Formation and Removal of α-Synuclein Aggregates in Cells Exposed to Mitochondrial Inhibitors*

Mitochondrial dysfunction has been associated with Parkinson's disease. However, the role of mitochondrial defects in the formation of Lewy bodies, a pathological hallmark of Parkinson's disease has not been addressed directly. In this report, we investigated the effects of inhibitors of the mitochondrial electron-transport chain on the aggregation of α-synuclein, a major protein component of Lewy bodies. Treatment with rotenone, an inhibitor of complex I, resulted in an increase of detergent-resistant α-synuclein aggregates and a reduction in ATP level. Another inhibitor of the electron-transport chain, oligomycin, also showed temporal correlation between the formation of aggregates and ATP reduction. Microscopic analyses showed a progressive evolution of small aggregates of α-synuclein to a large perinuclear inclusion body. The inclusions were co-stained with ubiquitin, 20 S proteasome, γ-tubulin, and vimentin. The perinuclear inclusion bodies, but not the small cytoplasmic aggregates, were thioflavin S-positive, suggesting the amyloid-like conformation. Interestingly, the aggregates disappeared when the cells were replenished with inhibitor-free medium. Disappearance of aggregates coincided with the recovery of mitochondrial metabolism and was partially inhibited by proteasome inhibitors. These results suggest that the formation of α-synuclein inclusions could be initiated by an impaired mitochondrial function and be reversed by restoring normal mitochondrial metabolism.

Many neurodegenerative diseases are associated with characteristic intracellular protein inclusions in the affected brain regions (1,2). Lewy body (LB) 1 is one of the inclusions that is associated with Parkinson's disease (PD) and related neurologic diseases, including dementia with Lewy bodies and Lewy body variant of Alzheimer's disease (3). Aggregation of ␣-synuclein (␣-syn) appears to be essential for the LB formation since fibrillar aggregates of ␣-syn are major components of LBs (4,5), thus the diseases that are characterized by LBs are collectively referred to as ␣-synucleinopathies. ␣-Syn itself forms typical amyloids in solution by a nucleation-dependent mechanism (6 -8). Importantly, two autosomal dominant mutations of the ␣-syn gene were linked to rare familial earlyonset PD (9,10), and the mutant proteins tend to aggregate more rapidly than the wild type (11)(12)(13). Transgenic flies modified to express human ␣-syn at various levels develop age-dependent pathological and behavioral changes that resemble human PD, including LB-like intraneuronal inclusions, loss of dopaminergic neurons, and a decline in locomotor activity (14). Also, transgenic mice that express human ␣-syn produced neuronal inclusions and dopaminergic presynaptic degeneration (15). In addition, the extent of formation of these LB-like inclusions has been shown to correlate with the expression level of ␣-syn in transgenic animals (15). These findings provide compelling evidence that the aggregation of ␣-syn is directly linked to LB formation, and therefore to the pathogenesis of ␣-synucleinopathies, especially PD. Although substantial progress has been made regarding structural and morphological changes during the fibrillization of isolated ␣-syn, physiological factors that influence the cellular aggregation process remain unknown.
Mitochondrial dysfunction has been implicated in pathogenesis of PD and other neurodegenerative diseases (16,17). Several groups reported a systemic or local decrease of mitochondrial complex I activity (18 -21) or immunoreactivity (22) in idiopathic PD. Another line of evidence for mitochondrial defects in PD came from studies where cybrids that contained mitochondria from PD patients showed reduced complex I activity (23,24). In fact, 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine-induced parkinsonism is attributed to the inhibition of complex I by its metabolite, 1-methyl-4-phenylpyridinium (MPP ϩ ), which leads to reduction in mitochondrial ATP production (25,26). Moreover, in a recent study administration of rotenone, an inhibitor of mitochondrial complex I, recapitulated the major pathological and behavioral features of PD, including selective loss of nigrostriatal dopaminergic neurons and ␣-syn-positive LB-like inclusions, indicating that the inhibition of complex I in the electron-transport chain (ETC) might be sufficient to trigger PD pathogenesis in a rat model (27). Although mitochondrial defects are clearly associated with PD, a direct role of mitochondrial dysfunction in LB formation has not been addressed. Here we provide evidence that the functional state of mitochondria might be a key determinant for aggregation of ␣-syn. In addition, the aggregates appeared to be eliminated by restoring normal mitochondrial metabolism, suggesting the presence of cellular mechanisms that actively disassemble and degrade protein aggregates.

