Parkinson's Disease-associated α-Synuclein Is a Calmodulin Substrate

α-Synuclein is a neuronal protein thought to be central in the pathogenesis of Parkinson's disease (PD) because it comprises the fibrillar core of Lewy bodies, one of the histologically defining lesions of PD, and because mutations in α-synuclein cause autosomal dominant PD. Although its physiologic role is uncertain, α-synuclein is a synaptic protein that may contribute to plasticity. We produced synuclein with incorporated photoprobes to identify and purify novel synuclein-interacting proteins both to begin to clarify the physiology of synuclein and to identify factors that may regulate synuclein conformation. We detected several cross-links and purified and identified one as calmodulin (CaM). CaM binds to both wild type and PD-associated mutant α-synucleins in a calcium-dependent manner. We further demonstrate that CaM and α-synuclein interact in intact cells in a calcium-dependent manner and that activated CaM accelerates the formation of synuclein fibrils in vitro. We hypothesize that the known calcium control of synuclein function is mediated through CaM interaction and that CaM potentially alters synuclein conformation.

␣-Synuclein is a neuronal protein thought to be central in the pathogenesis of Parkinson's disease (PD) because it comprises the fibrillar core of Lewy bodies, one of the histologically defining lesions of PD, and because mutations in ␣-synuclein cause autosomal dominant PD. Although its physiologic role is uncertain, ␣-synuclein is a synaptic protein that may contribute to plasticity. We produced synuclein with incorporated photoprobes to identify and purify novel synuclein-interacting proteins both to begin to clarify the physiology of synuclein and to identify factors that may regulate synuclein conformation. We detected several cross-links and purified and identified one as calmodulin (CaM). CaM binds to both wild type and PD-associated mutant ␣-synucleins in a calcium-dependent manner. We further demonstrate that CaM and ␣-synuclein interact in intact cells in a calcium-dependent manner and that activated CaM accelerates the formation of synuclein fibrils in vitro. We hypothesize that the known calcium control of synuclein function is mediated through CaM interaction and that CaM potentially alters synuclein conformation.
Although many neurodegenerative diseases are characterized by specific patterns of neuronal loss, distinct clinical presentations, and unique histopathological features, many of the diseases including Parkinson's, Alzheimer's, trinucleotide repeat diseases, and spongiform encephalopathies display proteinaceous aggregates, inclusions, or plaques as characteristic or defining features upon pathologic examination of involved tissue (1). These plaques or inclusions are composed of neuronal proteins or fragments thereof, which have adopted altered conformations that have resulted in the formation of fibrillar amyloid plaques or inclusions having a high degree of ␤-pleated sheet secondary structure.
Parkinson's disease (PD) 1 is the second most common neurodegenerative disease with a prevalence of about 2% of those 65 years of age and older. The etiology of PD is at least partly genetic (2,3), and kindreds having autosomal dominant and autosomal recessive forms of PD have been identified (4).
In rare families, either of two identified mutations in the ␣-synuclein gene can cause autosomal dominant early onset PD. ␣-synuclein was initially identified as an abundant nerve terminal protein in the Torpedo electric organ and thought to be transiently associated with neuronal membranes (5,6). Since then it has been "rediscovered" in a variety of assays (7)(8)(9).
Heightened interest in ␣-synuclein developed when ␣-synuclein gene mutations were found to be responsible for some cases of autosomal dominant Parkinson's disease (10 -12), and the protein was found to be a major component of Lewy bodies (LBs) (13)(14)(15), the fibrillar neuronal proteinaceous inclusions that are a histologically defining feature of Parkinson's disease.
Ultrastructurally, Lewy bodies have a dense fibrillar core, and the fact that they stain with thioflavin S (16,17) indicates that the constituent proteins have a high degree of ␤-pleated sheet secondary structure. The fibrillar core contains synuclein (18,19). LBs are seen in sporadic and familial PD, and in vitro evidence exists that PD-associated synuclein mutations increase the rate of synuclein fibrilization (20 -23).
Intriguingly, ␣-synuclein-containing glial cytoplasmic inclusions are found in oligodendrocytes in multiple system atrophy (24 -26), and in neurodegeneration with brain iron accumulation type I, neurons in the globus pallidus display Lewy bodylike ␣-synuclein containing inclusions (27). Taken together, these findings suggest that altered synuclein conformation is linked to neurodegeneration.
Overexpression of human ␣-synuclein as a transgene in mice (28,29) or in Drosophila (30) results in formation of synucleincontaining inclusions that in many respects resemble Lewy bodies, in alterations of dopamine neurons, and in movementrelated phenotypes. Deletion of the ␣-synuclein gene in mice does not result in a phenotype resembling PD. Instead, dopaminergic neurons prepared from ␣-synuclein-deficient mice display altered dopamine release in one experimental paradigm of synaptic plasticity (31). It therefore appears more likely that the synuclein mutations are associated with a gain of a toxic function, and it is not exclusively the loss of synuclein function that causes PD in the affected patients. One effect of PD mutations may be to increase the propensity to form fibrils, again suggesting that synuclein conformations are related to PD.
