α-Synuclein disrupts stress signaling by inhibitingpolo-like kinase Cdc5/Plk2Shaoxiao Wanga, Baoshan Xua,1, Liang-Chun Lioub, Qun Renb, Shile Huanga, Yan Luoa,2, Zhaojie Zhangb,and Stephan N. Witta,3

aDepartment of Biochemistry and Molecular Biology, Louisiana State University Health Sciences Center, Shreveport, LA 71130; and bMicroscope Core Facility,University of Wyoming, Laramie, WY 82071

Edited by Gregory A. Petsko, Brandeis University, Waltham, MA, and approved August 21, 2012 (received for review April 13, 2012)

Parkinson disease (PD) results from the slow, progressive loss ofdopaminergic neurons in the substantia nigra. Alterations inα-synuclein (aSyn), such as mutations or multiplications of thegene, are thought to trigger this degeneration. Here, we showthat aSyn disrupts mitogen-activated protein kinase (MAPK)-controlled stress signaling in yeast and human cells, which resultsin inefficient cell protective responses and cell death. aSyn is a sub-strate of the yeast (and human) polo-like kinase Cdc5 (Plk2), andelevated levels of aSyn prevent Cdc5 from maintaining a normallevel of GTP-bound Rho1, which is an essential GTPase that regu-lates stress signaling. The nine N-terminal amino acids of aSyn areessential for the interaction with polo-like kinases. The resultssupport a unique mechanism of PD pathology.

aging | neurodegeneration | proteinopathy

Parkinson disease (PD), which is the second most commonneurodegenerative disorder, results from the progressive

degeneration of dopaminergic neurons in part of the midbraincalled the substantia nigra pars compacta (1). A hallmark of thisdisease is the formation of protein inclusions (Lewy bodies) inthe cytoplasm of affected neurons, and the principal componentof Lewy bodies is the protein α-synuclein (αS) (2). Age-dependentaccumulation of αS, perhaps due to lysosome dysfunction (3), isthought to trigger sporadic PD, whereas missense mutations (4–6)or multiplication (7) of the αS gene trigger early-onset PD.Because αS molecules in Lewy bodies are phosphorylated at Ser-129 (8), a search has been underway to identify the kinase re-sponsible for this phosphorylation and its role in the pathobiologyof αS. Polo-like kinase 2 (Plk2/Cdc5, human/yeast) was recentlyidentified as a suppressor of αS toxicity in yeast, worm, and ratPD models (9), and Plk2 phosphorylates αS in vitro and in vivo atSer-129 (10).Plks are highly conserved serine/threonine kinases that possess

an N-terminal kinase domain and one or two C-terminal polo-boxdomains that interact with substrates and direct the kinase tovarious cellular loci (11). Plks regulate the cell cycle and cyto-kinesis in dividing cells, whereas their function in nondividingcells, such as neurons, is less well understood. Recent studies haveshown that in neurons, Plk2 regulates the activity of two smallguanosine triphosphatases (GTPases), Ras and Rap, by phos-phorylating their activators (GEFs, guanine nucleotide exchangefactors) and inactivators (GAPs, GTPase-activating proteins)(12, 13). In yeast, Cdc5 also regulates the activity of the smallGTPase Rho1 by phosphorylating its GEFs (Tus1 and Rom2)and GAP (Sac7) (14).Rho1, which is a member of the Ras-like family of small

GTPases, controls actin organization, cell wall biogenesis, po-larized secretion, and cytokinesis in yeast cells. Rho1 is the mainsignaling node in the yeast cell wall integrity (CWI) pathway,which is a MAPK cascade that helps cells monitor and respondto cell wall stress (15). This pathway is activated by elevatedtemperatures (37–39 °C), mating pheromone, hypo-osmotic shock,and various compounds (caffeine, calcofluor white, and Congored) (15). In response to these stresses, membrane-associated

Rho1 switches on and then activates protein kinase C (Pkc1).Activated Pkc1 triggers a signaling cascade that activates thetranscription factor Rlm1, which controls the transcription of cellwall biogenesis genes.In this study, the mechanism of toxicity of αS was probed by

using yeast, and the findings were validated in human neuroblas-toma cells. In yeast, αS decreases the level of GTP-Rho1, whichdisrupts stress signaling from the membrane to the nucleus andmakes cells hypersensitive to stress; αS also disrupts stress sig-naling in human cells. We propose that αS inhibits Plks fromphosphorylating and activating the downstream regulatory pro-teins Rho GEFs and/or Rho GAPs that, in turn, decreases thetotal cellular level of GTP-Rho and disrupts stress signaling.

