Hsc70 Ameliorates the Vesicle Recycling Defects Caused by Excess α-Synuclein at Synapses

α-Synuclein overexpression and aggregation are linked to Parkinson’s disease (PD), dementia with Lewy bodies (DLB), and several other neurodegenerative disorders. In addition to effects in the cell body, α-synuclein accumulation occurs at presynapses where the protein is normally localized. While it is generally agreed that excess α-synuclein impairs synaptic vesicle trafficking, the underlying mechanisms are unknown.


Introduction
␣-Synuclein is a presynaptic protein whose aberrant aggregation causes neurodegeneration in Parkinson's disease (PD), dementia with Lewy bodies (DLB), and several variants of Alzheimer's disease (Singleton et al., 2003;Lee and Trojanowski, 2006;Cookson and van der Brug, 2008;Ingelsson, 2016). Inherited forms of PD are linked to multiplication of the ␣-synuclein gene (SNCA), as well as several point mutations (e.g., A30P, E46K, A53T), which result in atypical aggregation of ␣-synuclein protein throughout neurons (Singleton et al., 2003;Lee and Trojanowski, 2006). ␣-Synuclein aggregation causes synaptic dysfunction, mitochondrial damage, and axonal transport deficits (Ingelsson, 2016). Though the mechanisms of ␣-synuclein toxicity are beginning to emerge, particularly with respect to events in the soma, much less is known about how increased ␣-synuclein levels affect synapses where the protein normally functions.
Lamprey giant reticulospinal (RS) synapses provide an excellent model for assessing how excess ␣-synuclein affects vertebrate synapses. RS synapses are ideal for these studies because they are amenable to acute perturbations of presynaptic processes Walsh et al., 2018), thus permitting a direct evaluation of the effects of excess ␣-synuclein without inducing molecular compensation that occurs after overexpression (Nemani et al., 2010;Scott et al., 2010). Furthermore, the large size of vesicle clusters at RS synapses facilitates detailed ultrastructural analyses of synaptic vesicle trafficking events, allowing us to determine the underlying mechanisms (Morgan et al., 2004(Morgan et al., , 2013Busch et al., 2014). We previously reported that acute introduction of excess human ␣-synuclein inhibited synaptic vesicle recycling at lamprey synapses, and the results were consistent with effects on both clathrin-mediated endocytosis (CME) and possibly compensatory bulk endocytosis (Busch et al., 2014;Medeiros et al., 2017Medeiros et al., , 2018. Similarly, excess ␣-synuclein also inhibited vesicle endocytosis at mammalian calyx of Held synapses (Xu et al., 2016;Eguchi et al., 2017).
We report here a mechanism by which excess ␣-synuclein induces synaptic vesicle recycling defects and present a novel strategy for ameliorating these defects. When introduced acutely, ␣-synuclein selectively impaired CCV uncoating during synaptic vesicle recycling, leading to a depletion of the synaptic vesicle cluster (Medeiros et al., 2017). We further show that human ␣-synuclein, lamprey ␥-synuclein, and several PD-linked mutants directly associate in vitro with Hsc70, the chaperone protein that uncoats CCVs at synapses, thus identifying an interaction that may affect synapses in vivo. Indeed, excess ␣-synuclein reduced Hsc70 availability at stimulated synapses, suggesting Hsc70 sequestration as a possible mechanism underlying the synaptic defects. Consequently, co-injection of exogenous Hsc70 and ␣-synuclein ameliorated the synaptic vesicle trafficking defects. Thus, Hsc70 is an in vivo target of excess ␣-synuclein at synapses, and increasing Hsc70 function reverses the deleterious impacts.

Acute perturbations and electron microscopy
All animal procedures were approved by the Institutional Animal Care and Use Committee at the MBL in accordance with standards set by the National Institutes of Health. Lampreys (Petromyzon marinus; 11-13 cm; five to seven years old of either sex) were anesthetized in 0.1 g/l MS-222 (Western Chemical Inc.). Spinal cord pieces (2-3 cm) were dissected and pinned ventral side up in a Sylgard-lined dish. Axonal microinjections were performed as described in Walsh et al., (2018). First, human ␣-synuclein was diluted in lamprey internal solution (180 mM KCl and 10 mM HEPES K ϩ ; pH 7.4) to a pipet concentration of 130 -160 M. In some experiments, recombinant bovine Hsc70 (27 M) was included in the injection pipet either alone or together with ␣-synuclein. Proteins were then loaded into glass microelectrodes (20 -25 M⍀) and microinjected into giant RS axons using small pulses of N 2 (5-20 ms, 30 -50 psi, 0.2 Hz) delivered through a picospritzer. Fluorescein (10 kDa) or tetramethylrhodamine (70 kDa) dextrans (100 M; Thermo Fisher), approximating the molecular weights of ␣-synuclein and Hsc70, respectively, were included in the pipets and coinjected to visualize the proteins' diffusion rates in the axons. Axonal injections resulted in a 10 -20ϫ dilution of the proteins. Thus, the final axonal concentration of exogenous human ␣-synuclein was estimated to be ϳ7-13 M, and the final concentration of exogenous bovine Hsc70 was ϳ1-3 M. Axons were subsequently stimulated with action potentials (20 Hz, 5 min) using current