Identification of Protein Interfaces between α-Synuclein, the Principal Component of Lewy Bodies in Parkinson Disease, and the Molecular Chaperones Human Hsc70 and the Yeast Ssa1p*

Background: The mechanism by which molecular chaperones sequester α-synuclein is unknown. Results: We identify the surface interfaces involved in Hsc70 and Ssa1p interaction with α-synuclein. Conclusion: Hsc70 and Ssa1p bind α-synuclein like a tweezer through the two tips of their client protein binding sites. Significance: Elucidating how molecular chaperones bind proteins involved in neurodegenerative diseases is relevant to design therapeutic tools. Fibrillar α-synuclein (α-Syn) is the principal component of Lewy bodies, which are evident in individuals affected by Parkinson disease (PD). This neuropathologic form of α-Syn plays a central role in PD progression as it has been shown to propagate between neurons. Tools that interfere with α-Syn assembly or change the physicochemical properties of the fibrils have potential therapeutic properties as they may be sufficient to interfere with and/or halt cell-to-cell transmission and the systematic spread of α-Syn assemblies within the central nervous system. Vertebrate molecular chaperones from the constitutive/heat-inducible heat shock protein 70 (Hsc/p70) family have been shown to hinder the assembly of soluble α-Syn into fibrils and to bind to the fibrils and very significantly reduce their toxicity. To understand how Hsc70 family members sequester soluble α-Syn, we set up experiments to identify the molecular chaperone-α-Syn surface interfaces. We cross-linked human Hsc70 and its yeast homologue Ssa1p and α-Syn using a chemical cross-linker and mapped the Hsc70- and Ssa1p-α-Syn interface. We show that the client binding domain of Hsc70 and Ssa1p binds two regions within α-Syn similar to a tweezer, with the first spanning residues 10–45 and the second spanning residues 97–102. Our findings define what is necessary and sufficient for engineering Hsc70- and Ssa1p-derived polypeptide with minichaperone properties with a potential as therapeutic agents in Parkinson disease through their ability to affect α-Syn assembly and/or toxicity.


Fibrillar ␣-synuclein (␣-Syn) is the principal component of Lewy bodies, which are evident in individuals affected by Parkinson disease (PD). This neuropathologic form of ␣-Syn plays a central role in PD progression as it has been shown to propagate between neurons.
Tools that interfere with ␣-Syn assembly or change the physicochemical properties of the fibrils have potential therapeutic properties as they may be sufficient to interfere with and/or halt cell-to-cell transmission and the systematic spread of ␣-Syn assemblies within the central nervous system. Vertebrate molecular chaperones from the constitutive/heatinducible heat shock protein 70 (Hsc/p70) family have been shown to hinder the assembly of soluble ␣-Syn into fibrils and to bind to the fibrils and very significantly reduce their toxicity. To understand how Hsc70 family members sequester soluble ␣-Syn, we set up experiments to identify the molecular chaperone-␣-Syn surface interfaces. We cross-linked human Hsc70 and its yeast homologue Ssa1p and ␣-Syn using a chemical cross-linker and mapped the Hsc70-and Ssa1p-␣-Syn interface. We show that the client binding domain of Hsc70 and Ssa1p binds two regions within ␣-Syn similar to a tweezer, with the first spanning residues 10 -45 and the second spanning residues 97-102. Our findings define what is necessary and sufficient for engineering Hsc70-and Ssa1p-derived polypeptide with minichaperone properties with a potential as therapeutic agents in Parkinson disease through their ability to affect ␣-Syn assembly and/or toxicity.
Various factors, including genetic susceptibility and environmental influences, can trigger the assembly of the protein ␣-synuclein (␣-Syn) 3 from its soluble state into fibrils with high ␤-sheet content (1,2). The latter form is one of the principal components of intracellular Lewy bodies, whose presence in the central nervous system is a defining feature of Parkinson disease (PD) and other synucleinopathies (3). Indeed, symptoms of PD are apparent after Ͼ70% of dopaminergic termini and/or neurons have been lost, with some of the remaining neurons containing filamentous ␣-Syn.