ResultsαS Toxicity Depends on the Integrity of Its N Terminus and onMembrane Binding. To determine whether αS toxicity is due tomembrane binding, two different C-terminal membrane-targetingsequences (mts) that should increase the amount αS on theplasma membrane were tested. Mts1 (SNSVCCTLM) and mts2(GSGGCCLLS) are derived from the yeast proteins Ste18 andRas2, respectively; in vivo, these tags are palmitoylated, and theresultant palmitoylated protein localizes to the plasma membrane(16). Cells expressing untagged αS exhibited impaired growthcompared with vector control cells, whereas cells expressing αS-mts1 or αS-mts2 exhibited an even more pronounced growthdefect than cells expressing the untagged protein (Fig. 1A). Cellsexpressing GFP-mts1 or GFP-mts2 exhibited nearly the samegrowth pattern as vector control cells. More dramatic results wereobtained with the A30P mutant, which is mainly cytosolic andless toxic than αS in yeast (17). Cells expressing A30P or A30P(Δ2–10) showed no growth defects compared with vector controlcells, whereas A30P-mts2 but not A30P(Δ2–10)-mts2 severelyinhibited growth (Fig. 1B). Western blot analysis showed similarexpression levels of all proteins (Fig. 1C). Fluorescence andelectron microscopy revealed that the mts tag promoted theaccumulation of A30P and A30P(Δ2–10) at the plasma mem-brane and the bud neck (Fig. 1D and Fig. S1). Overall, αS/A30Ptoxicity increases with increased plasma membrane binding.Additionally, the loss of toxicity of the membrane-tethered syn-ucleins upon deletion of only nine N-terminal amino acids indi-cates that the N terminus of membrane-bound αS inhibits some

Author contributions: S.W. and S.N.W. designed research; S.W., B.X., L.-C.L., Q.R., and Y.L.performed research; S.W., B.X., S.H., Z.Z., and S.N.W. analyzed data; and S.N.W. wrotethe paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1Present address: Stowers Institute for Medical Research, 1000 E 50th Street, Kansas City,MO 64110.

2Present address: National Key Laboratory of Biotherapy, Sichuan University, Yao-huaRoad, Chengdu, Sichuan 610041, People’s Republic of China.

3To whom correspondence should be addressed. E-mail: switt1@lsuhsc.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1206286109/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1206286109 PNAS Early Edition | 1 of 6

BIOCH

EMISTR

Y

other membrane protein or membrane-bound αS adopts a toxicconformation only in the context of the full-length protein.

αS Disrupts the Yeast Cell Wall Integrity Pathway. We also discov-ered that yeast cells expressing αS but not αS(Δ2–10) were hy-persensitive to elevated temperatures and caffeine, each resultingin impaired growth (Fig. 2A; see SI Text and Fig. S2). On thebasis of these findings, we hypothesized that membrane-boundαS disrupts the yeast cell wall integrity (CWI) pathway—theMAPK cascade that helps cells monitor and respond to cell wallstress (15). This hypothesis was tested in various assays.

