PNAS PLUS

α-Synuclein assembles into higher-order multimers upon membrane binding to promote SNARE complex formation Jacqueline Burréa,1, Manu Sharmaa,1, and Thomas C. Südhofa,b,2 a

Department of Molecular and Cellular Physiology and bHoward Hughes Medical Institute, Stanford University Medical School, Stanford, CA 94305

Physiologically, α-synuclein chaperones soluble NSF attachment protein receptor (SNARE) complex assembly and may also perform other functions; pathologically, in contrast, α-synuclein misfolds into neurotoxic aggregates that mediate neurodegeneration and propagate between neurons. In neurons, α-synuclein exists in an equilibrium between cytosolic and membrane-bound states. Cytosolic α-synuclein appears to be natively unfolded, whereas membrane-bound α-synuclein adopts an α-helical conformation. Although the majority of studies showed that cytosolic α-synuclein is monomeric, it is unknown whether membrane-bound α-synuclein is also monomeric, and whether chaperoning of SNARE complex assembly by α-synuclein involves its cytosolic or membrane-bound state. Here, we show using chemical cross-linking and fluorescence resonance energy transfer (FRET) that α-synuclein multimerizes into large homomeric complexes upon membrane binding. The FRET experiments indicated that the multimers of membrane-bound α-synuclein exhibit defined intermolecular contacts, suggesting an ordered array. Moreover, we demonstrate that α-synuclein promotes SNARE complex assembly at the presynaptic plasma membrane in its multimeric membrane-bound state, but not in its monomeric cytosolic state. Our data delineate a folding pathway for α-synuclein that ranges from a monomeric, natively unfolded form in cytosol to a physiologically functional, multimeric form upon membrane binding, and show that only the latter but not the former acts as a SNARE complex chaperone at the presynaptic terminal, and may protect against neurodegeneration. Parkinson’s disease

α

influences lipid packing (37, 38), and induces vesicle clustering (39). Moreover, membranes were found to be important for the neuropathological effects of α-synuclein (40–44). However, the relation of membrane binding to the in vivo function of α-synuclein remains unexplored, and it is unknown whether α-synuclein binds to membranes as a monomer or oligomer. Thus, in the present study we have investigated the nature of the membrane-bound state of α-synuclein and its relation to its physiological function in SNARE complex assembly. We found that soluble monomeric α-synuclein assembles into higher-order multimers upon membrane binding and that membrane binding of α-synuclein is required for its physiological activity in promoting SNARE complex assembly at the synapse. Results Membrane-Bound Multimeric α-Synuclein Chaperones SNARE Complex Assembly. Promotion of SNARE complex assembly by α-synuclein

requires binding of α-synuclein to the N terminus of the v-SNARE protein synaptobrevin-2 (3, 45), suggesting that α-synuclein promotes SNARE complex assembly in its membrane-bound state. Whether membrane binding is actually required for this activity, however, remained unknown. To address this question, we used a reconstituted system composed of purified recombinant proteins (3) (Fig. 1A). In these experiments, we measured SNARE complex assembly in the absence and presence of α-synuclein under Significance

| synapse | SNARE proteins | membrane fusion

-Synuclein is an abundant presynaptic protein that physiologically acts to promote soluble NSF attachment protein receptor (SNARE) complex assembly in vitro and in vivo (1–3). Point mutations in α-synuclein (A30P, E46K, H50Q, G51D, and A53T) as well as α-synuclein gene duplications and triplications produce early-onset Parkinson’s disease (PD) (4–10). Moreover, α-synuclein is a major component of intracellular protein aggregates called Lewy bodies, which are pathological hallmarks of neurodegenerative disorders such as PD, Lewy body dementia, and multiple system atrophy (11–14). Strikingly, neurotoxic α-synuclein aggregates propagate between neurons during neurodegeneration, suggesting that such α-synuclein aggregates are not only intrinsically neurotoxic but also nucleate additional fibrillization (15–18). α-Synuclein is highly concentrated in presynaptic terminals where α-synuclein exists in an equilibrium between a soluble and a membrane-bound state, and is associated with synaptic vesicles (19–22). The labile association of α-synuclein with membranes (23, 24) suggests that binding of α-synuclein to synaptic vesicles, and its dissociation from these vesicles, may regulate its physiological function. Membrane-bound α-synuclein assumes an α-helical conformation (25–32), whereas cytosolic α-synuclein is natively unfolded and monomeric (refs. 25, 26, 31, and 32; however, see refs. 33 and 34 and Discussion for a divergent view). Membrane binding by α-synuclein is likely physiologically important because in in vitro experiments, α-synuclein remodels membranes (35, 36), www.pnas.org/cgi/doi/10.1073/pnas.1416598111

Physiologically, α-synuclein promotes soluble NSF attachment protein receptor (SNARE) complex assembly during synaptic exocytosis. Pathologically, however, α-synuclein forms neurotoxic aggregates that promote neurodegeneration and represent hallmarks of Parkinson’s disease and other synucleinopathies. α-Synuclein exists in a monomeric unfolded state in solution and in an α-helical folded state upon binding to membranes. Yet the relation between these conformational states and their physiological and pathological roles remain unknown. Here, we demonstrate that α-synuclein multimerizes during membrane binding and that the membrane-bound, multimeric form of α-synuclein mediates SNARE complex assembly in presynaptic terminals. Our data delineate a folding pathway for α-synuclein that ranges from a monomeric unfolded form in cytosol to a physiologically functional multimeric form that is membrane bound and chaperones SNARE complex assembly, and that may protect against neurodegeneration. Author contributions: J.B., M.S., and T.C.S. designed research; J.B. and M.S. performed research; J.B., M.S., and T.C.S. analyzed data; and J.B., M.S., and T.C.S. wrote the paper. The authors declare no conflict of interest. 1

Present address: Appel Institute for Alzheimer’s Disease Research, Brain and Mind Research Institute, Weill Cornell Medical College, New York, NY 10021.

2

To whom correspondence should be addressed. Email: [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1416598111/-/DCSupplemental.

PNAS Early Edition | 1 of 10

NEUROSCIENCE

Contributed by Thomas C. Südhof, August 28, 2014 (sent for review April 21, 2014)

A

Buffer

PCPS

PCPS-Syb2 PCPS liposomes +/Syb2WT or Syb2ΔN α−Synuclein SNAP-25 Synt-1

Buffer

PCPS-Syb2WT

PCPS

+ αSyn

- αSyn

+ αSyn

In pu Pr t eIm m IP In pu Pr t eIm m IP

- αSyn

In pu Pr t eIm m IP In pu Pr t eIm m IP

IP: Synt-1

- αSyn

+ αSyn

In pu Pr t eIm m IP In pu Pr t eIm m IP

B Synt-1 SNAP-25 Syb2WT α−Syn

SNAP-25

*

160

80

120

D

*

*

60 80 40 40

-

+

− +

WT

Syb2ΔN

20

10

0

T

W T

Bu ff PC e r P P S Sy CP b2 SW

P Sy CP b2 S-

* *

200

100

***

α-Syn + + +

F 300

+

Syb2

PC PS

W T

-

− +

α−Synuclein 30

*** WT

Syb2

0 100 50

Syb2ΔN − +

Bu

ffe r

0 αSyn

− +

− +

PC Sy P S b2 -

SNARE complexes

+

− +

PC PS

SNARE complexes

-

0

Protein levels (% of - αSyn and buffer)

αSyn

− +

P Sy CP b2 S-

PC PS

Bu ffe r

Bu ffe r

E

− +

Bu

0

α-Syn − +

ffe r

20

PC PS PC PS -S yb 2

Immunoprecipitated protein (% of - αSyn and buffer)

Syb2WT 100

Immunoprecipitated protein (% of - αSyn and buffer)

C

Fig. 1. α-Synuclein multimerization upon lipid membrane binding activates SNARE complex assembly. (A) Diagram of the experimental design. (B–D) Binding of α-synuclein to both phospholipid membranes and synaptobrevin is required for α-synuclein to enhance SNARE complex assembly. Recombinant SNARE proteins were incubated with or without addition of α-synuclein in presence or absence of charged liposomes [composition: 30% phosphatidylserine (PS), 70% phosphatidylcholine (PC)]. SNARE complexes were immunoprecipitated with antibodies to syntaxin-1 or preimmune sera (“Pre-Imm,” as a control), and coimmunoprecipitated synaptobrevin-2 (Syb2), SNAP-25, and α-synuclein were analyzed by quantitative immunoblotting. (E and F) Independently, SNARE complex assembly in the same experiment as B–D was quantitated using SDS-resistant complexes on SDS/PAGE. For input protein levels, see Fig. S1. All data shown are means ± SEM (*P < 0.05; ***P < 0.001 by Student t test; n = 3).

