Amyloid Fibril Formation of α-Synuclein Is Accelerated by Preformed Amyloid Seeds of Other Proteins

α-Synuclein is one of the causative proteins of familial Parkinson disease, which is characterized by neuronal inclusions named Lewy bodies. Lewy bodies include not only α-synuclein but also aggregates of other proteins. This fact raises a question as to whether the formation of α-synuclein amyloid fibrils in Lewy bodies may occur via interaction with fibrils derived from different proteins. To probe this hypothesis, we investigated in vitro fibril formation of human α-synuclein in the presence of preformed fibril seeds of various different proteins. We used three proteins, Escherichia coli chaperonin GroES, hen lysozyme, and bovine insulin, all of which have been shown to form amyloid fibrils. Very surprisingly, the formation of α-synuclein amyloid fibril was accelerated markedly in the presence of preformed seeds of GroES, lysozyme, and insulin fibrils. The structural characteristics of the natively unfolded state of α-synuclein may allow binding to various protein particles, which in turn triggers the formation (extension) of α-synuclein amyloid fibrils. This finding is very important for understanding the molecular mechanism of Parkinson disease and also provides interesting implications into the mechanism of transmissible conformational diseases.

␣-Synuclein is one of the causative proteins of familial Parkinson disease, which is characterized by neuronal inclusions named Lewy bodies. Lewy bodies include not only ␣-synuclein but also aggregates of other proteins. This fact raises a question as to whether the formation of ␣-synuclein amyloid fibrils in Lewy bodies may occur via interaction with fibrils derived from different proteins. To probe this hypothesis, we investigated in vitro fibril formation of human ␣-synuclein in the presence of preformed fibril seeds of various different proteins. We used three proteins, Escherichia coli chaperonin GroES, hen lysozyme, and bovine insulin, all of which have been shown to form amyloid fibrils. Very surprisingly, the formation of ␣-synuclein amyloid fibril was accelerated markedly in the presence of preformed seeds of GroES, lysozyme, and insulin fibrils. The structural characteristics of the natively unfolded state of ␣-synuclein may allow binding to various protein particles, which in turn triggers the formation (extension) of ␣-synuclein amyloid fibrils. This finding is very important for understanding the molecular mechanism of Parkinson disease and also provides interesting implications into the mechanism of transmissible conformational diseases.
The causative factors of various neurodegenerative disorders, such as Parkinson disease (PD), 2 Alzheimer disease, and some forms of spongiform encephalopathy, are associated with the formation of abnormally aggregated proteins (1). The aggregates, known as amyloid fibrils, are formed from proteins and peptides and are characterized by their unique fibrous appearance and ␤-rich structure (2)(3)(4). To date, amyloid fibrils have been isolated from a number of pathological specimens, and it is becoming apparent that the ability to form fibrous aggregates is not merely a characteristic of a specific set of proteins but may be a common characteristic of proteins in general (5)(6)(7).
␣-Synuclein is one of the causative proteins of PD. PD forms neuronal inclusions named Lewy bodies or Lewy neurites, the major component of which is the protein, ␣-synuclein. Although its physiological role is not well known in neuronal cells, it has been observed that most of the ␣-synuclein accumulated in Lewy bodies and Lewy neurites is in fibril form in PD. Three different missense mutations in the ␣-synuclein gene, A53T (8), A30P (9), and E46K (10), have been identified in three instances of autosomal-dominantly inherited, early-onset PD. ␣-Synuclein is a soluble 14-kDa protein and exists abundantly in neuronal cells in a largely unfolded form (11). The primary structure of ␣-synuclein possesses several unique characteristics; that is, seven imperfect repeats of KTKEGV at the N-terminal and middle regions and a region rich in acidic amino acids at the C-terminal region (12).
In general, amyloid fibril formation proceeds according to two processes; the first is a nucleus formation step followed by an extension of fibrils from the nucleus. Usually, the nucleation process is very slow and rate-determining, and the nucleus-dependent extension process is much faster. The mechanism of amyloid fibril formation of ␣-synuclein in vitro has been studied extensively by Fink and Uversky (14) and Uversky (13) and other researchers (15)(16)(17). From the studies it was revealed that the hydrophobic repeat region between 65 and 95 and the negative charges at the C-terminal region play a key role in forming fibril structures. A molecular species that is relatively compact but still unfolded was found to be involved in the fibril formation. Furthermore, ␣-synuclein was shown to undergo significantly enhanced fibril formation under numerous conditions where this species was abundant, e.g. in the presence of metal ions (18), pesticides (19), moderate concentrations of trimethylamine-N-oxide (20), and organic solvents (21) and at high temperatures (22) or low pH (23). In contrast, fibril formation was slowed or inhibited under conditions favoring either more folded conformations (20,21) or the fully unfolded form (e.g. by oxidation of its methionines (24) or by stabilization of off-pathway oligomers via nitration of tyrosine (25)).
