purple) fibril seeds alone (␣-syn f seeds, top) and with AEP digestion (␣-syn AEP seeds, bottom) and monitored by ThT fluorescence ( obs ϭ 480 nm). The unseeded data are reproduced in both panels for comparison. Each condition has at least four replicates each. c, corresponding SDS-PAGE (4 -12%) visualized by Coomassie Blue staining of pelleted samples (100,000 ϫ g) taken post aggregation. d, representative TEM images taken after 20 h for each condition as indicated. Scale bar is 100 nm.

C-terminal ␣-syn truncations
1-122 segment originating from cysteine cathepsins (Fig. 4d), it proved difficult to conclusively identify the 8-kDa fragment from lysosomal degradation of ␣-syn fibrils. However, through proteolysis of soluble ␣-syn, LC-MS analysis suggested that it is most likely to be the fragment containing residues 1-90, generated solely by CtsB activity (Fig. S5). This is a newly identified ␣-syn truncation and further work is needed to establish its importance. Interestingly, the CTSB gene encoding CtsB has been recently identified as a PD risk factor (32), up-regulated with increased activity in DLB (33). Moreover, limited proteolysis of ␣-syn fibrils formed in vitro by CtsB have been shown to enhance aggregation of endogenous ␣-syn in cells (31), providing evidence for a potential role for CtsB in PD pathogenesis.
In addition to C-terminal truncations, four N-terminal truncations (Phe-4/Met-5, Met-5/Lys-6, Ser-9/Lys-10, and Gly-14/ Val-15) were observed in both SNCA A53T and N27 samples (Figs. 4c and 5e). Only the N-terminal truncation at residue 5 is found in LBs; thus, the relevance of the other sites remains to be defined. Because it has been shown that removal of 10 and 30 residues in the N terminus modulates fibril formation (15), these truncations will also likely influence ␣-syn aggregation. Hence, the interplay of N-and C-terminal truncations (e.g. residues 10 -122) in the mechanism of ␣-syn amyloid assembly warrants further investigation.
A pathological implication for ␣-syn⌬C species was clearly demonstrated by the seeding potency of C-terminal truncations 1-103 and 1-122 generated via AEP digestion of preformed fibrils (Fig. 6b). We propose that under disease conditions, the burden of ␣-syn accumulation and aggregation overwhelms lysosomal degradation, leading to incomplete proteolysis of fibrils and the buildup of ␣-syn⌬C species. If released from the lysosome, ␣-syn⌬C species would recruit and propagate aggregation of cytosolic ␣-syn. This proposed mechanism is supported by observations of lysosomal rupture induced by ␣-syn (46 -48). It has been shown that ␣-syn aggregates colocalize with galectin-3, a marker of endolysosomal membrane rupture (47), although the molecular mechanisms causing lysosomal rupture remain to be elucidated. Of note, in vivo studies have also suggested that intercellular transmission of ␣-syn aggregates occurs via the endolysosome pathway (38,39). One study proposes ␣-syn fibrils traffic inside lysosomes in tunneling nanotubes, contributing to intercellular transfer of ␣-syn fibrils (49). Taken together, these events substantiate the connection between lysosomes and amyloid propagation in PD.
With these identified ␣-syn⌬C species, quantification of their physiological concentrations would be of particular interest, as some are readily found in healthy tissue (50). One intriguing implication is how they could play a role in the reciprocal relationship between levels of ␣-syn and the lysosomal hydrolase glucocerebrosidase (GCase) (51), which is currently recognized as the most prevalent genetic risk factor for the development of PD and DLB (52,53). Although the specific role of GCase activity in modulating ␣-syn levels remains controversial, studies have shown that enhancing GCase activity is a potential therapeutic strategy for synucleinopathies (54 -57). Interestingly, reduced levels of GCase are reported to facilitate cell-to-cell transmission of ␣-syn (58). Because GCase binds to the C-terminal region of ␣-syn, from residues 118 to 137 (59 -61), the physical interaction between full-length ␣-syn and GCase would inhibit C-terminal truncation via steric hindrance, which would have a protective effect in reducing the presence of potent seeds for amyloid propagation. On the other hand, once truncations such as those observed in this study are formed (1-122, 1-114, 1-103, and 1-90) as well as others found in LBs (1-119, 1-115, and 1-101), this association would be mitigated. Further work is needed to establish the importance of this physical interaction in the context of PD where ␣-syn truncations are enriched.
Based on our results, a targeted therapeutic strategy is to decrease the production of ␣-syn⌬C species in lysosomes. Promising results from deletion of AEP were recently reported to reduce aggregation and dopaminergic cell death in mice overexpressing human ␣-syn (28). Importantly, we have previously shown that with sufficient amounts of CtsL, this cysteine cathepsin is fully capable of cannibalizing ␣-syn fibrils (27,29). However, from our studies in SNCA A53T mouse brain samples, it is clear that there is limited CtsL activity in lysosomes. Thus, one therapeutic approach might be to ameliorate amyloid load by enhancing CtsL activity in the lysosome.