three conditions: Assembly of recombinant soluble t- and v-SNARE protein fragments (syntaxin-1 and SNAP-25) without or with negatively charged liposomes [composed of 30% phosphatidylserine (PS) and 70% phosphatidylcholine (PC)], and assembly of soluble t-SNARE fragments (syntaxin-1 and SNAP-25) with negatively charged liposomes containing reconstituted fulllength synaptobrevin-2. SNARE complex assembly was monitored both by coimmunoprecipitations of SNAREs (Fig. 1 B–D) and by SDS/PAGE of nonboiled samples (Fig. 1 E and F and Fig. S1) because SNARE complexes are SDS resistant in the absence of boiling (46). For the 2 of 10 | www.pnas.org/cgi/doi/10.1073/pnas.1416598111

latter experiments, we included as a further negative control Nterminally truncated synaptobrevin-2 that still assembles into SNARE complexes but does not bind α-synuclein (3, 45). We found that, in the absence of liposomes, α-synuclein did not detectably enhance SNARE complex assembly (Fig. 1 B–F). However, α-synuclein significantly increased SNARE complex assembly when negatively charged liposomes were added to the SNARE fragments; this increase was observed even when soluble fragments of all three SNAREs were used. In contrast, negatively charged liposomes alone had no effect on SNARE complex assembly in the absence of α-synuclein (Fig. 1 B–F). Using reconstituted full-length synaptobrevin-2 as opposed to the synaptobrevin-2 fragment alone already increased SNARE complex assembly in the absence of α-synuclein, presumably because the immobilized synaptobrevin-2 has a higher effective concentration than a soluble synaptobrevin-2 fragment (Fig. 1 B–F). Nevertheless, α-synuclein again enhanced SNARE complex formation even in this experimental configuration. Thus, only membrane-bound multimeric α-synuclein increases SNARE complex assembly. Importantly, when we used N-terminally truncated synaptobrevin-2 (which lacks 28 N-terminal residues and does not bind α-synuclein) in these experiments instead of wild-type synaptobrevin-2, α-synuclein had no effect on SNARE complex assembly under any condition (Fig. 1 E and F). Because α-synuclein still binds to negatively charged liposomes even in the absence of synaptobrevin binding (3, 45), these results together show that α-synuclein does not simply enhance SNARE complex assembly by binding to the liposomes and crowding out the SNAREs, but performs an active role in accelerating SNARE complex assembly by binding simultaneously to the liposomes and to synaptobrevin-2. α-Synuclein Multimerizes upon Membrane Binding. Extensive studies showed that soluble α-synuclein is monomeric (refs. 25, 26, 31, and 32; however, see refs. 33 and 34 for a divergent view), but it is unknown whether membrane-bound α-synuclein is also monomeric. To address this question, we chose chemical cross-linking experiments because it is difficult to measure multimerization of native, unmodified proteins on a membrane with other techniques (e.g., gel filtration, scanning transmission electron microscopy, or analytical ultracentrifugation). Because chemical cross-linking experiments depend on the chance event that reactive groups in a protein complex are separated from each other by a precise distance corresponding to the cross-linking agent, chemical cross-linking experiments are inherently inefficient and depend on the size of the cross-linking agent (47, 48). Moreover, at high concentrations, chemical cross-linking agents nonspecifically link proteins to each other in a diffusion-controlled reaction that is used for tissue fixation. Thus, analysis of protein complexes by chemical cross-linking requires tests of a range of cross-linker concentrations, and use of multiple cross-linking agents with different linker lengths (Figs. 2 and 3, and Fig. S2). We found that, in mouse brain homogenates obtained in the absence of detergents, four different cross-linking agents [glutaraldehyde, dimethyl suberimidate (DMS), dimethyl adipimidate (DMA), and dimethyl pimelimidate (DMP)] caused cross-linking of native α-synuclein into higher order multimers at different concentrations (Fig. 2A and Fig. S2A). Low-percentage SDS/PAGE identified multimers containing eight or more α-synuclein subunits (Fig. 2B and Fig. S2B). α-Synuclein multimerization was not a biochemical artifact of cross-linking, as no multimers of control proteins were observed, demonstrating specificity (Fig. 2F and Fig. S2H). When we examined α-synuclein in brain cytosol (which lacks membranes), we detected no cross-linking with any chemical agent (Fig. 2C and Fig. S2C). This result suggests that the chemical cross-linking of α-synuclein in brain homogenates required the presence of membranes for α-synuclein binding. Indeed, simply adding to the brain homogenate Triton X-100, a detergent that dissolves lipid membranes, blocked α-synuclein cross-linking Burré et al.

A

Brain homogenate (15% gel)

150 100 75 50 37

Tetramer

Tetramer

Dimer

Dimer

Monomer

Monomer

Octamer

Octamer

Tetramer

Tetramer

37

Dimer

Dimer

150 100 75 50 37

Tetramer

Tetramer

Dimer

Dimer

Monomer

Monomer

Tetramer

Tetramer

Dimer

Dimer

Monomer

Monomer

Tetramer

Tetramer

Dimer

Dimer

Monomer

Monomer

25 20 15

B Brain homogenate (10% gel)

C Brain cytosol

150 100 75 50

25 20

(15% gel)

15

D Brain homogenate + Triton X-100 (15% gel)

150 100 75 50 37 25 20 15

E Brain cytosol + membrane (15% gel)

150 100 75 50 37 25 20 15

F

Glutaraldehyde kDa 101 88

Brain homogenate (10% gel)

36

kDa 150 100 75

GDI

GDI

53

101 88 50

Glutaraldehyde

Synt-1+ SNAP-25 Synt-1

37

SGT

SGT+ Hsc70 75 50 SGT

37 25

α-Synuclein Multimerizes on Phospholipid Surfaces in an Antiparallel Orientation. Although chemical cross-linkers can probe for for-

Synt-1

50

SNAP-25

SNAP-25+ Synt-1

*

SNAP-25

Fig. 2. Native α-synuclein multimerizes on the surface of brain membranes. (A–E) Brain homogenates containing intact membranes (A and B; 900 μg of protein), brain cytosol (C; 150 μg of protein), brain homogenates whose membranes were solubilized by addition of Triton X-100 (D; 900 μg of protein), and brain cytosol with readdition of brain membranes (E; 150 μg of cytosolic protein plus 750 μg of membrane protein) were exposed to increasing concentrations of chemical cross-linking agents [Left, glutaraldehyde (0–0.01%); Right, dimethyl suberimidate (DMS) (0–25 mM)], and equal volumes were analyzed by immunoblotting (arrowheads, α-synuclein oligomers as determined by SDS/PAGE). For additional cross-linking experiments, quantitations, cross-linker concentrations used, size calibrations, and controls, see Fig. S2. (F) Control cross-linking experiments. Brain homogenates were exposed to same increasing concentrations of glutaraldehyde and analyzed as described above (GDI, guanine-nucleotide dissociation inhibitor; SGT, small glutamine-rich tetratricopeptide repeat-containing protein; Synt-1, syntaxin-1; arrowheads, protein monomers and homooligomers or heterooligomers as determined by SDS/PAGE migration analysis; asterisks indicate nonspecific immunosignals).

(Fig. 2D and Fig. S2D). Moreover, addition of brain membranes to brain cytosol in the ratio present in brain homogenate, restored cross-linking of α-synuclein (Fig. 2E and Fig. S2E). Note that some cross-linking agents appear to increase the size of α-synuclein, probably because they covalently attach phospholipids to α-synuclein, and that cross-linked multimeric α-synuclein seems to exhibit an increased immunoreactivity, likely because cross-linked α-synuclein adheres better to blotting membranes (Fig. 2 and Fig. S2) (49). Burré et al.

