p25 (cid:1) Stimulates (cid:1) -Synuclein Aggregation and Is Co-localized with Aggregated (cid:1) -Synuclein in (cid:1) -Synucleinopathies*

Aggregation of the nerve cell protein (cid:1) -synuclein is a characteristic of the common neurodegenerative (cid:1) -synucleinopathies like Parkinson’s disease and Lewy body dementia, and it plays a direct pathogenic role as demonstrated by early onset diseases caused by missense mutations and multiplication of the (cid:1) -synuclein gene. We investigated the existence of (cid:1) -synuclein pro-aggregatory brain proteins whose dysregulation may contribute to disease progression, and we identified the brain-specific p25 (cid:1) as a candidate that preferentially binds to (cid:1) -synuclein in its aggregated state. Function-ally, determined by densitomet- of on a of the

The group of ␣-synucleinopathies is dominated by the frequent neurodegenerative disorders Parkinson's disease (PD), 1 Lewy body dementia (LBD), a Lewy body variant of Alzheimer's disease, and multiple system atrophy (MSA) (1)(2)(3)(4). Their unifying hallmark is the development of aggregates of the 140-amino acid ␣-synuclein (AS) protein, which is deposited in intracellular inclusions. The inclusions comprise the Lewy body-type of inclusions in the neuronal cell body, the Lewy neurites in axons, and the oligodendrocytic glial cytoplasmic inclusions in MSA. AS can play an active role in the degenerative processes as evidenced by studies of families with autosomal dominant early onset PD and LBD caused by missense mutations in the AS gene (5)(6)(7) and the overexpression of the wild type protein caused by gene multiplications of the AS locus (8 -10). However, the role of AS in the sporadic diseases is less clear. The frequency of AS aggregation in human diseases has given rise to extensive studies of the process in vitro. These studies have revealed that this aggregation represents a nucleation-dependent process (11). The transition from monomeric AS to filamentous aggregates is characterized by a lag phase during which a build-up of soluble nucleation-competent oligomeric AS species takes places. The rapid filament growth does not occur until these structures have reached a critical concentration. This process is stimulated by several factors like the pathogenic missense mutations A30P and A53T (12)(13)(14)(15), increased concentration of AS (16), decreased pH (16), elevated temperature (16), proteolytic truncations of the acidic C terminus (17,18), molecular crowding (19), phosphorylation of Ser-129 (20), and oxidative modifications like nitrations (21) and dopamine conjugation (22). The reverse inhibitory effects on AS aggregation have also been demonstrated, because the aggregation-incompetent synucleins, ␤-synuclein and ␥-synuclein, may block the process (23,24).
The mechanisms governing AS aggregation in the sporadic ␣-synucleinopathies remain unexplained, as is the case for tau and A␤ in sporadic Alzheimer's disease, but they are likely to involve perturbations in age-dependent, genetic, and environmental balances of pro-and anti-aggregative factors. We identified p25␣ as an AS filament-binding protein, which in substoichiometric amounts stimulates AS aggregation. p25␣ was originally co-purified with a tau kinase preparation from bovine brain (25) as a protein localized to oligodendrocytes (26). Functionally, p25␣ is subject to phosphorylation by several kinases (25,27,28), and it acts as a microtubule-associated protein that causes the formation of aberrant bundles of micro-tubules (29). In the ␣-synucleinopathies, PD, LBD, and MSA, the cellular expression of p25␣ is abnormal in the degenerating oligodendrocytes and neurons where it co-localizes with AS in Lewy bodies, Lewy neurites, and glial cytoplasmic inclusions. This suggests that by bringing the pro-aggregative p25␣ into contact with AS, abnormal p25␣ expression in neurons and AS in oligodendrocytes may actively contribute to the degenerative process. This would seem to represent a novel prodegenerative factor in the group of ␣-synucleinopathies.
Recombinant Human p25␣-The human p25␣ cDNA was amplified by reverse transcription-PCR from a human fetal brain mRNA library (Clontech) using the following primers, p25␣ 5Ј, 5Ј-CACCCATGGCTG-ACAAGGCCAA-3Ј, and p25␣ 3Ј, 5Ј-CACGGATCCCTACTTGCCCCCT-TGCAC-3Ј. For the expression of a hexahistidine-p25␣ fusion protein, the PCR fragment was inserted into the pT7-PI vector (35). After sequencing of the vector, the protein was expressed and purified on a Talon metal affinity resin (Clontech) according to the manufacturer's recommendations. For expression of native p25␣, the PCR fragment was inserted into the pET-11d vector (Novagen, San Diego) and sequenced. For protein purification, E. coli BL21 (DE3) cells (Stratagene, La Jolla, CA) were transformed, pelleted, and lysed by sonication on ice in buffer A (50 mM NaH 2 PO 4 , pH 8.2). The soluble proteins were heated to 100°C for 10 min whereupon the heat-denatured proteins were removed by centrifugation. Heat-stable proteins were loaded on a Poros HS50 cation column (PerSeptive Biosystems, Foster City, CA), which was eluted by a double linear gradient first into buffer B (1 M NaCl, 50 mM NaH 2 PO 4 , pH 8.2) and subsequently into buffer C (1 M NaCl, 50 mM NaH 2 PO 4 , pH 12). p25␣ eluted early after the pH was raised above pH 8.2. The final purification and buffer exchange were performed on a GF-75 gel filtration column (Amersham Biosciences) that had been pre-equilibrated with 120 mM NaCl, 20 mM sodium phosphate, pH 7.4 (PBS), supplemented with1 mM dithioerythritol. Dithioerythritol was included to avoid unspecific dimerization via the three free Cys residues. The purified p25␣ was Ͼ95% pure as determined by densitometric scanning of Coomassie Blue-stained gels. The identity of human p25␣ was ascertained by 15 cycles of Edman degradation on a P1000A protein sequencer (Agilent, Palo Alto, CA) of the intact protein with perfect agreement to residues 2-16 of the published sequence (Swiss-Prot accession number O94811).
Biophysical Characterization-Urea denaturation experiments were carried out at p25␣ concentrations of ϳ2.5 M in 2 mM dithioerythritol, PBS, pH 7.4, at 25°C. Fresh 10 M urea stock solutions were prepared on a daily basis. All CD studies were performed on a Jasco J-715 spectropolarimeter (Jasco Spectroscopic Co., Tokyo, Japan) with a Jasco PTC-348W temperature control unit. Spectra were recorded in a 0.1-cm path length cuvette with resolution at 0.2 nm, bandwidth at 1.0 nm, sensitivity at 50 millidegree, response at 2.0 s, and speed of 20 nm/min at 25°C. Three scans were averaged to yield the final spectrum. Protein concentrations were 20 (far-UV CD, 250 -205 nm) and 200 M (near-UV CD, 320 -250 nm).
p25␣-1 Antibody Production-Rabbits were immunized with a fusion protein of a hexahistidine tag linked to the N terminus of p25␣, and serum was affinity-purified on p25␣ immobilized to CNBr-activated Sepharose (Amersham Biosciences) followed by protein A chromatography. The resulting rabbit IgG p25␣-1 was dialyzed against 1 mM EDTA, PBS, pH 7.4, and stored at Ϫ20°C.