injections (30 -60 nA; 1 ms) to induce synaptic vesicle exocytosis/endocytosis. Spinal cords were fixed immediately after stimulation (3% glutaraldehyde, 2% paraformaldehyde in 0.1 M Na cacodylate; pH 7.4), processed for EM, sectioned at 70 nm, and counterstained with uranyl acetate and lead citrate, as described previously (Morgan et al., 2013;Busch et al., 2014;Walsh et al., 2018). Images were obtained at 37,000ϫ or 59,000ϫ magnification using a JEOL JEM 200CX electron microscope. We collected EM data on n ϭ 22-33 synapses from at least two axons from two lampreys per condition and confirmed that the phenotypes reported were consistent between axons/animals. Images were collected at distances surrounding the injection site (20 -150 m) where the protein concentration was measurable based on the diffusion of the co-injected fluorescent dye (i.e., the experimental condition), as well as distances farther from the injection site (150 -700 m) where no protein had diffused (i.e., the controls). Thus, each EM experiment was internally controlled, as shown in Figure 1B, which is necessary due to the variability in the sizes of synaptic vesicle clusters between axons and animals. 3D reconstructions were generated from five serial electron micrographs using Reconstruct software (Fiala, 2005), as described previously (Busch et al., 2014;Medeiros et al., 2017). One image per synapse, taken at or near the center of the active zone, was selected for morphometric analysis. A researcher blinded to the experimental conditions performed the morphometric analyses on all synaptic membranes within a 1-m radius of the active zone using FIJI 2.0.0 (Morgan et al., 2004(Morgan et al., , 2013Busch et al., 2014). These included synaptic vesicles, plasma membrane, cisternae, and Clathrin-coated pits (CCPs) and clathrin-coated vesicles (CCVs). Synaptic vesicles were defined as small, clear round vesicles Ͻ100 nm in diameter, while "cisternae" were defined as larger vesicles that were Ͼ100 nm in diameter, as in our previous studies (Busch et al., 2014;Medeiros et al., 2017). Plasma membrane evaginations were determined by drawing a straight line from the edge of the active zone to the nearest position on the axolemma, on both sides of the synapse, and then measuring the curved distance between these points; the mean value per synapse was recorded. CCPs and CCVs were staged as described previously (Morgan et al., 2004). In addition, the SV distribution was determined using a script written in Python (https://github.com/audreytmedeiros/Morgan-Lab), which measured the distance from the center of each SV to the nearest point on the active zone. After obtaining measurements for each organelle, a total membrane analysis was performed on each synapse to determine how synaptic membranes were redistributed with each perturbation. Here, SV and CCP/V membrane areas were calculated by multiplying the surface area of a sphere (4r 2 ) by the number of each type of vesicle at each synapse. Plasma membrane and cisternae areas were obtained by multiplying the length of membrane evaginations and summed cisternae perimeters, respectively, by the section thickness (70 nm). Graphing and statistical analyses, including Student's t tests and ANOVA, were performed in Origin 7.0 (OriginLab Corp). Data were reported as the mean value per section per synapse.

GST pull downs
Protein lysates from rat brains or lamprey central nervous system (CNS) (brains and spinal cords) were prepared in HKET buffer (25 mM HEPES K ϩ , 150 mM KCl, 1 mM EDTA, and 1% Triton X-100; pH 7.4) containing protease inhibitors. GST pull downs were performed using 50 -100 g of GST-tagged proteins and either 1-5 mg of rat brain or lamprey CNS extracts or 100 g of purified recombinant proteins. After performing the pull downs, the bound proteins were run on 10% SDS-PAGE gels and transferred to nitrocellulose membranes. For Western blotting, antibodies were diluted in TBST (20 mM Tris pH 7.6, 150 mM NaCl, and 0.1% Tween 20) with 1% dry milk or 5% BSA. Primary antibodies were used at 1:1000 and included: mouse monoclonal anti-␤1/␤2-adaptins (clone 100/1; Sigma), mouse monoclonal anti-clathrin heavy chain (clone 23; BD Biosciences), mouse monoclonal anti-dynamin (clone 41; BD Biosciences), mouse monoclonal anti-synaptojanin-1 (clone 26; BD Biosciences), mouse monoclonal anti-auxilin (gift from Ernst Ungewickell; Ungewickell et al., 1995), rabbit polyclonal anti-Hsc70 (ARP48445; Aviva Systems Biology), and rabbit polyclonal anti-Hsc70 (SPA-816; Enzo Life Sciences). Secondary antibodies were used at 1:1000 -1:4000 and included HRP-conjugated goat anti-rabbit, anti-mouse, and antirat IgGs (H ϩ L; Thermo Scientific). The Hsc70 antibodies were validated in lamprey by the appearance of a single protein band of the correct molecular weight matching that in rat brain and by the elimination of that band in secondary only control experiments. Protein bands were detected using Pierce ECL Western blotting substrate (Thermo Scientific). Band intensities were measured using FIJI 2.0.0 software and statistically analyzed in Prism 8.0 (GraphPad Software, Inc.). Data shown in biochemistry figures indicate mean Ϯ SEM from n ϭ 3-5 experiments.

