Tubulin Seeds (cid:1) -Synuclein Fibril Formation*

Increasing evidence suggests that (cid:1) -synuclein is a common pathogenic molecule in several neurodegenerative diseases, particularly in Parkinson’s disease. To understand (cid:1) -synuclein pathology, we investigated molecules that interact with (cid:1) -synuclein in human and rat brains and identified tubulin as an (cid:1) -synuclein binding-/associated protein. Tubulin co-localized with (cid:1) -synuc-lein in Lewy bodies and other (cid:1) -synuclein-positive pathological structures. Tubulin initiated and promoted (cid:1) -synuclein fibril formation under physiological conditions in vitro . These findings suggest that an interaction between tubulin and (cid:1) -synuclein might accelerate (cid:1) -synuclein aggregation in diseased brains, leading to the formation of Lewy bodies. The non- (cid:1) -amyloid (A (cid:1) ) 1 component of Alzheimer’s disease or NAC, originally detected in an amyloid-enriched fraction, to be a fragment of its precursor, NACP, cloning sion was induced by isopropyl-1-thio- (cid:1) - D -galactopyranoside. The recombinant (cid:2) -synuclein was purified using two-dimensional HPLC methods as described previously (15). The identity was confirmed and the purity was shown to be (cid:1) 98% by SDS-PAGE, immunoblotting, MALDI/TOF-mass spectrometry, and microsequencing analyses (15). and tubulin was purified from the microtubule fraction by phosphocellulose (Whatman P11) column chromatography (18). The purity of the tubulin was determined to be (cid:1) 98% by SDS-PAGE and immunoblot analyses. Reassembly of microtubules was performed at 37 °C for 20 min followed by 10 min of incubation with 10 (cid:3) M taxol for microtubule stabilization. m M pH 1 m M EGTA, 1% Nonidet P-40, and 0.5% sodium deoxycholate). PQE3, EQV1, and anti- (cid:2) -tubulin (mouse monoclonal clone DM 1A, Sigma) were added and incubated at 4 °C for 2 h. Antigen-antibody complexes were collected with protein A-Sepharose (Sigma). The beads were washed five times by centrifugation with 1% Triton X-100 in PBS and then boiled for 5 min in SDS-sample buffer. Following SDS-PAGE and immunoblotting, the signals were detected with chemilumines- cence reagents (PerkinElmer Life Sciences) and Hyper-ECL film (Am-ersham Biosciences) as described (9, 10). Anti- (cid:1) -tubulin (clone TUB 2.1, Sigma), TAU-2 (NeoMarkers), and anti-MAP5 (identical to MAP1b) (Chemicon International) antibodies were also employed. It was con- firmed that EQV1 and PQE3 did not cross react with tubulin,

The non-␤-amyloid (A␤) 1 component of Alzheimer's disease amyloid, or NAC, originally detected in an amyloid-enriched fraction, was shown to be a fragment of its precursor, NACP, by cloning of the full-length cDNA (1). Later, NACP turned out to be a human homologue of Torpedo synuclein (2). Therefore, it is also referred to as human ␣-synuclein (3). ␣-Synuclein is abundant in presynaptic terminals of neurons (4). Recently, two missense mutations in the ␣-synuclein gene (5) were discovered in certain pedigrees with familial Parkinson's disease and were shown to segregate with the illness (6,7). Shortly thereafter, ␣-synuclein was identified as the major filamentous component of Lewy bodies (LBs) in Parkinson's disease (8,9) and of cytoplasmic inclusions in multiple system atrophy (MSA) (10 -12).
Thus, ␣-synuclein appears to be a common pathogenic molecule in these diseases.
Although the physiological role of ␣-synuclein is unknown, ␣-synuclein has the property of forming fibrils by itself in vitro, and mutations of ␣-synuclein accelerate the fibril formation (13,14). However, the vast majority of cases of neurodegenerative diseases associated with LBs or with ␣-synuclein pathology, such as Parkinson's disease, dementia with Lewy bodies (DLB), MSA, and the LB variant of Alzheimer's disease, are sporadic, where wild-type ␣-synuclein has shown to be abnormally accumulated as fibrillar structures. It is therefore likely that at some stage(s) in the fibril formation of ␣-synuclein, either the nucleation and/or the elongation steps should be somehow accelerated in diseased brains, or alternatively, some degradation process(es) of abnormal structures of ␣-synuclein might be defective in those patients (15).