However, the actual in vivo mechanism of ␣-synuclein fibril formation in PD is still unclear because of the existence of a diverse population of proteins in the cell. Lewy bodies contain aggregates of not only ␣-synuclein protein but also various other proteins, such as Tau protein and A␤ peptide, which are also related to neurodegenerative disorders (26 -29). This fact prompted an idea that perhaps various amyloid-like fibrils formed in the cell might act as a "nucleus" and that ␣-synuclein protein may be attracted to this nucleus to then form ␣-synuclein amyloid fibrils. In support of this, cell culture experiments showed that the presence of abnormal Tau protein leads to ␣-synuclein amyloid fibril formation (30). Here, we have studied amyloid fibril formation of ␣-synuclein in the presence of other different amyloid fibril seeds in vitro. We used fibrils of Escherichia coli chaperonin GroES (31), hen lysozyme (32), and bovine insulin (33) as seeds for ␣-synuclein fibril formation. Interestingly, the formation of ␣-synuclein amyloid fibril was accelerated markedly in the presence of preformed amyloid seeds of these different proteins. However, a lag time before extension phase still remained, in contrast to experiments using ␣-synuclein as seed, indicating that there were variances in the ability to promote fibrillation between these protein fibrils. Furthermore, from fluorescence microscopy experiments using fluorescent dyes and transmission electron microscopy (TEM) experiments using gold-attached immunoglobulins, we demonstrate that the preformed seeds of different proteins were incorporated into the completed ␣-synuclein amyloid fibrils. These findings indicated that the natively unfolded structure of ␣-synuclein may allow binding to a variety of protein particles, which in turn triggers the formation (extension) of ␣-synuclein amyloid fibrils. The present results are quite important for understanding the molecular mechanism of not only PD but also other transmissible amyloidogenic diseases.

EXPERIMENTAL PROCEDURES
Expression and Purification of ␣-Synuclein-Human ␣-synuclein gene was amplified from cDNA of human brain (Cap site cDNA dT: Nippon gene) by PCR, subcloned into the NdeI and XhoI multicloning site of the expression vector pET23a(ϩ) (Novagen), and expressed in E. coli BLR(DE3) (Novagen). Cells were suspended in purification buffer (50 mM Tris-HCl, pH 7.5, containing 1 mM EDTA, 0.1 mM dithiothreitol, and 0.1 mM phenylmethylsulfonyl fluoride), disrupted using sonication, and centrifuged (14,000 rpm, 30 min). Streptomycin sulfate (final 2.5%) was added to the supernatant to remove nucleic acids. After removal of nucleic acids by centrifugation, the supernatant was heated to 90°C for 15 min and then centrifuged. In this step ␣-synuclein remained in the supernatant. The supernatant was precipitated by the addition of solid ammonium sulfate to 70% saturation, centrifuged, and dialyzed overnight and then applied onto a Resource-Q column (Amersham Biosciences) with 50 mM Tris-HCl buffer, pH 7.5, containing 0.1 mM dithiothreitol and 0.1 mM phenylmethylsulfonyl fluoride as running buffer. Samples were eluted with a linear gradient of 0 -1 M NaCl. Protein concentration of ␣-synuclein was determined by using a molar absorption coefficient of ⑀ 280 nm 0.1% ϭ 0.354 (34). Amyloid Fibril Formation and Thioflavin-T Binding Assay-␣-Synuclein (15 mg/ml) was incubated in 25 mM Tris-HCl buffer, pH 7.5, containing 1 M NaCl at 37°C without agitation, and fibril formation was monitored by thioflavin-T (ThioT, Wako) binding assay (35) using a HITACHI F-4500 fluorescence spectrophotometer. Aliquots were mixed with 25 M ThioT in PBS buffer, and fluorescence intensities were monitored at 480 nm with excitation at 450 nm. Fibrils of GroES were formed by incubation at 15 mg/ml protein concentration in 25 mM Tris-HCl buffer, pH 7.5, containing 1.6 M Gdn-HCl at 37°C for 20 days (31). Fibrils of hen lysozyme (Nacalai Tesque) were formed by incubation at 15 mg/ml protein concentration in HCl solution, pH 2.0, at 65°C for 7 days (32). Fibrils of bovine insulin (Sigma) were formed by incubation at 10 mg/ml protein concentration in 25 mM Gly-HCl buffer, pH 2.0, at 37°C for 10 days (33). Mixed amyloid fibrils of ␣-synuclein and GroES were formed by incubation of 8.0 mg/ml ␣-synuclein and 6.0 mg/ml GroES in 25 mM Tris-HCl buffer, pH 7.5, containing 0.9 M Gdn-HCl at 37°C for 15 days.