We next asked whether chemical cross-linking of α-synuclein could also be obtained in native brain tissue. We cut brain slices from mice immediately after they had been killed, and performed chemical cross-linking experiments on these acute slices (Fig. 3). We observed a similar α-synuclein cross-linking pattern in the slices as in brain homogenates (Fig. 3A). Again, the specificity of the cross-linking reactions was validated with control proteins with which we observed only cross-linking when expected, as for example for the SNARE proteins SNAP-25 and syntaxin-1 (Fig. 3B). Together, these experiments suggest that α-synuclein forms multimers when bound to phospholipid membranes. To test whether these multimers are homomultimers, we examined recombinant α-synuclein and pure liposomes. We found that purified recombinant α-synuclein could be efficiently cross-linked into higher-order multimers upon incubation with phospholipid membranes (Fig. 4). Cross-linking was observed only with negatively charged phospholipids, but not in buffer alone, or with neutral liposomes composed of 100% PC (Fig. 4). This result is consistent with the binding of α-synuclein only to negatively charged but not to neutral phospholipids (19, 20). Moreover, we found that α-synuclein truncations and mutations associated with PD reduce, but do not block α-synuclein cross-linking, in agreement with the generally modest effect of these mutations on phospholipid binding (Fig. S3 and refs. 22, 45, 50, and 51). Again, to exclude nonspecific aggregation, which is occasionally observed with recombinant α-synuclein, we performed parallel cross-linking experiments without liposomes or with uncharged liposomes, but detected only background levels of cross-linking (Fig. S3 D and E).

mation of multimers, they do not reveal a particular multimer configuration. To confirm multimerization of α-synuclein upon membrane binding by an independent method, and to investigate the configuration of α-synuclein molecules in the multimers, we used fluorescence resonance energy transfer (FRET) experiments. We introduced single cysteine residues into recombinant α-synuclein at defined positions, labeled the cysteines with fluorescent Alexa 488- or Alexa 546-maleimide, and purified the labeled proteins (Fig. 5 A and B). Liposome-binding experiments revealed that the cysteine modifications did not alter phospholipid binding except for the fluorescent modification of G41C, which produced a modest decrease in phospholipid binding (Fig. 5C and Fig. S4A). We incubated pairs of Alexa 488- and Alexa 546-modified α-synucleins with liposomes composed either of 30% PS and 70% PC (negatively charged liposomes), or of 100% PC (neutral liposomes, used as a negative control), and measured the fluorescence emission spectrum with an excitation at 490 nm. FRET manifests as a decrease in donor (Alexa 488) and an increase in acceptor (Alexa 546) emission, and can be quantified as donor fluorescence quenching by calculating the ratio of total to donor fluorescence. This experimental design allowed us to test the effect of liposomes on FRET between any given pair of Alexa 488and Alexa 546-labeled α-synucleins. In these experiments, we argued that multimerization of α-synuclein should only occur in the presence of negatively charged phospholipids because α-synuclein only binds to charged phospholipids (19, 20, 27–30). Thus, the effect of the neutral liposomes—if any—provides a tool to distinguish specific from nonspecific effects. If FRET can be specifically observed with a particular combination of labeled α-synucleins (i.e., is only detectable in the presence of negatively charged liposomes), such FRET must be due to multimerization of α-synuclein. Moreover, if some pairs of labeled α-synucleins do and some do not produce FRET, the mapping of the fluorescent probe position onto the α-synuclein sequence gives insight into the orientation of the α-synuclein molecules in the oligomers. PNAS Early Edition | 3 of 10

PNAS PLUS

DMS

NEUROSCIENCE

Glutaraldehyde kDa

(15% gel)

Tetramer Dimer

25 20

Monomer

15 150 100 75

Brain slices

(10% gel)

DMS

Octamer Tetramer

50 37

B

Dimer

15

slices

10

(15% gel)

5 0 40

Tetramer

30

Brain slices

20

(10% gel)

10 0

SGT

Monomer

75 50

25

150 100 75

Tetramer

50

*

Dimer

Syb2 Monomer

15

10 5 0 20

Tetramer

15 10 5 0

0

1

2

3

DMS (mM)

SNAP-25 + Synt1 + Syb2 SNAP-25 + Synt1

kDa 75

SGT+ Hsc70

50

SNAP-25 37

SGT

75 50 37 25 20

15

DMS

50 36

Dimer

20

Monomer

37

0 0.005 0.01 0.03

kDa 101 88

rab3A 36

mer

Glutaraldehyde (%)

kDa 101

53

*Di-

25 20

DMS

88

Tetramer

15

DMS

GDI

kDa 150 100 75 50 Brain 37

Dimer

20

α-Synuclein levels (% of input)

Brain slices

kDa 150 100 75 50 37

α-Synuclein levels (% of input)

Glutaraldehyde

A

25

Synt1 Monomer

150 100 75 50 37

*

SNAP-25

SNAP-25 + Synt1 + Syb2 SNAP-25 + Synt1 Synt-1

Fig. 3. α-Synuclein forms multimers in brain slices. (A) Acute brain slices from wild-type mice were exposed to increasing concentrations of cross-linking agents [Left, glutaraldehyde (0–0.03%); Right, dimethyl suberimidate (DMS) (0–3 mM)]. Brain slices were homogenized in sample buffer after cross-linking, and equal volumes of cross-linked proteins were analyzed by immunoblotting (Upper, 15% gels; Lower, 10% gels). The arrowheads indicate α-synuclein multimers as determined by SDS/PAGE migration analysis; the asterisks indicate nonspecific immunosignals (means ± SEMs; n = 3). (B) Control cross-linking experiments. Acute brain slices were exposed to same increasing concentrations of the cross-linker DMS and analyzed as described above (GDI, guaninenucleotide dissociation inhibitor; SGT, small glutamine-rich tetratricopeptide repeat-containing protein; Syb2, synaptobrevin-2; Synt-1, syntaxin-1).

We found that, of 15 potential FRET pairs tested, 6 pairs produced significant FRET (Fig. 5 D and E, and Fig. S4B). These data confirm that α-synuclein multimerizes on the phospholipid surface, and suggest that it does so in a defined orientation. Because the cross-linking experiments suggested that α-synuclein multimerizes on the lipid surface into at least octamers (Fig. 2), multimers are not simply dimers. A potential concern about our FRET experiments is that the α-synuclein added to liposomes may so densely populate the liposome surface with α-synuclein molecules that even α-synuclein molecules that are not multimerized may produce a FRET signal. However, this concern is inconsistent with the selective FRET observed: only some combinations of labeled α-synuclein molecules produced FRET, whereas others did not (Fig. 5 D and E, and Fig. S4B). To further address this concern, we performed titrations of a constant amount of donor-labeled α-synucleins (L8C or A140C) with increasing concentrations of acceptorlabeled α-synuclein (Fig. 6 A–D). These experiments showed that combinations that exhibited no FRET in the experiments of Fig. 5 were also unable to produce FRET at any concentration tested, whereas the positive control pair (L8C–L8C) exhibited robust FRET in a concentration-dependent manner if the acceptor was present in a ratio of at least 0.5 (Fig. 6 A–D). Thus, the FRET we observe is not due to nonspecific crowding of α-synuclein on the phospholipid surface. α-Synuclein is thought to adopt a bent, hairpin-like conformation on the lipid surface that is composed of two α-helices connected with a flexible linker (27–29, 52). The pattern of positive vs. negative FRET pairs of labeled α-synuclein molecules allows mapping the interactions of α-synuclein molecules in the membrane-bound multimer (Fig. 6E). This map reveals a striking organization of interactions, suggesting a model for the arrangement of α-synuclein molecules in the multimer (Fig. 6F). 4 of 10 | www.pnas.org/cgi/doi/10.1073/pnas.1416598111

α-Synuclein Associates with Synaptic Vesicles Docked at the Presynaptic Plasma Membrane. Is α-synuclein present on all synaptic vesicles,

or is it preferentially recruited to vesicles that are about to fuse to promote SNARE complex assembly? To address this question, we separated synaptic vesicles into two synaptic vesicle pools using sucrose gradient centrifugation (53), namely a free synaptic vesicle pool and a synaptic vesicle pool that is associated with the presynaptic plasma membrane. We then probed each gradient fraction for markers of integral synaptic vesicle proteins and presynaptic plasma membrane proteins. Integral synaptic vesicle proteins such as synaptophysin revealed a bimodal distribution across the gradient, reflecting free synaptic vesicles in fractions 6–13, and a plasma membrane-bound synaptic vesicle pool in fractions 23–32 (Fig. 7). Out of the three SNARE proteins, only synaptobrevin-2 revealed the same bimodal distribution consistent with its localization to synaptic vesicles, whereas the plasma membrane SNARE proteins SNAP-25 and syntaxin-1 were detected mainly in the plasma membrane-bound vesicle pool. Strikingly, α-synuclein was not associated with the free synaptic vesicle pool, but was detected only in fractions containing plasma membrane-bound synaptic vesicles (note that fractions 1–4 do not contain synaptic vesicles; thus, immunosignals in these fractions represent soluble protein). When we compared the distribution of α-synuclein with the distribution of SNARE complexes across the gradient, we found that α-synuclein exactly matched their pattern. Thus, α-synuclein is not associated with free synaptic vesicles, but only with synaptic vesicles that are docked at the presynaptic plasma membrane where SNARE complex assembly occurs. Discussion α-Synuclein is an abundant presynaptic protein that physiologically acts as a nonclassical chaperone for SNARE complex assembly, and pathologically contributes to the pathogenesis of PD and Burré et al.