Clathrin uncoating assays
Clathrin cages were assembled with 1 M recombinant bovine brain clathrin and 0.1 M auxilin, as described in Sousa et al. (2016). CCVs were freshly purified from bovine brains as described previously (Keen et al., 1979;Nandi et al., 1982). To visualize the clathrin cages and CCVs, freshly glow-discharged copper grids (EM Sciences) were floated onto a drop of each sample for 5 min, followed by six washes in distilled H 2 O, counterstaining in 1% uranyl acetate for 3 min in the dark. After drying, the grids were imaged on a JEOL JEM 200CX electron microscope at 100 kV using 100,000ϫ magnification. Clathrin disassembly from clathrin cages and purified CCVs was measure in vitro by light-scattering experiments conducted in an Applied Photosystems stopped-flow fluorometer with excitation/emission wavelengths of 395 nm as described in Sousa et al. (2016). Briefly, clathrin cages or purified CCVs corresponding to 0.3 M clathrin heavy chain with 1 mM ATP and 0.45 M auxilin in 20 mM imidazole, pH 6.8, 10 mM (NH 4 ) 2 SO 4 , 25 mM KCl, and 2 mM MgAc 2 were reacted with an equal volume of 4 M Hsc70 in the same buffer. Background scattering determined from reactions without cages or CCVs was subtracted from measured scattering values, which were normalized by dividing by the starting scattering value so that the initial scattering in all reactions was 1.0. In some reactions 20 M recombinant human ␣-synuclein (rPeptide) was added to the Hsc70 syringe. Various components were also omitted when indicated as controls. Data were plotted using Origin 7.0 software.

Immunofluorescence (IF) at lamprey synapses
Recombinant human ␣-synuclein was microinjected into giant RS axons as described above for the EM experiments, after which spinal cords were stimulated using high K ϩ (50 mM) Ringer for 10 min (Wickelgren et al., 1985). In some experiments, spinal cords were stimulated using action potentials (20 Hz, 5 min). Following stimulation, spinal cords were fixed in 4% paraformaldehyde in 0.1 M PBS, pH 7.4, for 3 h, washed in 0.1 M PBS, and incubated for 1 h in blocking buffer (10% normal goat serum; Thermo Fisher Scientific) containing 0.3% Triton X-100. Primary antibody incubations were overnight at 4°C, followed by 5 ϫ 1 h washes in wash buffer (20 mM Na phosphate buffer, 450 mM NaCl, and 0.3% Triton X-100; pH 7.4). Primary antibodies included: mouse monoclonal anti-SV2 antibody, which was deposited to DSHB by K. M. Buckley (1:100;DSHB;Buckley and Kelly, 1985), rabbit polyclonal anti-Hsc70 (ARP48445; Aviva Systems Biology, Corp.), and rat monoclonal anti-Hsc70 (SPA-815; Enzo Life Sciences). Secondary antibodies used were Alexa Fluor 594 goat anti-mouse IgG (H ϩ L; 1:200), Alexa Fluor 488 goat anti-rabbit IgG (H ϩ L; Ther-moFisher), or DyLight 488 goat anti-rat IgG (H ϩ L; 1:100; Thermo Fisher). After immunostaining, synapses were imaged within intact whole mounted spinal cords using a Zeiss LSM510 Meta confocal on an Axioskop 2FS microscope. Images were acquired using a Zeiss 40ϫ, 0.8 NA Achroplan objective with 3ϫ optical zoom. All analyses on synapses were performed in FIJI 2.0.0 as follows. Giant RS synapses were first identified using the SV2 labeling, and then the associated Hsc70 puncta (Ն2ϫ background intensity) that were located on and around the synapses were identified. Those that were overlapping or touching the SV2 puncta were considered to be associated with the synaptic vesicle cluster while those within a 1-m radius of the synapses were located within the endocytic periactive zone. The percentage of synapses containing Hsc70 puncta within each axon, as well as the average number of Hsc70 puncta per synapse, was calculated. All graphing and statistical analyses were performed in GraphPad Prism 8.0 software, including outlier and normality tests (Shapiro-Wilk; D'Agostino and Pearson), as well as statistical comparisons using Student's t test or ANOVA.

Excess ␣-synuclein impairs clathrin uncoating in vivo during synaptic vesicle recycling
We previously reported that acute introduction of excess human ␣-synuclein to lamprey RS synapses caused a reduction in the number of synaptic vesicles, which was compensated by an increase in the size of plasma membrane evaginations, as well as greater numbers of irregular membranous cisternae and clathrin-coated structures [i.e., the total number of CCPs and CCVs combined; Busch et al., 2014]. Many of the cisternae originated from the plasma membrane and had CCPs budding from them, suggesting that they were plasma membrane extensions and/or bulk endosomes derived through compensatory endocytosis, though we cannot rule out an effect on recycling endosomes (Chanaday et al., 2019). The phenotype reported is consistent with an impairment of synaptic vesicle recycling via inhibition of CME and possibly bulk endocytosis. Because CME is a predominant mechanism for locally recycling synaptic vesicles at many synapses (Heuser and Reese, 1973;Granseth et al., 2006;Heerssen et al., 2008;Walsh et al., 2018), and because clathrin-mediated vesicle budding is also critical for other modes of SV recycling such as ultrafast and bulk endocytosis (Watanabe et al., 2014;Gan and Watanabe, 2018;Chanaday et al., 2019), we set out to determine in this study exactly how excess ␣-synuclein impacts clathrin-mediated processes at synapses and to identify the underlying mechanisms.
To identify how excess ␣-synuclein impairs CME, we first needed to determine which stage or stages were preferentially affected. Briefly, clathrin-mediated vesicle recycling is initiated when clathrin adaptors, AP180 and AP2, recruit clathrin triskelia to the plasma membrane, promoting their assembly into coats ( Fig. 1A; Morgan et al., 1999;Saheki and De Camilli, 2012). After maturation of the CCP, vesicle fission occurs through the actions of dynamin, a large GTPase that is abundant at synapses (Takei et al., 1995;Antonny et al., 2016). Synaptojanin is recruited to the neck of the CCP to assist during fission and subsequently in CCV uncoating (Cremona et al., 1999;Chang-Ileto et al., 2011). After vesicle fission is complete, the free CCVs are uncoated by the actions of the chaperone protein Hsc70 and its co-chaperone, auxilin (Ungewickell et al., 1995;Morgan et al., 2001). Uncoated vesicles are then refilled with neurotransmitter molecules (Farsi et al., 2018) and returned to the vesicle cluster for subsequent bouts of exocytosis.