Construction of Recombinant Adenoviral
Vector and Plasmids-Human wild type ␣-syn cDNA was amplified by polymerase chain reaction using the primer sets, 5Ј-ATCAGATCTGCCATGGATGTATTCATGAA-AGGA and 5Ј-ATCGTCGACGGATGGAACATCTGTCAGCA, and pRK172/␣-syn (a gift from Peter T. Lansbury, Jr., Harvard Medical School, Boston, MA) as a template. The cDNA was digested with BglII and inserted into the BamHI site of the transfer vector, pQBI-AdCMV5 (Qbiogene, Carlsbad, CA). In vivo homologous recombination and subsequent amplification of the recombinant virus were carried out in 293A cells according to the manufacturers instructions (Qbiogene). Final stock of the amplified recombinant virus (adeno/␣-syn), which was aliquoted and stored in Ϫ80°C, had the titer of ϳ2.0 ϫ 10 8 pfu/ml. cDNAs for ␤and ␥-syns were amplified by PCR from a human brain cDNA pool (CLONTECH, Palo Alto, CA) using primer sets, 5Ј-ACTG-GATATCGCCAGGATGGACGTGTTCATG and 5Ј-ACTGTCTAGACGC-CTCTGGCTCATACTCCTG for ␤-syn, and 5Ј-ACTGGATATCGCCACC-ATGGATGTCTTCAAGAAGG and 5Ј-ACTGTCTAGAGTCTCCCCCAC-TCTGGGCCTC for ␥-syn. ␣-Syn cDNA was amplified by PCR from pRK172/␣-syn using a primer set, 5Ј-ACTGGATATCGCCACCATGGA-TGTATTCATGAAAG and 5Ј-ACTGTCTAGAGGCTTCAGGTTCGTAG-TCTTG. The PCR products were inserted to an EcoRV/XbaI-cut pCDNA3.1/Myc-His vector (Invitrogen, Carlsbad, CA). The sequences of the inserts in all the plasmids were confirmed by sequencing.
Expression of Syn Proteins and Induction of Aggregation-Transformed African monkey kidney cell line COS-7 was maintained in Dulbecco's modified Eagle's medium (Invitrogen, Rockville, MD) with 10% fetal bovine serum (HyClone Laboratories, Inc., Logan, UT) in a 37°C, 5% CO 2 incubator. For the infection with adeno/␣-syn, cells were split on 100-mm tissue culture dishes a day prior to infection to obtain about 90% confluency on the day of infection. For infection, 1 ϫ 10 8 pfu of adeno/␣-syn in 1 ml of medium was added to each dish. After 90 min incubation at 37°C, 9 ml of fresh medium was added to the cells and incubated overnight. Cells were then split onto 60-mm dishes or coverslips in 12-well plates for further experiments. For the comparison of ␣-, ␤-, and ␥-syn, cells were transfected with pCDNA␣-synMH, pCDNA␤-synMH, or pCDNA␥-synMH using LipofectAMINE Plus (Invitrogen) according to Lee et al. (28). For the inhibition of ETC, cells were treated with rotenone (0.1 M in Me 2 SO; Sigma) or oligomycin (0.1 g/ml in Me 2 SO; Sigma) 2 days post-infection and harvested or fixed for immunostaining at the times indicated (0 -3 days). For the wash-out experiments, the medium from the cells treated with rotenone or oligomycin was replaced with fresh inhibitor-free medium after 48 h or at the time indicated.