(15% gel)

Tetramer Dimer

Tetramer

50

α-Syn

150

(6% TA gel)

75

α-Syn

(10% gel) 50

100 75

α-Syn

(6% TA gel)

Tetramer

150 100 75

25 20 15

Buffer

α-Synuclein levels (% of input)

E

Buffer 120 100 80 60 40 20 0

0

Glutaraldehyde

***

Tetramer Dimer

Tetramer Dimer

Monomer

25 20 15

Monomer

Buffer

Charged liposomes

Tetramer

DMS

DMS

kDa 150 100 75 50 37

Neutral liposomes

Neutral liposomes

Dimer

Negatively charged liposomes

D

***

4 5 0 .01 .02 .03 .04 .05 1 2 3 0 0 0.0 0.0 0.0 0.0 0.0 0 0 0 Glutaraldehyde (%) Glutaraldehyde (%)

F

Buffer

α-Synuclein levels (% of input)

Glutaraldehyde

kDa 150 100 75 50 37

12-mer 11-mer Decamer Nonamer Octamer Hexamer Tetramer

250

Negatively charged liposomes

C

Dimer

37

12-mer 11-mer Decamer Nonamer Octamer Hexamer

250

Octamer Hexamer Tetramer

100

Dimer

37

Monomer

150

Octamer Hexamer

75

Tetramer Dimer

25 20

Monomer

100

α-Syn

α-Syn

kDa 150 100 75 50 37

(15% gel)

25 20

150

(10% gel)

DMS

B

100 80 60 40 20 0

Neutral liposomes

Dimer

0

5

Neutral liposomes Charged liposomes

*

10 15 20 25 0 DMS (mM)

Tetramer

5

***

10 15 20 25 DMS (mM)

Fig. 4. Reconstitution of α-synuclein multimerization on charged liposomes. (A and B) Recombinant α-synuclein was incubated with negatively charged liposomes (composition: 30% PS, 70% PC) and exposed to increasing concentrations of the chemical cross-linkers glutaraldehyde (A; 0–0.05%) or DMS (B; 0–25 mM). Cross-linked species were analyzed by 15% and 10% SDS/PAGE (Top and Middle) or 6% Tris-acetate PAGE (Bottom). (C and D) Recombinant α-synuclein was incubated in buffer or with neutral liposomes (composition: 100% PC), and cross-linked as described above. Cross-linked species were analyzed by 15% SDS/PAGE. (E and F) Quantitations of dimerization and tetramerization of α-synuclein upon chemical cross-linking in buffer, on neutral or negatively charged liposomes, using glutaraldehyde (E) or DMS (F). Multimerization was plotted relative to α-synuclein input levels and was measured by quantitative immunoblotting (means ± SEMs; n = 3; *P < 0.05; ***P < 0.001).

other neurodegenerative disorders (1–18). α-Synuclein is largely cytosolic, but readily binds to membranes, and associates with synaptic vesicles in the presynaptic terminal (22, 23). In the present study, we have asked three questions that are central to an understanding of the role of α-synuclein in health and disease: (i) Does α-synuclein function as a SNARE complex assembly chaperone in its soluble or membrane-bound form? (ii) Is α-synuclein bound to membranes a monomer, or does it form a multimer upon membrane binding? (iii) Is α-synuclein normally associated with all synaptic vesicles, or only with those vesicles that are docked and engaged in SNARE complex formation? Our data provide initial answers to these questions. We demonstrate that α-synuclein acts as a SNARE chaperone only when bound to membranes, that membrane binding causes α-synuclein to form higher-order multimers larger than octamers that have a defined orientation, and that α-synuclein is only stably associated with vesicles that are docked to the presynaptic plasma membrane and are thought to be engaged in SNARE complex formation. Thus, our results suggest a physiological folding pathway for α-synuclein whereby unstructured soluble α-synuclein not only adopts an α-helical conformation upon membrane binding, but simultaneously associates into multimers on the membrane surBurré et al.

face (Fig. 8). These multimers are the functionally active form of α-synuclein in promoting SNARE complex assembly. Consistent with this function, α-synuclein is stably bound to synaptic vesicles only when these vesicles are docked and are thought to be in the process of forming SNARE complexes. We believe that these observations not only define an unanticipated multimeric state of α-synuclein that is selectively formed upon membrane binding but also localize the point of action of α-synuclein to the site of synaptic vesicle fusion (Fig. 8). Our results raise new questions. First, how does α-synuclein work? The multimers cluster synaptobrevin-2/VAMP2 and thus increase the local concentration of synaptobrevin molecules, while at the same time covering a fraction of the lipid surface. Our data show that synaptobrevin binding by α-synuclein is essential for α-synuclein to enhance SNARE complex formation (Fig. 1 E and F), and membrane binding is also required even when synaptobrevin is not actually in the membrane (again, Fig. 1). Thus, a plausible hypothesis is that multimeric α-synuclein promotes SNARE complex assembly by increasing the local synaptobrevin concentration, which would be expected to enhance SNARE complex assembly via mass action law. PNAS Early Edition | 5 of 10

NEUROSCIENCE

α-Syn

kDa 150 100 75 50 37

PNAS PLUS

Glutaraldehyde

A

B

N

Helix 1

18

C

Helix 2 96

41

C

140

kDa 36 28

Cysteine substitutions Alexa-488 & -546 modifications

80

Lipid-bound protein (%)

α-Synuclein

M 1 L8 C-4 G C- 88 41 48 K 9 C- 8 4 A1 6 C 88 40 - 4 8 C 8 M -4 1C 8 L8 -5 8 4 G C-5 6 4 4 K9 1C 6 A 1 6C- 546 4 0 54 C 6 -5 46

A

60

*

40

*

20 0

M

1C WT L 8 -48 G C- 8 41 4 8 K C- 8 A196C 488 40 -4 C 8 M -4 8 1C 88 L 8 -54 G C- 6 41 5 K9 C- 4 6 5 A1 6C 46 40 -54 C 6 -5 46

19

Fluorescence (A.U.)

M1CA140C

10 520

540

560

580

0 600 500

12 8

20

4

10 520

540

12

560

580

0 600 500

8

20

4

10 520

540

560

580

0 600 500

12 8

20

4

10 520

540

560

580

0 600 500

Wavelength (nm)

E

560

580

600

Donor Donor + Acceptor

G41CG41C

520

540

560

580

600

Donor Donor + Acceptor

K96CA140C 520

540

40 Donor Donor + Acceptor 30

16

0 500

540

40 Donor Donor + Acceptor 30

16

0 500

520

40 Donor Donor + Acceptor 30

16

0 500

L8CA140C

560

580

Fluorescence (A.U.)

20

4

Charged liposomes (PC:PS) Neutral liposomes (PC)

Donor Donor + Acceptor

Fluorescence (A.U.)

8

0 500

Fluorescence (A.U.)

M1CL8C

L8CL8C

12

Fluorescence (A.U.)

M1CM1C

Donor 40 Donor + Acceptor 30

16

600

Donor Donor + Acceptor

Fluorescence (A.U.)

Fluorescence (A.U.)

Charged liposomes (PC:PS) Neutral liposomes (PC)

Fluorescence (A.U.)