As in our prior studies, we microinjected recombinant monomeric human ␣-synuclein into lamprey giant axons, thereby delivering the protein directly to presynapses (Fig.  1B). The axons were subsequently stimulated (20 Hz, 5 min), fixed, and processed for standard transmission electron microscopy (Walsh et al., 2018). After injection, the final axonal concentration of ␣-synuclein was estimated to be ϳ7-13 M, which is approximately two to four times greater than measurements of endogenous ␣-synuclein at synapses and commensurate with overexpression levels in mammalian PD models and human patients (Singleton et al., 2003;Miller et al., 2004;Nemani et al., 2010;Scott et al., 2010;Westphal and Chandra, 2013). Images of untreated control synapses were collected from the same axon but from regions beyond where the ␣-synuclein protein had diffused (Fig. 1B), thus providing an internal control for each experiment, which is important because of the natural variability in the sizes of synaptic vesicle clusters between axons and animals. At control synapses, local synaptic vesicle recycling was efficient enough to maintain a large synaptic vesicle cluster, and very few CCP/Vs were observed ( Fig. 2A,D). In contrast, after injection of ␣-synuclein, synapses were dramatically altered due to deficits in synaptic vesicle recycling. Specifically, synapses treated with ␣-synuclein exhibited fewer synaptic vesicles, expanded plasma membrane evaginations, increased numbers of large atypical cisternae, and abundant clathrin-coated structures (Fig. 2B,E), as previously reported (Busch et al., 2014;Medeiros et al., 2017). Cisternae were classified as any irregular-shaped vesicles with a diameter Ͼ100 nm (Busch et al., 2014;Medeiros et al., 2017). These cisternae often had CCPs budding from them and could sometimes be traced back to the plasma membrane (Busch et al., 2014). Strikingly, when we further examined the clathrin-coated structures, we observed atypical clusters of free CCVs at synapses treated with excess ␣-synuclein, a phenotype typically associated with defective CCV uncoating (Fig. 2C, arrows; Cremona et al., 1999;Morgan et al., 2001). 3D reconstructions generated from serial micrographs revealed the gross alterations in synaptic structure caused by ␣-synuclein (Fig. 2D,E). Whereas control synapses exhibited only a few CCPs and CCVs close to the plasma membrane, the ␣-synuclein treated synapses exhibited dozens of free CCVs that were dispersed throughout the synaptic area ( Fig. 2D,E, insets).
To determine whether the Hsc70/␣-synuclein interaction is direct and to further map it, the GST pull downs were repeated using purified, recombinant bovine Hsc70. Hsc70 consists of a highly conserved N-terminal nucleotide-binding domain (NBD), and a substratebinding domain (SBD) with a more variable 10 kDa C-terminal region (Fig. 3E). While the affinity of Hsc70 for typical client polypeptides is normally modulated by nucleotides (ATP/ADP), we performed the pull downs without nucleotides because prior studies reported more effective binding and sequestration of ␣-synuclein by Hsc70 in fibrillation assays under nucleotide-free conditions (Pemberton et al., 2011;Redeker et al., 2012). Under these conditions, GST-tagged human ␣-synuclein and its NTD pulled down bovine Hsc70, indicating a direct interaction (Fig. 3F, top; GST: 8.3 Ϯ 1.6 AU; GST-Syn: 68.1 Ϯ 5.3 AU; GST-NTD: 67.9 Ϯ 3.9 AU; n ϭ 3; ANOVA p ϭ 5 ϫ 10 Ϫ5 ; Tukey's post hoc). Similarly, lamprey ␥-synuclein and its NTD also pulled down bovine Hsc70 in these direct binding assays (Fig. 3G, top; GST: 1.5 Ϯ 0.4 AU; GST-Syn: 48.9 Ϯ 11.6 AU; GST-NTD: 62.8 Ϯ 9.6 AU; n ϭ 3; ANOVA p ϭ 0.006; Tukey's post hoc). To further map the interaction, we additionally tested the binding of ␣-synuclein to several Hsc70 truncation mutants. One truncation consisted of only the NBD (a.a. 1-383), and the other contained the NBD and SBD but lacked the C-terminal 10 kDa region (Hsc70⌬C; a.a. 1-554; Fig. 3E; Flaherty et al., 1990;Wilbanks et al., 1995). Both truncation mutants are well expressed, folded, stable, soluble, and exhibit ATP and ADP binding affinities and ATP hydrolysis activity like full length Hsc70 (Wilbanks et al., 1995;Zhang and Zuiderweg, 2004;Jiang et al., 2005). Neither human ␣-synuclein, lamprey ␥-synuclein, nor the NTDs pulled down the Hsc70 truncation mutants that were missing the C-terminal fragments, indicating a role for the C terminus in mediating or stabilizing the interaction (Fig. 3F,G, middle and bottom). Taken together, these data indicate that the N terminus of ␣-synuclein interacts directly with Hsc70 via an interaction that is conserved between synuclein orthologs and disrupted by the deletion of the C terminus of Hsc70.