With respect to the amyloidogenesis of Alzheimer's disease, it was demonstrated in vitro that a seed of NAC can accelerate A␤ fibril formation, and conversely, a seed of A␤ can promote NAC fibril formation (16). Similarly, heterogeneous molecules could also be involved in the formation of ␣-synuclein fibrils, leading to pathological structures of ␣-synuclein such as LBs.
In this study, we performed a biochemical investigation of molecules that interact with ␣-synuclein in the human brain, and we identified tubulin as one of the ␣-synuclein binding/ associated proteins. This interaction was confirmed by co-immunoprecipitation experiments. Further, ␣-synuclein was co-purified with microtubules. Double labeling immunofluorescence revealed that tubulin co-localized with ␣-synuclein-positive pathological structures such as LBs, Lewy-related neurites in Parkinson's disease and DLB, and glial cytoplasmic inclusions (GCIs) in MSA. In vitro studies demonstrated that ␣-synuclein fibril formation was initiated and accelerated in physiological media containing a small amount of tubulin.
To our knowledge, this is the first demonstration of the initiation and promotion of ␣-synuclein fibrillogenesis by an intrinsic neural protein under physiological conditions. We speculate that the interaction between tubulin and ␣-synuclein may be linked to ␣-synuclein-associated neurodegenerative diseases.
Preparation of Human Brain Extract-A human brain neocortex (75 g) was homogenized in 9 volumes of 20 mM phosphate buffer, pH 8.0, containing 150 mM NaCl, 1 mM EDTA, 1 g ml Ϫ1 pepstatin, and 10 g ml Ϫ1 leupeptin. Insoluble materials were precipitated by centrifugation at 140,000 ϫ g at 4°C for 60 min. Soluble proteins from three cycles of homogenization and centrifugation were pooled.
Isolation and Characterization of ␣-Synuclein-binding Proteins from a Human Brain Extract-An affinity column was prepared by the conjugation of cyanogen bromide-activated Sepharose-4B resin (Pharmacia) and recombinant ␣-synuclein (27 mg) according to the manufacturer's instructions. Soluble proteins from the human brain were loaded onto the column, and the column was then washed with a large volume of loading buffer, 50 mM PBS, pH 7.0, at 4°C. The adsorbed proteins were eluted with 50 mM sodium citrate, pH 3.0, containing 0.5 M NaCl. An identical control experiment was performed using a glycine-Sepharose-4B column. ␣-Synuclein-binding proteins were purified from the adsorbed fraction obtained from the ␣-synuclein affinity column by standard gel-cut methods followed by a reversed phase HPLC and characterized by immunoblotting, mass spectrometry, and microsequencing analyses.
Isolation of Microtubules and Tubulin from Rat Brain-Microtubules were isolated from rat brains by two cycles of a temperature-dependent assembly and disassembly method (17) in PIPES buffer (0.1 M PIPES, pH 6.8, 1 mM EGTA, 1 mM MgCl 2 ) containing 1 mM GTP, 1 g ml Ϫ1 pepstatin, and 10 g ml Ϫ1 leupeptin, and tubulin was purified from the microtubule fraction by phosphocellulose (Whatman P11) column chromatography (18). The purity of the tubulin was determined to be Ͼ98% by SDS-PAGE and immunoblot analyses. Reassembly of microtubules was performed at 37°C for 20 min followed by 10 min of incubation with 10 M taxol for microtubule stabilization.
In Vitro Fibrillization-␣-Synuclein (100 -700 M) in 25 l of PBS, pH7.4, was incubated with a fixed amount of tubulin (1 M) at 37°C without agitation for 0 -15 days with or without 1 M each MgCl 2 and GTP. Control experiments were performed using either incubation with BSA or incubation in the absence of tubulin or ␣-synuclein. The resulting fibrils were monitored by light scattering at OD 400 nm (19,20) using 1 l of the reaction mixture with a Gene Spec III spectrophotometer (Hitachi) and examined by immunoelectron microscopy (immuno-EM).

RESULTS
Tubulin Is a Binding/Associated Protein of ␣-Synuclein-To detect molecules that interact with ␣-synuclein in the human brain, we subjected human brain proteins to ␣-synuclein affinity column chromatography. Employing a recombinant ␣-synuclein column, several proteins were isolated as possible ␣-synuclein-binding/associated proteins (Fig. 1a). After purification of the proteins bound to the ␣-synuclein column by a reversed phase HPLC, several proteins were determined from their amino acid sequences. One of them with an apparent molecular weight of ϳ50 kDa (Fig. 1a, arrowhead) was sequenced as MRE-CISIHVG for the N terminus, and VGINYQPPTVVPGGDLAK for a peptide from the Achromobacter lyticus protease digest. In a data base search (SWISS-PROT), both sequences were found to belong to human ␣-tubulin. To confirm the interaction between ␣-synuclein and ␣-tubulin, immunoprecipitation was carried out, and ␣-tubulin was co-precipitated with endogenous ␣-synuclein in the cytoplasmic fraction of the rat brain (Fig. 1b) and vice versa (Fig. 1c). ␤-Tubulin was also co-precipitated using anti-␣-synuclein antibodies (not shown), indicating the interaction of ␣-synuclein with ␣-␤ tubulin heterodimers.