Iodination of p25␣-Purified recombinant human p25␣ (6 g) was iodinated essentially as described previously for tau (36), yielding a tracer with a specific activity of 4.8 ϫ 10 5 Ci/mol. ␣-Synuclein Aggregate Analyses-Aggregates of recombinant AS-(1-140), AS-(1-125), AS-(1-110), AS-(1-95), and A␤-(1-40) were formed as described for AS-(1-140), AS-(1-95), and A␤-  and quantified by Coomassie Blue-stained SDS-PAGE after their isolation from the pellet following density gradient centrifugation (34). Aggregate-binding proteins, isolated after aggregate co-sedimentation, were identified after nonequilibrium pH-gradient gel electrophoresis (NEPHGE) for the resolution of basic protein (37) followed by mass spectrometric tryptic peptide mapping of individual protein spots (34). Photoaffinity Cross-linking of ␣-Synuclein Aggregate-bound Proteins-A chemical cross-linking assay was established in order to investigate protein-protein interactions involving aggregated AS. For this purpose, we used the cross-linker sulfosuccinimidyl-2-(p-azidosalicylamido)ethyl-1,3Ј-dithiopropionate (SASD) (Pierce). SASD is a heterobifunctional cross-linking reagent that contains a photoreactive crosslinking group and an amine-reactive group and produces a 18.9-Å bridge between conjugated molecules (38). The bridge of SASD contains a disulfide bridge that provides cleavability after conjugation. Its photosensitive part can be radiolabeled with 125 I prior to the conjugation reaction. After cleavage by a reducing agent, the radioactive label will remain attached to the protein conjugated by photoactivation and thus allow identification of the labeled protein. SASD radiolabeling was performed by using the oxidizing agent, IODO-GEN (1,3,4,6-tetrachloro-3␣,6␣-diphenylglycouracil, Pierce), which was plated on the surface of a glass tube prior to iodination. The procedures for radiolabeling and cross-linking were as follows. 20 g of IODO-GEN was dissolved in 100 l of chloroform and added to a glass tube. The chloroform was slowly evaporated under N 2 . SASD was dissolved in dimethyl sulfoxide to 50 mM and further diluted in PBS, pH 7.4, to a final concentration of 0.5 mM. Next, 200 l of the SASD solution and 100 Ci of Na 125 I were added to the IODO-GEN-coated tube and incubated for 30 s at room temperature. Removing the SASD solution from the IODO-GEN-coated tube and supplementing it with 20 mM KI to quench any residual iodinating activity terminated the iodination reaction. The 125 I-SASD was subsequently incubated with the purified AS aggregates (200 g) in PBS, pH 7.4, for 1 h at room temperature to create the 125 I-SASDprotein complex. The SASD-labeled aggregates were then placed on a 40% sucrose cushion and centrifuged at 15,000 rpm for 30 min to remove free 125 I and 125 I-SASD, which stayed in the supernatant. The 125 I-SASD-AS aggregate tracer, with a specific activity of 1 ϫ 10 5 Ci/mol AS monomer in the aggregate, was resuspended from the pellet in PBS and used immediately. The SASD-modified AS aggregate tracer was then incubated with p25␣ for 30 min at 37°C during continuous shaking. All the above procedures were carried out in the dark, and all reaction vessels were covered by aluminum foil. After incubation, the reaction mixtures were exposed to UV irradiation (Desega Uvis lamp, Copenhagen, Denmark) at 254 nm and at 10 cm from the source for 10 min at room temperature to activate the photoreactive cross-linker. The disulfide bridge in SASD was cleaved by 2% (v/v) ␤-mercaptoethanol (Applichem, Darmstadt, Germany), and samples were analyzed by SDS-PAGE and autoradiography. 125 I-p25␣ Solid Phase Binding Assay to ␣-Synuclein-(1-140) Aggregates-Aggregates of recombinant AS (25 g/ml) in 200 mM NaHCO 3 , pH 9.6, were sonicated and immobilized on Polysorp microtiter plates (Nunc, Copenhagen, Denmark) for 2 h on ice, and residual proteinbinding sites were blocked by incubation with 5% bovine serum albumin (Sigma) for another 2 h. After rinsing, the wells were incubated with ϳ50 pM 125 I-p25␣ in the presence of various concentrations of unlabeled competitor peptides in binding buffer (150 mM NaCl, 2 mM MgCl 2 , 1 mM EGTA, 0.01% bovine serum albumin, 20 mM Hepes, pH 7.4) for 16 h at 4°C. Unbound ligand was removed by rinsing the wells three times with 250 l of binding buffer, and bound tracer was quantified by ␥-counting (Packard Cobra II, Albertville, MN) after release with 250 l of 10% SDS.
Determination of ␣-Synuclein/p25␣ Aggregate Formation-Monomeric AS (340 M) was incubated in the absence and presence of p25␣ (3.4 and 10.2 M) in 0.1% ␤-mercaptoethanol, 120 mM NaCl, 50 mM Tris, pH 7.4, at 37°C. After 3 days, insoluble material of the 25-l sample was isolated by density gradient centrifugation. The pellets were resuspended in 30 l of 8 M urea, 4% SDS overnight at 37°C, and supplemented with 30 l of dithioerythritol-containing SDS-loading buffer. The samples were subjected to SDS-PAGE and analyzed by Coomassie staining.
Thioflavin T Fluorescence Assay-Samples were prepared as above, and samples were taken at different time points. Fluorescence measurements were performed at final concentrations of 10 M protein and 20 M thioflavin T in 90 mM glycine-NaOH, pH 8.5, using a Wallac Victor 3 1420 (PerkinElmer Life Sciences) multilabel counter (excitation at 450 nm, emission at 486 nm) with 1-s integration.