We repeated the pull downs using the ␣-synuclein point mutants that are linked to PD: A30P, E46K, and A53T (Fig.  4A). All three mutants bound directly to bovine Hsc70 (Fig.  4B-D). While A30P and A53T did not exhibit any statistically significant differences in binding efficacy, as compared to wild-type ␣-synuclein, E46K binding to Hsc70 was slightly reduced (

␣-Synuclein does not affect Hsc70-mediated clathrin disassembly in vitro
Taken together, the above results showing the in vivo clathrin uncoating defect and in vitro interaction between Hsc70 and ␣-synuclein suggest that when in excess, ␣-synuclein may interact with Hsc70 and alter its CCV uncoating function at synapses. Among the possible mechanisms, ␣-synuclein could interfere with Hsc70 uncoating activity and/or alter Hsc70 availability at synapses. We began testing these possibilities by examining whether ␣-synuclein affects the ability of Hsc70 to pro-mote clathrin cage disassembly in vitro using a light scattering assay as previously described (Sousa et al., 2016). In one set of experiments, we used empty clathrin cages assembled in vitro (Fig. 5A). In a typical reaction, Hsc70 binds the clathrin cages and subsequently uncoats clathrin, as indicated by a transient increase in light scattering followed by an exponential decrease in scattering intensity as the clathrin is disassembled (Fig. 5A, blue trace). Even when ␣-synuclein was in excess of Hsc70, it had no effect on the disassembly of clathrin cages (Fig. 5A, red trace).
It is well established that ␣-synuclein interacts in vitro with small vesicles containing anionic lipids and in vivo with synaptic vesicles, and such vesicle interactions promote folding of the NTD into an ␣-helix (Maroteaux et al., 1988;Davidson et al., 1998;Fortin et al., 2005;Burré et al., 2015). Thus, because ␣-synuclein effects on un- Figure 3. ␣-Synuclein interacts directly with Hsc70, the chaperone protein that uncoats CCVs during synaptic vesicle recycling. A, Domain diagrams of full-length human ␣-synuclein, lamprey ␥-synuclein and their respective NTDs. B, GST pull downs from rat brain lysates revealed no detectable interactions between ␣-synuclein and several major components of CME including AP2, clathrin, dynamin, synaptojanin, and auxilin. SM ϭ starting material. Blots shown are representative of n ϭ 3 experiments. C, In contrast, human ␣-synuclein and its NTD pulled down Hsc70 from rat brain and lamprey CNS lysates. D, Similarly, lamprey ␥-synuclein pulled down Hsc70 from rat brain lysates, demonstrating conservation of the interaction. E, Domain diagrams of bovine Hsc70 and several truncations used in the experiments. F, G, In direct binding assays, both human ␣-synuclein and lamprey ␥-synuclein, and their NTDs, pulled down recombinant bovine Hsc70. No interactions were detected with either NBD or Hsc70⌬C, indicating a role for the C terminus. In panels C, D, F, G, bars represent mean Ϯ SEM from n ϭ 3 independent experiments. n.s. ϭ not significant by ANOVA (p Ͼ 0.05).
coating may require membrane binding, we repeated the in vitro clathrin uncoating experiments using CCVs purified from bovine brains, which contained the underlying endocytic vesicle membranes (Fig. 5B). Under these conditions, some binding of ␣-synuclein to purified CCVs occurred, as indicated by an initial increase in light scattering (Fig. 5B, green trace). As with the clathrin baskets, Hsc70 also bound and disassembled purified CCVs, although the decay in light scattering was less pronounced due to the presence of the vesicular membranes in the reaction (Fig. 5B, blue trace). Excess ␣-synuclein also had no obvious effect on Hsc70-mediated uncoating of purified CCVs (Fig. 5B, red trace). Using these highly optimized, reduced conditions, we thus conclude that ␣-synuclein does not directly interfere with the kinetics or core process of Hsc70-mediated clathrin uncoating in vitro.

Excess ␣-synuclein inhibits Hsc70 availability at synapses
Compared to the in vitro assays, the complex in vivo environment of synapses permits a more physiologic partitioning of ␣-synuclein and Hsc70 between cytosol and synaptic membranes, raising the possibility of observing more functional interactions. We therefore examined whether excess ␣-synuclein affects the localization of Hsc70 in vivo at lamprey synapses. To do so, we performed whole mount immunostaining in the lamprey spinal cord for Hsc70 and SV2, a marker of synaptic vesicle clusters, and then imaged the Hsc70 in the vicinity of the synapses. Giant RS synapses are large en passant synapses (1-2 m in diameter; 1000 -2000 synaptic vesicles), which reside along the periphery of the giant RS axons (Fig. 1B, inset). They comprise large synaptic vesicle clusters at the active zone, which are surrounded by a distinct