We and others (4 -6) recently contributed evidence for intercellular propagation of ␣-Syn assemblies in vivo. This process has been proposed to contribute to the progressive spreading of ␣-synucleinopathies throughout the nervous system (7,8). Thus, tools that inhibit ␣-Syn assembly into fibrils, which we recently showed to be the most toxic species (9), or change the physicochemical properties of the fibrils in such a way that the cell-to-cell propagation of ␣-Syn aggregates is affected, are most likely expected to have therapeutic potential in PD.
Molecular chaperones facilitate the folding of newly synthesized polypeptides (10). They also inhibit the aggregation of misfolded and stress-induced unfolded proteins by binding to exposed hydrophobic regions on non-native proteins and holding the proteins (11,12). The expression of members of the Hsp70 family (13)(14)(15)(16) is induced when a cellular environment is under stress such as heat shock, ischemia, or other stresses (17). This prompted us and others to document the effect of Hsc/p70 on ␣-Syn assembly in vitro and in vivo (18,19). Hsc/p70 have been shown to inhibit ␣-Syn fibril formation. Furthermore, a reduction in the toxicity of fibrillar ␣-Syn upon coating the fibrils with Hsc70 was observed. These findings encouraged us to consider the use of Hsc/p70 as a potential therapeutic tool in PD (18,20).
Here, using chemical cross-linking with the homobifunctional NHS-ester BS2G, we show evidence for an Hsc70-␣-Syn complex. We map the surface interface between Hsc70 and ␣-Syn after identification of the cross-linked polypeptides by nanoLC-MS/MS LTQ-Orbitrap analyses.
As the introduction to the body of an endogenous protein as a therapeutic agent may evoke an autoimmune response as described (21)(22)(23)(24)(25), we also documented the interaction of a polypeptide that may have an effect on ␣-Syn assembly and/or the toxicity similar to Hsc70 although exogenous to the human body: the yeast homologue of Hsp70, Ssa1p.
We show here that similarly to Hsc/p70, Ssa1p inhibits ␣-Syn assembly by sequestering it in an assembly incompetent state. We determined the affinity of Ssa1p for ␣-Syn and show that the apparent dissociation constant is comparable to that we measured between human Hsc70 and soluble ␣-Syn. Finally, we show evidence for an Ssa1p-␣-Syn complex using chemical cross-linking with the homobifunctional NHS-ester BS2G, and we map the surface interface between Ssa1p and ␣-Syn after identification of the cross-linked polypeptides by nanoLC-MS/MS LTQ-Orbitrap analyses.
Assembly of ␣-Syn into Fibrils-Soluble wild-type ␣-Syn was assembled in 50 mM Tris-HCl, pH 7.5, 150 mM KCl, in the absence or presence of Ssa1p, with or without ADP-MgCl 2 or ATP-MgCl 2 at 37°C under magnetic stirring. The assembly of ␣-Syn with each Ssa1p concentration was performed in triplicate. Aliquots (90 l) were removed at time intervals, spun at 40,000 ϫ g, 20°C, and 20 min in a TL100 tabletop ultracentrifuge (Beckman Coulter). The proteins within the supernatant and pellet fractions were analyzed by SDS-PAGE and quantified following staining/destaining using NIH ImageJ software (available at rsb.info.nih.gov/ij). The nature of the oligomeric species was assessed using a Jeol 1400 transmission electron microscope (Jeol, Ltd.) following adsorption of the samples onto carbon-coated 200-mesh grids and negative staining with 1% uranyl acetate. The images were recorded with a Gatan Orius CCD camera (Gatan).
Peptide Preparation-The protein bands corresponding to ␣-Syn, Hsc70, Ssa1p, and ␣-Syn after treatment with BS2G, Hsc70, or Ssa1p after treatment with BS2G, and the Hsc70-␣-Syn BS2G cross-linked complex and Ssa1p-␣-Syn BS2G crosslinked complex were excised from the SDS-PAGE separation. Each protein band was subjected to enzymatic digestion using the Progest robot (Genomic Solutions, Chemsford, MA). Briefly, protein bands were extensively washed with acetronitrile and 25 mM NH 4 HCO 3 , before reduction and alkylation of cysteine residues in the presence of 10 mM DTT and 55 mM of iodoacetamide. Trypsin (Promega Gold; Promega, Madison, WI) digestion was performed overnight at room temperature in 25 mM NH 4 HCO 3 . The volume of the trypsin solution (10 ng/l) was adapted to the size of the protein band excised from the gel (25 l per 5 mm 3 ). Peptides were extracted by incubation during 2 h at room temperature after addition of two volumes of 60% CH 3 CN, 0.1% HCOOH. The extracted peptides were vacuum dried, dissolved in water and stored at Ϫ20°C until MS analysis.