␣and ␤-Synucleins Are Associated with Microtubules-To prove the binding or association of ␣-synuclein to microtubules, we purified microtubules from rat brains by two cycles of polymerization and depolymerization (17,18). First, the presence of ␣-tubulin, ␤-tubulin, tau, and MAP5 (MAP1b) was confirmed in the purified microtubules by immunoblot analyses (Fig. 2a). ␣-Synuclein was also detected in the same sample with three kinds of anti-␣-synuclein antibodies (MDV2, EQV1, and PQE3) that interact with epitopes covering the entire ␣-synuclein molecule (Fig. 2, b and c). Doublet-like bands with molecular mass of about 19 kDa were detected with MDV2 (anti-␣-and ␤-synuclein), and a band corresponding to the upper one was After purification of the upper band (arrowhead), amino acid sequencing revealed the protein to be ␣-tubulin (see text). b, ␣-tubulin was co-immunoprecipitated (IP) from the cytosolic fraction of rat brain with anti-␣-synuclein antibodies PQE3 and EQV1. The immunoprecipitated proteins were separated by 12% SDS-PAGE, analyzed by immunoblot (IB), and detected with anti-␣-tubulin antibody (1:5,000). c, ␣-synuclein was co-immunoprecipitated with anti-␣-tubulin antibody and detected with ␣-synuclein-specific antibodies PQE3 (1:10,000) and EQV1 (1:10,000). detected with REE1 (anti-␤-synuclein) antibodies (Fig. 2c), indicating that both ␣and ␤-synucleins are present in the purified microtubule fraction. These two bands disappeared when these antibodies were preadsorbed with each of the respective peptide antigens, showing the specificity (Fig. 2c).
The association of ␣-synuclein with microtubules was then examined by EM. Microtubules with a diameter of 24 -26 nm were decorated by gold particles in immuno-EM using either anti-␣-tubulin (Fig. 3a), or anti-␤-tubulin (not shown) antibody. Tau is a well-characterized microtubule-associated protein (MAP). Therefore, TAU-2 was used as a positive control to show its interaction with purified microtubules (Fig. 3b). In contrast to this, anti-A␤ was used as a negative control for this experiment (Fig. 3f). All of the anti-␣-synuclein antibodies (MDV2 (Fig. 3c), EQV1 (not shown), PQE3 (Fig. 3d), and anti-␤synuclein antibody, REE1 (Fig. 3e)) showed positive immunoreactions with purified microtubules. These tubules were not labeled when the primary antibodies for ␣and ␤-synucleins were preadsorbed with the respective antigen (not shown). These results suggest that both ␣and ␤-synucleins are associated with purified microtubules.
Tubulin Stimulates ␣-Synuclein Fibril Formation-To prove the involvement of tubulin in the fibril formation of ␣-synuclein, solutions with ␣-synuclein concentrations ranging from 0 to 700 M were incubated in the presence or absence of 1 M tubulin in PBS at 37°C. Purified tubulin was analyzed by SDS-PAGE (silver stain) and immunoblotting and was confirmed to be free from microtubule-associated proteins, such as tau or MAP5 (MAP1b) (not shown). It is possible that oligomers of tubulin are more effective in acting as seeds for ␣-synuclein fibrillogenesis. To test this potential, microtubule assembling cofactors, such as MgCl 2 and GTP, were added in a separate reaction at minimal concentrations (the commonly used concentrations times 10 Ϫ3 ), and the fibril formation was found to be accelerated by these cofactors (Fig. 4, a and b). ␣-Synuclein was assembled into fibrillar structures only in the presence of 1 M tubulin as shown by light scattering (Fig. 4, a-c) and by EM (Fig. 4, d-f). Over the course of several hours filaments 10 nm in diameter and about 1 m in length were formed (Fig.  4d). ␣-Synuclein fibrils thus produced were examined by immuno-EM with anti-␣-synuclein antibodies MDV2 (not shown), EQV1 (Fig. 4e), and PQE3 (not shown), which were previously used to study LBs in Parkinson's disease (9) and GCIs in MSA (10). These filaments were also decorated with anti-␣-tubulin (Fig. 4f) and anti-␤-tubulin (not shown), but not with negative controls, such as anti-␤-synuclein REE1 (not shown) or anti-A␤ (not shown). The dimensions and morphology of these structures are similar to those of ␣-synuclein filaments in the brain sections of Parkinson's disease (9) (Fig. 4g). ␣-Synuclein filaments were produced only in the presence of 1 M tubulin (Fig.  4, a-d) not in solutions containing ␣-synuclein alone ( Fig. 4c; C1) or in other controls ( Fig. 4c; C2, C3, and C4). It is notable that the additives alone were not able to stimulate production of ␣-synuclein fibrils (not shown) but enhanced the fibril production in the presence of tubulin. The morphology of the fibrils produced in the presence of tubulin with additives was identical (not shown) to the morphology of fibrils produced without additives (Fig. 4, d-f). These results indicate that tubulin is able to induce and promote ␣-synuclein fibril formation under physiological conditions in vitro.