Lysosome isolation from wt, ϳSNCA A53T , and SNCA A53T mice
Mice with a human SNCA A53T transgene (34) were housed and bred under NHGRI Animal Care and Use Committeeapproved protocols. Mice were euthanized when progressive neurodegenerative symptoms develop, typically at ϳ16 months of age. Symptoms were identified by clinical deterioration beginning with weight loss and progressing to abnormal gait and posture. When they become unable to ambulate, we euthanize them. Age-matched nonsymptomatic SNCA A53T littermates as well as nontransgenic mice were euthanized in parallel. A total of six brains, two from each group were used for lysosome isolation. Samples were kept on ice at all times. Brains were gently homogenized using a Dounce homogenizer (ϳ50 strokes) and spun at 500 ϫ g, and soluble lysates were combined with OptiPrep (lysosomal enrichment kit, Pierce) to a final concentration of 15% and placed on top of a discontinuous density gradient with the following steps from top to bottom: 17, 20, 23, 27, and 30%. After centrifugation for 2 h at 145,000 ϫ g, the top fraction containing the lysosomes was collected. The lysosomal fraction was diluted at least three times with PBS and pelleted by centrifugation for 30 min at 16,100 ϫ g. The pellets were washed once with PBS and then resuspended in 50 mM sodium acetate (NaOAc), 20 mM NaCl, pH 5 buffer and centrifuged again at 16,100 ϫ g for 30 min. Pellets were stored at Ϫ80°C. To generate lysosomal extracts, purified lysosomes were frozenthawed four times. Total protein concentrations in the resulting lysosome lysates were determined using the Bradford assay (detergent compatible kit from Pierce Biotechnology): For biological replicate 1: wt (0.54 mg/ml), ϳSNCA A53T (0.75 mg/ml), and SNCA A53T (0.5 mg/ml) and biological replicate 2: wt (1.53 mg/ml), ϳSNCA A53T (0.9 mg/ml) and SNCA A53T (0.58 mg/ml).

Cathepsin activity assays
Cathepsin activity was measured using substrates (Ac-RR-AMC for CtsB, MCA-GKPILEFRKL(Dnp)-D-R-NH 2 for CtsD, Ac-FR-AMC for CtsL, and z-AAN-AMC) at a final concentration of 200 M (for CtsB, CtsL, and AEP) and 20 M (for CtsD) with 1 g total protein lysosomal extract in pH 5 buffer. All buffers contained 5 mM DTT. Reactions were performed in polypropylene 384-well flat-bottom microplates (Greiner Bio-One) containing 50-l solution incubated at 37°C using a microplate reader (Tecan Infinite M200 Pro). For each biological replicate set, a total of three repetitions were performed. CtsB, CtsL, and AEP (excitation and emission wavelengths at 360 and 460 nm, respectively) and CtsD fluorescence (excitation and emission wavelengths at 328 and 393 nm, respectively) were recorded as a function of time (0 -90 min).

Degradation reactions of recombinant ␣-syn
In microcentrifuge tubes (1.5-ml Protein LoBind tubes, catalog no. 022431081, Eppendorf), soluble ␣-syn and preformed ␣-syn fibrils (15 and 30 M) were incubated with lysosomal extracts. The amounts of lysosomal extracts were 1-5 and 2-30 g for soluble and fibrillar ␣-syn, respectively. For inhibition experiments, lysosomal extracts were preincubated with 1 M of each inhibitor for 30 min at room temperature prior to the addition of ␣-syn. Limited proteolysis of ␣-syn f (100 M) was performed using 200 nM of AEP. All reactions were carried out in a total volume of 100 l reaction buffer (50 mM NaOAc, 20 mM NaCl, 5 mM DTT, pH 5), agitated at 600 rpm at 37°C for 20 h in a Mini-Micro 980140 shaker (VWR).

LC-MS
Proteolyzed samples (5-20 l) were separated using an HPLC (Agilent 1100 series, Agilent Technologies) on a reverse phase C18 column (Zorbax, 2.1 ϫ 50 mm, 3.5 m, Agilent Technologies) and introduced into the mass spectrometer as described (65,66). Fibrillar samples were preincubated to a final concentration of 2 M guanidium hydrochloride before injection. Positive ion electrospray ionization (ESI) mass spectra for intact protein (peptides) of ␣-syn were obtained with an Agilent G1956B mass selective detector (MSD) equipped with an ESI interface (Agilent Technologies). The HPLC systems and MSD C-terminal ␣-syn truncations were controlled and data were analyzed using LC/MSD Chem-Station software (Rev. B.04.03, Agilent Technologies).