NanoLC-Linear Ion Trap (LTQ)-Orbitrap Mass Spectrometry Analysis-Tryptic peptide digests were analyzed by nanoLC-MS/MS using an EASY-nLC II high performance liquid chromatography (HPLC) system (Proxeon, Thermo-Scientific, Waltham, MA) coupled to the nanoelectrospray ion source of an LTQ-Orbitrap Velos mass spectrometer (Thermo Scientific). Peptide separation was performed on a reversedphase C18 nano HPLC column (100 m inner diameter, 5-m C18 particles, 15-cm length, NTCC-360/100-5) from Nikkyo Technos (Nikkyo Technos Co., Ltd., Tokyo, Japan). The peptides were loaded at a pressure-dependent flow rate corresponding to a maximum pressure of 200 bars and eluted at a flow rate of 300 nl/min using a gradient of 5 to 35% solvent B for 60 min, followed by a washing step at 100% solvent B. Solvent A was 0.1% formic acid in water, and solvent B was 0.1% formic acid in 100% acetonitrile. NanoLC-MS/MS experiments were conducted in the data-dependent acquisition mode. The mass of the precursors was measured with a high resolution (60,000 full weight at half maximum) in the Orbitrap. The 20 most intense ions, above an intensity threshold of 5000 counts, were selected for CID fragmentation and analysis in the LTQ.
NanoLC-MS/MS data were processed automatically using the Thermo Proteome Discoverer software (version 1.3) and the Sequest search engine with the following chemical modifications: methionine oxidation, cysteine carbamidomethylation, and monolink or loop link modification of lysine residues with BS2G-d0 or BS2G-d4. NanoLC-MS data were further ana-lyzed as described previously (31). Briefly, nanoLC-MS data were deisotoped using the Decon2LS software (http://omics. pnl.gov/software/Decon2LS.php). The resulting csv files were further analyzed with viper (http://omics.pnl.gov/software/ VIPER.php) (32) to identify, within the nanoLC-MS analysis, the d0/d4 peptide pairs corresponding to mass differences of 4.0247 Da as a signature of peptides that reacted with at least one molecule of BS2G-d0/d4. A list of peptide pairs with a maximum mass tolerance of 10 ppm was generated. Mass deviation and peptide elution time were used to filter the list of peptide pairs, corresponding to candidate cross-linked peptides or mono-linked or loop-linked peptides. The list of light and heavy precursor masses was used to identify the BS2G-d0/d4 peptides using GPMAW (33) and XQUEST (34) software. MS/MS spectra of the candidate cross-linked peptides were further analyzed manually to confirm their identification.

RESULTS
Interaction between Hsc70 and Soluble ␣-Syn-We recently demonstrated that Hsc70 sequesters ␣-Syn in an assembly incompetent state (18). We show here ( Fig. 1A) that the Hsc70-␣-Syn complex is cross-linked by the homobifunctional NHSester BS2G. BS2G reacts with the ⑀-amino group of lysine residues and the ␣-amino group from protein N termini and, to a lesser extent, with the hydroxyl groups of serine, threonine, and tyrosine residues (35) and has a spacer arm 7.7 Å long. Thus, the residues involved in the interaction within the partner protein complex must be ϳ7.7 Å away from each other for cross-linking to occur. The affinity of Hsc70 for its client polypeptides is modulated by nucleotides (ATP or ADP). The most efficient sequestering activity was observed for Hsc70 in the absence of nucleotide. Hsc70 inhibits to a lesser extent ␣-Syn assembly in the presence of ATP (18). This is why cross-linking was performed in the absence of nucleotide. No inhibition ␣-Syn assembly is observed in the presence of ADP (18). Consistent with that no Hsc70-␣-Syn complex with apparent molecular mass of 100 kDa was observed in the presence of BS2G and ADP.