The most striking feature of the primary amino acid sequence of ␣-synuclein is the repetition 8 times of an 11-amino acid motif that is predicted to form amphipathic ␣-helices (7). Contrary to the predictions, recombinant or purified ␣-synuclein is considered "natively unfolded" and has little identifiable periodic secondary structure as assessed by circular dichroism (32,33) or NMR spectroscopy (34). However, when incubated with small unilamellar liposomes of certain lipid (mimicking those found in synaptic vesicles), SDS micelles, or certain helix-forming solvents, the protein undergoes a conformational change, becoming largely helical, and associates with the liposomes (33,35,36). An NMR study showed that it is indeed the N-terminal portion of the protein that can become helical and membrane-associated (34). Despite these findings and the synaptic localization of ␣-synuclein (5), immunoelectron (37) and immunofluorescence microscopic and cell fractionation experiments (7,38) have failed to demonstrate a definite or stable association with synaptic vesicles or any other subcellular membrane. At this point, it is not clear which subcellular membranes ␣-synuclein may associate with.
Although purified ␣-synuclein initially has little in the way of periodic secondary structure, in vitro studies with purified or recombinant synuclein have demonstrated that synuclein can spontaneously form fibrils having a high degree of antiparallel ␤-pleated sheet structure (16,21,39) and that PD-associated synuclein mutations accelerate fibril formation (20,22,23,40).
Given the apparent paradox that pure synuclein spontaneously forms fibrils or, in the presence of certain lipid membranes, becomes largely helical and binds to membranes but that in cells most synuclein appears to be natively unfolded and not membrane bound, we hypothesize that synuclein conformation and/or membrane association may be regulated by synuclein-binding proteins.
Although much work has begun to elucidate the mechanism by which purified ␣-synuclein forms fibrils and of some stimuli that may trigger aggregation of ␣-synuclein, relatively little is understood about how or whether ␣-synuclein interactors alter fibril formation.
We employed a photocross-linking strategy (45) to detect proteins in brain extracts present at endogenous levels and in direct contact with ␣-synuclein. We purified and identified one such protein as calmodulin (CaM). We show that CaM and ␣-synuclein interact in a calcium-dependent manner both in vitro and in vivo and that activated CaM accelerates the kinetics of synuclein fibrilization.

EXPERIMENTAL PROCEDURES
In Vitro Transcription/Translation and Photocross-linking-In vitro transcription using T7 or T3 RNA polymerase (Promega) was as described (46). mRNAs were translated for 15-40 min at 26°C in a wheat germ translation system supplemented with [ 35 S]Met (Amersham Biosciences; 1100 Ci/mmol). Photocross-linking using trifluoromethyldiazirinobenzoic acid (TDBA)-Lys-tRNA was as described (45). After translation, 500 ml of buffer (20 mM HEPES, pH 7.4, 150 mM KCl) was added to translation reactions (100 -150 ml) containing His-tagged ␣-synucleins. 25 ml of TALON Co 2ϩ -agarose beads (Clontech) equilibrated in binding buffer was added, and binding was allowed to proceed for 1 h at room temperature. Beads were washed free of the majority of the wheat germ lysate protein with two 1-ml washes of binding buffer (20 mM HEPES, pH 7.4, 150 mM KCl). Beads were resuspended in a volume equal to the original translation volume in binding buffer. Typically 5 ml of this suspension was used for each cross-linking assay. After cross-linking, beads were sedimented, after which supernatants were aspirated. Laemmli sample buffer containing 50 mM EDTA was added to the beads to elute bound ␣-synuclein and photocross-links. After boiling for 5 min, samples were subjected to SDS-PAGE and fluorography.
In experiments not utilizing bead-isolated ␣-synuclein, components indicated in the figure legends were added directly to 5-ml translation reactions and subjected to photoactivation. After cross-linking, 10 volumes of 10% trichloroacetic acid was added, and proteins were precipitated. Pellets were rinsed with acetone before being resuspended in Laemmli sample buffer. After heating, samples were subjected to SDS-PAGE and fluorography.
Mutagenesis of synuclein was achieved using the QuikChange kit (Stratagene), and all mutations were confirmed by DNA sequencing.
Preparation of BBC-Meninges and large vessels were peeled from three calf brains (ϳ1 kg of tissue). After rinsing in PBS, brains were homogenized in a blender in 2 liters of 20 mM HEPES, 150 mM KCl, pH 7.5.
The homogenate was filtered through cheese cloth three times and then centrifuged to remove unbroken cells, nuclei, and mitochondria (Sorvall GS-3 rotor, 8500 rpm, 1 h, 4°C). The resulting supernatant was clarified by centrifugation (Beckman Ti 50.2 rotor, 37,000 rpm, 60 min, 4°C, ϳ110,000 ϫ g av ). The protein concentration was 4 mg/ml by the Bio-Rad DC protein assay kit using bovine serum albumin as a standard.
Purification of 15-19-kDa Cross-links from Brain Cytosol-Bovine brain cytosol (500 ml) was heated to 70°C in a circulating water bath. 100-ml aliquots were heated in flasks for 5 min after the thermometer in the sample reached 70°C. Flasks were rapidly cooled in liquid nitrogen and then kept on ice. Aggregated protein was removed by ultracentrifugation (Beckman Ti50.2 rotor, 37,000 rpm, 35 min, 4°C). 400 ml of heat-treated cytosol was incubated with 10 ml of SP-Sepharose fast flow (Amersham Biosciences) previously equilibrated in 20 mM HEPES, 150 mM KCl, pH 7.5, in a batch for 12 h at 4°C. 360 ml of the SP-Sepharose supernatant was added to a vessel containing 7 ml of Q Sepharose FF (Amersham Biosciences) equilibrated in 20 mM HEPES, 150 mM KCl, pH 7.5. After 1 h at room temperature, the flow-through was removed, and the resin was washed with 100 ml of 20 mM HEPES, 400 mM KCl. The resin was then packed into a column (7-ml bed volume) and was washed with 20 ml of 20 mM HEPES, pH 7.4, 400 mM KCl. Bound proteins were eluted with a linear gradient of KCl (0.5-1.0 M) in ϳ12 column volumes. 28 3.5-ml fractions were collected. Fractions were assayed for the ability to cross-link to ␣-synuclein. Peak activity eluted at around 515 mM KCl.