Samples of fibril seed were prepared as follows. Fibrils prepared as above were ultracentrifuged at 110,000 ϫ g for 1 h. Pellets were resuspended in a 1.5-ml plastic tube with Milli-Q water and sonicated for 30 s (0.5-s pulse; Sonics & Materials Inc.). The concentrations of seed samples (1.4 -2.5 mg/ml) were adjusted to give the same fluorescence intensities in ThioT assays before the addition to ␣-synuclein solutions (1 ml). ␣-Synuclein amyloid fibrils were added as seed in 0.3, 0.5, and 0.8% (v/v) concentrations. GroES fibrils, lysozyme fibrils, and insulin fibrils were seeded in 1, 5, and 10% (v/v) concentrations. The concentration of seed protein was determined after solubilization in 6 N NaOH using the Bio-Rad protein assay kit with each native protein as a standard.
CD Measurements-CD spectra were measured using a Jasco J-700 spectropolarimeter equipped with a constant temperature cell holder at 25°C. Water-washed pellets of amyloid fibril (0.6 mg/ml) were resuspended in 25 mM Tris-HCl, pH 7.5, and far-UV spectra were recorded using a 1-mm light path length cell. For comparison, native ␣-synuclein (0.2 mg/ml) was also measured.
Fluorescence microscopy measurements were performed using a Nikon ECLIPSE E-600 microscope equipped with Y-FL fluorescence instrument and super high pressure mercury lamp power supply HB-10103AF (Nikon). Amyloid fibrils were diluted 40-fold with distilled water and stained overnight in the presence of 25 M ThioT before measurements. The observation was done using a Plan Fluor series objective lens (Nikon) and B-2A filter block (Nikon) whose excitation and observation wavelengths were 450 -490 nm and Ͼ520 nm, respectively.
Atomic force microscopy measurements were performed using a Digital Instruments Nanoscope IV scanning microscope (model MMAFM-2) at 25°C. Measurements were performed using air tapping and fluid tapping modes. For the air tapping mode measurements, 20 ml of 40-fold diluted fibril solution was put onto freshly cleaved mica, incubated for 20 min, and then washed with 100 l of water and dried. For the fluid tapping mode measurements, fibril preparations were diluted 7.5-fold with distilled water, and 15 l of the solution was placed on the mica cell, incubated for 20 min, and then washed with 15 l of water, and finally 100 l water was reapplied.
Visualization of Amyloid Fibrils Formed by ␣-Synuclein and Various Protein Seeds-␣-Synuclein and proteins used for amyloid seeds (␣-synuclein, GroES, lysozyme) were each labeled before the experiments with Cy5 (Amersham Biosciences) and fluorescein 5-isothiocyanate (FITC) (Sigma), respectively, according to the manufacturer's protocols. The degree of Cy5 incorporation to ␣-synuclein was determined to be 0.1 mol of Cy5/mol of protein by UV absorption, and the degrees of FITC labeling for ␣-synuclein, GroES, and lysozyme were 0.8, 0.6, and 0.3 mol of FITC/mol of protein, respectively. Seed fibrils were then prepared from the FITC-labeled proteins and collected by ultracentrifugation at 110,000 ϫ g for 1 h. Pellet samples were resuspended in Milli-Q water, sonicated for 30 s, and used as seeds. Fifteen l of Cy5labeled ␣-synuclein (5 mg/ml) was placed on non-fluorescent glass slides (Matsunami Glass Industry Ltd., Japan), and 0.3-1 l of FITClabeled preformed amyloid samples (␣-synuclein, GroES, and lysozyme) were added as a seed fibril and then covered with a glass coverslip. Fibril formation of Cy5-labeled ␣-synuclein was observed on these sealed glass plates during incubation at 4 or 25°C for several days using fluorescence microscopy.