Tubulin Co-localizes with Pathological Structures of ␣-Synuclein-We subsequently investigated the in vivo co-localization of tubulin and ␣-synuclein in LBs in the dorsal motor nucleus of the vagus nerve of a patient with Parkinson's disease and cervical sympathetic ganglia of that with DLB, as well as in GCIs of that with MSA using double labeling immunofluorescence methods. The rhodamine-labeled anti-␣-synuclein antibody (PQE3) decorated LBs, pale bodies, Lewy-related neurites, and GCIs intensely (Fig. 5, a, d, and g). FITC-tagged anti-␣-tubulin stained these structures as well (Fig. 5, c, f, and i). These signals overlapped completely or partially when merged (Fig. 5, b, e, and h). ␤-Tubulin was also detected in these pathological structures by a similar method (not shown).

DISCUSSION
The discovery of two missense mutations in the ␣-synuclein gene in certain autosomal dominant familial Parkinson's disease pedigrees (6, 7) suggested a pivotal role of ␣-synuclein in the onset and progression of Parkinson's disease. This has been substantiated by the following findings. (i) Anti-␣-synuclein antibodies detect the major filamentous component of LBs, Lewy-related neurites, and pale bodies (a possible early phase of LB-formation) (9) in Parkinson's disease and DLB (8,9) and LB variant of Alzheimer's disease (23,24). (ii) Insoluble ␣-synuclein filaments are recovered from DLB brains (25). (iii) Recombinant ␣-synuclein forms LB-like filaments (26), and familial Parkinson's disease mutations accelerate the fibril formation (13,14). Because familial Parkinson's disease patients are rare, unknown epigenetic factors may induce fibril formation of wild-type ␣-synuclein, leading to pathological structures such as LBs. In this study, we searched for potential intrinsic factors that may initiate and/or promote fibril formation of ␣-synuclein using an ␣-synuclein column, and we found that tubulin was one of the ␣-synuclein-binding/associated proteins. Co-localization of tubulin with ␣-synuclein in LBs and other pathological structures suggested that tubulin might be such a factor.

FIG. 4. A small amount of tubulin can initiate and promote fibril formation of ␣-synuclein under physiological conditions in vitro.
Solutions of ␣-synuclein (0 -700 M) in 25 l of PBS were incubated at 37°C with a fixed concentration (1 M) of tubulin. In a separate reaction, 1 M each MgCl 2 and GTP were used as cofactors to induce formation of oligomers of tubulin (see text). Control experiments were performed by incubating either ␣-synuclein or tubulin alone, with or without BSA (300 or 1 M). The ordinate represents the relative amount of ␣-synuclein aggregates shown as A 400 nm , and the abscissa in a represents time, and in b and c the concentration of ␣-synuclein. The value represents the average of three determinations. a, the amount of ␣-synuclein aggregates was measured by quasi-elastic light scattering methods using 1 l of each reaction mixture at 400 nm with a spectrophotometer (Hitachi, Gene Spec III). ␣-Synuclein and tubulin at a molar ratio of 300:1 in the presence or absence of cofactors, including four controls (C1-C4), were incubated. The light scattering was monitored at various times (0, 6,12,24,48 nant ␣-synuclein and human brain proteins, ␣-tubulin was identified as an ␣-synuclein-binding/associated protein. Both the ␣ and ␤ subunits of tubulin were equally co-immunoprecipitated with endogenous ␣-synuclein in the cytoplasmic fraction of rat brain, which is consistent with a notion that ␣-tubulin forms a heterodimer with ␤-tubulin. The interaction between ␣-synuclein and tubulin was confirmed by copurification of both ␣and ␤-synucleins with microtubules. Immuno-EM studies clearly showed the association of both ␣and ␤-synucleins with purified microtubules as has also been shown for other microtubule-associated proteins such as tau. These results suggest that tubulin may be a physiological binding partner of ␣-synuclein. However, until the binding affinities are known, it cannot be certain that this interaction between tubulin and ␣-synuclein is physiologically meaningful. Further studies will be needed to know the physiological role of ␣-synuclein in association with tubulin.