Brain Tissue-Samples of substantia nigra from sporadic PD (n ϭ 4), p25␣ Stimulates Aggregation of ␣-Synuclein LBD (n ϭ 3), and control subjects (n ϭ 4) were obtained from the Netherlands Brain Bank and the National Health and Medical Research Council South Australian Brain Bank. Samples of the cerebral cortex, temporal cortex, and basal ganglia were obtained from MSA cases (n ϭ 3) from the NHMRC South Australian Brain Bank. The brain samples from this bank were bisected, and one-half was fixed with 4% formaldehyde and 2% picric acid, and the other was sliced, fresh frozen, and stored at Ϫ80°C. Control samples were obtained among patients without previous neurological disease before death and no pathological changes in sections of the substantia nigra pars compacta. In the human autopsies, the PD diagnosis was based on loss of neuromelanin pigment and an excessive number of Lewy bodies in the remaining substantia nigra pars compacta neurons in eosin-stained sections as described previously in detail (see Refs. 39 and 40). To verify the distribution of p25 in optimally treated brain tissue, material from four brains was collected from adult male Wistar rats processed for immunohistochemistry (40). Immunohistochemistry of Paraffin Sections-Paraffin sections (6 m) were obtained from pathologically confirmed PD (n ϭ 4), LBD (n ϭ 4), MSA (n ϭ 4), and age-matched control cases (n ϭ 3) without significant brain pathology. The sections were deparaffinized, treated with antigen retrieval (boiling in 1 mM EDTA, pH 8.0, for 10 min), and placed for 10 min in 3% H 2 O 2 to bleach the endogenous peroxidase reactivity. The sections were blocked with 20% normal horse serum for 60 min and then incubated overnight at 4°C with affinity-purified rabbit p25␣-1 antibody (1:1000) and biotinylated donkey anti-rabbit IgG (1:200, The Jackson Laboratory) for 2 h. Following incubation with streptavidinbiotin-peroxidase complex for 1 h, the sections were developed using 3,3-diaminobenzidine tetrahydrochloride as chromogen. Control staining was done by omitting primary antibodies. The rat brains were cut at 40 m and reacted by floating freely. All sections were incubated overnight at 4°C with polyclonal rabbit anti-p25␣-1 antibody diluted 1:200 or sheep anti-human ␣-synuclein antibody (Abcam, Cambridge, UK) diluted 1:200. Specific binding was verified by using either biotinylated swine anti-rabbit antibody diluted 1:500 (Dakopatts, Copenhagen, Denmark) or biotinylated rabbit anti-sheep antibody diluted 1:500 (Abcam, Cambridge, UK), followed by amplification with biotinylated tyramide using TSA Indirect antibody (PerkinElmer Life Sciences) diluted 1:100 for 5 min followed by horseradish peroxidase-conjugated streptavidin-biotin complex for 30 min. The sections were developed as described for the human specimens.
Immunofluorescence Double Staining of Paraffin Sections-To achieve double labeling, human brain sections were pretreated as described above and simultaneously incubated overnight with affinity-purified sheep anti-AS antibody (1:300) in combination with affinity-purified rabbit p25␣-1 antibody (1:200), and subsequently for 2 h in fluorescent dye-tagged secondary antibodies. Secondary antibodies were Cy3-conjugated anti-sheep IgG and Cy2 conjugated anti-rabbit IgG (1:100, The Jackson Laboratory). Control staining was performed by omitting the primary antibodies. Following the completion of fluorescence immunostaining, sections were mounted on gelatin-coated slides, and buffered glycerol was used as coverslips. Sections were examined by using a Bio-Rad confocal laser-scanning microscope and software package (MRC 1024, Bio-Rad). Lewy bodies were identified in relevant brain regions of PD or LBD cases as neuronal inclusions immunopositive for AS. A Lewy body (usually Ͻ30 m) was scanned along the z axis (through the thickness of the section) at 2-m increments to determine whether it was completely represented within the thickness of the section (50 m) and its equatorial plane. This plane was defined at the z axis coordinate where it exhibited the clearest central halo or largest circumference. Images were captured at ϫ2400 magnification.
Immunolabeling of Isolated Lewy Bodies-Lewy body and Lewy neurites were isolated essentially as described previously (33,41). The pellet, enriched in Lewy bodies and Lewy neurites along with other cellular components, was smeared on slides and air-dried. Following a 10-min fixation with 4% formaldehyde and 2% picric acid in phosphate buffer, pH 7.4, the smears were immunostained for AS (sheep anti-AS, 1:300) and p25 antibody (rabbit p25␣-1, 1:400). Following labeling with Cy3-conjugated anti-sheep IgG and Cy2-conjugated anti-rabbit IgG (1:100, The Jackson Laboratory, West Grove, PA), the slides were placed on a coverslip using buffered glycerol and were examined by using a Bio-Rad confocal laser-scanning microscope as described above. Control staining was done by omitting the primary antibodies.
Immunoelectron Microscopy-Immunoelectron microscopic localization of recombinant p25␣ on AS fibrils was performed essentially as described previously for the demonstration of the binding of 20 S proteasomes to AS fibrils (34). In brief, 0.6 g of recombinant p25␣ was incubated with 35 g of AS filaments resuspended at a total volume of 20 l for 16 h at 4°C, whereupon the filaments with associated p25␣ were isolated by density gradient centrifugation. These filaments were resuspended in 70 l of distilled water and applied onto carbon-coated nickel grids. The primary antibodies used were rabbit p25␣-1 IgG and nonimmune rabbit IgG, both at 0.1 mg/ml.
Quantitative Electron Microscopic Determination of ␣-Synuclein Aggregate Formation-Quantitative electron microscopy was used to determine whether p25␣ stimulated the formation of AS filaments or other kinds of aggregates, e.g. amorphous aggregates, and to quantify the profilamentous effect. Monomeric AS (300 M) was incubated alone and with 3 M p25␣ in 120 mM NaCl, 50 mM Tris, pH 7.4, at a total volume of 200 l at 37°C with five independent samples from each situation. Insoluble material from 25 l of each sample was isolated by density gradient centrifugation after 3 days of incubation, and the pellets were resuspended in 50 l of H 2 O, whereupon 3 l of the resuspended aggregates were pipetted onto carbon-coated nickel grids and subjected to negative staining (34). The ultrastructure of negatively stained grids was examined with a Philips CM10 electron microscope (Eindhoven, the Netherlands) using the AnalySIS software version 3.1 (Soft Imaging System, Mü nster, Germany). All measurements were performed blindly at a magnification of ϫ3600. Each grid was examined in a meandering fashion using systematic, uniformly random sampling where the step length was 1700 m in the x direction and 85 m in the y direction. We used point counting with a 14 ϫ 9 grid and area per test point equal to 46.3 m 2 to estimate the area fraction of the fibers as shown om Equation 1, where ⌺P(fiber) is the test point hitting the AS fibers and ⌺P(grid) the test point hitting the grid. This method assumes that the fibers are round in order to avoid the effect of overprojection.