The Interaction between Ssa1p and Soluble ␣-Syn-In a manner similar to what we observed upon incubation of Hsc70 with ␣-Syn, when Ssa1p and ␣-Syn are incubated in the presence of BS2G, an abundant Ssa1p-␣-Syn complex with an apparent molecular mass of ϳ100 kDa forms (Fig. 1B), suggesting that Ssa1p binds ␣-Syn.
The effect of Ssa1p on ␣-Syn assembly was assessed using a sedimentation assay and electron microscopy (EM). Over time, ␣-Syn readily assembles into high molecular weight assemblies in vitro, which sediment upon centrifugation. Assembly follows a typical sigmoidal pattern, with nucleation, growth, and steady state phases. In the absence of Ssa1p, the nucleation phase of ␣-Syn at 200 M is ϳ200 min. Upon addition of increasing concentrations of Ssa1p, the length of the nucleation phase increases, indicating that the yeast chaperone slows down ␣-Syn assembly (Fig. 1C). EM images of the samples at 1200 min show that there are less fibrils and more non fibrillar oligomeric species upon increasing the concentration of Ssa1p from 0 to 10 M Ssa1p (Fig. 1D). The lag time preceding assembly of ␣-Syn fibrils in the presence of increasing concentrations of Ssa1p was measured and plotted as a function of Ssa1p concentration and appeared to increase nearly linearly with increasing Ssa1p concentrations (Fig. 1E). These results there- fore show that Ssa1p affects ␣-Syn assembly in a manner similar to Hsc70 (18).
Similarly to Hsc70, Ssa1p also binds to preformed ␣-Syn fibrils. Indeed, as shown in Fig. 1F, Ssa1p is found in the pellet after coincubation with preformed ␣-Syn fibrils, whereas Ssa1p is in the supernatant of control samples spun under identical conditions in the absence of ␣-Syn fibrils. Binding of Ssa1p to fibrillar ␣-Syn reduces significantly the toxicity of ␣-Syn fibrils as assessed using the MTT assay and cell survival upon treatment of the cells for 24 h with the assemblies (Fig. 2A). This parallels a reduced activation of caspase-3 (Fig. 2B).
Identification of the Hsc70 or Ssa1p and ␣-Syn Cross-linked Polypeptides-The analytical strategy used to characterize the polypeptides involved in the Hsc70-␣-Syn and Ssa1p-␣-Syn interaction was described previously (31) and relies on the use of a mixture of the non-deuterated (d0) and deuterated (d4) forms of BS2G. Untreated, BS2G-treated Hsc70, Ssa1p and ␣-Syn and the BS2G cross-linked Hsc70-␣-Syn and Ssa1p-␣-Syn complexes resolved by SDS-PAGE were cleaved by trypsin and the protein sequence coverage determined by the nanoLC-MS/MS analyses described under "Experimental Procedures." The sequence coverage was 94, 98, and 100% for untreated Hsc70, Ssa1p, and ␣-Syn, respectively, 78, 92, and 100% for Hsc70, Ssa1p, and ␣-Syn treated with BS2G, respectively, and 100 and 78% for ␣-Syn and Hsc70 and 73 and 81% for ␣-Syn and Ssa1p, in the cross-linked Hsc70-␣-Syn (Fig. 3A) and Ssa1p-␣-Syn complexes, respectively (Fig. 3B). The modified peptides were detected by MS using the 4.0247 atomic mass unit mass difference conferred by the binding of the nondeuterated or deuterated cross-linkers (36,37). The exact mass measure-ments we obtained allowed the identification of all of these peptides using GPMAW (33) and XQUEST (34) SEPTEMBER 21, 2012 • VOLUME 287 • NUMBER 39

JOURNAL OF BIOLOGICAL CHEMISTRY 32635
We conclude from these observations that ␣-Syn interacts exclusively with the client binding domain of the molecular chaperones Hsc70 and Ssa1p. We further conclude that ␣-Syn interacts with a subset of lysines from the client polypeptide binding domain of Hsc70 (four of 25) and Ssa1p (seven of 25). The latter findings highlight the specificity of the interaction between Hsc70 or Ssa1p and ␣-Syn.