Additionally, immuno-TEM measurements using gold-conjugated antibodies were performed. First, the mixed amyloid fibrils prepared according to the protocol described above were incubated with 1000fold diluted anti-GroES antibody (from rabbit) (Stressgen Biotechnologies Corp.) and 100-fold diluted anti-␣-synuclein (116 -131) antibody (from goat) (Biogenesis, Ltd.) at 25°C overnight. After ultracentrifugation at 110,000 ϫ g for 1 h, the pellet samples were resuspended in 300 ml of PBS buffer, pH 8.2, containing two secondary antibodies, antirabbit IgG antibody (from goat) conjugated with 40-nm colloidal gold and anti-goat IgG antibody (from rabbit) conjugated with 10-nm colloi-dal gold (British BioCell International, Inc.). After incubation at 25°C for 1 h, the sample was centrifuged and washed with phosphate-buffered saline throughout and then negatively stained with 2% (w/v) uranyl acetate. The location of GroES and ␣-synuclein were visualized by attachment of 40-and 10-nm colloidal gold particles, respectively. No localization was observed using only the secondary antibodies conjugated with colloidal gold particles in control experiments under these conditions.
Small Angle X-ray Scattering Measurements-Small angle x-ray scattering experiments were performed with the optics and detector system installed at Beamline BL-40B2 of the synchrotron radiation facility SPring-8, Hyogo, Japan. The wavelength of incident x-ray beam was 1.5 Å. An imaging plate area detector R-AXIS IV (Rigaku, Japan) was used to detect the scattered x-ray signal. The distance between the sample and the detector was 1000 mm. Protein solutions of 10 -20 mg/ml were placed in a mica cell of 1-mm path length. Irradiation time of the sample to X-rays was typically 5-10 min according to the beam intensity. A concentric image of scattered x-ray radiation was digitized with 0.1-mm resolution. Signals from points equidistant from the center of the concentric image were integrated and averaged as a signal of the corresponding scattering angle. Data from scattering angles of up to 12.1°w ere collected. The radius of gyration, R g , was determined using Guinier's approximation (36), where I(Q) is the scattering intensity, and Q is the momentum of transfer (Q ϭ 4sin /, where 2 is the scattering angle and is the incident x-ray wavelength). Fitting of the above equation was performed in QR g regions from 0.8 to 1.3.

RESULTS
Amyloid Fibril Formation of ␣-Synuclein-It has been shown that ␣-synuclein forms amyloid fibrils in vitro that closely resemble those found in Lewy bodies (37,38). The addition of multivalent cations significantly accelerated the fibrillation (18). This is attributed to the preferential binding of cations that neutralize negative charges localized in the C-terminal region of ␣-synuclein, which allows formation of a compact conformation that is presumably favorable for fibrillation (23,39). Similar effects were also observed at low pH (23). In small-angle x-ray scattering experiments, we observed a gradual decrease of the R g value (corrected for variances in protein concentrations) from 39.0 to 35.4 Å upon the addition of increasing NaCl concentrations (0 -1 M) (data not shown). This value (39.0 Å) in the absence of NaCl was consistent with the previously reported value (40 Å) at pH 7.5 and 100 mM NaCl (23,40). However, no changes in CD and fluorescence spectra were observed under similar conditions, indicating that the conformation of ␣-synuclein was still unfolded (data not shown). Also, we detected an accelerated fibrillation of ␣-synuclein in the presence of NaCl (data not shown). Additionally, under conditions of constant agitation in the presence of NaCl, fibrillation was greatly accelerated (essentially completed within 40 h) (data not shown). In this study, we ultimately performed all of the fibril formation experiments of this protein in the presence of 1 M NaCl at 37°C without agitation because this condition allowed more reproducible and reliable results to be obtained. As shown in Fig. 1a, fibril formation of ␣-synuclein showed a lag time of about 370 h, after which fibril extension was observed, and plateaus in ThioT fluorescence increase were seen after another 40 h. As shown in Fig. 1b, although native ␣-synuclein existed in a largely unfolded form under physiological conditions (11, 13), a typical ␤-structure was observed in CD spectra of amyloid fibrils. The amyloid fibrils were also examined by using fluorescence microscopy ( Fig. 1c), TEM (Fig. 1d), and atomic force microscopy (Fig. 1, e and f). All these observations demonstrated that the sample was a typical amyloid fibril whose morphology was the same as that of other studies reported previously (16,38,41,42). From the TEM and atomic force microscopy measurements in the present study, the fibrils were determined to be 1000 -5000 Å in length and 170 Å in diameter, with a twisted structure resembling a left-handed spiral. The helical pitch (periodicity) of this twist was about 570 Å. Although they vary somewhat depending upon the method of measurements and the conditions of fibrils formation, these values were similar to the reported values for amyloid fibrils of human (41)(42)(43) and mouse ␣-synuclein (44) and also consistent with those of the SH3 domain of phosphatidylinositol 3Ј-kinase (45) and GroES (31).