Identification of p25␣ as an ␣-Synuclein
Filament-binding Protein-Cellular proteins binding to aggregated AS may represent targets or modifiers of the toxic effect of the aggregates (34) or modulators of the aggregative process. We used in vitro formed AS aggregates to search for such proteins in rat brain cytosol, deploying an AS aggregate sedimentation technique (34). A range of proteins co-sedimented with the AS aggregates upon cytosol incubation with the AS aggregates, and two-dimensional NEPHGE was used to resolve basic proteins. Fig. 1A (right panel) presents a part of a silver-stained gel that demonstrates a strong protein spot with a weaker, slightly more acidic isoform that was only present when the cytosol was incubated with the AS aggregates. This protein was absent in the cytosol control ( Fig. 1A, left panel) and the AS aggregate control (data not shown). This spot was subjected to mass spectrometric peptide mapping (Fig. 1C), and the spectrum identified the protein as the rat homologue of human p25␣ with the identified peptides covering 33% of the entire protein (Table I). Rat p25␣ identification was further confirmed by immunoblotting with the rabbit anti-p25 peptide IgG (Fig. 1D), which was raised against a peptide in bovine p25␣ (25). The AS aggregate-binding property was not restricted to rat p25␣, as incubation of a human brain extract with AS aggregates enabled the co-sedimentation of human p25␣ (Fig. 1B). p25 is a brain phosphoprotein (see Ref. 25; data not shown) and rat brain cytosol comprises multiple pI isoforms of p25 as compared with the single immunoreactive spot for recombinant human hexahistidine-p25␣ (Fig. 1D, upper versus lower panel). All these isoforms could associate with AS aggregates as demonstrated by co-sedimentation (Fig. 1D, middle panel).
AS aggregates display amyloid-type characteristics that were accounted for by structural determinants in the N-terminal repeat region (42,43). This allowed the assembly of protein aggregates from full-length AS-(1-140) and C terminally truncated AS-(1-95), which lacks the entire acidic C terminus, as demonstrated in Fig. 1E. p25␣ was bound only by the aggre- A, rat brain cytosol was incubated in the absence and presence of 35 g of ␣-synuclein (AS) aggregates, whereupon the samples were subjected to density gradient centrifugation. Pellets were resolved by nonequilibrium pH-gradient two-dimensional gel electrophoresis and subjected to silver staining. Electrophoresis resolved the basic proteins in the pH range 7-11, which left the acidic AS out of the gel. Left panel, part of the gel containing the cytosolic sample with some insoluble proteins stained at the bottom. Right panel, similar part of the gel containing AS aggregate ϩ cytosol sample. Arrow indicates AS aggregate-associated protein subjected to mass spectrometric peptide mapping. Arrowhead marks putative, more acidic isoform of the same protein. Anode and cathode localizations are presented on top of the gels. The fractionation corresponding to molecular weight (MW) is displayed to the left. B, immunoblot demonstration of the binding of human brain p25␣ to AS aggregates. Human brain detergent extract was co-incubated with AS aggregates as in A. Pellets were subjected to reducing SDS-PAGE and immunoblotting using rabbit anti-p25␣ antibody p25␣-1. Lane 1, 10% of inputs of brain extract. Lane 2, pellet from extract without aggregates. Lane 3, pellet from extract plus AS aggregates containing associated human brain proteins. Molecular size markers in kDa are indicated to the left. C, mass spectrum of tryptic digests of protein spot indicated by arrow in A. The gel spot was reduced and alkylated with iodoacetamide prior to digestion. The abscissa shows the mass divided by charge (m/z) of individual ions, and the ordinate shows the signal intensity corresponding to the ions. The peptide pattern was searched against the NCBI p25␣ Stimulates Aggregation of ␣-Synuclein gates of full-length AS but not by the C terminally truncated AS (Fig. 1E). The lack of a role of the amyloid structure generated by the N-terminal part of AS was corroborated by the absence of binding to aggregates formed from the amyloidogenic peptide A␤-(1-40) (Fig. 1E). p25␣ bound preferentially to AS in its aggregated state as demonstrated by the inability of p25␣ to bind AS aggregates in the presence of a 32-fold excess of monomeric AS competing with the binding of p25␣ (Fig. 1F). Accordingly, p25␣ binds to aggregate-selective determinants within the C terminal segment of AS.
Characterization of Recombinant Human p25␣-The p25␣ protein was first identified as a bovine brain-specific phosphoprotein (25), and it has recently been attributed to tubulinassembling properties (29). Sequence analysis demonstrates that p25␣ belongs to the highly conserved p25 gene family present in mammals, flies, nematodes, and even tetrahymenae (44). The human genome contains at least three p25-like genes, here designated as p25␣, p25␤, and p25␥, but previously named 25-kDa brain-specific protein (Swiss-Prot accession number O94811), brain-specific protein (Swiss-Prot accession number P59282), and protein CGI-38 (45, 46) (Swiss-Prot accession number Q9BW30), among which only the p25␣ protein has been detected. These three gene products display a high degree of sequence identity in their C termini, but their N termini differ, with the ␣-form containing a unique 43-amino acid insertion ( Fig. 2A).
We cloned the human p25␣ from a human fetal brain cDNA library and made a deletion mutant p25␣-(⌬3-43) truncated for the ␣-specific N-terminal insertion. Both proteins were expressed in E. coli and purified to more than 95% purity by a protocol based on their heat stability, ion exchange chromatography, and gel filtration (Fig. 2B, left panel). The recombinant proteins displayed a tendency to form disulfide-bridged dimers (Fig. 2B, right panel), and subsequent analyses were performed under reducing conditions. Rabbits were immunized with recombinant human hexahistidine-tagged p25␣ purified by immobilized metal affinity chromatography. The purification pro-tocol for the hexahistidine p25␣ differed from the protocol for purifying the untagged p25␣, which reduced the risk for raising antibodies against possibly contaminating proteins in the untagged p25␣ preparation. The immune serum was subsequently affinity purified onto immobilized recombinant untagged human p25␣. The resulting rabbit IgG, p25␣-1 was specific, as demonstrated by the binding to a single band of 27 kDa in PC12 cells transfected with a human p25␣ vector, but not the empty vector (data not shown), to human brain p25␣ (Fig. 1B) and to the 27-kDa recombinant p25␣ band, which could be inhibited by preincubation of the antibody with recombinant human p25␣ (Fig. 2C, lanes 1 and 2). The p25␣-1 antibody also bound to a less abundant ϳ55-kDa species in the reduced recombinant p25␣ preparation (Fig. 2C, lanes 1-4). However, this immunoreactive species is of very low abundance as no such bands were detectable by sensitive silver staining methods (Fig. 2B). These bands may represent dimeric p25␣ species bonded by a reduction-insensitive bond, which may bind the p25␣-1 antibody avidly. A similar band was detected in rat brain cytosol (Fig. 2C, lanes 5-8) but was absent in human brain extract (Fig. 1B). The antibody also bound to the N-terminally truncated recombinant p25␣-(⌬3-43) peptide (data not shown).
p25␣ is a heat-stable protein, which would suggest that the protein is either (a) folded but very thermostable, or (b) natively unfolded. To gain structural insight into recombinant human p25␣, the purified protein was analyzed by far-UV CD spectroscopy (Fig. 2D). Deconvolution of the spectrum by the k2d program (kal-el.ugr.es/k2d/k2d.html) predicted 15% ␣-helix, 30% ␤-helix, and 55% random coil, but the fit was rather poor (data not shown), suggesting unusual features in its protein structure. However, p25␣ clearly has some degree of organized structure, because incubation of the protein in 5 M urea (well above the midpoint of denaturation, which is around 3.7 M urea, data not shown) caused a marked loss of spectral intensity (Fig. 2D). This was corroborated by the fluorescence spectrum, where the emission intensity peaks at 330 nm and undergoes a dramatic red shift to around 355 nm in the presence of 5 M urea that is typical of a buried Trp residue (Fig. 2E). The above data clearly indicate that human p25␣ possesses both a secondary and a tertiary structure, which are lost at high denaturant concentrations.