D

12

10

4 0 500

520

540

16 12

0 600 500 40 Donor Donor + Acceptor 30 560

580

520

540

560

580

600

Wavelength (nm)

1

600

520

540

20

560

580

600

Donor Donor + Acceptor

16 12

0 500

520

540

40

560

580

600

Donor Donor + Acceptor

30 20

8 10

4 0 500 20

520

540

16

0 600 500 40 Donor Donor + Acceptor 30 560

580

4 0 500

520

540

560

580

600

Donor Donor + Acceptor

20 10 520

540

560

580

0 600 500

520

540

560

580

600

Wavelength (nm)

2 1

1C M -M1 1 M C- C 1C L8 M -G C 1 4 M C-K 1C 1C 9 -A 6C 1 L 8 40 L8 C - L C C 8C L8 -G4 C 1 L 8 -K C C 9 G - A 6C 41 1 4 K9 C-G 0 C 6 4 K9 C-G 1C K9 6C 41C 6 -K A1 C- 96 A C A1 40C 140 40 -G C C 41 -A C 14 0C

0 M

1C M -M1 M 1C- C 1C L8 M -G C 1 4 M C-K 1C 1C 9 -A 6C 1 L8 40 L8 C-L C C 8C L 8 -G4 C 1 L8 - K C C 9 G -A 6 C 41 14 K9 C-G 0C 6 4 K9 C-G 1C K9 6C 41C 6 -K A1 C-A 96C 4 1 A1 0C 40 40 -G C C 41 -A C 14 0C

580

Neutral liposomes (PC)

0 M

560

Donor Donor + Acceptor

3 RD+A / RD

2

540

10

4

Wavelength (nm)

Charged liposomes (PC:PS)

520

20

8

0 500

Donor Donor + Acceptor

20

8

A140C- 12 G41C 8

3 RD+A / RD

40 Donor Donor + Acceptor 30

16

Fig. 5. α-Synuclein multimerizes on the phospholipid membrane surface in a defined orientation. (A) Fluorescent α-synuclein labeling scheme. Single-cysteine substitutions were introduced into α-synuclein at position 1 (M1C), 8 (L8C), 41 (G41C), 96 (K96C), and 140 (A140C) for modification with Alexa 488- or Alexa 546-maleimide. (B) SDS/PAGE analysis of purified Alexa 488- or Alexa 546-labeled recombinant α-synuclein proteins (3 μg of protein per lane). (C) Liposome binding of wild-type and mutant α-synuclein. Recombinant α-synuclein was incubated with negatively charged liposomes (30% PS, 70% PC) and subjected to a flotation assay. Eight fractions were collected from top to bottom, and α-synuclein was quantified in each fraction (Fig. S4A). Liposome binding is expressed as the amount of α-synuclein present in the liposome fractions of the flotation gradient in percent of the total α-synuclein. (D and E) Fluorescence resonance energy transfer (FRET) between donor and acceptor α-synuclein. Alexa 488-labeled donor α-synuclein (2.5 μg) was incubated for 2 h with Alexa 546-labeled acceptor or unlabeled α-synuclein (2.5 μg) in the presence of 100 μg of charged (red spectra) or neutral liposomes (blue spectra). Emission spectra (D) were used for calculation of FRET signals (E). For further spectra, see Fig. S4B. Data in C and E are means ± SEM (*P < 0.05 by Student t test; n = 3).

Second, how do our results relate to recent prominent studies on the multimerization of α-synuclein? Initially, soluble α-synuclein was described as a natively unfolded monomer (25, 26). This view was challenged when α-synuclein purified from human erythrocytes was found to be a stably folded tetramer, as judged by analytical ultracentrifugation (33). Confusingly, however, in the same study electron microscopy failed to detect a monodisperse population of α-synuclein tetramers, and high-resolution NMR studies also did not observe a stable tetramer of α-synuclein, but revealed a dynamic multimerization of α-synuclein molecules at very high concentrations (54). Subsequent work, moreover, demonstrated that both recombinant α-synuclein and native cytosolic α-synuclein purified from brain were monomeric and natively unfolded, suggesting that, at least in brain (where α-synuclein is most highly expressed and its pathological role in neurodegeneration is most important), soluble α-synuclein is a natively unfolded monomer (31, 32). Indeed, the multimers that we identified here form only upon membrane binding and are very different from the stable tetramer 6 of 10 | www.pnas.org/cgi/doi/10.1073/pnas.1416598111

proposed for α-synuclein purified from erythrocyte cytosol (33). We find no evidence for a stable tetramer (Figs. 2–4) but observe that the membrane-induced multimerization produces large α-synuclein assemblies containing more than eight subunits (Figs. 2–4). Moreover, we find that α-synuclein multimers are absent from cytosol—simple disruption of membranes by detergents disrupts all multimers (Fig. 2D)—whereas the previous postulate was that α-synuclein tetramer is a soluble cytosolic species. Thus, our data are fully consistent with the studies that concluded that α-synuclein is a natively unfolded monomer in solution and are inconsistent with the notion of a stable tetramer. Third, the arguably most important question raised by our data is how the α-synuclein multimers that are physiologically assembled on phospholipid membrane surfaces and promote SNARE complex assembly relate to pathological oligomers and fibrils. Does binding and multimerization of α-synuclein on synaptic vesicles induce aggregation or protect against aggregation? Although the relatively high concentration of α-synuclein on the synaptic vesicle membrane may increase its propensity to aggregate, the formation Burré et al.

2

600

0

0 0. 2 0. 4 0. 6 0. 8 1

550 Wavelength (nm)

L8C-546 (fold L8C-488)

L8C-A140C 0x 0.2x 0.4x 0.6x 0.8x 1x

10

3 2 1 0

550 Wavelength (nm)

0 0. 2 0. 4 0. 6 0. 8 1

0 500

4 RD+A / RD

20

600

A140C-546 (fold L8C-488)

A140C-L8C

8 4 0 500

550 Wavelength (nm)

D

4 3 2 1 600

0

0 0. 2 0. 4 0. 6 0. 8 1

0x 0.2x 0.4x 0.6x 0.8x 1x

12

RD+A / RD

A140C-546 (fold L8C-488)

A140C-A140C 0x 0.2x 0.4x 0.6x 0.8x 1x

12 8 4 0 500

E

4

550 Wavelength (nm)

600

Interaction Map

F

Helix 1 Helix 2 N 18 96 41 18 96 41 Helix 2 Helix 1 N

3 2 1 0

Materials and Methods 0 0. 2 0. 4 0. 6 0. 8 1

Fluorescence (A.U.)

3 1

0 500

C 16

Fluorescence (A.U.)

RD+A / RD

10

RD+A / RD

Fluorescence (A.U.)

20

*

4

A140C-546 (fold L8C-488)

Model

C C 140 140

C

N

Helix 1

N

Helix 1 Helix 2

Fluorescence (A.U.)

0x 0.2x 0.4x 0.6x 0.8x 1x

Helix 2

B

5

30

Helix 2

C

C 1 xileH

N

1 xileH Helix 2

N C

Fig. 6. α-Synuclein multimers on the phospholipid membrane surface adopt an antiparallel configuration. (A–D) FRET between donor and acceptor α-synuclein. Alexa 488-labeled donor α-synuclein L8C (A and B; 2.5 μg) or A140C (C and D; 2.5 μg) was incubated for 2 h in the presence of 100 μg of charged liposomes with the indicated increasing amounts of Alexa 546labeled acceptor α-synuclein L8C (A and C ) or A140C (B and D), or with unlabeled α-synuclein to balance total α-synuclein amounts. Emission spectra (Left) were used for calculation of FRET signals (Right; means ± SEM; *P < 0.05; n = 3). (E and F) Modeling of α-synuclein multimers based on FRET experiments. E shows schematic drawing of two α-synuclein proteins (red lines, positive FRET signals; blue lines, no FRET signals. F shows the model of

Burré et al.

Fractionation of Brain Tissue and Synaptic Vesicle Isolation. One mouse brain was homogenized in 2 mL of homogenization buffer (PBS, pH 7.4, for glutaraldehyde cross-linking experiments, 150 mM NaCl, 50 mM Hepes-NaOH, pH 8.5, for imidoester cross-linking experiments) including EDTA-free protease inhibitors (Roche). For experiments with brain homogenate, samples were directly used. For further fractionation, nuclei were removed by centrifugation for 15 min at 4 °C and 1,000 × gav. Supernatants were then separated into cytosolic and membrane fractions using ultracentrifugation for 1 h at 280,000 × gav and 4 °C. Supernatants were removed, and pellets were homogenized in an equal volume of homogenization buffer. Separation was repeated three times to remove any trace residues of cytosol in membrane fractions, and vice versa. To solubilize membranes, brain homogenate was incubated with 0.5% Triton X-100 for 2 h at 4 °C. Insoluble material was pelleted by centrifugation for 20 min at 10,000 × gav and 4 °C. Isolation of synaptic vesicles and separation of free and docked vesicle populations was performed as previously described (53). Recombinant Protein Expression and Purification. Full-length human α-synuclein cDNA was inserted into modified pGEX-KG vectors (GE Healthcare), containing an N-terminal TEV protease recognition site. α-Synuclein contained an extra N-terminal glycine after cleavage with TEV protease in addition to the normal N terminus of α-synuclein. Mutant α-synuclein constructs were generated by site-specific mutagenesis, according to the manufacturer’s protocol (Stratagene). All proteins were expressed as GST fusion proteins in bacteria (BL21 strain), essentially as described (3). Bacteria were grown to OD 0.6 (measured at

a possible arrangement of α-synuclein multimers on the surface that would be consistent with the FRET signals.