Seeding Effects on ␣-Synuclein Amyloid Fibril Formation-In many cases it has been well established that amyloid fibril formation proceeds in a biphasic manner; that is, nucleation followed by nucleation-dependent extension. In this mechanism the rate-determining step is the nucleation reaction. If preformed seeds of the same protein are added, the seed acts as a nucleus (starting material) for fibril extension, and the extension occurs immediately with no lag time in typical amyloidogenic proteins (46,47). We first confirmed this seed-dependent amyloid fibril formation of ␣-synuclein. As shown in Fig. 2a, when seeds of preformed ␣-synuclein amyloid fibril were present, fibrillation of ␣-synuclein was accelerated greatly; the 370-h lag time observed in the absence of seeds was abolished in the presence of 0.3, 0.5, 0.8, 1% (v/v) seeds, reflecting a typical seed (nucleus)-dependent fibril formation mechanism. Although a slightly increased extension rate was seen with increasing amounts of seeds added (0.05-0.07 h Ϫ1 ), the fibril extension rate was almost the same as that of spontaneous fibril formation (0.05 h Ϫ1 ). Next, we examined whether ␣-synuclein fibril formation was accelerated by seeds of different proteins. We used three proteins, co-chaperonin GroES (31), hen-egg lysozyme (32), and bovine insulin (33) as seed proteins, all of which have been demonstrated to form amyloid fibrils (diameter of ϳ200 Å, several thousands Å long) in vitro under certain conditions. Surprisingly, all three types of seed accelerated the fibril formation of ␣-synuclein significantly, as shown in Fig. 2, b, c, and d. The lag time before fibril extension became shorter with increasing amounts of preformed seeds (1, 5, 10% v/v) for all three proteins. For instance, in experiments where preformed GroES seeds were added, the lag time was reduced from 350 h (0% seeds) to 310 h (1% seeds), 140 h (5% seeds), and 85 h (10% seeds). Although the seeding effect was different depending on the seed protein added, the lag phase still remained for all cases even in the presence of 10% seeds, indicating that those preformed seeds were less suitable for extension than seeds of ␣-synuclein. The extension rates (0.04 -0.06 h Ϫ1 ) were almost the same as that (0.05 h Ϫ1 ) of spontaneous fibril formation of ␣-synuclein itself. It should be noted that no effects were observed in the fibril formation reaction when the same amounts of native or heat-aggregated GroES and lysozyme proteins were added instead of preformed seeds (data not shown). As shown in Fig. 2, e-h, the structural characteristics of amyloid fibrils formed in the presence of various seed types were very similar to that of fibrils formed from ␣-synuclein-preformed seeds as observed by TEM, with the notable exception of fibril diameter. The diameter values of amyloid samples prepared from GroES fibril seeds (Fig. 2f), lysozyme fibril seeds (Fig. 2g), and insulin fibril seeds (Fig. 2h) were 190 (Ϯ15), 160 (Ϯ10), and 140 (Ϯ10) Å, respectively. These values were significantly different from the value of 170 (Ϯ10) Å of fibril samples prepared from ␣-synuclein fibril seeds (Fig. 2e) and spontaneous fibrillation (Fig. 1d). Interestingly, each value was almost the same as that of the protein fibril seed (diameters: GroES fibrils, 200 (Ϯ20) Å; lysozyme, 160 (Ϯ10) Å; insulin, 140 (Ϯ10) Å). This finding suggested that the different protein seeds might affect the morphology of the ␣-synuclein amyloid fibrils, especially fibril diameter; in other words, ␣-synuclein was quite adaptable during amyloid fibril formation, and some structural properties of the initiating fibril seed might be retained and propagated in the newly formed amyloid fibrils.
pass filter (Fig. 3, a, b, and c). When the same view was observed with a Cy5 filter, Cy5-labeled ␣-synuclein was observed as amyloid fibrils extending from these FITC-labeled seeds with a clear red color. Soluble Cy5-labeled ␣-synuclein was also seen in pale red as a background (Fig.  3, d, e, and f). With this filter, the FITC-labeled seed fibril was also illuminated slightly. This result suggests that ␣-synuclein could elongate in a bidirectional manner from the seeded nucleus, similar to that of A␤ (48) and islet amyloid peptide (49) but different from that of sup35 (50).