Characterization of the Interaction between ␣-Synuclein and p25␣-The binding of brain p25␣ to AS aggregates (Fig. 1, A and B) could in principle be mediated via unidentified cytosolic linker proteins or require specific, yet concealed post-translational modifications of p25␣. Different experimental approaches were used to demonstrate a direct binding between purified recombinant human p25␣ and purified AS aggregates as follows: first, co-sedimentation analysis (Fig. 3A); second, immunoelectron microscopic demonstration of recombinant The peaks in the mass spectrum shown in Fig. 1C were assigned to the monoisotopic mass values of tryptic peptides in rat p25␣. The total sequences coverage was 33%. The peak at m/z 2211.1 is due to a tryptic autodigest peptide. nr protein data base using the Mascot search program (Matrix Science, UK) and showed a significant correlation with rat p25␣ (peak assignment in Table I). p25␣ was identified in three independent experiments using different rats. D, rat brain cytosol and AS aggregate-binding proteins from cytosol were isolated by centrifugation, and the pellets were resolved by two-dimensional gel electrophoretic analysis as in A, but subjected to immunoblotting with the p25␣-1 antibody. Input demonstrates multiple pI isoforms in cytosol. The sample to be incubated with the aggregates was  p25␣ Stimulates Aggregation of ␣-Synuclein p25␣ decorating AS filaments (Fig. 3B); third, photoaffinity labeling of p25␣ by conjugates between AS aggregates and the 125 I-labeled cleavable cross-linker SASD (Fig. 3E); and fourth, competition analysis of the binding of 125 I-p25␣ to immobilized AS aggregates (Fig. 3F).
It is demonstrated in Fig. 3A by means of co-sedimentation analysis that p25␣ binds to the insoluble AS aggregates. Similar results were obtained when using the N-terminally truncated p25␣-(⌬3-43) peptide (data not shown). Immunogold electron microscopy was used to ascertain that the p25␣ binding did indeed take place onto fibrillar types of AS aggregates (Fig. 3B). AS fibrils were incubated with p25␣ and probed with p25␣-1 IgG (Fig. 3B-1) and nonimmune rabbit IgG (Fig. 3B-2) followed by incubation with anti-rabbit IgG conjugated to 5 nM gold particles. Evidently, fibril gold labeling was only seen after incubation with the anti-p25␣ antibody.
The 18.9-Å heterobifunctional cleavable cross-linker SASD was further employed to demonstrate the tight interaction between p25␣ and AS aggregates by photoaffinity labeling. The 125 I-labeled aggregates were quite resistant to depolymerization by short term incubation in SDS-PAGE loading buffer but could be depolymerized by prolonged incubation in 4 M urea, 2% SDS (Fig. 3C, lanes 1 versus 2) as demonstrated previously (34) for unlabeled AS aggregates. Activation of the photoreactive cross-linker by UV light made the AS aggregates insoluble in 4 M urea, 2% SDS (Fig. 3C, lane 3) because of the formation of intermolecular covalent bonds between different AS monomers in the aggregate. The SASD molecule contains a disulfide bridge, which, upon reduction, cleaves the cross-linker and leaves the 125 I label on the structure targeted by the photoreactive group. Hence, reduction of the labeled insoluble aggregates increased their solubility in urea/SDS and revealed the presence of 125 I-labeled AS monomers (Fig. 3C, lanes 3 versus  4). The labeled AS aggregates were used as a probe to label aggregate-binding ligands as demonstrated by the labeling of the AS aggregate-binding antibody FILA-1 but not the nonimmune IgG control (Fig. 3D, lanes 5 and 7). By this technique, FIG. 3. Characterization of the direct binding of p25␣ to ␣-synuclein. A, purified recombinant human p25␣ (p25␣) was incubated in the presence and absence of 10 g of purified AS aggregates (A) and analyzed for co-sedimentation. p25␣ content in 10% of input and total pellet fractions was analyzed by immunoblotting. B, electron microscopic demonstration of p25␣ binding to AS filaments. AS filaments were incubated with recombinant p25␣, whereupon the p25␣-AS filament complexes were isolated by density gradient centrifugation. Purified filaments were incubated with p25␣-1 IgG (panel 1) and nonimmune IgG (panel 2), and subsequently anti-rabbit IgG was conjugated to 5-nm gold particles and analyzed by negative staining electron microscopy. A bar representing 100 nm is presented to the right. p25␣ Stimulates Aggregation of ␣-Synuclein both p25␣ and the deletion mutant p25␣-(⌬3-43) were photoaffinity-labeled upon incubation with the 125 I-SASD conjugated AS aggregates (Fig. 3E, lanes 5 and 7).
A recently developed AS aggregate solid phase binding assay (34) allowed a detailed quantitative analysis of the binding of p25␣ to AS aggregates (Fig. 3F). The 125 I-labeled p25␣ tracer migrated predominantly as a single 25-kDa band, which comigrated with Coomassie Blue-stained recombinant p25␣ (Fig.  3F, inset, lane 2 versus 1). The binding of 50 pM 125 I-p25␣ to immobilized aggregated AS was inhibited by fluid phase AS aggregates with an IC 50 of about 10 nM and to the same extent as when using unlabeled p25␣ and p25␣-(⌬3-43) (Fig. 3D). Monomeric AS displayed an ϳ10-fold higher IC 50 than aggregated AS (Fig. 3F). However, the molar concentration of the aggregates could not be expressed precisely due to their heterogeneous homopolymeric nature, and they are accordingly expressed by the concentration of their monomeric content. The selectivity of p25␣ for aggregated AS on a molar basis is hence even larger, and it is likely to lie in the range of 10 2 -10 3 , given that the aggregates contain 20 -100 monomeric subunits. Accordingly, p25␣ binds to monomeric AS but exhibits a higher affinity for aggregated AS.