We built a three-dimensional model of Hsc70 using the structures of the ATPase and peptide binding domains, amino acid residues 1-537 of bovine Hsc70 (Uniprot entry P19120), which shares 100% primary structure identity with human Hsc70, and the NMR solution structure of the C-terminal domain (amino acids 537 to 610) of human Hsp70 (Uniprot entry P08107, sharing 69% sequence identity with Hsc70), Pro- The asterisks indicate the fragments with the BS2G-d4 cross-linker. C, the identified fragments are indicated on the cross-linked sequences. The ␣and ␤-sequences correspond to the 11-21 ␣-Syn and the 491-497 Ssa1p peptides, respectively. This cross-link involves residues Lys-12 and Lys-494 from ␣-Syn and Ssa1p, respectively. D-F, identification of the cross-link between peptide 97-102 from ␣-Syn and peptide 548 -567 from Ssa1p. D, mass spectrum of the quadruple charged cross-linked peptide with m/z 737.3781 and 738.3834 for the BS2G-d0 and BS2G-d4 peptides, respectively. E, fragmentation spectrum of the precursor ion at m/z 738.63 corresponding to the second isotope of the BS2G-d4 peptide. The identified fragments are indicated, together with their charge state. The asterisks indicate the fragments with the BS2G-d4 cross-linker. F, the identified fragments are indicated on the cross-linked sequences. The ␣and ␤-sequences correspond to the 97-102 ␣-Syn and the 548 -567 Ssa1p peptides, respectively. This cross-link involves residues Lys-97 and Lys-556 from ␣-Syn and Ssa1p, respectively.

TABLE 1 List of the identified Hsc70-␣-synuclein cross-linked peptides
For each identified cross-linked peptide in Hsc70-␣-Syn BS2G complexes, the table gives the experimental mass (M exp), the charge state (z), the m/z ratio for both the BS2G-d0 and the BS2G-d4 peptides, the amino acid segment, the cross-linked lysine residue (XL site), and the amino acid sequence for both the Ssa1p peptide and the ␣-Syn peptide. DM (ppm) corresponds to the mass deviation between the experimental mass and the theoretical mass, given by XBobCat software (34).

DISCUSSION
␣-Syn assembly into high molecular weight species is tightly associated to the pathogenesis of PD. We and others previously showed that members of the molecular chaperone family such as Hsc70, Hsp70, and Hsp90 inhibit ␣-Syn assembly into fibrils in vitro, reduce the size of ␣-Syn aggregates in vivo, and protect against ␣-Syn toxicity (4 -6). Extracellular fibrillar ␣-Syn assemblies have also been shown to be uptaken by cells exposed to these aggregates, to seed the aggregation of endogenous ␣-Syn, and to propagate from cell-to-cell (4). It has been hypothesized that the latter phenomena might contribute to the progressive spread of aggregated ␣-Syn throughout the nervous system in synucleinopathies (7,8). As molecular chap-erones sequester ␣-Syn in an assembly incompetent state and as they reduce the toxicity of fibrils most likely because they change their physicochemical properties upon binding to the fibrils, there is a possibility that they could be effective therapeutic agents in PD (18,20). We therefore assessed the surface interface between Hsc70 and ␣-Syn using chemical cross-linking and nanoLC-MS/MS LTQ-Orbitrap analyses to design through future investigations a minimal synthetic molecular chaperone containing what is necessary and sufficient for interacting with ␣-Syn. Indeed, the thorough identification of client  Ssa1p-␣-Syn BS2G complexes, the table gives the experimental mass (M exp), the charge state (z), the m/z ratio for both the BS2G-d0 and the BS2G-d4 peptides, the amino acid segment, the cross-linked lysine residue (XL site), and the amino acid sequence for both the Ssa1p peptide and the ␣-Syn peptide. DM (ppm) corresponds to the mass deviation between the experimental mass and the theoretical mass, given by XBobCat software (34).  protein binding sites was successful in designing the GroELderived minichaperone that has been shown to have high chaperone activity, independent of the central cavity or allosteric behavior. As the introduction of an endogenous protein or a fragment thereof within the body may evoke an autoimmune response, we also mapped the surface interface between the yeast homologue of Hsc/p70, Ssa1p, and ␣-Syn.