Additional immuno-TEM experiments using gold-attached immunoglobulin were also performed to further clarify our results. For this experiment soluble ␣-synuclein and GroES proteins were mixed and incubated in buffer, pH 7.5, containing 0.9 M Gdn-HCl at 37°C for several weeks to form amyloid fibrils. It was confirmed that under these conditions ␣-synuclein was able to form amyloid fibrils similar to those prepared in the absence of Gdn-HCl (data not shown). If amyloid fibril containing a mixture of ␣-synuclein and GroES proteins were formed under these conditions, antibodies against ␣-synuclein and GroES would bind to the same fibril and be detected by the presence of conjugated 10-and 40-nm colloidal gold particles, respectively. As shown in Fig. 4, TEM measurements indicated that both 10-and 40-nm gold particles were observed attached to a single amyloid fibril (Fig. 4b). In the absence of specific primary antibody, no co-localization of gold particles with fibrils was observed (not shown). These results, taken together with the results from the fluorescence microscopy experiments in Fig. 3, clearly indicated that different proteins (seed) could be incorporated into a single amyloid fibril of ␣-synuclein. In other words, ␣-synuclein is able to form amyloid fibrils from a seed made from a different protein.

DISCUSSION
␣-Synuclein has been identified as a causative protein of Parkinson disease (12,51). It is conceivable that its participation in the formation of amyloid fibrils is involved in the onset of the disease. It has been shown that not only ␣-synuclein (Fig. 1), but also several proteins unrelated to amyloid diseases, may form amyloid-like fibrils under certain conditions in vitro (5). Very recently, it was reported that the structures of amyloid fibrils are very similar, consisting of parallel ␤-strands that form a pair of ␤-sheets arranged in a cross-␤ structure (52)(53)(54). This ability to form this type of structure may be a generic feature of polypeptide chains (55), although the specific amino acid sequence of the chain affects both the propensity to form fibrils and the way a given molecule is arranged within the fibrils. This implication together with the fact that an abnormal Tau protein (30) and tubulin (56) lead to ␣-synuclein amy-loid fibril formation in vivo prompted us to consider the possibility that a fibril formed from different protein may act as a seed of ␣-synuclein fibrillogenesis.
To examine this hypothesis, we studied the in vitro amyloid fibril formation of ␣-synuclein in the presence of fibrils formed from various other proteins, some implicated in amyloid disease (lysozyme and insulin) and another that is not (GroES). Interestingly, as shown in Fig. 2, ␣-synuclein amyloid fibril formation was accelerated in a dose-dependent manner by the addition of preformed protein fibril seeds, although the effect was slightly different for each protein seeded. The fibril extension rates were almost the same in all cases. However, a critical difference in the fibril formation profile in experiments using preformed ␣-synuclein seeds (Fig. 2a) compared with profiles of analogous experiments using GroES fibrils as seeds, for example, was that the lag phase, although shortened, still remained in the latter experiments. The results of Fig. 3 demonstrated that the ␣-synuclein amyloid fibril extended from the preformed seeds of heterogeneous proteins (GroES and lysozyme). Also, the results of Fig. 4 showed that ␣-synuclein could form mixed amyloid fibrils with heterogeneous protein GroES. It should be noted here that there are no homologous sequence regions that may be observed in comparisons of the primary structures of ␣-synuclein, GroES, lysozyme, and insulin proteins. It was reported that A␤ peptide fibril formation was accelerated by seeding of heterologous peptide (islet amyloid polypeptide); however, the peptide shares sequence similarity with the region of the amyloid core of A␤ peptide fibril (57). Also, polyQ amyloidogenic protein Sup35 was shown to promote fibril formation rapidly only by the addition of polyQ-containing proteins but not non-polyQ amyloidogenic proteins (58). The findings shown in Figs. 2, 3, and 4 clearly indicated that the amyloid fibril formation of ␣-synuclein could extend from preformed amyloid seeds of completely diverse proteins, in sharp contrast to the results for A␤ peptide and Sup35. With regard to the formation of mixed fibrils in Fig. 4, the fact that GroES fibrillogenesis was faster than ␣-synuclein fibrillogenesis in the presence of 0.9 M Gdn-HCl, taken together with the ability of GroES fibrils to promote ␣-synuclein fibrillation (Fig. 2b), strongly suggests that the initially formed GroES fibrils acted as seeds for ␣-synuclein. However, the specific fibril formation mechanism may not follow the typical seed-dependent extension mechanism, from evidence shown in Fig. 2a. If the added seeds were the exact same structure as that of ␣-synuclein amyloid fibril, the fibril extension should start with no lag time (59). Also, native or amorphous aggregates of the various proteins we used did not accelerate the formation of ␣-synuclein amyloid fibrils under the conditions we used. From these findings we conclude that ␣-synuclein may be capable of adapting and binding to amyloid seeds of other different proteins but not to completely structured or highly disordered protein aggregates. The first conformational adaptation to other different protein amyloid seeds presumably occurs relatively slowly, but once ␣-synuclein polypeptide binds to the surface of the seeds the binding of the next ␣-synuclein polypeptide (i.e. extension reaction) is faster because the interaction is converted to one between identical polypeptides. This unusual phenomenon of cross-seeding may be due in part to an intrinsic characteristic of ␣-synuclein, namely, its "native unfolded" structure (11). The expanded and flexible characteristics of ␣-synuclein might be conducive to mixed fibril formation. Also, our recent finding that fibril extension from preformed seeds is faster in the presence of an unfolded/expanded molecular species rather than a compact species (31) supports the results of the present experiments.