The basis for the aggregate selectivity within the AS primary structure was investigated using aggregates prepared from AS molecules truncated for the last 30 residues AS-(1-110) and 15 residues AS-(1-125). To ensure equal concentrations of the aggregates prior to analysis, samples from the purified aggregate stock solutions were depolymerized and subsequently compared by Coomassie Blue staining of SDS-PAGE (data not shown). AS-(1-110) aggregates displayed a 100-fold higher IC 50 than the wild type peptide (Fig. 3F), whereas the IC 50 for binding to AS-(1-125) was indistinguishable from the wild type AS. Moreover, ␤-synuclein, whose C terminus resembles AS in terms of content and spacing of their negatively charged residues, and a synthetic peptide AS-(109 -140), corresponding to AS C terminus, were unable to inhibit the tracer binding (Fig.  3F). The high affinity binding of p25␣ to AS aggregates accordingly relies on the presentation of a segment within the AS C terminus close to residues 110 -125 in the context of an AS aggregate.
p25␣ Stimulates ␣-Synuclein Aggregation-The binding of p25␣ to monomeric AS raised the possibility that p25␣ may affect the process of AS aggregation. This question was first investigated by an aggregate-sedimentation assay where AS was incubated in the absence and presence of substoichiometric concentrations of p25␣ followed by isolation of the AS aggregates by centrifugation (Fig. 4A). At time 0, a small amount of AS was detected in the pellet, which represents the background level of the assay when using 340 M AS. A slight, spontaneous aggregation of AS was detected after 72 h of incubation. However, co-incubation of the AS with 3 and 10 M recombinant human p25␣ produced a dose-dependent increase in the amount of aggregated AS. p25␣ was recovered in the pellet with the AS aggregates (Fig. 4A) but remained in the supernatant in the absence of AS (data not shown). p25␣ also stimulated the sedimentation of AS peptides carrying either of the two PD-causing mutations A30P and A53T (data not shown).
Thioflavin-T fluorescence has been used extensively to monitor the development of amyloid-type aggregates. Supplementing AS with p25␣ produced a time-and dose-dependent increase in the thioflavin-T fluorescence measured after incubation for 1 and 3 days (Fig. 4B). The dose dependence is clearly demonstrated after incubation for 1 day where 3% p25␣, but not 1%, stimulated the aggregation of 100 and 200 M AS. The largest relative p25␣-stimulated increase in AS aggregation was observed by using lower concentrations of AS (100 and 200 M), where 1-3% p25␣ caused an ϳ5-6-fold increase as compared with the autoaggregation of AS. By comparison, the autoaggregation at 340 M AS was so large that the p25␣ increase only amounted to ϳ2-fold. The data using 100 and 200 M AS clearly demonstrate that supplementing with 1 and 3% p25␣ reduces the lag phase of the aggregation process. Quantitative electron microscopy was used to confirm that the thioflavin-T positive aggregates were of a filamentous nature like those present in Lewy bodies and Lewy neurites. For this analysis, 340 M AS was incubated in the absence and presence of 1% p25␣ (n ϭ 5) for 3 days, whereupon the insoluble material Points represent mean of triplicates. One of five representative experiments is presented. C, AS (340 M) aggregation performed for 3 days in the absence and presence of 1% p25␣. Filaments from five individual samples were purified by density gradient centrifugation, resuspended in PBS, applied onto grids for quantitative negative stain electron microscopy, and analyzed in systematic random measuring mode. Upper and lower left panels demonstrate representative electron micrographs obtained in the absence and presence of p25␣. A 1-m bar in the lower left panel applies to the high magnification images in the upper and lower left panels, and 5-m bars are present in the low magnification images in the middle upper and lower images. Right panel shows grid percentage covered with filaments in the absence (AS) and presence of 1% p25␣ (AS ϩ p25␣). There is a significant difference (p ϭ 1.2 ϫ 10 Ϫ4 ) in area fraction of filaments between negatively stained samples of AS and AS incubated with 1% recombinant p25␣ as visualized by electron microscopy.
p25␣ Stimulates Aggregation of ␣-Synuclein was applied onto grids. The negatively stained grids were examined by systematic random sampling to determine the fraction of the grid covered by filaments. Fig. 4C (left panels) demonstrates examples of the images of AS aggregates captured in the absence and presence of 1% p25␣ where the high magnification images clearly show the fibrillar structure of the assemblies. No filaments were present in the absence of AS (data not shown). Quantitation of the mean area fraction covered by AS filaments demonstrated an increase from 12 to 43%, which was statistically significant (Fig. 4C, right panel). No detectable difference of the filaments formed in the absence and presence of p25␣ was observed. Accordingly, substoichiometric amounts of p25␣ can stimulate the AS aggregation via a process resembling the disease-associated aggregation.
p25␣ Accumulates in Pathological ␣-Synuclein Inclusions-In the rat brain, p25␣ is distributed to cells with a morphology corresponding to that of oligodendrocytes and choroid plexus epithelial cells (data not shown). This is consistent with early observations in rat (26). We had the impression that virtually all oligodendrocytes present in both gray and white matter regions expressed p25␣. Similarly, in human brains the vast majority of oligodendrocytes was also immunopositive for p25␣. The immunoreaction product within oligodendrocytes of rat brain and human control cases distributed predominantly to the perinuclear cytoplasm leaving their nuclei unlabeled, whereas peripheral processes displayed a more punctate pattern evident both by histochemical staining and immunofluorescence microscopy (Fig. 5, A and B). Cellular labeling of oligodendrocytes was never observed when the primary antibody was omitted from the immunoreaction (Fig. 5C). MSA cases, however, showed a population of oligodendrocytes with a p25␣ distribution in the perinuclear cytoplasm (Fig. 5E, arrowhead) that was more robust than the slight labeling of the perinuclear cytoplasm in the oligodendrocytes (Fig. 5E, arrows) that resembled the controls. The p25␣ distribution appeared to be expanded in the group of oligodendrocytes with the most intense p25␣ immunoreactivity. A parallel to the cellular distribution of AS in oligodendrocytes during pathological conditions was made in the brains of MSA cases in which oligodendrocytes are known to express AS in glial cytoplasmic inclusions (3,47). AS indeed labeled several oligodendrocytes of MSA (Fig. 5D), leaving oligodendrocytes of PD, LBD, and normal cases unstained (data not shown). The use of doublelabeling fluorescence and confocal microscopy revealed a clear co-distribution of p25␣ and AS in many oligodendrocytes (Fig.  5, G-I). Quantification of images like Fig. 5, G and H, revealed   FIG. 5. p25␣ and ␣-synuclein are co-expressed in glial cytoplasmic inclusions in multiple system atrophy. A, p25␣ in human oligodendrocytes of the normal brain white matter. Note the labeling of the thin perinuclear cytoplasm (arrow). B, p25␣ in human oligodendrocytes of the normal brain white matter shown at high power magnification. C, omitting the primary anti-p25␣ antibody of the immunoreaction abolishes labeling of oligodendrocytes. D, AS in human oligodendrocytes of multiple system atrophy case. Section shows labeled oligodendrocytes (arrowhead) of white matter underlying cerebral cortex. Subcellular AS distribution within individual oligodendrocytes is heterogeneous but leaves the nucleus unlabeled. E, p25␣ in human oligodendrocytes of same multiple system atrophy case. Numerous labeled oligodendrocytes are seen. Subcellular p25␣ distribution within individual oligodendrocytes is heterogeneous but leaves the nucleus unlabeled. It varied from labeling of thin perinuclear cytoplasm in normal and apparently normal oligodendrocytes (arrow) to a robust labeling of expanded cytoplasm in degenerating cells (arrowhead). F, p25␣ in human oligodendrocytes of multiple system atrophy case shown at high power magnification highlighting principal difference in subcellular p25␣ distribution within apparently normal (arrow) and pathological (arrowhead) oligodendrocytes. G-I, double-labeling fluorescence revealing co-distribution of AS (G) and p25␣ (H) in glial cytoplasmic inclusions (arrowhead in D and E) of multiple system atrophy case. I, superimposing the images reveals co-localization of p25 and AS in oligodendrocytes. H and I, presumably normal oligodendrocytes without glial cytoplasmic inclusions identified as p25␣-labeling of thin perinuclear cytoplasmic region (arrows in E) and devoid of AS immunoreactivity (arrows in I). Scale bars ϭ 10 m (A, C-E, and G-I), 2.5 m (B and F). p25␣ Stimulates Aggregation of ␣-Synuclein that about 60% of the AS-positive inclusions were p25␣-positive. The number of AS-positive oligodendrocytes was, however, significantly lower compared with the number of oligodendrocytes labeled with p25␣.