Live cell imaging of N27 dopaminergic cells
N27 dopaminergic cells were maintained in RPMI media containing 10% fetal bovine serum and supplemented with 2% penicillin/streptomycin at 37°C in 5% CO 2 atmosphere. N27 cells at 70 -80% confluency were trypsinized by the addition of 0.05% trypsin-EDTA for 5 min. Resuspended cells (50 -80 l) and fresh media (300 l) were added to Labtek 8-well chambers precoated with poly-L-lysine (P8920, Sigma). Exogenous soluble and fibrillar ␣-syn in pH 5 buffer was diluted with media to desired final concentrations (100 nM and 1 M) and added prior to imaging. Samples were imaged using a 20ϫ air objective on a Zeiss 780 confocal microscope (NHLBI Light Microscopy Core) and maintained at 37°C in CO 2 atmosphere.

Immunofluorescence analysis of N27 cells
N27 cells were plated on no. 1.5 glass coverslips in 6-well plates and allowed to recover for 24 h. Cells were then incubated for 48 h at 37°C in fresh media containing the desired concentration of soluble (exchanged into pH 5 buffer) or fibrillar (preformed in pH 5 buffer, bath sonicated 15 min before use) ␣-syn. Cells were fixed with 2% paraformaldehyde in PBS buffer for 15 min and washed with PBS. Cells were then permeabilized with 0.5% Triton X-100 and 3% BSA in PBS for 7 min and blocked with 0.2% Triton X-100 and 3% BSA in PBS for 1 h. Finally, cells were stained for 1 h with primary antibody (mouse CtsL (ab6314) and rabbit ␣-syn (ab51252), diluted 1:50 -1:100 in blocking buffer), washed with blocking buffer, stained for 1 h with secondary antibody (goat ␣-mouse Alexa Fluor 532 (A-11002) and donkey ␣-rabbit Alexa Fluor 488 (ab150073), diluted 1:1000 in blocking buffer), and washed with blocking buffer. All procedures were performed at room temperature. Coverslips were stored in PBS at 4°C in the dark before imaging. Samples were imaged using a UPLSAPO 100ϫ/1.35 NA silicon oil objective (Olympus) on an OlympusIX73 inverted microscope fitted with a Thorlab Confocal Laser Scanner (CLS-SL) fiber coupled to a multichannel CMLS-E laser source. Alexa Fluor 488 and 532 emission were excited at 488 and 532 nm and collected using 512 Ϯ 25 and 582 Ϯ 75 nm band-pass filters, respectively, by two independent high-sensitivity GaAsP photomultiplier tubes. A 45-m pinhole was used and the scale was 0.1 m/pixel. Two independent treatments were imaged and analyzed. Images were analyzed with Fiji (67).

Lysosome isolation from rat N27 dopaminergic cells
N27 cells from 31 T75 flasks containing 10 ml of cell culture at 70 -80% confluency was scraped using a cell scraper and centrifuged at 3220 ϫ g for 10 min. Cell pellet was resuspended in PBS and gently homogenized using a Dounce homogenizer (200 strokes). Lysosome purification proceeded as described for mice lysosome isolation. The protein concentration in the resulting lysates was determined to be 0.52 mg/ml using the Bradford assay (detergent compatible kit from Pierce Biotechnology).

Aggregation kinetics
Aggregation was performed in sealed black polypropylene 384well flat-bottom microplates (781209, Greiner Bio-One) in the absence or presence of preformed ␣-syn f seeds and continuously shaken linearly (1.0 mm, 173.9 rpm) at 37°C using a microplate reader (Tecan Infinite M200 Pro). Seeds were added (2.5, 5, and 10 l from aggregation or digestion reactions performed at 100 M as described above) to a solution (␣-syn ϭ 50 M in 20 mM sodium phosphate and 100 mM NaCl, pH 7.4) to a final volume of 40 l. Each well also was supplemented with a 2-mm glass bead. The final ␣-syn concentration is 40 M. ThT (10 M) fluorescence (excited and monitored at 415 and 480 nm, respectively) was recorded as a function of time. A total of three independent experiments was performed with at least four replicates on each plate.

TEM
Samples (10 l) were put on TEM grids (400-mesh Formvar and carbon-coated copper, Electron Microscopy Sciences) for approximately 2 min and wicked away by filter paper. Addition of 10 l of deionized water was then applied and wicked away immediately. A solution of 1% uranyl acetate (10 l) is placed on the grid for 2 min, wicked away, and air-dried. TEM was performed using a JEOL JEM 1200EX transmission electron microscope (accelerating voltage 80 keV) equipped with an AMT XR-60 digital camera (NHLBI EM Core Facility).