Mass Spectrometry-Bands were excised from Coomassie Bluestained gels and digested with trypsin and processed for mass spectrometric analysis (47). Briefly, the peptide mixture was partially fractionated on a Poros 50 R2 RP microtip, and resulting peptide pools were analyzed by matrix-assisted laser-desorption/ionization reflectron timeof-flight MS using a Reflex III instrument (Brü ker Franzen; Bremen, Germany), and by electrospray ionization tandem MS on an API 300 triple quadrupole instrument (PE-SCIEX; Thornhill, Canada), modified with an ultrafine ionization source as described (48). Selected mass values from the matrix-assisted laser desorption time-of-flight experiments were taken to search a protein nonredundant data base (NCBI, Bethesda, MD) using the PeptideSearch (49) algorithm. Triple quadrupole MS/MS spectra were inspected for yЉ ion series, and the information was transferred to the PepFrag (50) program for use as a search string.
Production of Recombinant ␣-Synuclein-␣-Synuclein cDNAs were cloned into pET 21D (Novagen) and the plasmids were expressed in BL21DE3 E. coli. Cultures of 750 ml were grown to midlog phase and isopropyl-1-thio-␤-D-galactopyranoside was added to 0.4 mM. After 2 h, cells were pelleted, washed in PBS, and finally resuspended in 50 ml of 20 mM HEPES, 100 mM KCl, pH 7.2. The resuspended bacteria were heated to 90°C for 5 min. Aggregated protein was removed by centrifugation (Beckman Ti 60 rotor, 32,000 rpm, 30 min, 4°C). The supernatant was at least 90% pure synuclein as assessed by SDS-PAGE. Contaminating nucleic acids and proteins were removed from the lysate by ion exchange chromatography on Q-Sepharose Hi-Trap columns (Amersham Biosciences). ␣-Synuclein eluted at ϳ250 mM KCl. Synucleincontaining fractions were pooled and chromatographed on a Superose 12 column (Amersham Biosciences) in 20 mM HEPES, pH 7.4, 100 mM KCl. Peak fractions were identified by Western blotting.
Chemical Cross-linking with Disuccinimidyl Suberate (DSS) in Intact Cells-DSS (Pierce) (25 mM stock in Me 2 SO) was prepared fresh for each experiment and was used at 1 mM. Cross-linking was allowed to proceed for 15-20 min at 37°C. The reaction was then quenched by the addition of 0.1 volumes of 1 M glycine buffered in 100 mM NaHCO 3 , pH 8.
SK-N-SH neuroblastoma cells were rinsed three times with PBS containing calcium to remove serum proteins and amino acids that would quench the DSS. DSS (1 mM final), a calcium ionophore, A23187 (20 mM final from 2 mM stock in Me 2 SO), and calmidizol (10 -20 mM from 5 mM stock in Me 2 SO, from Sigma) were added to prewarmed 10-cm dishes as indicated in the figures. After 20 min, the cross-linker was quenched as described above. After 10 min on ice, cells were rinsed with PBS, and each dish was lysed in 200 ml of lysis buffer (20 mM HEPES, pH 7.4, 150 mM NaCl, 1% Triton X-100, 0.1% SDS). Nuclei and debris were removed by centrifugation at 1500 ϫ g av at 4°C for 10 min. Anti-synuclein monoclonal antibody (Transduction Laboratories) was added to 6 mg/ml and incubated overnight at 4°C. Immune complexes were collected by the addition of 10 ml of Protein G-Sepharose. After 1 h, immunoprecipitates were washed three times in lysis buffer and once in water. Immunoprecipitated protein was eluted by the addition of Laemmli sample buffer. Samples were electrophoresed through 12% gels and transferred to nitrocellulose, and blots were probed with monoclonal anti-CaM antibodies (Upstate Biotechnology, Inc., Lake Placid, NY) at 1 mg/ml in 5% milk/PBS overnight at 4°C. Blots were developed using the ECL-Plus Chemiluminescence kit (Amersham Biosciences).
CaM-Sepharose Chromatography-Wild type ␣-synuclein (120 ml of 0.5 mM synuclein in PBS plus 0.1 mM CaCl 2 ) was incubated with 15 ml of CaM-Sepharose (Amersham Biosciences) for 30 min at room temperature. The flow-through was collected, and beads were washed three times with 200 ml of binding buffer. Specifically bound synuclein was eluted with 200 ml of PBS containing 5 mM EDTA. Fractions were subjected to trichloroacetic acid precipitation and synuclein content was determined by Western blotting.
Dansyl Labeling of CaM and Fluorescence Measurements-CaM (1.5 mg) was dissolved in 400 ml of 0.2 M NaHCO 3 , pH 9.0. 20 ml of 10 mg/ml dansyl chloride in dimethyl formamide was added. After mixing, derivatization was allowed to proceed for 1 h on ice. Unincorporated dye was removed by gel filtration (PD-10 column (Amersham Biosciences) developed in PBS) and then by dialysis. Spectra were obtained in a PerkinElmer LS50B fluorimeter. Excitation was at 338 nm, and emission spectra were collected from 425 to 600 nm with an excitation slit width of 7.5 nm and an emission slit width of 5.0 nm unless otherwise stated. Dansyl-CaM was used at approximately 1 mM (Fig. 6a) or 0.086 mM ( Fig. 6, b and c).