ATP Assay and Western Blotting-Cells grown in 60-mm tissue culture dishes were harvested in 0.5 ml of phosphate-buffered saline (PBS) with protease inhibitor mixture (Sigma) and quickly frozen in dry ice/methanol bath. After thawing on ice, cells were diluted 100-fold with the same buffer. ATP assay was performed following manufacturers instructions using luminescence-based ATP assay kit (Calbiochem-Novabiochem Corp., San Diego, CA). Data was normalized by protein contents. Similar results were obtained after the data was normalized by the number of cells. Mean and S.E. were calculated and graphed with GraphPad Prism version 3.02 for Windows (GraphPad Software, San Diego, CA). One-way ANOVA with Dunnett's post-test was performed using GraphPad Instat version 3.05 (GraphPad Software). For the Western analysis, 5 l of 10% Triton X-100 was added to 45 l of cells to make a final 1% Triton X-100 and vortexed. The extraction was done for 15 min on ice and the cell extract was centrifuged at 14,000 ϫ g for 10 min. The supernatant (Triton-soluble fraction) was saved and the pellet (Triton-insoluble fraction) was resuspended in 1 ϫ sample buffer (62.5 mM Tris, pH 6.8, 10% glycerol, 2% SDS, 2.5% ␤-mercaptoethanol, 0.025% bromphenol blue) and sonicated. Both fractions were loaded onto 12% SDS-PAGE, and the Western blotting was performed as described elsewhere (29). Antibodies used were LB509 (Zymed Laboratories Inc., South San Francisco, CA) for ␣-syn and anti-Myc antibody (Invitrogen) for epitope-tagged proteins.
Fluorescent Cell Staining-The procedure for the staining is described elsewhere except for a few modifications (28). All the steps were done at room temperature unless stated otherwise. Briefly, cells on coverslips were fixed in 4% paraformaldehyde in PBS for 30 min and permeabilization with 0.1% Triton X-100 in PBS for 5 min. Coverslips were blocked in blocking solution (PBS, 5% bovine serum albumin, 3% goat serum) for 30 min and incubated with LB509 alone or in combination with anti-ubiquitin antibody (Chemicon International, Inc., Temecula, CA) or anti-20 S proteasome ␣-subunit antibody (Calbiochem-Novabiochem Corp.) in blocking solution for another 30 min. For the co-staining with anti-␥-tubulin antibody (Sigma) or anti-vimentin antibody (Sigma), a polyclonal antiserum against ␣-syn, 7071 (a gift from Peter T. Lansbury, Jr.), was used. After extensive washing for 1-2 h in PBS, Cy3-conjugated goat secondary antibodies (Jackson Immu-noResearch Laboratories, Inc., West Grove, PA) in blocking solution was added for 30 min. For thioflavin S staining, fixed cells were incubated with 0.05% thioflavin S (Sigma) for 8 min and washed three times with 80% ethanol for 5 min each before the antibody incubations (30). Terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) was carried out using the kit from Roche Molecular Biochemicals (Indianapolis, IN) for 1 h at 37°C and washed with PBS. The nuclei was stained with Hoechst 33258 (2 g/ml; Molecular Probes, Inc., Eugene, OR) in PBS for 10 min and washed three times with PBS for 30 min before mounting with Prolong anti-fade reagent (Molecular Probes, Inc.). Images were analyzed using the DeltaVision deconvolution microscope (Applied Precision, Inc., Issaquah, WA) in the Cell Science Imaging Facility at Stanford University.

