Functional cooperation of α-synuclein and VAMP2 insynaptic vesicle recyclingJichao Suna,1, Lina Wanga,1, Huan Baob, Sanjay Premia, Utpal Dasc, Edwin R. Chapmanb,d, and Subhojit Roya,b,2

aDepartment of Pathology and Laboratory Medicine, University of Wisconsin–Madison, Madison, WI 53705; bDepartment of Neuroscience, University ofWisconsin–Madison, Madison, WI 53705; cDepartment of Neuroscience, University of California, San Diego, La Jolla, CA 92093; and dHoward HughesMedical Institute, University of Wisconsin–Madison, Madison, WI 53705

Edited by Nancy Y. Ip, The Hong Kong University of Science and Technology, Hong Kong, China, and approved April 30, 2019 (received for review February21, 2019)

The function of α-synuclein (α-syn) has been long debated, andtwo seemingly divergent views have emerged. In one, α-syn bindsto VAMP2, acting as a SNARE chaperone—but with no effect onneurotransmission—while another posits that α-syn attenuatesneurotransmitter release by restricting synaptic vesicle mobilizationand recycling. Here, we show that α-syn–VAMP2 interactions arenecessary for α-syn–induced synaptic attenuation. Our data connectdivergent views and suggest a unified model of α-syn function.

alpha synuclein | Parkinson’s disease | synaptic transmission

The normal function of the small presynaptic proteinα-synuclein (α-syn) is of exceptional interest, not only in the

context of neurodegeneration, but also as a cytosolic regulatorof neurotransmission (1). Over the years, two seemingly di-vergent views have emerged. In one, α-syn binds to VAMP2(synaptobrevin-2) and chaperones SNARE complexes, butwith no effect on neurotransmission (2). Alternatively, we andothers have advocated the concept that α-syn is a physiologicattenuator of neurotransmitter release, based on evidence thatmodest α-syn overexpression attenuates synaptic vesicle (SV)recycling and exocytosis (3–7). Furthermore, we proposed amodel where α-syn helps in physiologic clustering of SV pools,restricting their egress to the presynaptic plasma membrane,thus attenuating SV recycling (6). Here, we asked if the twoseemingly divergent views can be reconciled.

Results and DiscussionAlthough previous studies have documented α-syn–VAMP2binding, some have reportedly failed to detect such associa-tions (8). Interaction at synapses is also unclear. In support ofref. 2, our coimmunoprecipitation (co-IP) experiments showthat full-length α-syn [amino acids 1 to 140 (α-syn 1–140)] bindsVAMP2, whereas a deletion lacking the reported VAMP2-bindingregion (α-syn 1–95, see ref. 2) does not (Fig. 1 A and B). To validateα-syn–VAMP2 interaction at synapses, we used the bimolecularfluorescence complementation (BiFC) assay, where candidate in-teracting proteins are tagged to N- and C-terminal Venus fragments(VN and VC) that are reconstituted upon interaction of the proteinpartners (6). Indeed, there was robust complementation ofVAMP2:VN and α-syn 1–140:VC (but not VAMP2:VN and α-syn1–95:VC) in HEK cells and presynaptic boutons (Fig. 1 C–E)—the latter indicating synaptic α-syn–VAMP2 interactions.A previous study showed that overexpression of an α-syn se-

quence lacking the C terminus (α-syn 1–110) also attenuated SVrecycling (4), leading to the notion that VAMP2 binding may notbe critical for synaptic function. However, we noticed that theα-syn 1–110 sequence used in ref. 4 has a 15-aa region (α-syn 96–110) that overlaps with the reported VAMP2-binding site thatstarts at amino acid 96 (see Fig. 1F). Accordingly, we first askedif α-syn 1–95—a sequence that definitively lacks the VAMP2-bindingsite—can attenuate SV recycling in pHluorin assays (6) that reportexo/endocytic cycles as fluctuations of SV luminal fluorescence (see