For measurements of dansyl-CaM binding to synuclein, synuclein was added to cuvettes containing dansyl-CaM at a final concentration of 86 nM in PBS with 2.5 mM CaCl 2 . Emission spectra were collected, and fluorescence intensity at 486 nm was recorded for each sample. For construction of the Scatchard plot in Fig. 6, the fluorescence measured at 486 nm of dansyl-CaM in the absence of synuclein was subtracted from the reading obtained for each sample containing synuclein. Neither the synuclein alone nor unlabeled CaM showed any fluorescence at 486 nm. The fraction of bound dansyl-CaM at each concentration was determined from the fractional fluorescence increase at 486 nm using the following relationship: . The concentration of free synuclein was determined by subtracting the bound concentration from the total synuclein concentration in each sample. These numbers allowed for construction of a saturation curve and a Scatchard plot from which the dissociation constant was estimated (y ϭ Ϫ0.0561x ϩ 4.5944). The least squares method was used to fit the data from the Scatchard plot to a straight line.

RESULTS
Experimental Strategy-To identify ␣-synuclein-interacting molecules, we employed a photocross-linking approach. The technology, developed by investigators primarily interested in protein translocation across membranes, has been successfully utilized to identify proteins comprising the translocation pores in ER (45,51) and mitochondrial membranes (52). In addition, it has been useful in tracing the temporal changes in molecular interactions that newly synthesized proteins undergo as they execute their folding program (53)(54)(55).
In this approach (see Fig. 1), the "bait" is translated in vitro in the presence of [ 35 S]Met and lysyl-tRNA, where the amino group of the lysine residue has been selectively modified with the photoactivable cross-linker TDBA. The result is that crosslinker-modified lysyl residues are incorporated into the in vitro translated protein, as dictated by the distribution of lysine codons in the mRNA. After in vitro translation, covalent attachment of the bait to interacting molecules is caused by UV light-induced activation of the cross-linker. UV activation of the photoprobe generates a highly reactive carbene that will react with nearly any chemical bond. The carbene intermediate persists for ϳ1 ns and is therefore diffusion-limited. If water is next to the cross-linker, it will be quenched. If, on the other hand, a protein is in direct contact with the cross-linker-modified lysine upon activation, the bait will be cross-linked to the target molecule. The size of the probe is about 5 Å. After translation, cross-links are detected by SDS-PAGE followed by fluorography. The molecular weight of the in vitro translated bait is known, and cross-linking of proteins to the radiolabeled bait results in decreased electrophoretic mobility on SDS-PAGE from which the approximate molecular weight of the cross-linked protein can be inferred.
A salient feature of the approach is that it allows for temporal and spatial control of cross-linking. When compared with conventional cross-linkers, this approach is more specific, and cross-linking cannot occur through intermediate proteins.
Compared with other methods of detecting protein-protein interactions, this approach has the advantage that low affinity or transient interactions can be detected, and the biochemical milieu of the reaction can be varied. Additionally, the assay does not rely on overexpression systems. Therefore, all proteins in the system are present at endogenous levels in the extracts used. Also, since the full repertoire of cellular protein is present during the assay, it is less likely that nonspecific interactions, such as those that can be detected when two purified proteins are assayed together, will be detected.
Identification of ␣-Synuclein Cross-links in Brain Extracts-mRNAs encoding human ␣-synuclein where the C-ter-

␣-Synuclein and Calmodulin
minal 20 amino acids were replaced with eight histidine codons were translated in a wheat germ lysate supplemented with [ 35 S]Met and TDBA-Lys-tRNA. After translation, ␣-synuclein was bound to Co 2ϩ -chelating agarose beads. Beads were washed to remove the majority of the contaminating wheat germ proteins. Bovine brain cytosol was added to the washed ␣-synuclein-containing beads. After incubation, samples were subjected to cross-linking as indicated in Fig. 2. Beads were washed, and then both cross-linked and uncross-linked ␣-synuclein were released from the beads by the addition of Laemmli sample buffer containing EDTA. Cross-links were analyzed by SDS-PAGE on 12% gels followed by fluorography. The radiolabel is in ␣-synuclein, and bands migrating more slowly than ␣-synuclein represent cross-links of brain cytosolic proteins to ␣-synuclein. The uncross-linked ␣-synuclein runs near the dye front at ϳ14 kDa. Three cross-links were seen with estimated molecular masses of 30 -35 kDa. After subtracting the mass of synuclein itself, the cross-links represent proteins of ϳ16 -19 kDa. The cross-links were dependent on the addition of brain cytosol, the presence of ␣-synuclein mRNA (not shown), and irradiation. In human ␣-synuclein, the wildtype amino acid at position 53 is Ala, whereas in rodents and birds, it is Thr. However, in humans, Thr at position 53 is a Parkinson's disease-linked mutation. Whereas peptide sequence from the bovine ␣-synuclein is known, the full cDNA sequence is not known. Since it is therefore unclear whether residue 53 of the bovine protein is Ala or Thr, we tested both versions in our assay. The cross-links were observed to the WT (Fig. 2, lanes 1-3) and the mutant A53T (Fig. 2, lanes 4 -6) ␣-synuclein proteins.