Beside its distribution to oligodendrocytes and the choroid plexus of the adult rat brain, p25␣ was also confined to the small group of neurons in the supraoptic nucleus. This was the sole p25␣-positive neuronal population observed during the screening of the rat brain (Fig. 6A, inset). In these neurons, p25␣ was distributed to the entire perikaryal cytoplasm, and the nucleus and peripheral processes were unlabeled (Fig. 6A) similar to what has been observed in human oligodendrocytes (Fig. 5E). Likewise, p25␣ was expressed in many neuronal cell populations during the prenatal development of the rat brain. 2 This demonstrates that neurons, at least in the rat brain, have the potential for p25␣ expression. The human normal control brain sections examined did not contain the supraoptic nucleus, but other normal neuronal populations were not expressing p25␣ (data not shown). Histochemically, p25␣ was occasionally demonstrated in Lewy bodies of substantia nigra pars compacta of LBD (data not shown) and PD (Fig. 6B), where the periphery of the brain stem Lewy body in the melanized neuron displayed a p25␣ immunoreactivity with a frequency of 1-5 labeled Lewy bodies per section. With the p25␣-1 antibody, the Lewy body labeling intensity was generally low but most in-tense in their peripheral portion (Fig. 6B). Lewy body labeling appeared irrespective of whether melanin was removed before immunostaining or not. Confocal analysis of double fluorescence-labeled sections from cortical LBD tissue demonstrated a diffuse p25␣ immunoreactivity within the Lewy body in addition to the p25␣-stained oligodendroglial processes (Fig. 6, C-E), which were absent when the primary antibody was omitted (Fig. 6, F-H). The double immunofluorescence analysis revealed that ϳ5% of the AS-positive nigral and 10% of cortical Lewy bodies in PD and LBD were p25␣-positive.
The emergence of procedures to isolate AS-containing inclusions from human PD, MSA, and LBD tissue have allowed detailed structural and biochemical analyses of these structures. Fig. 7, A and B, demonstrates that p25␣ is present in both Lewy bodies and Lewy neurites isolated from LBD brain tissue, and the p25␣ in the inclusions co-localizes to a large extent with the AS as demonstrated by double immunofluorescence confocal laser-scanning microscopy. The negative controls without primary antibodies showed no staining (data not shown) as demonstrated for the tissue staining with the same antibodies in Fig. 6G. 100% of the AS-positive isolated glial cytoplasmic inclusions and 80% of isolated AS-positive Lewy bodies and Lewy neurites were p25␣-positive. Biochemical analysis revealed that p25␣ in both isolated Lewy bodies and glial cytoplasmic inclusions appeared as a single 27-kDa band with no signs of aggregation and degradation when analyzed by reducing SDS-PAGE and immunoblotting (Fig. 7C).
Accordingly, p25␣ and AS are co-expressed in the inclusions that develop during the degenerative process of neurons in PD and LBD and oligodendrocytes in MSA. Most interestingly, p25␣ and AS are not normally co-expressed in the adult nervous system. This suggests that abnormal expression of p25␣ in neurons and AS in oligodendrocytes may actively contribute to the degenerative process by bringing the pro-aggregative p25␣ into contact with AS.

DISCUSSION
The ability of mutant AS to induce autosomal, dominant PD and LBD is evident from the rare familial cases. The role of wild type AS in sporadic PD and LBD, however, is less clear, although these diseases demonstrate a strong aggregation of the wild type AS protein in intracellular Lewy body and Lewy neurite inclusions. Recent studies (8,9) have established a direct link between the expression of the wild type AS protein, its aggregation, and the induction of early onset PD and LBD as demonstrated in families with triplication of the AS gene. The affected members among the Swedish-American kindred had an ϳ40% increased level of AS protein (9), and an increased level of aggregated AS was demonstrated in the Iowa kindred (10). This indicates that the level of AS in the normal brain is close to the point where AS may aggregate spontaneously, as an elevation of about 50% triggers early onset disease. This gives credit to the hypothesis that aggregate-stimulating factors are inducing and contributory players in the pathogenesis of the spontaneous ␣-synucleinopathies, which also comprise MSA, which displays a strong ␣-synuclein aggregation in oligodendrocytes.
In our search for aggregate-stimulating factors, we looked for proteins that bound preferentially to AS in its aggregated state, and we identified p25␣. We demonstrated that an increased or ectopic expression of p25␣ may represent one mechanism in sporadic diseases that can drive the conversion of monomeric AS into filamentous aggregates. cence, where the AS tendency to auto-aggregation is less pronounced. Second, when analyzed in situ, p25␣ accumulates in the majority of the AS-aggregate-containing glial cytoplasmic inclusions in MSA and cortical and nigral Lewy bodies in PD and LBD. This was demonstrated in the analysis of purified inclusions from nonfixed and nonfrozen tissue, a procedure that increases the sensitivity of detection. The tau and histone proteins have also been shown to stimulate aggregation of AS at substoichiometric concentrations, but their presence in AS aggregate-containing inclusions is rarer (48,49).
Human p25␣ is predominantly expressed in oligodendrocytes, which is in agreement with early observations in rat (26) where it appears to be a very good marker for these cells as it does not stain myelin but merely the cellular soma and some processes. It is therefore not surprising to find accumulations of p25␣ in the AS aggregate containing glial cytoplasmic inclusion in MSA as compared with the neuronal Lewy bodies, but it may play a similar AS pro-aggregatory role in this disease. However, we demonstrate a dramatic change in the p25␣ expression pattern in MSA, where a large accumulation of p25␣ takes place in the expanded cell bodies of the apparently dystrophic oligodendrocytes containing glial cytoplasmic inclusions. Neuronal p25␣ expression is not a common finding in rat and human brain tissue, but the neuronal phenotype is not incompatible with p25␣ expression as we detected neuronal expression of p25␣ in the nucleus supraopticus and during prenatal brain development of rat (data not shown). This suggests that an abnormal expression of the protein occurs in the affected nerve cells in PD and Lewy body dementia. Dysregulation of p25␣ expression could accordingly be a contributory factor in cases of sporadic ␣-synucleinopathies, which makes studies of the regulation of its gene highly desired along with studies of p25␣ in other diseased states.