RESULTS
To investigate the role of mitochondrial dysfunction in ␣-syn aggregation, human wild type ␣-syn was expressed in COS-7 cells using a recombinant adenoviral vector system, which yielded homogeneous expression in nearly 100% of the cells. Cells were then treated with two ETC inhibitors, rotenone, a complex I (NADH-coenzyme Q reductase) inhibitor, and oligomycin, a complex V (F 0 F 1 -ATP synthase) inhibitor. At each time point, cell extracts were fractionated to the Triton X-100soluble and -insoluble fractions and analyzed by Western blot analysis. After 48 h of incubation in the presence of rotenone, the cells produced visible amounts of SDS-resistant ␣-syn aggregates, and by 72 h, most of the ␣-syn was in aggregated form ( Fig. 1A). Vehicle alone had almost no effect on ␣-syn aggregation during the time course we examined (Fig. 1C). Rotenone did not influence the aggregation of ␣-syn in cell-free assays, suggesting that the effect of rotenone was not targeted directly to the protein itself (Fig. 1D). Oligomycin, although affecting a different target in the ETC, induced the aggregation of ␣-syn with a similar time course, supporting the idea that the ETC inhibition leads to the aggregation of ␣-syn (Fig. 1B).
There are other proteins, such as ␤-syn, ␥-syn, and synoretin, that are closely related to ␣-syn (31,32). Despite the sequence homology within the synuclein family of proteins, only ␣-syn is found in LBs (33). Furthermore, in vitro studies showed that ␣-syn is much more prone to aggregate than other syns and does not cross-seed ␤or ␥-syn (34,35). To compare the effects of ETC inhibition on ␣-, ␤-, and ␥-syns, cells that transiently expressed Myc/His-tagged synucleins were treated with rotenone, and the aggregation was monitored in the Triton-insoluble fractions. The antibody against Myc-tag was used to quantify the amounts of monomer and aggregates of each protein. Consistent with the previous in vitro studies with purified recombinant proteins (34,35), only ␣-syn produced aggregates at day 2 despite the fact that the expression level of ␤-syn was much higher than that of ␣-syn. This data suggests that ␣-syn has highest aggregation tendency in the syn family in response to altered mitochondrial metabolism (Fig. 2).
In PD and other related synucleinopathies, the aggregation of ␣-syn is often manifested by the presence of large cytoplasmic inclusion bodies called LBs (3). We, therefore, investigated the formation of inclusion bodies in rotenone-treated cells microscopically (Fig. 3). Before rotenone treatment (t ϭ 0 in Immunohistochemical studies showed that LBs contain not only ␣-syn, but also ubiquitin and subunits of proteasome (33,36). To determine the presence of these proteins in the inclusion, we co-stained the rotenone-treated cells with antibodies against ␣-syn and ubiquitin or the ␣-subunit of 20 S proteasome. Both anti-ubiquitin and anti-␣-subunit antibodies co-stained with ␣-syn antibody in the perinuclear inclusions (Fig.  4, A and B). Thus, the perinuclear inclusions that were ␣-synpositive contained ubiquitin and the proteasome subunit. In contrast, small aggregates that were dispersed in the cytoplasm contained neither ubiquitin nor the proteasome subunit (Fig. 4). We have not detected any small ␣-syn aggregates that were positive for either of these proteins.
Ubiquitin was also detected in the SDS-resistant, high molecular weight protein complex by Western blotting (data not shown). However, whether ␣-syn was directly ubiquitinated is not clear because of several reasons. First, ␣-syn aggregates and ubiquitin-positive aggregates can be separate entities, which happen to run at a similar rate on the SDS gel. Coimmunostaining of the perinuclear inclusions often showed segregation of these proteins (Fig. 4A). Second, ␣-syn may not be ubiquitinated directly, but simply co-aggregates with other ubiquitinated proteins. Analysis of monomeric proteins dissociated from aggregates will provide the direct answer to this problem.
Recently, the formation of perinuclear inclusions called aggresomes were described as a general response of cells to an accumulation of misfolded proteins (37). These aggresomes contain ␥-tubulin, since they form near the microtubule organizing center, and are often surrounded by vimentin filaments (37).
The ␣-syn-positive inclusions in rotenone-treated cells also contain ␥-tubulin and are surrounded by vimentin (Fig. 4, C and  D). However, the cells with inclusions did not show a significant rearrangement of vimentin filaments (Fig. 4D), as was the  case for typical aggresome-containing cells (37).