As shown by the TEM measurements in Fig. 2, morphologies of the ␣-synuclein amyloid fibrils eventually formed from the different protein seeds were diverse with regard to the fibril diameter value; the diameter of the resultant fibrils was the same as that of the initially seeded-protein fibrils and not that of orthodox ␣-synuclein fibrils. Because the rates of extension were all similar to that for homologous ␣-synuclein fibril growth (Fig. 2, b-d), one would expect similar dimensions in the diameter of the finished fibrils. This was found not to be the case. We were not able to explain this discrepancy decisively at present. However, as shown in Fig. 2, b-d, the interval after which initiation of fibril extension was observed (the lag time) was quite different depending on the seeds added, indicating that the initial adaptation time to the "template" face may be different depending on the seeds. After the first adaptation, the next binding step of ␣-synuclein, i.e. extension, would proceed according to a common mechanism in all cases. To propose a plausible explanation, if the rate of this extension reaction of ␣-synuclein was dictated by a specific interaction, such as hydrophobic or charged amino acid residues in the repetitive sequences (12), the extension rate would not be influenced by the differences in the overall shape of the interface. TEM-observed structures of the mixed fibrils from ␣-synuclein and GroES proteins shown in Fig. 4 were also quite different from those in Fig. 2, although in this case the conditions we applied during the mixed fibril formation were different. Judging solely from our present experimental data, we were not able to conclusively determine if the final fibril structure was affected by the seeded fibril and/or faithfully retained the properties of the initial seed fibril. However, it is interesting to note that the fibril morphology of prion protein may be affected by the conformation of prion fibril that was initially seeded, as reported in other studies (60). Also recently, the structural properties of amyloid fibril were suggested to be versatile during A␤ peptide fibril formation in heterologous peptide seeding experiments (57) and also in repeated seeding and extension experiments of ␤2-microglobulin (61) as well as extension of this protein under pressure (62). To clarify this point in the present study, further studies regarding structural details are necessary.