PNAS Early Edition | 7 of 10

PNAS PLUS

of α-helices upon lipid binding may impede misfolding of α-synuclein into β-sheet–containing conformations, and thereby hamper pathological oligomerization and aggregation into fibrils. Regardless of which of these possibilities will turn out to be true, our data support the notion that α-synuclein is subject to constant conformational switches associated with binding to and dissociation from synaptic membranes in vivo. Our study also has several limitations. Given that we were able to demonstrate the multimerization of α-synuclein upon membrane binding by two independent methods that have little in common (chemical cross-linking with multiple different agents and FRET with fluorescently labeled α-synuclein), and that both of these methods showed the absolute requirement for membrane binding in order for multimerization to occur, we think that the possibility for artifacts is low. Moreover, we found similar cross-linking patterns in brain slices as with purified α-synuclein binding to liposomes (Figs. 3 and 4), demonstrating that the α-synuclein multimers are present in vivo and that they constitute homomultimers. However, we do not know the tertiary structure of α-synuclein on the membrane surface. Although the FRET data are intriguing in suggesting an ordered array, they obviously do not conclusively characterize such an array at the atomic level. Moreover, we do not yet understand the dynamics of the α-synuclein multimers on the lipid surface and have no insight into how their assembly and dissociation are regulated. Independent of the answers to these open questions, the multimerization of α-synuclein on the lipid surface that we reveal here suggests a mode of action connecting protein and lipid interactions to each other, and broadly resembles that of other lipidbinding proteins such as BAR-domain proteins (55, 56). In summary, viewed together our data describe a folding pathway whereby monomeric cytosolic α-synuclein physiologically folds into α-helical multimers upon phospholipid-membrane binding in presynaptic terminals (Fig. 8). The membrane-bound multimeric α-synuclein is composed of ordered array and promotes SNARE complex assembly, whereas the monomeric α-synuclein lacks such activity, suggesting a dynamic, activity-dependent fluctuation in the state and conformation of α-synuclein that contributes to maintaining nerve terminal function.

NEUROSCIENCE

L8C-L8C

A

A

35

33

31

29

27

25

23

21

19

17

15

13

9

11

7

5

1

3

Gradient fraction

α-Syn Syp1

Syb2

SNAP-25 Synt1 SNARE complexes Free synaptic vesicles 12 10 8 6 4 2 0 12 10 8 6 4 2 0 12 10 8 6 4 2 0 1

α−Synuclein Free synaptic vesicles

Synaptophysin-1

Synaptobrevin-2

6

12 10 8 6 4 2 0 11 16 21 26 31 36 1 Fraction

Docked synaptic vesicles

SNAP-25 Free synaptic vesicles

C

0.6

Sy Free/docked synaptic vesicles na Sy ptop na hy pto sin bre -1 v SN in-2 Sy AP-2 SN α− ntax 5 AR Sy inE c nuc 1 om lein ple xe s

Protein levels (% of total protein)

B

12 10 8 6 4 2 0 12 10 8 6 4 2 0

0.5 0.4

Syntaxin-1

0.3 0.2 0.1

SNARE complexes

6

11 16 21 26 31 36 Fraction

0

** *

* *** **

Fig. 7. α-Synuclein associates with synaptic vesicles docked at the presynaptic plasma membrane. (A) Separation of synaptic vesicles by sucrose gradient centrifugation into free (fractions 6–13) and docked vesicle pools (fractions 23–32). Equal volumes of each fraction were analyzed by immunoblotting (Syb2, synaptobrevin-2; α-Syn, α-synuclein; Synt1, syntaxin-1; Syp1, synaptophysin 1). (B and C) Quantitation of the distribution of the indicated proteins across the sucrose gradient (B) and of relative protein levels in the free vs. docked synaptic vesicle pool (C; means ± SEM; *P < 0.05; **P < 0.01; ***P < 0.001 by Student t test; n = 3).

600 nm), and protein expression was induced with 0.05 mM isopropyl β-Dthiogalactoside for 6 h at room temperature. Bacteria were harvested by centrifugation for 20 min at 2,100 × g, and pellets were resuspended in solubilization buffer [PBS, 0.5 mg/mL lysozyme, 1 mM PMSF, DNase, and an EDTA-free protease inhibitor mixture (Roche)]. Cells were broken by sonication, and insoluble material was removed by centrifugation for 30 min at 7,000 × gav and 4 °C. Proteins were affinity-purified using glutathione Sepharose bead (GE Healthcare) incubation overnight at 4 °C, followed by TEV protease (Invitrogen) cleavage overnight at room temperature. His-tagged TEV protease was removed by incubation with Ni-NTA (Qiagen) overnight at 4 °C. SNARE proteins were expressed as described (3). Synaptobrevin-2ΔN lacks 28 N-terminal residues and does not bind to α-synuclein (3). Protein concentrations were assessed using the bicinchoninic acid method according to the manufacturer’s protocol (Thermo Scientific). Liposome Preparation and Liposome-Binding Assay. Liposomes were prepared as previously described (3). For lipid-binding assays, 1 mg of brain PC (Avanti Polar Lipids) or 0.7 mg of PC and 0.3 mg of brain PS (PCPS) (Avanti Polar Lipids) in chloroform were dried in a glass vial under a nitrogen stream. Residual chloroform was removed by lyophilization for 2 h. Small unilamellar vesicles were formed by sonicating 1 mg/mL of lipids in PBS on ice. For lipidbinding studies, 5 μg of protein of α-synuclein were incubated with 100 μg of liposomes for 2 h at room temperature. Samples were then either subjected to a liposome flotation assay (3) or cross-linking experiments. Chemical Cross-linking. Cross-linking experiments were performed using glutaraldehyde (TCI America), or the imidoesters DMA, DMP, and DMS (all from Thermo Scientific). Glutaraldehyde cross-linking experiments were carried out in PBS, pH 7.4, imidoester cross-linking experiments in 50 mM Hepes buffer, pH 8.5, with 150 mM NaCl. Thirty microliters of brain homogenate (900 μg),

8 of 10 | www.pnas.org/cgi/doi/10.1073/pnas.1416598111

30 μL of brain lysate (900 μg), 15 μL of brain cytosol (150 μg) plus 15 μL of buffer, 15 μL of brain cytosol (150 μg) plus 15 μL of brain membranes (750 μg), or 30 μL of recombinant myc-tagged α-synuclein (2 μg in buffer or on liposomes) were incubated for 5 min with 10 μL of glutaraldehyde. For imidoester cross-linking experiments, samples were incubated with 5 μL of cross-linker in Hepes buffer, pH 8.5, for 2 h at room temperature. Reaction was stopped by addition of 10 μL of 1 M Tris, pH 8.8. For cross-linking experiments in brain slices, equal tissue amount of 250-μm brain slices from wild-type mice were incubated in 500 μL of PBS with glutaraldehyde for 2 h at room temperature, or in 500 μL of Hepes, pH 8.5, with DMS for 2 h at room temperature. Reaction was stopped by addition of 20 μL of 1 M Tris, pH 8.8. Supernatants were removed, and brain slices were sonicated in 100 μL of 2× Laemmli sample buffer containing 100 mM DTT. For cross-linking experiments in vitro, 5 μg of recombinant purified α-synuclein were incubated with 100 μg of liposomes in 100 μL of buffer, or in 100 μL of buffer for 5 min with 10 μL of glutaraldehyde or for 2 h with 10 μL of DMS. Reaction was stopped by addition of 10 μL of 1 M Tris, pH 8.8. Samples were boiled before separating via SDS/PAGE. FRET Experiments. One hundred micromolar GST-fusion protein of α-synuclein containing a cysteine (M1C, L8C, G41C, K96C, or A140C) were captured on GS4B beads (GE Healthcare). GST-α-synuclein was reduced with 1 mM DTT for 20 min at 4 °C. Beads were washed four times with PBS containing protease inhibitors and protein was labeled with 2 mM Alexa 488 C5 maleimide or Alexa 546 C5 maleimide (Invitrogen) overnight at 4 °C in the dark. Beads were washed four times with PBS to remove residual unbound dye, and α-synuclein was eluted from the GST moiety using TEV protease overnight at room temperature. His-tagged TEV protease was removed using Ni-NTA agarose (Qiagen). For FRET experiments, 2.5 μg of Alexa 488-labeled α-synuclein and 2.5 μg of Alexa 546-labeled α-synuclein were incubated with and without 100 μg of

Burré et al.