␣-Syn aggregates in LBs are known to have a structure similar to amyloids, which are characterized by an ordered array of crossed ␤-sheet structure (6,7). This structure can be selectively recognized by thiazole dyes, such as thioflavin T and S, good indicators of amyloid-like conformation (38,39). To obtain conformational information, we double-stained rotenone-treated cells with an anti-␣-syn antibody and thioflavin S. As described above, two forms of aggregates can be detected after rotenone treatment, small cytoplasmic aggregates and large perinuclear inclusions. The perinuclear inclusions in rotenone-treated cells were strongly stained with thioflavin S, while the small aggregates were thioflavin S-negative (Fig. 5).
To evaluate the functional defects of mitochondria in the cells exposed to the ETC inhibitors, we measured ATP levels as a marker of mitochondrial metabolism after rotenone or oligomycin treatment. When the cells were treated with rotenone, the reduction in ATP preceded and paralleled the aggregation of ␣-syn up to 72 h, with reductions of 23, 87, and 94% at 24, 48, and 72 h, respectively (Fig. 6A). Oligomycin treatment resulted in a similar time course of ATP reduction, which again paralleled the aggregation of ␣-syn (Fig. 6B). These data suggest that functional defects in the ETC may cause the aggregation of ␣-syn. To test whether mitochondria can recover after removal of the ETC inhibitors, the cells were cultured with rotenone or oligomycin for 48 h, then the medium was replaced with the inhibitor-free medium (Fig. 7A). ATP levels were slowly restored in the wash-out phase. At 48 h of wash-out, the rotenone-treated cells had almost fully recovered from the ATP deficit while the oligomycin-treated cells had recovered by 50%, suggesting a functional recovery of mitochondria (Fig. 6, C  and D).
To determine whether the aggregation of ␣-syn can be reversed by the removal of the ETC inhibitors, the extent of SDS-resistant ␣-syn aggregates were investigated in the washout experiment (Fig. 7A). During this period, the ␣-syn aggregates gradually diminished, and by 48 h of wash-out, most of the aggregates had disappeared (Fig. 7, B and C). Disappearance of the aggregates paralleled the recovery of mitochondrial metabolism, which was monitored by the cellular ATP level. Even after longer aggregation period, which produced more ␣-syn aggregates, replacement with rotenone-free medium resulted in a progressive removal of the aggregates (Fig. 7D). This phenomenon could be due to the death of the cells with ␣-syn aggregates. However, TUNEL staining showed that the number of dying cells was minimal (4 -8% of total cells), with no correlation between ␣-syn aggregation and apoptotic cell death during the entire time course of the experiment (data not shown). An independent assessment of cell death by nuclear morphology produced similar results (data not shown). Therefore, the disappearance of the ␣-syn aggregates is not due to the  Fig. 1. C and D, the cells were treated as in Fig. 1, and then the medium was replaced with fresh inhibitor-free medium after 48 h (see Fig. 7A). Four independent experiments were performed, and ATP was measured in duplicate (in some experiments, in triplicate) in each set of experiments. Error bar represent S.E. *, indicates a significant difference from the control (0 h), as assessed by one-way ANOVA (p Ͻ 0.05); #, indicates that there is a significant difference between before and after the wash-out (p Ͻ 0.05). selective loss of the cells with the aggregates. Rather, there may be an active mechanism for the removal of the pre-existing aggregates in the cells. Disappearance of aggregates during the wash-out phase was also analyzed microscopically (Fig. 7E). Consistent with the Western analysis, replacement with rotenone-free medium resulted in a disappearance of small dispersed aggregates. However, the number of cells with the large perinuclear inclusions (about 8% of total cells) remained throughout the wash-out phase up to 2 days, suggesting that these large inclusions are more resistant to the aggregatecleanup mechanism than the small aggregates.
Disappearance of the aggregates in the wash-out phase was not accompanied by a concomitant increase of monomeric ␣-syn. This suggests an elimination of aggregates by degradation rather than by simple disassembly. To investigate the role of the proteasome system in the degradation of ␣-syn aggregates, we studied the effects of proteasome inhibitors on the disappearance of the aggregates during the rotenone wash-out.