We bound our substrate to beads so that we can wash away the wheat germ cytosol and replace it with brain cytosol. To determine whether immobilization altered the spectrum of detectable cross-links, we analyzed the cross-links to soluble synuclein (Fig. 2, lanes 7 and 8). After in vitro translation of the ␣-synuclein mRNA in the presence of [ 35 S]Met and TDBA-Lys-tRNA, cycloheximide was added to 0.5 mM to stop translation. Next, 20 ml of bovine brain cytosol or buffer was added directly to 5-ml aliquots from the translation as indicated in Fig. 2. After 10 min at room temperature, samples were irradiated as indicated. Samples were trichloroacetic acid-precipitated and resuspended in sample buffer, after which cross-links were analyzed by fluorography after SDS-PAGE. As was the case with the experiment done on beads, the ϳ30 -35-kDa crosslinks were also evident when synuclein was in solution. However, additional cross-links with apparent molecular masses of ϳ45 and ϳ85 kDa were also observed. The simplest explanation for the difference is that immobilization on Co 2ϩ chelating agarose beads sterically blocks these larger proteins from cross-linking to ␣-synuclein.
Purification of the Cross-links from Bovine Brain Cytosol-We used the cross-linking assay to purify the 15-19-kDa proteins from bovine brain cytosol using a standard biochemical approach. Since immobilization of the synuclein on beads resulted in only detecting the smaller cross-links, gave cleaner gels, and resulted in greater cross-linking efficiencies, we used this version of the assay for our purification. Aliquots from each step of the purification or each column fraction were assayed for the ability of proteins to cross-link to in vitro translated ␣-synuclein. Fig. 3a shows the results of the cross-linking assay, whereas Fig. 3b shows the Coomassie-stained gels of the same fractions. The 15-19-kDa proteins remain soluble after heating the cytosol and did not bind to SP-Sepharose but were quantitatively absorbed to Q-Sepharose. After washing at 0.4 M KCl, the bound proteins were eluted with a 0.5-1.0 M KCl gradient. The peak of activity eluted at 500 -525 mM KCl. Coomassie staining of the peak fractions revealed a major band in the 15-19-kDa range (Fig. 3b). The band indicated by an asterisk was excised from the gel, digested with trypsin, and processed for mass spectrometric analysis. The protein was identified as calmodulin and was verified by comparing the computer-generated fragment ion series of the predicted tryptic peptide with the experimental MS/MS data.
Purified CaM Cross-links to ␣-Synuclein in a Calcium-dependent Manner-To confirm that we had excised and identified the protein that was giving rise to the observed cross-links, we added commercially purified CaM (Ͼ98% pure) to crosslinking assays. His-tagged WT ␣-synuclein was translated in the presence of radiolabel and TDBA-Lys-tRNA and isolated on beads, and purified CaM was added at the concentrations indicated in Fig. 4a prior to cross-linking. Purified CaM indeed cross-links to ␣-synuclein and results in a pattern of three cross-links similar to that seen when complete BBC is assayed (Fig. 4a, lanes 3 and 4 -7), suggesting that all three cross-links observed in the brain cytosol arise from the interaction with CaM. Half-maximal cross-linking intensity occurred at ϳ0.5 mM CaM, although cross-links were detectable at lower concentrations. It should be pointed out that cross-linking could significantly underestimate the apparent affinity of CaM for synuclein (see Fig. 6). Additionally, when the calmodulin inhibitor calmidizol is added prior to cross-linking, cross-links to CaM are not observed when using either crude BBC (lane 9) or purified CaM (lane 8). The CaM preparation used has a sufficient amount of the protein in the calcium-bound conformation to allow for visualization of the cross-link in the absence of added calcium (lane 10). However, the addition of Ca 2ϩ to the reaction resulted in a more intense cross-link (lanes 11-13).

FIG. 2. Detection of photocross-links between brain cytosolic
proteins and human ␣-synuclein. mRNAs encoding octahistidine tagged WT or A53T mutant human ␣-synucleins were translated in vitro in a wheat germ lysate supplemented with TDBA-Lys-tRNA and [ 35 S]Met. After translation, synuclein with the photoprobes incorporated was isolated on chelating Co 2ϩ -agarose beads as described under "Experimental Procedures" (lanes 1-6), or translation mixtures were used directly (lanes 7 and 8). Brain cytosol was added as indicated (lanes 2 and 3, 5 and 6, and 7 and 8) either to washed synucleincontaining beads or to crude translation mixtures (lanes 7 and 8). After incubation for 5-10 min at room temperature, samples were irradiated to induce cross-linking (lanes 1, 3, 4, 6, and 8). Synuclein and synuclein photoadducts were released from the beads by the addition of Laemmli sample buffer containing EDTA. Where no beads were used (lanes 7 and 8), proteins were trichloroacetic acid-precipitated and resuspended in sample buffer. Cross-links were analyzed by fluorography after SDS-PAGE. Synuclein runs near the dye from at ϳ14 kDa (all lanes), and cross-links (indicated by asterisks) between brain proteins and radiolabeled ␣-synuclein are seen at ϳ35 kDa only in the presence of cytosol and upon irradiation (lanes 3 and 6). Larger cross-links are also observed (lane 8) when synuclein is not bound to the beads.
Cross-linking to ␣-synuclein requires the presence of calcium. Fig. 4b shows that the CaM cross-links normally seen in BBC cytosol are not observed if the cytosol is EGTA-treated and subjected to dialysis prior to use in the cross-linking assay. The addition of CaCl 2 to the calcium-depleted cytosol restores the CaM cross-links, showing that the interaction is calciumdependent.
Our assay utilizes a slightly truncated and His-tagged version of synuclein. We therefore wanted to know whether fulllength, nontagged synuclein can interact with CaM. To this end, full-length WT ␣-synuclein was translated in a wheat germ lysate supplemented with [ 35 S]Met and TDBA-Lys-tRNA (Fig. 5a). After translation, purified CaM was added to 1 mM prior to cross-linking. The pattern of three CaM cross-links is indistinguishable from that obtained with octahistidine-tagged and immobilized synucleins, indicating that His tagging and bead binding did not introduce readily apparent artifacts into our assay. Also, the cross-links were not observed in the presence of calmidizol (not shown) and were blocked by EGTA.