Synaptobrevin/ VAMP

Cytosol Cytosol

Synaptic vesicle

Cytosol

Cytosol

Soluble natively unfolded monomeric α-synuclein N

C C N

C

Docking & Priming

N N

Fusion C

C N

C

Syntaxin-1

Multimeric membrane-bound helical α-synuclein

C

N N

SNAP-25

partial

full SNARE complex assembly

Presynaptic plasma membrane

Synaptic cleft

Fig. 8. Scheme of a physiological folding pathway for α-synuclein. Soluble α-synuclein is natively unstructured and monomeric (Left). Upon binding to synaptic vesicles during docking and priming of the vesicles, which involves partial SNARE complex assembly, α-synuclein undergoes a conformational change, folds into a “broken” amphipathic α-helix and multimerizes. As a result of membrane binding, α-synuclein becomes competent to promote SNARE complex assembly during docking and priming of synaptic vesicles.

liposomes in 100 μL of PBS for 2 h at room temperature in the dark. For control experiments containing only donor α-synuclein, acceptor α-synuclein was replaced with unlabeled α-synuclein to keep the total α-synuclein concentration constant. For titration experiments, 2.5 μg of donor α-synuclein was incubated with 0, 0.5, 1, 1.5, 2, or 2.5 μg of acceptor α-synuclein. Total amount of α-synuclein was kept constant by including unlabeled α-synuclein. Emission spectra were measured using a FluoroMax-4 spectrofluorometer (Horiba Jobin Yvon; excitation: 490 nm; slit width, 5 nm; emission: 500–650 nm; slit width, 5 nm). FRET signals were measured by calculating ratios of RD+A (i.e., ratio of fluorescence of donor plus acceptor α-synuclein at 573 nm and of fluorescence of donor plus acceptor α-synuclein at 519 nm, each subtracted by fluorescence of acceptor α-synuclein) and RD (i.e., ratio of fluorescence of donor α-synuclein at 573 nm and of fluorescence of donor α-synuclein at 519 nm; Fig. 6 A–D, Right). SNARE Complex Assembly Assays. Five micrograms of each soluble SNARE protein, or 10 μg of reconstituted synaptobrevin-2, were incubated with or without addition of 25 μg of α-synuclein, and in the presence of 400 μg of liposomes or an equivalent volume of buffer (PBS, pH 7.4) overnight at 4 °C. For analysis of SNARE complexes as high–molecular-mass bands, 2× SDS sample

1. George JM, Jin H, Woods WS, Clayton DF (1995) Characterization of a novel protein regulated during the critical period for song learning in the zebra finch. Neuron 15(2): 361–372. 2. Iwai A, et al. (1995) The precursor protein of non-A beta component of Alzheimer’s disease amyloid is a presynaptic protein of the central nervous system. Neuron 14(2): 467–475. 3. Burré J, et al. (2010) α-synuclein promotes SNARE-complex assembly in vivo and in vitro. Science 329(5999):1663–1667. 4. Polymeropoulos MH, et al. (1997) Mutation in the alpha-synuclein gene identified in families with Parkinson’s disease. Science 276(5321):2045–2047. 5. Krüger R, et al. (1998) Ala30Pro mutation in the gene encoding α-synuclein in Parkinson’s disease. Nat Genet 18(2):106–108. 6. Zarranz JJ, et al. (2004) The new mutation, E46K, of α-synuclein causes Parkinson and Lewy body dementia. Ann Neurol 55(2):164–173. 7. Kiely AP, et al. (2013) α-Synucleinopathy associated with G51D SNCA mutation: A link between Parkinson’s disease and multiple system atrophy? Acta Neuropathol 125(5): 753–769. 8. Proukakis C, et al. (2013) A novel α-synuclein missense mutation in Parkinson disease. Neurology 80(11):1062–1064. 9. Singleton AB, et al. (2003) α-Synuclein locus triplication causes Parkinson’s disease. Science 302(5646):841. 10. Ibáñez P, et al. (2004) Causal relation between α-synuclein gene duplication and familial Parkinson’s disease. Lancet 364(9440):1169–1171. 11. Spillantini MG, et al. (1997) α-Synuclein in Lewy bodies. Nature 388(6645): 839–840. 12. Wakabayashi K, Matsumoto K, Takayama K, Yoshimoto M, Takahashi H (1997) NACP, a presynaptic protein, immunoreactivity in Lewy bodies in Parkinson’s disease. Neurosci Lett 239(1):45–48.

Burré et al.

Gel Electrophoresis and Protein Quantitation. SDS/PAGE. Protein samples were separated by SDS/PAGE and transferred onto nitrocellulose membranes. Blots were blocked in Tris-buffered saline containing 0.1% Tween 20 (Sigma) and 3% fat-free milk for 2 h at room temperature. The blocked membrane was incubated in blocking buffer containing primary antibody for 1 h, followed by five washes. Membranes were incubated in blocking buffer containing horseradish peroxidase (HRP)-conjugated secondary antibody (MP Biomedicals; 1:5,000) for 2 h at room temperature. HRP immunoblots were developed using enhanced chemiluminescence (GE Healthcare). Tris-acetate PAGE. For resolution of high mass range, proteins were separated on 6% Tris-acetate gels and transferred onto PVDF membranes according to the manufacturer’s protocol (Invitrogen). After transfer, membranes were destained in methanol, and were blocked and incubated with primary antibody as described above. All quantitative immunoblotting experiments were performed with iodinated secondary antibodies (PerkinElmer, 1:1,000) overnight at room temperature as described. 125I blots were exposed to phosphorimager screens (GE Healthcare) overnight and scanned using a Typhoon scanner (GE Healthcare), followed by quantification with ImageQuant software (GE Healthcare). Primary Antibodies. Monoclonal antibodies used were α-synuclein (610786; BD Transduction), GDI (cl. 81.2; SYSY), neuroligin-1 (cl. 4C12; SYSY), rab3A (cl. 42.4; SYSY), SNAP-25 (cl. 71.1, SYSY; SMI81, Covance), synaptobrevin-2 (cl. 69.1; SYSY), and syntaxin-1 (HPC-1; SYSY). Polyclonal antibodies used were α-synuclein (U1127), syntaxin-1 (438B), and SGT (CHAT33). Statistical Analyses. Curves obtained from cross-linking experiments were analyzed by two-way repeated-measure ANOVA, using GraphPad Prism. Coimmunoprecipitation experiments are shown as recovered protein (relative to the input), which was first normalized to the immunoprecipitated protein and then normalized to the control. All other data shown are means ± SEMs and were analyzed by Student’s unpaired two-tailed t test to compare the data groups. ACKNOWLEDGMENTS. We thank Dr. Axel Brunger (Stanford University) for helpful discussions. This work was supported by National Institutes of Health Grants AG010770 and NS077906 (to T.C.S.).