Treatment of cells with rotenone for 56 h resulted in high molecular weight aggregates. While continued incubation for another 16 h in the presence of rotenone produced more aggregates, a parallel incubation in the absence of rotenone resulted in a disappearance of pre-existing aggregates (Fig. 8). However, addition of proteasome inhibitors MG132, ALLN, or lactacystin in the wash-out medium partially blocked the disappearance of the aggregates, suggesting that the proteasome system might be at least partly responsible for the degradation of ␣-syn aggregates.

DISCUSSION
Our study provides the first cell-based evidence that mitochondrial dysfunction may result in ␣-syn aggregation. One of the outcomes of ETC inhibition is an increased production of free radicals, hence increased oxidative stress (17). In recent studies, several groups have shown that the aggregation of ␣-syn could be promoted by oxidative and nitrative stresses (30,40,41). Indeed, oxidizing and nitrating agents induced dityrosine cross-linking of recombinant ␣-syn and stabilized preassembled aggregates (42). Furthermore, accumulation of nitrated ␣-syn was demonstrated in the inclusions of PD, dementia with Lewy bodies, Lewy body variant of Alzheimer's disease, and multiple system atrophy, implicating the role of oxidative stress in LB formation in ␣-synucleinopathies (43). Another outcome of mitochondrial dysfunction is a defect in energy production (ATP generation). In our study, we showed tight temporal correlations between ATP level and ␣-syn aggregation both in depletion and recovery phases, suggesting that an impaired energy supply may also play a role in ␣-syn aggregation.
Protein aggregation is considered to be a manifestation of a disturbed cellular protein-folding homeostasis, which is maintained by at least two defense mechanisms against damaged (misfolded) proteins: degradation by the ubiquitin-proteasome (Ub-Pr) system and the chaperone-mediated refolding system (44). Impairment of these systems, most of which are dependent on ATP, will cause accumulation of the misfolded proteins. Therefore, while oxidative stress can increase the rate of protein misfolding, the concomitant reduction in ATP levels can decrease the rate at which the cells rescue or remove the misfolded proteins, thus resulting in protein aggregates. Interestingly, unlike globular proteins, ␣-syn in isolation does not appear to have any stable structure (45). To initiate the aggre- gation process, ␣-syn has to undergo a structural transition to form an aggregation-prone, partially folded intermediate, which is equivalent to the misfolded proteins in the aggregation process of globular proteins. In fact, the presence of the partially folded intermediate and the stabilization of this conformation in the ␣-syn aggregation process were demonstrated in recent studies (46,47). Furthermore, the degradation of ␣-syn is mediated by the ubiquitin-proteasome system (48,49). These results suggest that the aggregation of ␣-syn might be under control of the same defense mechanism as globular proteins. Therefore, the correlation between the ATP level and ␣-syn aggregation shown in our study suggests that the reduction in energy production, hence the impaired defense mechanism against misfolded proteins, is likely to contribute to the aggregation of ␣-syn. During the rotenone treatment, chymotrypsin-like activity of proteasome in the cell lysates was measured using a fluorogenic peptide substrate. In this assay, the proteasome activity was only slightly decreased by about 10% in the rotenone-treated cells compared with the vehicle-treated cells. 2 It is unlikely that this slight decrease in catalytic activity of 20 S proteasome core is responsible for the protein aggregation, given that this activity is quite resistant to ATPdepletion when peptidic substrate is used (50,51). Rather, other ATP-dependent processes in Ub-Pr system, such as uniquitination cascade and regulatory activity of 19 S proteasome, might be impaired. Further characterization of these processes should unveil the role of energy metabolism and Ub-Pr system in ␣-syn aggregation.