Because the brain is a rich source of tubulin, we speculate that the interaction between ␣-synuclein and free tubulin may occur in vivo and that under pathological conditions tubulin might initiate and promote the fibril formation of ␣-synuclein, leading to pathological structures such as LBs. To test this hypothesis we performed in vitro studies, which showed that ␣-synuclein fibril formation was in fact initiated and promoted by tubulin. Most importantly, these ␣-synuclein fibrils were produced only in the presence of a small amount of tubulin under physiological conditions. Immuno-EM studies revealed that the dimensions and morphology of these fibrils made in vitro closely resembled those of the ␣-synuclein filaments observed in autopsied brain sections of Parkinson's disease and were obviously different from those of microtubules. These results indicate that tubulin is capable of seeding the fibril formation of ␣-synuclein.
We then investigated the in vivo co-localization of tubulin and ␣-synuclein in autopsied brains using double labeling immunofluorescence, and we showed that both ␣-synuclein and ␣-␤ tubulin epitopes are co-localized in LBs, pale bodies, Lewy-related neurites, and GCIs in the central and peripheral nervous systems. These results indicate that tubulin co-localizes with pathological structures of ␣-synuclein in Parkinson's disease, DLB, and MSA.
Tau is a well-characterized MAP, and abnormal aggregates of tau, or tau pathology, are thought to be an early event in the development of neurofibrillary lesions (27,28). The C terminus of ␣-synuclein is reported to be reactive with the microtubule binding region of tau (29) and MAP5 (MAP1b) (22). In our in vitro experiments (Fig. 4) purified tubulin produced ␣-synuclein fibrils in the absence of tau and MAP5 (MAP1b). More importantly, tau does not co-localize with ␣-synuclein in LBs in general (30), but tubulin does as shown in Fig. 5, suggesting that the interaction between ␣-synuclein and tubulin may not be mediated by tau.
The precise mechanism of neurodegeneration in Parkinson's disease brain is unknown. Accumulating evidence suggests that the aggregation of ␣-synuclein may play a critical role in the pathogenesis of Parkinson's disease (31), although the mechanisms by which ␣-synuclein is preferentially aggregated in the Parkinson's disease brain remained elusive. As shown here, tubulin is apparently able to initiate the polymerization of ␣-synuclein, resulting in the formation of ␣-synuclein fibrils, which may eventually lead to pathological structures such as LBs in diseased brains. Thus, it is possible that some epigenetic elements (e.g. drugs, chemicals, additives in food, or environmental toxins) may affect the assembly/disassembly equilibrium of microtubules. Abnormally increased free tubulin thus produced may trigger ␣-synuclein fibril formation. If so, those microtubule-disrupting elements can be risk factors for ␣-synuclein-associated degenerative diseases such as Parkinson's disease, DLB, MSA, and LB variant of Alzheimer's disease. Further investigations of the role of tubulin in the aggregation of ␣-synuclein may clarify the mechanisms of neurodegeneration in ␣-synuclein-related diseases.
FIG. 5. ␣-Tubulin co-localizes with ␣-synuclein in cytoplasmic inclusions. A double labeling immunofluorescence study of brain tissues demonstrated co-localization of ␣-tubulin and ␣-synuclein (PQE3) epitopes. a, d, and g, PQE3 is labeled with rhodamine and shown in red. c, f, and i, ␣-tubulin is detected by the FITC-tagged secondary antibody and shown in green. b, e, and h, when the two signals are merged, the overlap is indicated by yellow signals. Dual immunolabeling was demonstrated in Lewy bodies (arrows), pale bodies (double arrows), Lewyrelated neurites (arrowheads) in the dorsal motor nucleus of the vagus nerve of a patient with Parkinson's disease (a-c), and in the cervical sympathetic ganglia of that with dementia with Lewy bodies (d-f), as well as in glial cytoplasmic inclusions (arrows) of that with multiple system atrophy (g-i). Bar represents 10 m.