Based on the results in the present study, a possible mechanism of ␣-synuclein amyloid fibril formation may be summarized as shown in Fig. 5. Spontaneous fibril formation (Fig. 5a) is initiated by compaction of ␣-synuclein molecules whose negative charges at C-terminal region are neutralized by the presence of NaCl. This compact molecule is still unfolded. Formation of amyloid fibril nuclei is very slow, but once formed, fibril extension becomes fast. For homo-nuclei-seeded fibril formation (Fig. 5b), because the rate-determining step of the nucleus formation is bypassed by the presence of seed, fast fibril formation (extension) of ␣-synuclein occurs. In the presence of hetero-nuclei (Fig.  5c), the initial binding step of ␣-synuclein polypeptide to the surface of the hetero-nuclei is relatively slow, most likely due to a requirement for adaptation to a suitable conformation for the nuclei. However, once ␣-synuclein polypeptide binds to the surface of the hetero-nuclei and acts as an adaptor to the initial hetero-nuclei, fast fibril formation (extension) occurs as well as the case of Fig. 5, a and b. Therefore, the ␣-synuclein protein might be characterized to have a high adaptability in its conformation to be able to bind any other fibril seeds. Such characteristics may come from the intrinsic disordered nature of this protein under physiological conditions (11,13). It was reported for many proteins that an unfolded conformations was favorable for fibril formation (63). Furthermore, in our previous study characterizing the fibril formation mechanism of GroES protein in terms of its molecular compactness (31), it was also revealed that the fibril extension reaction preferred a more expanded molecular conformation. These facts indicate that unfolded and expanded molecular characteristics of polypeptides are very important for amyloid fibril formation. Finally, it is very interesting to consider the generality of the crossseeding effects observed in the present study. In this study, preformed GroES, lysozyme, or insulin amyloid fibril seeds were added into ␣-synuclein protein solution, and as a result, an acceleration of ␣-synuclein amyloid fibril formation was observed. Accordingly, one may consider the reverse scenario, the effects of the addition of FIGURE 5. Schematic representation of ␣-synuclein amyloid fibril formation mechanism in the presence or absence of seeds. a, amyloid fibril formation in the absence of seeds, i.e. spontaneous fibril formation. Expanded and unfolded monomer ␣-synuclein assumes compact but still unfolded species in the presence of 1 M NaCl. For long term incubation, amyloid nuclei of ␣-synuclein are formed. This nuclei formation is the rate-determining step. Once nuclei of ␣-synuclein are formed, fibril extension is fast, and mature fibrils are formed. Therefore, ␣-synuclein can form amyloid fibrils rapidly without lag times. b, amyloid fibril formation in the presence of preformed ␣-synuclein fibril seeds, i.e. homo-nuclei seeded fibril formation. Because the rate-determining step of the nuclei formation can be omitted, fast fibril extension of ␣-synuclein occurs. c, amyloid fibril formation in the presence of preformed different protein amyloid fibril seeds, i.e. hetero-nuclei-seeded fibril formation. The first binding step of ␣-synuclein polypeptide to the surface of the hetero-nuclei is relatively slow due to adaptation to a suitable conformation for the nuclei. But this step is faster apparently than spontaneous nucleus formation. Once ␣-synuclein polypeptide binds to the surface of the hetero-nuclei, which converts the hetero-nuclei to pseudohomo-nuclei, the fast fibril extension occurs. See "Discussion" for more detail.
␣-synuclein fibril seeds on the fibrillogenesis of other proteins. Because the fibril formation conditions of those proteins were different from that of ␣-synuclein, we could not examine every case in detail. However, when preformed ␣-synuclein seeds (final, 5% v/v (70 g)) were added to GroES protein solution (8 mg/ml) at pH 7.5 in the presence of 1.6 M Gdn-HCl, the lag time in the fibril formation was reduced from 90 to 70 h (data not shown). This finding strongly suggested that two-way cross-type seeding of various protein fibrils is possible. This result was also supported by in vivo experiments of the fibril formation of apoA-II protein using mice (64). ApoA-II is the second most abundant apolipoprotein in plasma high density lipoprotein after apoA-I and is a causative protein of hereditary renal amyloidosis (65). In mice, apoA-II protein is a precursor of amyloid fibrils in senile amyloidosis (66). When various preformed amyloid fibrils including GroES, lysozyme, and ␣-synuclein were injected into mice with amyloidogenic apoAII gene, depositions of apoAII amyloid fibrils were detected in the tissues of the mice (64). Furthermore, very recently it was also reported that fibril extension of amyloid protein A was accelerated upon the addition of naturally occurring fibrils of various different proteins in vivo (67). These results together with in the present in vitro experiments strongly demonstrate that preformed amyloid seeds of other proteins are able to induce amyloid fibril formation. If this conclusion might be applied to other amyloidogenic proteins, a very important implication is suggested in the mechanism of amyloidogenic disease. Although we do not know the precise structure of the amyloid fibrils prepared from the heterologous seeds, critical differences were observed in the fibril diameters between seeded and non-seeded fibrils in the present study. From a view point of structural changes influenced by fibril additives, it is interesting to note the report that changes in the specific structure of fibrils, caused either by mutation (68) or other extrinsic factors (69), are implicated in the onset of prion-like behavior of Sup35. Such changes in the fibril structure might be caused by fibril seeds of the same protein (in the case of Sup35) or different proteins (in the case of the present experiments). Although more detailed studies are necessary to clarify this intriguing issue, the results of our experiments contain implications that may be relevant to understanding the process by which various amyloid diseases adapt to different species and cellular environments to propagate themselves (70). In addition, our results suggest that such molecular mechanisms of amyloid fibril-mediated propagation may involve many diverse participants that are either originally present in, or subsequently incorporated into, the cellular environment.