Recently, a mechanism that explains the formation of protein inclusions in eukaryotic cells was proposed through the description of a structure called an aggresome. Aggresomes are formed near the microtubule organizing center by accumulation of small aggregates that are initially dispersed in the cytoplasm and delivered to the perinuclear region by dyneindependent retrograde transport on microtubules (37,52). In the present study, we showed that the ␣-syn-positive inclusions share some properties with the aggresomes. First, the temporal changes in the size and the distribution of ␣-syn aggregates, from small dispersed aggregates to large perinuclear inclusions, resemble those in aggresome formation. Second, the ␣-syn-positive inclusions contain ␥-tubulin, a marker of the microtubule organizing center. And finally, the inclusions are surrounded by what appear to be vimentin filaments, reminiscent of the "vimentin cages" that surround the aggresomes. These aggresome-like properties were also found in other inclusions that are relevant to neurodegenerative diseases, such as the inclusions of mutant Cu,Zn-superoxide dismutase and mutant huntingtin fragment (53,54).
Although small aggregates dispersed in the cytoplasm appeared to be precursors for the large perinuclear inclusions, thioflavin S staining showed that conformations of ␣-syn in these structures might be different: small aggregates were thioflavin S-negative, whereas large inclusions were thioflavin S-positive. This interesting distinction between small aggregates and large perinuclear inclusions suggests that the structural transition of ␣-syn into highly ordered ␤-sheet-rich conformation occurred after the aggregates accumulated in the perinuclear region. This thioflavin S reactivity can be explained by the "nucleated conformational conversion" model proposed by Serio et al. (55) in their work on the fibrillization of the N-terminal and middle regions of Sup35 (so called NM fragment), a yeast prion. In this model, elongation of crossed ␤-sheet amyloid is mediated by templated conformational conversion of less structured oligomeric intermediate. Less struc-tured ␣-syn oligomers (thioflavin S-negative) formed throughout the cytoplasm might be transported and concentrated in the perinuclear region where they might form nuclei for the polymerization and undergo a rapid, induced-fit type of structural conversion.
Another interesting observation in our study is the clearance of the ␣-syn aggregates with concomitant restoration of cellular energy metabolism. The dynamic nature of intracellular protein aggregates has also been demonstrated in transgenic mice that express mutant huntingtin with an expanded polyglutamine repeat under the control of an inducible promoter. The mice developed inclusion bodies when the mutant huntingtin expression was induced, but when the expression was subsequently turned off, the inclusion bodies disappeared (56). In our study, the disappearance of ␣-syn aggregates correlated with the recovery of cellular ATP level. This clearance might be simply due to a spontaneous disassembly reaction driven by a sudden decrease of misfolded proteins. However, our recent cell-free analysis showed that SDS-insoluble ␣-syn aggregates are chemically quite stable since they maintained the aggregated state in 6 M urea. 2 This data suggests that spontaneous dissociation of ␣-syn aggregates does not occur readily. Furthermore, the finding that the disappearance of the aggregates was not accompanied by a concomitant increase of monomeric ␣-syn argues against the simple disassembly. Our data, therefore, suggest the presence of cellular machinery that actively eliminate pre-existing aggregates. One possible mechanism might be the degradation by ubiquitin-proteasome system. Indeed, the finding that proteasome inhibitors partially blocked the disappearance of these aggregates during the rotenone wash-out phase supports the involvement of the proteasome system. Further characterization of our cell system should provide insights into how the proteasome system targets and eliminates the protein aggregates in cells and whether other cellular destruction systems, such as lysosomal pathway, are also involved.
In summary, this study provides evidence that the aggregation of ␣-syn is subject to dynamic regulation within the cells and that normal mitochondrial function is critical to prevent accumulation of the aggregates. It is, therefore, reasonable to hypothesize that altered mitochondrial activity may play a role in LB formation in PD and other related disorders. Since mitochondrial dysfunction is associated with other related neurodegenerative diseases, such as Alzheimer's disease, amyotrophic lateral sclerosis, and Huntingtin's disease (16), our data also have implications for the formation of the intraneuronal inclusions found in these diseases.