actin-rich periactive zone where clathrin-mediated synaptic vesicle recycling occurs ( Fig. 6A; Shupliakov et al., 2002;Morgan et al., 2004;Bourne et al., 2006;Saheki and De Camilli, 2012). To detect Hsc70, we used an Hsc70 antibody (Aviva ARP48445 rabbit polyclonal) that detects a single 70 kDa band in both lamprey and rat brain lysates by Western blotting (Fig. 6B). After immunostaining the spinal cords, the giant RS synapses were first identified using a monoclonal SV2 antibody, which labels presynaptic vesicle clusters in all vertebrates tested, including lampreys (Fig. 6C,D; Buckley and Kelly, 1985;Busch et al., 2014). The synapse-associated Hsc70 was subsequently evaluated. The Hsc70 antibody immunolabeled distinct puncta within axons and around RS synapses, as observed at mammalian synapses (Wilhelm et al., 2014), as well as diffuse patches (Fig. 6C,D). Although Hsc70 is one of the most abundant proteins in neurons, the IF signals at both lamprey and mammalian synapses is dimmer than expected, which may be due to some Hsc70 being tightly bound in chaperone protein complexes and therefore inaccessible for immunolabeling. At unstimulated control synapses, the immunolabeled Hsc70 puncta were fairly sparse and in low abundance (Fig. 6C, top   Figure 4. PD-associated ␣-synuclein mutants also interact directly with Hsc70. A, Diagram showing the locations of PD-linked point mutations A30P, E46K, and A53T, which occur within the alpha helical NTD of ␣-synuclein. B-D, In direct binding assays, GST-tagged A30P, E46K, and A53T all pulled down Hsc70 to a similar degree as wild-type ␣-synuclein. Only binding to E46K was slightly reduced. Bars represent mean Ϯ SEM from n ϭ 3-5 experiments. Asterisk indicates significance (p ϭ 0.04); n.s. ϭ not significant by ANOVA. Figure 5. ␣-Synuclein does not affect Hsc70-mediated clathrin disassembly in vitro. A, left, Clathrin cages were assembled in vitro using recombinant clathrin heavy chain and auxilin, as described in Sousa et al. (2016). Right, An in vitro light scattering assay showed that addition of 2 M Hsc70 exponentially reduced the light scattering intensity as clathrin cages were disassembled (blue trace). Addition of 10 M ␣-synuclein did not alter the rate of Hsc70-mediated clathrin disassembly (red trace). No change in light scattering was observed in baseline control measurements after addition of buffer (yellow trace) or 10 M ␣-synuclein alone (green trace). B, left, CCVs, which contain the underlying endocytic vesicle membranes, were freshly purified from bovine brains. Right, ␣-Synuclein alone slightly increased light scattering (green trace), which is likely due to its binding to the endocytic vesicles. Similar to the results with clathrin cages, introduction of 10 M ␣-synuclein did not significantly affect the dynamics of Hsc70-mediated CCV uncoating (blue vs red trace). row). In contrast, stimulation using high K ϩ induced an increase in Hsc70 puncta within or adjacent to the synaptic vesicle clusters, indicating enhanced Hsc70 availability at synapses (Fig. 6C, second row). The average size of Hsc70 puncta was 0.20 Ϯ 0.004 m 2 (n ϭ 174 puncta, n ϭ 132 synapses, n ϭ 33 axons). Next, as with the EM experiments, we injected human ␣-synuclein to a final axonal concentration of 7-13 M (see Materials and Methods). Measurements of endogenous ␣-synuclein range from 3-6 M (Westphal and Chandra, 2013) to 45 M (Wilhelm et al., 2014), suggesting that the total ␣-synuclein concentration after injection in lamprey axons could be anywhere between 10 -58 M, which is submolar to equimolar with measurements of endogenous Hsc70 in rat synaptosomes (55 M; Wilhelm et al., 2014). After introducing excess ␣-synuclein, the stimulationdependent increase in Hsc70 at presynapses was no longer observed, suggesting an in vivo interaction with ␣-synuclein (Fig. 6D). Next, we quantified all of the Hsc70 puncta associated with synapses, both those directly touching or overlapping the synaptic vesicle cluster, as well as those within a 1-m radius of the synaptic vesicle cluster, representing the endocytic periactive zone (Fig.  6A). In Figure 6E-H, the stippled regions of the bars indicate the proportion of Hsc70 puncta associated with the synaptic vesicle cluster (30 -77% across all conditions), while the unmarked regions indicate the Hsc70 puncta localized exclusively within the periactive zone. At lamprey synapses, many endocytic proteins involved in CME are localized to synaptic vesicle clusters, as well as the periactive zone (Evergren et al., 2004(Evergren et al., , 2006, and so both pools of Hsc70 may be involved in synaptic vesicle endocytosis. In unstimulated control axons, Ͻ20% of synapses per axon had clearly defined immunolabeled Hsc70 puncta associated with them (Fig. 6E). Stimulation induced