Fig. 1G). Interestingly, while α-syn 1–110 attenuated SV recycling,α-syn 1–95 had no effect (Fig. 1 H and I).In co-IP experiments, α-syn 1–140 and 1–110 bound VAMP2

with equal affinity (Fig. 2A), suggesting that α-syn 96–110 mightbe the VAMP2-binding domain. Indeed, scrambling the α-syn96–110 amino acids abrogated α-syn–VAMP2 interaction (Fig.2B). To narrow down the amino acid region required forbinding, we did alanine scanning of the α-syn 96–110 region,sequentially mutating amino acids starting at α-syn 96 to ala-nine—an inert methyl functional group mimicking secondarystructures of other amino acids (Fig. 2C). As shown in Fig. 2D,the first ∼9 amino acids starting at α-syn 96 seem critical forVAMP2 binding in this setting. However, we note that other C-terminal sequences may also be important, particularly if suchinteractions do not affect SV recycling. Finally, these α-syn–VAMP2 deletions and subtle mutations also abrogated α-syneffects on SV recycling (Fig. 2 E and F).Previously, we found that α-syn multimers appear to cluster

SVs (6), resembling phenotypes in yeast where α-syn inducedvesicle clusters (9). Here, we used a single-vesicle optical mi-croscopy system to directly visualize α-syn–induced clusteringof small synaptic-like vesicles in vitro (Fig. 2G; see ref. 10 fordetails of methods). As shown in Fig. 2 H and I, α-syn 1–140induced vesicle clustering, whereas any perturbation in theα-syn–VAMP2 binding region abrogated this effect. Synaptictargeting of α-syn deletions/mutations—determined by a quanti-tative ratiometric paradigm (11)—was comparable to wild-type(WT) α-syn (targeting of α-syn 1–110, α-syn 1–95, KKD, andScr-1 was ∼93%, 86%, 98%, and 95% of the WT protein,respectively; changes were statistically nonsignificant).Taken together, the data indicate that α-syn–VAMP2 binding

is essential for α-syn function and advocate an “interlocking model”where α-syn multimers on the SV surface interact with VAMP2 onadjacent SVs, helping to maintain physiologic SV clustering. Anunderstanding of normal α-syn function is likely critical to appre-ciate pathologic triggers in disease (12, 13). Degradation of α-synby chaperone-mediated autophagy (CMA) is strongly implicated insynucleinopathies, and interestingly, a pentapeptide region consis-tent with a CMA recognition motif—α-syn 95–99 (see ref. 14)—also lies within the VAMP2-binding site. Thus, this small regionin the C terminus of α-syn may be a key “hub” in pathophysiologictransition. Our findings link divergent views in the field and offer aunified model of α-syn function that provides a new platform forfuture studies probing the pathobiology of this enigmatic protein.

Author contributions: J.S., L.W., H.B., S.P., E.R.C., and S.R. designed research; J.S., L.W.,H.B., and S.P. performed research; J.S., H.B., and U.D. contributed new reagents/analytictools; J.S., L.W., H.B., and S.R. analyzed data; and J.S. and S.R. wrote the paper.

The authors declare no conflict of interest.

This open access article is distributed under Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND).1J.S. and L.W. contributed equally to this work.2To whom correspondence may be addressed. Email: roy27@wisc.edu.

Published online May 20, 2019.

www.pnas.org/cgi/doi/10.1073/pnas.1903049116 PNAS | June 4, 2019 | vol. 116 | no. 23 | 11113–11115

NEU

ROSC

IENCE

BRIEFRE

PORT

Dow

nloa

ded

by g

uest

on

Aug

ust 3

1, 2

020

Along with the paper in PNAS by Atias et al. (15), the collectiveevidence points to a scenario where VAMP2 and synapsin (anothercytosolic presynaptic protein with known roles in SV clustering)cooperate to help cluster SVs and regulate SV recycling. WhereasVAMP2 and α-syn directly induce SV clustering, synapsin assists

in clustering by enhancing α-syn targeting to SVs, perhaps byfacilitating the axonal transport of α-syn (see ref. 15 for details).

ACKNOWLEDGMENTS. This project was supported by NIHGrants P50AG005131—project 2 and R01AG048218 (to S.R.).

Co-IP (neuro2a)

VFPSynaptophysin

(mRFP-tagged) Overlay

VN

:VA

MP

2

+

-syn

(1

-95

):V

C

VN

:VA

MP

2

+

-syn

(1

-14

0):

VC

VN

:VA

MP

2

+

-syn

(1

-95

):V

C

VN

:VA

MP

2

+

-syn

(1

-14

0):

VC

BiFC in HEK cells

BiFC in neurons

96 -140

(VAMP-2 binding sequence)

1-95

(Memb-binding sequence)

1 140N C

95

2017

11

2017

2017

11

2017

myc

:α-s

yn 1

-140

+ V

AM

P2 (I

P m

yc A

b)

myc

:α-s

yn 1

-95

+ V

AM

P2 (I

P m

yc A

b)

myc

:α-s

yn 1

-140

+ V

AM

P2 (I

P Ig

G c

ont.