CaM Cross-links to Parkinson's Disease-associated Mutant Synucleins-The preceding experiments showed that CaM cross-links to both WT and A53T ␣-synucleins (Fig. 2). Fig. 5a shows that A30P ␣-synuclein also cross-links to CaM. ␣-Synuclein is a phosphoprotein with Ser 87 and Ser 129 being identified as the major sites of phosphorylation (56). Ser 129 is the major target of GRK5, a kinase that phosphorylates Gprotein-coupled receptors and synucleins (9). Mutating both residues to Ala to prevent phosphorylation or to Asp to mimic phosphorylation did not alter the ability to cross-link to CaM (Fig. 5b).
Recombinant ␣-Synuclein Binds CaM-To measure synuclein-CaM binding independently of cross-linking, we used an established fluorescence assay exploiting changes in the fluorescence of dansyl-labeled CaM upon Ca 2ϩ and substrate binding (57) (Fig. 6a). The yellow spectrum in Fig. 6a shows the emission spectrum of dansyl-CaM in the presence of EDTA. Upon the addition of CaCl 2 (dark blue), there is a slight increase in emission intensity, and the emission maximum is blue-shifted from 505 to 490 nm. The addition of recombinant SS 87,129 DD synuclein (0.16 mM (pink), 0.5 mM (purple), or 0.8 mM (red)) resulted in increased fluorescence emission in a dosedependent manner. The addition of 25 mM calmidizol (a competitive CaM inhibitor) but not 4 mM RNase A increased fluorescence intensity (not shown), indicating that the increase in fluorescence required a CaM substrate. When EDTA was added to a sample containing 0.16 mM synuclein (light blue), FIG. 3. Purification of ␣-synuclein cross-linking partners from brain cytosol. a, aliquots taken from sequential steps from the purification scheme were analyzed for the ability to cross-link to synuclein. The three cross-linking partners co-purify and are found in the supernatant after a heating step and did not bind to SP-Sepharose but did bind to Q Sepharose and eluted at around 500 -530 mM KCl. In lanes 3-7, aliquots corresponding to 5 ml of starting material were assayed. In lanes 8 -13, 5-ml aliquots of 2.6-ml fractions were assayed. b, Coomassie Blue-stained gel of fractions shown in a. In lane 1, 30 mg of brain cytosol was loaded, and the corresponding amount of material was loaded in lanes 2-5. In lanes 6 -11, 50 ml of 2.6-ml fractions was loaded after proteins were concentrated by trichloroacetic acid precipitation. The major band as marked by an asterisk in lane 7 was excised from a duplicate gel and identified by mass spectrometry as calmodulin. c, mass spectrum of trypsin digest of the excised band.
FIG. 4. Purified CaM cross-links to ␣-synuclein. a, to verify that the band excised from the gel and identified as CaM is indeed the protein giving rise to the ␣-synuclein cross-links, His-tagged WT ␣-synuclein mRNA was translated in vitro in the presence of [ 35 S]Met and TDBA-Lys-tRNA, and then synuclein was isolated on beads. After washing, brain cytosol, commercially purified bovine brain CaM, CaCl 2 , and calmidizol (a CaM antagonist) were added as indicated. After 10 min at room temperature, samples were subjected to irradiation to induce cross-linking as indicated. The three cross-links coming from brain cytosol (lane 3) were also observed when CaM alone was added (lanes 4 -7). Calmidizol prevented detection of the cross-links in brain cytosol (lane 8) and when purified CaM was used (lane 9). Residual Ca 2ϩ in the CaM preparation allowed for detection of cross-links in the absence of added CaCl 2 (lane 10), but supplementation with exogenous CaCl 2 resulted in a more intense cross-link (lanes 11-13; see text). b, calcium dependence of CaM-synuclein interaction. Brain cytosol was treated with EGTA and then dialyzed to remove EGTA and calcium. The treated cytosol was then assayed for the ability of CaM to cross-link to bead-isolated ␣-synuclein. there was a decrease in fluorescence and a reversal of the blue shifting of the emission maximum. Fig. 6b shows that A53T synuclein binds to dansyl-CaM in a saturable manner (Fig. 6b). To estimate the dissociation constant for the binding of dansyl-CaM to synuclein, increasing amounts of A53T ␣-synuclein were added to cuvettes containing dansyl-CaM at a final concentration of 86 nM (Fig. 6, b and  c). The fraction of free and bound dansyl-CaM at each concentration was determined as described under "Experimental Procedures," and from these numbers a Scatchard plot (Fig. 6c) was constructed. With the caveats discussed below, we estimate the K d to be about 20 nM, and the number of binding sites was estimated to be 82 nM, in good agreement with the 86 nM dansyl-CaM concentration determined by Lowry protein assay. Ideally, we would have used dansyl-CaM concentrations near or below the K d . However, we were not able to label the dansyl-CaM to high enough specific activity where we could use lower concentrations and reliably detect changes in fluorescence. Similarly, although we wished to include synuclein concentrations below 25 nM in the construction of the curves in Fig. 6, b and c, we were unable to measure changes in dansyl-CaM fluorescence reliably at synculein concentrations below 25 nM. Fig. 6d shows that recombinant WT ␣-synuclein binds to CaM-Sepharose in a calcium-dependent manner. Together, these results show that synuclein is a CaM substrate and that CaM-synuclein interaction can be detected by methods other than cross-linking.