a significant 3-fold increase in the percentage of where CME occurs (green). B, By Western blotting, the Hsc70 antibody used for these experiments (Aviva ARP48445) specifically recognized a single band at 70 kDa in both lamprey CNS and rat brain lysates, consistent with the expected molecular weight of Hsc70. C, Confocal images showing clusters of giant synapses immunostained for the synaptic vesicle-associated protein SV2 (red) and Hsc70 (green). Compared to unstimulated conditions, stimulation with high K ϩ increased Hsc70 availability at synapses, as evidenced by an increase in the number of visible Hsc70 puncta (arrows). D, In contrast, ␣-synuclein inhibited the stimulationdependent increase in Hsc70 at synapses. E, F, Graphs showing the percentage of synapses (per axon) with associated Hsc70 puncta, as well as the average number of Hsc70 puncta per synapse. Bars represent mean Ϯ SEM from n ϭ 76 -135 synapses, n ϭ 5-13 axons, n ϭ 5-7 animals/condition; ‫ء‬p Ͻ 0.05, ‫‪p‬ءء‬ Ͻ 0.01, ‫‪p‬ءءء‬ Ͻ 0.001, ‫‪p‬ءءءء‬ Ͻ 0.0001; n.s. ϭ not significant by ANOVA. G, H, Similar results were obtained using action potential stimulation (20 Hz, 5 min) and a different Hsc70 antibody (Enzo Life Sciences SPA815). Bars represent mean Ϯ SEM from n ϭ 31-47 synapses, n ϭ 3-4 axons, n ϭ 2-3 animals/condition; ‫ء‬p Ͻ 0.05, ‫‪p‬ءء‬ Ͻ 0.01; n.s. ϭ not significant by ANOVA. In panels E-H, the stippled regions of the bars represent the proportion of Hsc70 puncta overlapping or touching the SV cluster, while the clear regions indicate the proportion localized within the periactive zone. synapses with Hsc70 puncta, and this did not occur after introduction of excess ␣-synuclein into the axons (Fig. 6E; unstimulated control: 17.61 Ϯ 4.11%, n ϭ 85 synapses, n ϭ 9 axons; stimulated control: 62.06 Ϯ 7.49%, n ϭ 135 synapses, n ϭ 13 axons; unstimulated ␣-Syn: 29.39 Ϯ 8.76%, n ϭ 77 synapses, n ϭ 5 axons; stimulated ␣-Syn: 29.26 Ϯ 7.80%, n ϭ 76 synapses, n ϭ 6 axons; ANOVA p Ͻ 0.0005; Tukey's post hoc). We also quantified the average number of Hsc70 puncta per synapse under the same conditions and obtained similar results. At control synapses, the average number of Hsc70 puncta per synapse increased on stimulation, but this did not occur after ␣-synuclein treatment ( Fig. 6F; unstimulated control: 0.19 Ϯ 0.05 puncta, n ϭ 85 synapses, n ϭ 9 axons; stimulated control: 0.84 Ϯ 0.14 puncta, n ϭ 135 synapses, n ϭ 13 axons; unstimulated ␣-Syn: 0.30 Ϯ 0.09 puncta, n ϭ 77 synapses, n ϭ 5 axons; stimulated ␣-Syn: 0.37 Ϯ 0.13 puncta, n ϭ 76 synapses, n ϭ 6 axons; ANOVA p ϭ 0.0012, Tukey's post hoc). In an independent set of experiments, using action potential stimulation (20 Hz, 5 min as in the EM experiments) and a second Hsc70 antibody (Enzo SPA815 rat polyclonal), we observed similar results (Fig. 6G,H; % synapses w/Hsc70 puncta: unstimulated control: 20.48 Ϯ 5.79%, n ϭ 34 synapses, n ϭ 3 axons; stimulated control: 83.23 Ϯ 11.75%, ϭ 47 synapses, n ϭ 3 axons; unstimulated ␣-Syn: 40.74 Ϯ 11.75%, n ϭ 31 synapses, n ϭ 3 axons; stimulated ␣-Syn: 50.58 Ϯ 14.11%, n ϭ 43 synapses, n ϭ 4 axons; ANOVA p ϭ 0.03; Tukey's post hoc; #Hsc70 puncta/synapse: unstimulated control: 0.23 Ϯ 0.07%, n ϭ 34 synapses, n ϭ 3 axons; stimulated control: 1.08 Ϯ 0.19%, n ϭ 47 synapses, n ϭ 3 axons; unstimulated ␣-Syn: 0.52 Ϯ 0.18%, n ϭ 31 synapses, n ϭ 3 axons; stimulated ␣-Syn: 0.63 Ϯ 0.23%, n ϭ 43 synapses, n ϭ 4 axons; ANOVA p ϭ 0.08; Tukey's post hoc). Thus, excess ␣-synuclein sequesters Hsc70 in vivo and reduces its availability at stimulated synapses, Figure 7. Increasing Hsc70 alone has no effect on CME at synapses. A, B, Hsc70 does not dramatically alter synaptic morphology, as demonstrated by the large synaptic vesicle (SV) clusters, shallow plasma membrane evaginations (dotted lines), and few CCP/Vs (circles) in both control and Hsc70-treated synapses. Asterisks mark postsynaptic spines. C ϭ cisternae. Scale bars ϭ 500 nm. C, D, 3D reconstructions further highlight the similarities and show that CCPs (yellow spheres) and CCVs (white spheres) are in normal numbers and clustered around the plasma membrane (green slabs; see insets). Active zone is shown in red. Scale bars ϭ 500 nm. E-M, There is little effect of exogenous Hsc70 on clathrin-mediated synaptic vesicle endocytosis, as illustrated by normal number and distribution of SVs, PM evaginations, and CCPs/CCVs. Only the number of cisternae was greater, however, their size remains normal. Bars represent mean Ϯ SEM (per section, per synapse) from n ϭ 27-28 synapses, n ϭ 2 axons, n ϭ 2 animals/condition. Asterisk indicates significance (p Ͻ 0.05); n.s. ϭ not significant (p Ͼ 0.05) by Student's t test (E, H-K, M) or ANOVA (F, G, L).
suggesting a possible mechanism underlying the CCV uncoating defects during synaptic vesicle recycling.