α -syn

VAMP2

α -syn

VAMP2

IP

Input

0.0

0.5

1.0

Rel

ativ

e flu

ores

cenc

e

****

VN:VAM

P2

+ -syn (1

-140):V

C

VN:VAM

P2

+ -syn (1

-95):V

C

0.0

0.5

1.0

Rela

tive

fluor

esce

nce

****

VN:VAM

P2

+ -syn (1

-140):V

C

VN:VAM

P2

+ -syn (1

-95):V

C

PM

pH=5.5

pH=7.4

pH<5.5

0.0

0.2

0.4

F 6

00/

FNH

4Cl

***

Cont.

1-140

1-110

1-95

-synuclein

VCVN VCVNVNVN VC

BiFC

BifC constructs used:

1) VN:VAMP2 + -syn(1-140):VC

2) VN:VAMP2 + -syn (1-95):VC

VN VC

VAMP2 -synC N

VAMP2

-syn

N C1-110 attenuates SV recycling

(Nemani et al., 2010)

1 14011095

VAMP2-binding site

(Burre et al., 2010)

REST STIM

5 m 0 50 100

0.0

0.5

1.0

Time (s)

Nor

mal

ized

Del

ta F

-syn 1-110 (n=194)-syn 1-140 (n=323)control (n=429)

600 AP

NH4Cl

0 50 100

0.0

0.5

1.0

Time (s)

Nor

mal

ized

Del

ta F

-syn 1-95 (n=380)-syn 1-140 (n=323)control (n=429)

NH4Cl

600 AP

A

B

F

G H I

E

DC

Fig. 1. α-Syn sequence lacking the VAMP2-binding domain fails to attenuate SV recycling. (A and B) α-Syn sequence (A) and co-IP of α-syn–VAMP2 (B).Neuro2a cells were cotransfected with myc-tagged α-syn and VAMP2 and then immunoprecipitated with an anti-myc antibody. Note that α-syn 96–140 is theVAMP2-binding region (repeated twice). (C) Principle of our BiFC assay. (D) HEK cells were transfected with VN:VAMP2 and α-syn:VC (α-syn 1–140 or 1–95).Note punctate fluorescence with VN:VAMP2 + α-syn(1–140):VC, which was greatly attenuated with α-syn 1–95 sequence lacking VAMP2-binding domain;quantification of data shown Right (mean ± SEM; α-syn 1–140, 1 ± 0.06499, n = 38 from 3 independent experiments; α-syn 1–95, 0.3074 ± 0.01676, n = 39;****P < 0.0001). (E) Cultured hippocampal neurons were cotransfected with the constructs listed; note α-syn–VAMP2 Venus complementation at boutonswith WT α-syn; quantification shown Right (mean ± SEM from 2 independent experiments; α-syn 1–140, 1 ± 0.03949, n = 280; α-syn 1–95, 0.1691 ± 0.007379,n = 312; ****P < 0.0001). (F) α-Syn 1–110 contains the VAMP2-binding site that starts at amino acid 96. (G) Principle of pHluorin assay (Top) with repre-sentative images (Below). (H and I) Experiments in cultured hippocampal neurons (pHluorin). (H) While α-syn 1–110 attenuates SV recycling (Left), α-syn 1–95has no effect (Right) (note that some error bars are too small to see). (I) Quantification of data in H (mean ± SEM from at least 3 independent experiments);control, 0.4703 ± 0.03215, n = 16; α-syn 1–140, 0.3165 ± 0.02914, n = 13; α-syn 1–110, 0.3464 ± 0.05906, n = 8; α-syn 1–95, 0.4763 ± 0.0192, n = 17; **P = 0.0025,*P = 0.0459 (one-way ANOVA followed by Dunnett’s post hoc test).

11114 | www.pnas.org/cgi/doi/10.1073/pnas.1903049116 Sun et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 3

1, 2

020

1. Y. C. Wong, D. Krainc, α-Synuclein toxicity in neurodegeneration: Mechanism and

therapeutic strategies. Nat. Med. 23, 1–13 (2017).2. J. Burré et al., Alpha-synuclein promotes SNARE-complex assembly in vivo and in vitro.