Synuclein and CaM Interact in a Calcium-dependent Manner in Intact Cells-Next we addressed the question of whether ␣-synuclein and CaM expressed at endogenous, physiologic levels in nontransfected cells interact in intact cells in a calciumdependent manner. In preliminary experiments, we found that DSS, a membrane-permeable amine-reactive homobifunctional cross-linker, cross-links CaM to recombinant synuclein in a calcium-dependent manner (not shown). Therefore, SK-N-SH neuroblastoma cells were treated with DSS in the absence or presence of a calcium ionophore and with and without calmidizol, a CaM antagonist. After incubation with the crosslinker, the reaction was quenched with an excess of buffered glycine before cell lysis. Synuclein immunoprecipitates containing both uncross-linked and cross-linked ␣-synuclein were made and subjected to SDS-PAGE, and, after transfer to nitrocellulose, blots were probed with an anti-CaM monoclonal antibody. Two cross-links immunoreactive for both ␣-synuclein and CaM were observed in a DSS-and ionophore-dependent manner (Fig. 7, lane 3). Note that these cross-links were not observed in the presence of a calmodulin antagonist (lane 4). The ϳ37-kDa band migrates at the correct position for the Syn-CaM cross-link. One other cross-link migrating more slowly was also observed at ϳ66 kDa (marked by an asterisk). Since DSS potentially cross-links several proteins together, this cross-link could represent an a synucein-CaM complex of multiple stoichiometries. Alternatively, as yet unidentified proteins could be present in the complex, thereby giving rise to the larger cross-link.
Notice that uncross-linked CaM was co-immunoprecipitated with ␣-synuclein (lane 3). While it is not surprising that this only occurred in the presence of ionophore, the presence of CaM in the synuclein IP also required DSS and was not observed in the presence of ionophore (lane 1) or DSS (lane 2) alone. Calmidizol prevented the co-precipitation (lane 4).
There are at least two potential explanations for why DSS could increase the co-immunoprecipitation efficiency. First, whereas synuclein and CaM alone can interact, CaM might bind with increased efficiency or affinity to synuclein in the presence of a third factor or may bind better to oligomers of synuclein. Perhaps DSS stabilized either of these apparently labile complexes, thereby providing a better or more abundant CaM substrate. Whereas DSS may have entrapped some CaM in these complexes (perhaps giving rise to the band marked by an asterisk), CaM that bound but did not cross-link was recovered as co-precipitating CaM.
A more trivial explanation for why DSS allows for co-immunoprecipitation of synuclein is that, before lysis, DSS derivatized synuclein and that this results in creation of an enhanced CaM substrate. We ruled out this possibility by showing that DSS-derivatized synuclein does not bind to CaM-Sepharose and in fact is a worse CaM substrate (not shown). Thus, ␣-synuclein and CaM can interact in intact cells in a calciumdependent manner.
Calcium-activated CaM Accelerates Synuclein Fibril Formation-Although the function of ␣-synuclein is unknown, its conformational status is closely linked to Parkinson's disease.  8, and 12). After 10 min at room temperature, samples were irradiated as indicated. Cross-links were resolved by SDS-PAGE and detected by fluorography. CaM cross-linked to all of the synucleins tested, and cross-links were calcium-dependent. b, CaM cross-linked to both S87A,S129A and S87D,S129D mutant synucleins.
Purified ␣-synuclein assembles into amyloid fibrils that are morphologically and biochemically similar to those found in Lewy bodies. Because CaM often induces conformational changes in its substrates and because synuclein's secondary structure changes when it forms fibrils, we assessed whether Ca 2ϩ /CaM could alter the kinetics of ␣-synuclein fibril formation. ␣-Synuclein fibrils are easily separated from soluble monomeric synuclein by ultracentrifugation (23). Reactions were mixed according to the legend for Fig. 8. The synuclein used was precentrifuged to remove any preformed aggregates before the start of the assay. At each time point, aliquots were removed, diluted, and centrifuged to pellet fibrils. The amount of soluble synuclein remaining in the supernatant fractions was maximum (pale blue). Excitation was at 338 nm, and emission spectra were collected from 425 to 600 nm. Slit widths were 7.5 nm (excitation) and 5.0 nm (emission). b, ␣-synuclein binds dansyl-CaM in a saturable manner. The indicated concentrations of recombinant A53T ␣-synuclein were added to cuvettes containing dansyl-CaM at a final concentration of 86 and 2.5 mM CaCl 2 in PBS. The fluorescence emission at 486 nm was measured. Data from triplicate samples were averaged and plotted as F (fluorescence intensity with synuclein)/F o (fluorescence intensity in the absence of synuclein) versus synuclein concentration. Excitation was at 338 nm, and emission spectra were collected from 425 to 600 nm. Slit widths were 10 nm. Error bars denote the S.E. C, quantitative analysis of the binding of dansyl-CaM to A53T ␣-synuclein. The data from b above was used to create a Scatchard plot. For each concentration of synuclein, the fraction of dansyl-CaM bound to synuclein was calculated from the fractional increase in fluorescence as described under "Experimental Procedures." Data are expressed as the mean Ϯ S.E. from three independent experiments. We estimated the K d to be ϳ20 nM from the negative reciprocal of the slope of the Scatchard plot, and the number of binding sites was estimated to be 82 nM. d, WT ␣-synuclein binds to CaM-Sepharose. WT ␣-synuclein (120 ml of 0.5 mM) in PBS plus 0.1 mM CaCl 2 was incubated with 15 ml of CaM-Sepharose (Amersham Biosciences) for 30 min at room temperature. The flow-through was collected, and beads were washed three times with 200 ml of binding buffer. Specifically bound synuclein was eluted with 200 ml of PBS containing 5 mM EDTA. Fractions were subjected to