Discussion
While a number of studies have focused on the physiologic roles of ␣-synuclein at synapses (for review, see Sulzer and Edwards, 2019), the precise effects of excess ␣-synuclein and the underlying mechanisms are less clear. Data presented here identify loss of Hsc70 availability at synapses, and consequently its function, as one mechanism by which excess ␣-synuclein induces synaptic vesicle trafficking defects. Specifically, the sequestration of Hsc70 leads to an impairment of CCV uncoating at synapses, which consequently inhibits synaptic vesicle recycling. How might this work? A plausible explanation, shown in our working model, is that inhibiting CCV uncoating may trap clathrin and/or other limited coat proteins such as AP180 and AP2 within CCVs, making them unavailable for initiating subsequent rounds of endocytosis thereby leading to aberrant plasma membrane expan-sion and compensatory bulk endocytosis, resulting in formation of cisternae ( Fig. 9A; Morgan et al., 1999Morgan et al., , 2000Walsh et al., 2018). Similarly, acute inactivation of CME in other models also resulted in a compensatory increase in atypical cisternae (Heerssen et al., 2008;Kasprowicz et al., 2008). In these respects, the phenotypes produced by excess ␣-synuclein are suggestive of a pathologic gain of function. However, the plasma membrane expansion observed with excess ␣-synuclein would also be consistent with direct effects of ␣-synuclein accelerating vesicle fusion (Logan et al., 2017) and/or slowing early stages of clathrin-mediated synaptic vesicle endocytosis (Vargas Figure 8. Increasing exogenous Hsc70 largely reverses the ␣-synuclein-induced synaptic defects. A, B, Unlike synapses treated with ␣-synuclein alone (Fig. 2), those co-treated with Hsc70 and ␣-synuclein appear similar to control synapses, with large SV clusters and few cisternae or CCP/Vs (circles). C, D, 3D reconstructions reveal that synapses treated with Hsc70 and ␣-synuclein appear normal. Insets show the distributions of CCPs (yellow spheres) and CCVs (white spheres), which are sparse and clustered around the plasma membrane (green slabs). Active zone is shown in red. E-M, The CCV uncoating and vesicle recycling defects caused by ␣-synuclein were largely ameliorated by co-injection of Hsc70 as evidenced by normal numbers of SVs (E), cisternae (J), and CCPs/CCVs (K-L). Only the PM evaginations were larger (H). Notably, there was no longer any difference in the number of free CCVs after co-injection of Hsc70ϩ␣-synuclein, indicating a reversal of the uncoating defects (L, stage 4). Bars represent mean Ϯ SEM (per section, per synapse) from n ϭ 22-30 synapses, n ϭ 2 axons, 2 n ϭ animals/condition. Asterisk indicates significance (p Ͻ 0.05); n.s. ϭ not significant (p Ͼ 0.05) by Student's t test (E, H-K, M) or ANOVA (F-G, L). et al., 2014), which is in line with our current understanding of its normal physiologic function. Similar to the ␣-synuclein phenotype, impairments of CCV uncoating have also been observed at squid and mammalian synapses after directly perturbing Hsc70 recruitment to CCVs (Morgan et al., 2001;Leshchyns'ka et al., 2006). As with direct Hsc70 perturbations (Morgan et al., 2001), excess ␣-synuclein also induced a loss of synaptic vesicles and expansion of the plasma membrane, indicating effects on vesicle recycling ( Fig. 2; Busch et al., 2014;Medeiros et al., 2017). By this working model, replenishing Hsc70 at synapses should have the effect of restoring CCV uncoating, thereby freeing clathrin and other limited coat proteins to recycle synaptic vesicles more efficiently (Fig. 9B). Indeed, when exogenous bovine Hsc70 was introduced along with human ␣-synuclein, the CCV uncoating defects were completely rescued, and the synaptic vesicle clusters were restored to normal, indicating a significant improvement in synaptic vesicle recycling (Fig. 8). That the number and size of cisternae were also restored to control levels indicates their role in replenishing the vesicle cluster and suggests that they are affected by clathrin-dependent budding processes, although their identity remains unclear (e.g., recycling or bulk endosomes). The only morphologic deficits remaining after co-injection of ␣-synuclein and Hsc70 was an enlargement of the plasma membrane evaginations and a slight, but non-significant, dispersion of synaptic vesicles (Fig. 8). One possible ex-planation is that there may be a pool of ␣-synuclein bound tightly to the plasma membrane and vesicles, which does not release in the presence of exogenous Hsc70 and therefore continues to slow endocytosis. While our data are consistent with the working model shown in Figure 9, we also acknowledge the possibility of an alternative explanation that ␣-synuclein may inhibit vesicle recycling by some other undetermined mechanism, which is reversed by Hsc70 for example through its general chaperoning functions. Whatever the case, it is an exciting prospect that even substoichiometric amounts of Hsc70 can reverse the vast majority of the synaptic vesicle trafficking defects associated with excess ␣-synuclein. It will be important in future experiments to determine if the remaining morphologic alterations can be further ameliorated by increasing the concentration of co-injected Hsc70 or by introducing Hsp110, which was recently shown to mitigate ␣-synuclein pathology in mouse models (Taguchi et al., 2019) and is the most likely Hsc70 nucleotide exchange factor that regulates availability of free clathrin at synapses (Morgan et al., 2013).
Our results corroborate and extend previous reports of an interaction between ␣-synuclein and Hsc70 (Pemberton et al., 2011;Redeker et al., 2012). Hsc70 binds both soluble and fibrillar ␣-synuclein in the absence of nucleotides and inhibits fibril formation in vitro, leading to increased cell viability (Pemberton et al., 2011). Crosslinking studies using soluble ␣-synuclein and Hsc70 indicated two discrete regions in ␣-synuclein (a.a. 10 -45, human genetics indicates that defects in CME, and specifically in CCV uncoating, are at least susceptibility factors in PD and Parkinsonism, if not causal factors. Detection of mutations in auxilin that increase such susceptibility is especially notable because, unlike Hsc70 or other chaperones that have multiple cellular functions, auxilin's role appears to be limited to the single function of recruiting Hsc70 to CCVs to drive uncoating, so these observations strongly indicate that disruption of clathrin uncoating is a strong contributor to PD disease etiology. Thus, strategies that ensure proper CCV uncoating, for example by increasing Hsc70 availability or function, may hold promise for improving synaptic function and reducing neurodegeneration in PD and other ␣-synucleinassociated diseases.