We show here that ␣-Syn cross-links to Hsc70 and Ssa1p in the presence of the homobifunctional NHS-ester BS2G to yield Hsc70-␣-Syn and Ssa1p-␣-Syn complexes with apparent molecular masses of ϳ100 kDa. ␣-Syn binds to the client protein binding sites of Hsc70 and Ssa1p. The residues involved in the interaction were identified by nanoLC-MS/MS LTQ-Orbitrap analyses of the tryptic digest of the BS2G cross-linked Hsc70-␣-Syn and Ssa1p-␣-Syn complexes. The finding that Lys-97 from ␣-Syn is exclusively found within 7.7 Å distance (the length of BS2G spacer arm) from Lys-557 and Lys-561 from Hsc70 and Lys-556 from Ssa1p which all lie within the lid of the client binding domain of the molecular chaperones, whereas Lys-10, Lys-12, Lys-21, Lys-23, Lys-32, Lys-34, and Lys-45 from ␣-Syn is found cross-linked with Lys-497 and Lys-512 from Hsc70 and Lys-446, Lys-448, Lys-494, Lys-504, Lys-528, Lys-536, or Lys-556 from Ssa1p strongly suggests that one part of ␣-Syn is specifically docked to the Ssa1p client protein binding site, whereas the other is flexible and capable of interacting with several regions within this site. Two amino acid stretches within ␣-Syn located between residues 45 and 97 have hydrophobicity scores equivalent to DNAK client peptide NRLLLTG (Protein Data Bank code 1DKX), allowing them to bind the client binding sites of Hsc70 and Ssa1p. These stretches span residues 50 -56 and 69 -79 based on Kyte and Doolittle hydrophobicity profiles. The dynamics of ␣-Syn binding to Hsc70 and Ssa1p allows ␣-Syn Lys-97 to interact with Hsc70 Lys-557 and Lys-561 and Ssa1p Lys-556, whereas the rest of the N-terminal part of ␣-Syn interacts with either the back or the bottom tip of the client protein binding site of Hsc70 and Ssa1p leading to a scheme where Ssa1p and Hsc70 appear to bind ␣-Syn as a tweezer, through the two tips of their client protein binding site (Figs. 6 and 7).
Our findings together with those where antibodies directed against ␣-Syn assembly are used (39,40) lay the foundations for a therapeutic strategy targeting soluble ␣-Syn and inhibiting its assembly into fibrils or modifying the surface properties of fibrils in such a way that they become less toxic and/or intercellular propagation is perturbed. Indeed, our results are invaluable for the design of future therapeutic tools in PD as we describe within a molecular chaperone client binding domain what is necessary and possibly sufficient to bind ␣-Syn. This should allow the engineering of a subset of the client binding site of molecular chaperones from the Hsc/p70 family that bind ␣-Syn in a manner similar to the entire chaperone or with increased affinity. The agent should ideally be based on a nonhuman protein, as small as possible while still effective, to limit the chance of autoantibodies being produced leading to autoimmune disease, similar to that seen when the drug levodopa was first introduced to alleviate PD symptoms (22)(23)(24)(25). The proposed model is based on the identification of the residues involved in the interaction between the heat shock protein Ssa1p and ␣-Syn, probed by chemical cross-linking. Lys-10, Lys-21, Lys-23, and Lys-97 from ␣-Syn are exclusively cross-linked to Lys-448, Lys-528, and Lys-556 from Ssa1p, whereas Lys-12 from ␣-Syn is found cross-linked with Lys-446, Lys-494, Lys-504, or Lys-536 from Ssa1p. This suggest that after the docking of the regions centered on residues Lys-97 and Lys-556 in ␣-Syn and Ssa1p, respectively, the N-terminal part of ␣-Syn establishes interactions with either the back or the bottom tip of the client proteins binding site of Ssa1p. At one stage, Ssa1p appears to bind ␣-Syn as a tweezer, through the two tips of its client protein binding site.