Science 329, 1663–1667 (2010).3. K. E. Larsen et al., Alpha-synuclein overexpression in PC12 and chromaffin cells im-

pairs catecholamine release by interfering with a late step in exocytosis. J. Neurosci.

26, 11915–11922 (2006).4. V. M. Nemani et al., Increased expression of alpha-synuclein reduces neurotransmitter re-

lease by inhibiting synaptic vesicle reclustering after endocytosis. Neuron 65, 66–79 (2010).5. D. A. Scott, et al., A pathologic cascade leading to synaptic dysfunction in alpha-

synuclein-induced neurodegeneration. J. Neurosci. 30, 8083–8095 (2010).6. L. Wang et al., α-Synuclein multimers cluster synaptic vesicles and attenuate recycling.

Curr. Biol. 24, 2319–2326 (2014).7. D. Scott, S. Roy, α-Synuclein inhibits intersynaptic vesicle mobility and maintains

recycling-pool homeostasis. J. Neurosci. 32, 10129–10135 (2012).

8. D. C. DeWitt, E. Rhoades, α-Synuclein can inhibit SNARE-mediated vesicle fusion

through direct interactions with lipid bilayers. Biochemistry 52, 2385–2387 (2013).9. A. D. Gitler et al., The Parkinson’s disease protein alpha-synuclein disrupts cellular Rab

homeostasis. Proc. Natl. Acad. Sci. U.S.A. 105, 145–150 (2008).

10. J. Diao et al., Native α-synuclein induces clustering of synaptic-vesicle mimics viabinding to phospholipids and synaptobrevin-2/VAMP2. eLife 2, e00592 (2013).

11. D. Gitler et al., Molecular determinants of synapsin targeting to presynaptic termi-

nals. J. Neurosci. 24, 3711–3720 (2004).

12. A. D. Gitler, J. Shorter, Prime time for alpha-synuclein. J. Neurosci. 27, 2433–2434 (2007).13. J. Burré, M. Sharma, T. C. Südhof, Cell biology and pathophysiology of α-synuclein.

Cold Spring Harb. Perspect. Med. 8, a024091 (2018).

14. A. M. Cuervo, L. Stefanis, R. Fredenburg, P. T. Lansbury, D. Sulzer, Impaired degradation of

mutant alpha-synuclein by chaperone-mediated autophagy. Science 305, 1292–1295 (2004).15. M. Atias et al., Synapsins regulate α-synuclein functions. Proc. Natl. Acad. Sci. U.S.A.

116, 11116–11118 (2019).

myc 140

myc 110

myc 95

A Co-IP with -syn deletions1.....96 - KKDQLGKNEEGAPQE - 110.....140

AAAQLGKNEEGAPQEKKDAAAKNEEGAPQEKKDQLGAAAEGAPQEKKDQLGKNEAAAPQEKKDQLGKNEEGAAAA

] Alanine

mutants

KKD->AAAQLG->AAAKNE->AAAEGA->AAAPQE->AAA

C -syn 96-110 alanine mutants

Control

-syn 1-1

40

-syn Scrb

-1

-syn "K

KD"0.0

0.2

0.4

0.6

F 600

/FN

H4C

l

E F

2017

11

2017

myc

:α-s

yn 1

-140

+ V

AM

P2 (I

P m

yc A

b)

myc

:α-s

yn 1

-95

+ V

AM

P2 (I

P m

yc A

b)

myc

:α-s

yn 1

-140

+ V

AM

P2 (I

P Ig

G c

ont.)

myc

:α-s

yn 1

-110

+ V

AM

P2 (I

P m

yc A

b)

myc

:α-s

yn 1

-110

+ V

AM

P2 (I

P Ig

G c

ont.)

α-syn

VAMP2

IP

**

IP: a

nti

-myc -syn

VAMP2

Inp

ut -syn

VAMP2

myc

--s

yn 1

-140

+ V

AM

P2 (I

gG c

ntrl)

myc

--s

yn 1

-140

+ V

AM

P2m

yc-

-syn

KKD

->A

AA

+ V

AM

P2m

yc-

-syn

QLG

->A

AA

+ V

AM

P2m

yc-

-syn

KN

E->

AA

A

+ V

AM

P2m

yc-

-syn

EG

A->

AA

A

+ V

AM

P2m

yc-

-syn

PQ

E->

AA

A

+ V

AM

P2

D

1.....96 - KKDQLGKNEEGAPQE - 110.....140

Scrambled-2:

Scrambled-1:

1.....96 - NLGEKDEAQEQPGKK - 110.....140

1.....96 - LQNKGEPDQKKEAEG - 110.....140

WT:

B Co-IP with scrambled mutants

myc

:α-s

yn 1

-140

+ V

AM

P2 (I

P m

yc A

b)

myc

:α-s

yn 1

-140

+ V

AM

P2 (I

P Ig

G c

ont.)

myc

:α-s

yn S

crb

-1

+ V

AM

P2(IP

myc

Ab

)m

yc:α

-syn

Scr

b-2

+ V

AM

P2(IP

myc

Ab

)

myc:α-syn

VAMP2

IP

myc:α-syn

VAMP2

Input

2017

2017

2017

2017

NH4Cl

50 1000.0

0.5

1.0

Time (s)

Norm

alize

d De

lta F

control (n=190)-syn 1-140 (n=140)-syn Scr-1 (n=164)-syn “KKD” (n=170)

600 AP

0 25 50 75 100

125

150

175

200

225

250

275

0

100

200

300

400

# o

f ve

sicl

e cl

ust

ers

+ -syn 1-140+ -syn 1-110

+ -syn 1-95+ -syn KKD+ -syn Scr-1

No -syn

no -syn + -syn 1-140

+ -syn Scr-1 + -syn KKD

HG

+-sy

n 1-14

0

+-sy

n 1-110

No-sy

n

+-sy

n 1-95

+-sy

n KKD

+-sy

n Scr-1

0

25

50

75

100

# of

ves

icle

clu

ster

s(n

orm

aliz

ed, %

)

I

****

NS

Intensity of vesicle clusters

Fig. 2. Mapping of the α-syn–VAMP2 binding domain and requirement of α-syn–VAMP2 interactions for α-syn–mediated SV attenuation. (A and B) Bothα-syn 1–140 and 1–110 bind VAMP2 with equal affinity, and scrambling of amino acid sequences in α-syn 96–110 attenuates α-syn–VAMP2 binding (co-IP inneuro2a, repeated twice). (C and D) Sequential amino acids (from α-syn 96) were mutated to alanine, and association of these mutants (myc-tagged) withVAMP2 was evaluated (co-IP in neuro2a cells). Note that mutations in α-syn 96–104 show the greatest disruption (repeated twice). (E) Scrambled and KKDmutations in the α-syn 96–110 sequence abrogated α-syn–mediated synaptic attenuation, as determined by pHluorin assays in hippocampal neurons. (F)Quantification of data in E (mean ± SEM from at least 3 independent experiments); control, 0.423 ± 0.029, n = 6; α-syn 1–140, 0.178 ± 0.032, n = 6; KKD,0.453 ± 0.070, n = 5; Scr-1, 0.464 ± 0.041, n = 6; **P = 0.0017 (one-way ANOVA followed by Dunnett’s post hoc test). (G–I) Optical single vesicle clusteringexperiments were carried out as described in ref. 10. Briefly, VAMP2-containing synaptic-like vesicles were first immobilized on a glass slide assembled in amicrofluidic chamber, and then WT or mutant α-syn protein was added. After extensive washing (to remove unbound α-syn), DiI-labeled VAMP2-containingvesicles were added to the chamber, and clustering of the labeled vesicles was visualized by prism-type total internal reflection fluorescence microscopy (afterextensive washing to remove unbound vesicles). As shown in representative images (G) and quantitative data (H), α-syn induced vesicle clustering, anddeletions or subtle mutations in the VAMP2-binding site markedly abrogated the number of vesicle clusters. Mean ± SEM from 4 independent experimentswhere observer was blinded to the conditions; α-syn 1–140, 100%; α-syn 1–110, 83.15% ± 6.439%; no α-syn, 21.83% ± 7.437%; α-syn 1–95, 29.94% ± 8.332%;KKD, 33.89% ± 3.465%; Scr-1, 30.49% ± 8.138%; NS, nonsignificant; ****P < 0.001 (one-way ANOVA followed by Dunnett’s post hoc test). (I) Scatter plotsshowing number of vesicle clusters (on y axis) and fluorescence intensities (on x axis) of all Dil-labeled clusters, along with a smoothened curve through thedata points.

Sun et al. PNAS | June 4, 2019 | vol. 116 | no. 23 | 11115

NEU

ROSC

IENCE

BRIEFRE

PORT

Dow

nloa

ded

by g

uest

on

Aug

ust 3

1, 2

020