13. Arawaka S, Saito Y, Murayama S, Mori H (1998) Lewy body in neurodegeneration with brain iron accumulation type 1 is immunoreactive for α-synuclein. Neurology 51(3): 887–889. 14. Gai WP, Power JH, Blumbergs PC, Blessing WW (1998) Multiple-system atrophy: A new α-synuclein disease? Lancet 352(9127):547–548. 15. Desplats P, et al. (2009) Inclusion formation and neuronal cell death through neuronto-neuron transmission of α-synuclein. Proc Natl Acad Sci USA 106(31):13010–13015. 16. Volpicelli-Daley LA, et al. (2011) Exogenous α-synuclein fibrils induce Lewy body pathology leading to synaptic dysfunction and neuron death. Neuron 72(1):57–71. 17. Luk KC, et al. (2012) Pathological α-synuclein transmission initiates Parkinson-like neurodegeneration in nontransgenic mice. Science 338(6109):949–953. 18. Rey NL, Petit GH, Bousset L, Melki R, Brundin P (2013) Transfer of human α-synuclein from the olfactory bulb to interconnected brain regions in mice. Acta Neuropathol 126(4):555–573. 19. Davidson WS, Jonas A, Clayton DF, George JM (1998) Stabilization of α-synuclein secondary structure upon binding to synthetic membranes. J Biol Chem 273(16):9443–9449. 20. Jo E, McLaurin J, Yip CM, St George-Hyslop P, Fraser PE (2000) α-Synuclein membrane interactions and lipid specificity. J Biol Chem 275(44):34328–34334. 21. Nuscher B, et al. (2004) α-Synuclein has a high affinity for packing defects in a bilayer membrane: A thermodynamics study. J Biol Chem 279(21):21966–21975. 22. Jensen PH, Nielsen MS, Jakes R, Dotti CG, Goedert M (1998) Binding of α-synuclein to brain vesicles is abolished by familial Parkinson’s disease mutation. J Biol Chem 273(41): 26292–26294. 23. Kahle PJ, et al. (2000) Subcellular localization of wild-type and Parkinson’s diseaseassociated mutant α-synuclein in human and transgenic mouse brain. J Neurosci 20(17):6365–6373. 24. Fortin DL, et al. (2004) Lipid rafts mediate the synaptic localization of α-synuclein. J Neurosci 24(30):6715–6723.

PNAS Early Edition | 9 of 10

PNAS PLUS

buffer was added, and samples were separated by SDS/PAGE and immunoblotted. For analysis of SNARE complex assembly by coimmunoprecipitation, samples were incubated with polyclonal syntaxin-1 antibody (438B) and 50 μL of a 50% (vol/vol) slurry of protein A-Sepharose beads (GE Healthcare) for 2 h at 4 °C. Following five washes with 0.1% Triton X-100 in PBS, bound proteins were eluted with 2× SDS samples buffer and boiled for 20 min at 100 °C. Coprecipitated proteins were separated by SDS/PAGE, with 5% of the input in the indicated lane.

NEUROSCIENCE

Physiological function of α-synuclein

25. Weinreb PH, Zhen W, Poon AW, Conway KA, Lansbury PT, Jr (1996) NACP, a protein implicated in Alzheimer’s disease and learning, is natively unfolded. Biochemistry 35(43):13709–13715. 26. Kim J (1997) Evidence that the precursor protein of non-A β component of Alzheimer’s disease amyloid (NACP) has an extended structure primarily composed of random-coil. Mol Cells 7(1):78–83. 27. Chandra S, Chen X, Rizo J, Jahn R, Südhof TC (2003) A broken alpha-helix in folded alpha-Synuclein. J Biol Chem 278(17):15313–15318. 28. Jao CC, Der-Sarkissian A, Chen J, Langen R (2004) Structure of membrane-bound α-synuclein studied by site-directed spin labeling. Proc Natl Acad Sci USA 101(22): 8331–8336. 29. Ulmer TS, Bax A (2005) Comparison of structure and dynamics of micelle-bound human α-synuclein and Parkinson disease variants. J Biol Chem 280(52):43179–43187. 30. Lokappa SB, Ulmer TS (2011) α-Synuclein populates both elongated and broken helix states on small unilamellar vesicles. J Biol Chem 286(24):21450–21457. 31. Fauvet B, et al. (2012) α-Synuclein in central nervous system and from erythrocytes, mammalian cells, and Escherichia coli exists predominantly as disordered monomer. J Biol Chem 287(19):15345–15364. 32. Burré J, et al. (2013) Properties of native brain α-synuclein. Nature 498(7453):E4–E6, discussion E6–E7. 33. Bartels T, Choi JG, Selkoe DJ (2011) α-Synuclein occurs physiologically as a helically folded tetramer that resists aggregation. Nature 477(7362):107–110. 34. Dettmer U, Newman AJ, Luth ES, Bartels T, Selkoe D (2013) In vivo cross-linking reveals principally oligomeric forms of α-synuclein and β-synuclein in neurons and nonneural cells. J Biol Chem 288(9):6371–6385. 35. Jiang Z, de Messieres M, Lee JC (2013) Membrane remodeling by α-synuclein and effects on amyloid formation. J Am Chem Soc 135(43):15970–15973. 36. Varkey J, et al. (2013) α-Synuclein oligomers with broken helical conformation form lipoprotein nanoparticles. J Biol Chem 288(24):17620–17630. 37. Ouberai MM, et al. (2013) α-Synuclein senses lipid packing defects and induces lateral expansion of lipids leading to membrane remodeling. J Biol Chem 288(29):20883–20895. 38. Leftin A, Job C, Beyer K, Brown MF (2013) Solid-state ¹³C NMR reveals annealing of raft-like membranes containing cholesterol by the intrinsically disordered protein α-Synuclein. J Mol Biol 425(16):2973–2987. 39. Diao J, et al. (2013) Native α-synuclein induces clustering of synaptic-vesicle mimics via binding to phospholipids and synaptobrevin-2/VAMP2. eLife 2:e00592. 40. Narayanan V, Scarlata S (2001) Membrane binding and self-association of α-synucleins. Biochemistry 40(33):9927–9934.

10 of 10 | www.pnas.org/cgi/doi/10.1073/pnas.1416598111

41. Cole NB, et al. (2002) Lipid droplet binding and oligomerization properties of the Parkinson’s disease protein α-synuclein. J Biol Chem 277(8):6344–6352. 42. Lee HJ, Choi C, Lee SJ (2002) Membrane-bound α-synuclein has a high aggregation propensity and the ability to seed the aggregation of the cytosolic form. J Biol Chem 277(1):671–678. 43. Zhu M, Fink AL (2003) Lipid binding inhibits α-synuclein fibril formation. J Biol Chem 278(19):16873–16877. 44. Jo E, et al. (2004) α-Synuclein-synaptosomal membrane interactions: Implications for fibrillogenesis. Eur J Biochem 271(15):3180–3189. 45. Burré J, Sharma M, Südhof TC (2012) Systematic mutagenesis of α-synuclein reveals distinct sequence requirements for physiological and pathological activities. J Neurosci 32(43):15227–15242. 46. Hayashi T, et al. (1994) Synaptic vesicle membrane fusion complex: Action of clostridial neurotoxins on assembly. EMBO J 13(21):5051–5061. 47. Melcher K (2004) New chemical crosslinking methods for the identification of transient protein-protein interactions with multiprotein complexes. Curr Protein Pept Sci 5(4):287–296. 48. Migneault I, Dartiguenave C, Bertrand MJ, Waldron KC (2004) Glutaraldehyde: Behavior in aqueous solution, reaction with proteins, and application to enzyme crosslinking. Biotechniques 37(5):790–796, 798–802. 49. Newman AJ, Selkoe D, Dettmer U (2013) A new method for quantitative immunoblotting of endogenous α-synuclein. PLoS One 8(11):e81314. 50. Jo E, Fuller N, Rand RP, St George-Hyslop P, Fraser PE (2002) Defective membrane interactions of familial Parkinson’s disease mutant A30P α-synuclein. J Mol Biol 315(4): 799–807. 51. Bodner CR, Maltsev AS, Dobson CM, Bax A (2010) Differential phospholipid binding of α-synuclein variants implicated in Parkinson’s disease revealed by solution NMR spectroscopy. Biochemistry 49(5):862–871. 52. Ulmer TS, Bax A, Cole NB, Nussbaum RL (2005) Structure and dynamics of micellebound human α-synuclein. J Biol Chem 280(10):9595–9603. 53. Burré J, et al. (2006) Analysis of the synaptic vesicle proteome using three gel-based protein separation techniques. Proteomics 6(23):6250–6262. 54. Wang W, et al. (2011) A soluble α-synuclein construct forms a dynamic tetramer. Proc Natl Acad Sci USA 108(43):17797–17802. 55. Gallop JL, McMahon HT (2005) BAR domains and membrane curvature: Bringing your curves to the BAR. Biochem Soc Symp 72(72):223–231. 56. Daumke O, Roux A, Haucke V (2014) BAR domain scaffolds in dynamin-mediated membrane fission. Cell 156(5):882–892.

Burré et al.