trichloroacetic acid precipitation, and ␣-synuclein content was determined by Western blotting. detected by Western blotting (Fig. 8a). The decrease in synuclein solubility proceeded more rapidly in the presence of calcium-activated CaM (lanes 1-5) than in the presence of EGTA (lanes 6 -10). Western blotting with CaM antibodies shows that the CaM remained soluble. We were therefore not observing bulk protein precipitation or protein degradation. Synuclein, including higher molecular weight forms, was recovered from pellet fractions but was not quantitatively solubilized with SDS and recovered on gels, a known property of amyloid fibrils (not shown). Neither calcium (50 mM) nor EGTA (1 mM) alone had any effect on synuclein fibril formation under our conditions (not shown). Fig. 8b presents a quantitative analysis of the aggregation assay. Synuclein fibril formation is a nucleation-dependent reaction and proceeds through a protofibrillar intermediate. Further work will be devoted to figuring out precisely where and how CaM affects fibril formation. DISCUSSION We utilized a photocross-linking strategy to identify potential ␣-synuclein-binding proteins that may have eluded identification by other methodologies. We show that several brain cytosolic proteins are in contact with synuclein and identified CaM as an ␣-synuclein-binding protein. We employed a photocross-linking technique heretofore not utilized in investigating neurodegenerative diseases. With this technology, crosslinks are detected only if the two proteins are contacting one another during the instant the cross-linker is active. An additional advantage is that biochemical environment can be altered to assess what effect such alterations have on the interaction under investigation. Furthermore, any interactions detected are occurring in extracts made from tissues expressing endogenous levels of all proteins in question. We show that the interaction between synuclein and CaM is calcium-dependent and is unaffected by Parkinson's disease-related mutations in synuclein. Also, the interaction could be revealed in intact cells expressing endogenous levels of both CaM and ␣-synuclein in a calcium-dependent manner and could be blocked with a CaM inhibitor.
Whereas one recent report has identified a low affinity calcium binding domain in synuclein (58), this region, comprising the last ϳ30 residues of ␣-synuclein, is probably dispensable in our assay, since our His-tagged version of synuclein, which binds CaM, lacks the C-terminal 20 residues. Furthermore, there are no lysine residues in this region to give rise to the cross-links (see Fig. 1). Although some metal ions can alter synuclein secondary structure and propensity to aggregate, the Lansbury laboratory (32) has published that low millimolar concentrations (10 mM) of calcium ions do not directly alter synuclein secondary structure as assessed by CD spectroscopy. We have demonstrated that CaM and synuclein interact directly in a calcium-dependent manner. The apparent affinity constant of CaM for synuclein is ϳ20 nM. CaM and synuclein are abundant enough to make this a physiologically relevant interaction. Intracellular CaM concentrations have been estimated to be 1-10 mM, and CaM is estimated to be 0.1-1% of total protein in neurons (59). Since the BBC protein concentration is 4 mg/ml (which is significantly diluted compared with intracellular cytosol concentrations), the CaM concentration is likely to be 0.004 -0.04 mg/ml, which corresponds to at least 0.24 -2.4 mM. Therefore, CaM concentrations are high enough to allow for physiologically relevant interactions with ␣-synuclein.
Intriguingly, ␣-synuclein alters stimulation-dependent dopamine release, and calcium levels modulate this function of synuclein (31). Because synuclein is regulated by calcium and we independently purified CaM as a cross-linking partner of synuclein, we hypothesize that CaM may mediate the effect of calcium on the previously observed ability of synuclein to alter dopamine release.
Synucleins are also substrates for GRK 5, a G-protein-coupled receptor kinase (9). This kinase, which is involved in ligand-dependent receptor desensitization, is inhibited by CaM. Whereas CaM inhibits GRK5 autophosphorylation and phosphorylation of other substrates in the absence of synuclein, CaM actually stimulates kinase autophosphorylation and synuclein phosphorylation in the presence of synuclein. Thus, synuclein may act as a switch to convert CaM from an inhibitor to an activator of GRK.
The fact that we observed several species that were both CaM-and synuclein-immunoreactive (Fig. 7) only with calcium and not when the CaM inhibitor was used, indicates that CaM and synuclein may form higher order complexes in the presence of Ca 2ϩ . It is tempting to speculate that Ca 2ϩ /CaM drives assembly of synuclein-containing multimeric complexes or perhaps regulates the oligomerization status of synuclein. There is precedence for CaM driving oligomeric assembly of subunits of small conductance K ϩ channels (60). Our in vitro fibril formation data is at least consistent with this idea. Further work will be directed toward elucidating the mechanism by which CaM alters synuclein's ability to change conformation.  6 -10). Samples were incubated at 37°C without agitation. At the indicated times, aliquots were removed, diluted 10-fold, and centrifuged (Beckman TL100 rotor, 50,000 rpm, 4°C, 15 min, 100,000 ϫ g av ) to sediment synuclein fibrils. The fibril-depleted supernatants were analyzed for synuclein and CaM content by Western blotting. The synuclein stock was precentrifuged to remove any preformed fibrils or aggregates before the start of the experiment. b, quantitation of fibril formation assay. The amount of soluble synuclein remaining at each time point was quantitated by densitometry using Image J software and normalized to the initial amount of soluble synuclein present at the beginning of the experiment. The mean values obtained from three experiments are shown. The error bars denote the S.E.