The preferential p25␣ binding to aggregated AS is displayed by a 100 -1000-fold lower IC 50 for the binding of aggregated versus monomeric AS, assuming that the AS aggregates contain at least 20 -100 monomers. The interaction relies critically on the acidic C terminus of AS, as demonstrated by the strong inhibitory effect of truncation of the last 30 residues. This suggests a role for charge interaction with basic peptide motifs in stimulators of AS aggregation as p25␣, histones, and the AS-interacting microtubule-binding segments in tau are basic peptides (32,47), and a similar charge neutralization by Ca 2ϩ binding to this segment may explain the pro-aggregative effect of this ion (32). However, a simple charge interaction cannot explain the interaction as ␤-synuclein, which possesses an even more acidic C terminus, and a synthetic peptide corresponding to the C-terminal 30 residues of AS are unable to bind to p25␣. The solid phase binding assay demonstrated that AS aggregates formed from AS-(1-125) displayed full binding activity as compared with the 100-fold higher IC 50 for AS-(1-110) aggregates. These are the first data to demonstrate, albeit indirectly, an aggregate specific folding of the C terminus. We believe that future studies should pay attention to the structure of this segment in aggregates rather than focus entirely on the amyloid-type ␤-folded segment comprising the first ϳ100 residues detectable in both oligomers and filaments. The C-terminal segment in the aggregates may be exposing the FIG. 7. ␣-Synuclein and p25␣ co-localize in Lewy bodies and Lewy neurites isolated from Lewy body dementia brain tissue. Lewy bodies (A) and Lewy neurites (B) isolated from LBD brain tissue were analyzed for localization of ␣-synuclein (a-syn) and p25␣ (P25) by confocal laser-scanning microscopy. AS immunoreactivity, detected with sheep anti-␣-synuclein antibody, is presented in the left columns, and p25␣, detected by the rabbit p25␣-1 IgG, is presented in the middle columns. Merged images are presented in the right columns. 10-m scale bars are presented. C, immunoblot analysis of p25␣ in isolated brain inclusions. Extracts of human brain tissue homogenate (BH) (20 g; lane 1), purified Lewy bodies (LB) (5 g) from brain tissue affected by LBD (lane 2), and isolated glial cytoplasmic inclusions (GCI) (5 g) from tissue affected by MSA (lane 3) resolved by reducing SDS-PAGE, electroblotted, and probed with anti-p25␣ antibody p25␣-1. Molecular size markers in kDa are presented to the left. p25␣ Stimulates Aggregation of ␣-Synuclein toxic AS aggregate-specific determinants, which induce the specific neurodegeneration of the ␣-synucleinopathies in contrast to the tauopathies, for example, which display intracellular tau aggregates of the amyloid type that likely share some characteristics with the N-terminal amyloid-type part of AS. The interaction was also demonstrated by photoaffinity labeling using AS aggregates conjugated to 125 I-SASD. This technique represents a neat way to label a pathological human protein aggregate for subsequent use as a probe to identify its putative ligands. By using this technique, we have been able to label proteins in rat brain extracts (data not shown), which suggests that the method possesses a potential that justifies its further exploitation.
p25␣ was originally co-purified with a tau kinase preparation from bovine brain, identified as a brain-specific phosphoprotein (25), and later shown to be a substrate for phosphatidic acid-stimulated kinases (27) and protein kinase A and CDK5 (28). Accordingly, we found that rat brain p25 was present in several pI isoforms, which indicates phosphorylated species, but all these species bind to aggregated AS like nonphosphorylated recombinant human p25␣. This indicates that p25␣ phosphorylation is unimportant in the regulation of its binding to AS. Bovine p25␣ is a microtubule-associated protein that stimulates tubulin aggregation and probably undergoes significant conformational changes upon binding to monomeric tubulin, because both far-UV CD (29) and fluorescence spectra 3 show loss of a spectral signal upon complex formation. However, incubation of p25␣ with monomeric AS does not lead to spectral changes (data not shown). The minimal interpretation of this result is that p25␣ binds in a different manner to AS and tubulin. It is therefore likely that the mechanism used by p25␣ for aggregating tubulin differs from that employed on AS. The different p25 gene products display a high degree of amino acid sequence identity, apart from the N-terminal ␣-specific segment in p25␣, which corresponds to residues 3-43. Accordingly, they may share the ability to interact with ␣-synuclein. The ␤and ␥-p25 forms, formerly named brain-specific protein (Swiss-Prot accession number P59282) and protein CGI-38 (Swiss-Prot accession number Q9BW30) (45,46), have never been demonstrated at the protein level. However, they may be recognized by the p25␣-1 antibody, which also binds the truncated p25␣-(⌬3-43) peptide and is thus likely to cross-react with the p25 ␤and ␥-forms. This antibody only binds to the single p25␣ band in human brain extracts, which suggests that the other forms are not present at detectable levels. However, we cannot exclude that that the ␤and ␥-forms may be expressed in some parts of the brain during pathological conditions where they may contribute to AS aggregation.
The mechanism of the pathological AS aggregation remains unclear, but the process is likely to involve reversible initial steps where monomeric AS associates into smaller polymers prior to the development of the nucleation-competent oligomers and subsequent large filaments. Our data support a hypothesis where hetero-oligomers may play a role in this nucleation process. A simple scenario could be that p25␣ binds to monomeric AS and changes its conformation into the form present in the amyloid-type oligomers. The preference of p25␣ to bind to aggregated AS suggests that the p25␣-AS complex formed upon binding monomeric AS induces a secondary structure in AS similar to that present in nucleation-competent oligomers. This structure will enable the addition of additional monomers. Coupling of binding and structure induction means that part of the binding energy has to be channeled into conformational changes that may otherwise be unfavorable. This may explain the decreased affinity for monomers compared with oligomers, although other possibilities such as an increased binding surface in the oligomer-p25␣ complex cannot be ruled out at present. Accordingly, p25␣ stimulates the nucleation, not the fibrillization step, which is corroborated by the lowering of the AS concentration needed to trigger aggregation. This means that an inhibition of p25␣-stimulated aggregation will inhibit the build up of both oligomers and fibrils in contrast to a selective inhibition of fibrillization, where toxic oligomers may accumulate. This makes it worth investigating the mechanisms involved in inhibiting the pro-aggregatory function of p25␣ or its expression as novel pathways for neuroprotective strategies in the ␣-synucleinopathies.