Cells expressing αS but not αS(Δ2–10) were hypersensitiveto cercosporamide (Cer), which inhibits Pkc1 (18), and tolatrunculin-A (Lat-A), which inhibits F-actin polymerization(19), resulting in impaired growth (Fig. 2 B and C). The hyper-sensitivity of αS-expressing cells to heat and to these variouscompounds is consistent with the notion that αS disrupts theCWI pathway.To probe the CWI pathway for genes with synthetic lethal

interactions with αS at 30 °C, 15 deletion mutants in the CWIpathway were tested and enhanced toxicity of αS but not αS(Δ2–10) was found in wsc1Δ and mid2Δ and rom2Δ and tus1Δ(Fig. 2D and Fig. S3A). WSC1 and MID2 encode for membranebound sensors, and ROM2 and TUS1 encode for Rho1 GEFs.One deletion mutant, sac7Δ, was identified that partially pro-tected against αS-induced toxicity at 39 °C (Fig. S3B). SAC7encodes a Rho1 GAP, and deletion of this gene increases thetotal cellular level of GTP-Rho1 (20).

αS Decreases the Global Level of GTP-Rho1 by Disrupting Cdc5Function. Sakchaisri and colleagues discovered that Cdc5, whichlocalizes to the bud neck (21), phosphorylates Tus1 and thatphospho-Tus1 concentrates at the bud neck where it activatesRho1 to trigger cytokinesis (14). These latter authors proposedthat Cdc5 also controls the global level of GTP-Rho1 by its abilityto phosphorylate and, thus, activate or inactivate Rho1 GEFs orRho1 GAPs, respectively. Our hypothesis is that αS at high levelssaturates Cdc5, which prevents Cdc5 from phosphorylating andactivating Rho1 GEFs (or inactivating Rho1 GAPs). The net re-sult is that the global level of GTP-Rho1 decreases, which blockssignaling to the nucleus under conditions that would otherwiseactivate the CWI pathway. Several different types of assays wereused to test this hypothesis.First, to determine whether Cdc5 interacts with αS and αS(Δ2–

10), kinase and binding (pull-down) assays were conducted withpurified recombinant αS and immunoprecipitated hemagglutinin(HA)-tagged Cdc5. In the kinase assay, HA-Cdc5 and αS or αS(Δ2–10) (100 μM) were incubated with Mg+2/ATP, and then thereaction mixture was subjected to SDS/PAGE followed byWestern blot analysis by using a monoclonal antibody specific forαS Ser-129-(Pi). αS was phosphorylated by HA-Cdc5 at Ser-129,whereas αS(Δ2–10) was nearly devoid of phosphorylation at thisamino acid (Fig. 3A). In vivo, αS was phosphorylated at Ser-129,whereas αS(Δ2–10) was nearly devoid of phosphorylation at thisamino acid (Fig. 3B). In the binding assay, we tried withoutsuccess to pull down αS and αS(Δ2–10) by using HA-Cdc5 boundto agarose beads. Instead, a cross-linking assay was used whereHA-Cdc5 and αS or αS(Δ2–10) (50 μM) were incubated for 1 h at4 °C with the reversible cross-linker 3,3′-dithiobis-[sulfosuccini-midyl-propionate]. After centrifugation and washing, the samplewas subjected to SDS/PAGE followed by Western blot analysisusing antibodies against αS and the HA tag (Fig. 3C); this ex-periment revealed that αS but not αS(Δ2–10) cross-linked to HA-Cdc5. Our interpretation is that αS binds to and dissociates fromCdc5 with rapid kinetics and that the 9 N-terminal amino acidsfacilitate the interaction with Cdc5.Second, to determine whether αS inhibits the binding of Tus1 to

the polo box (14), we used GST attached to three different Cdc5constructs—the polo box (GST-PBD), the inactive polo box (GST-Pincer), and the kinase domain (GST-KD)—to pull down Tus1. Forthis assay, BY4741 cdc15-2 cells, which are arrested in anaphase at37 °C (14), were used. Lysates from cdc15-2 cells expressing Tus1-myc and αS or αS(Δ2–10) were incubated with the various purifiedGST constructs, and then after several washing steps samples weresubjected to SDS/PAGE followed by Western blotting with an anti-myc antibody (Fig. 3D). GST-PBD pulled down Tus1-myc fromcontrol cells (no αS expression), from cells expressing αS(Δ2–10),but not from cells expressing αS. A similar pattern was observedwith GST-Pincer, although in general less Tus1-myc was pulled

Fig. 1. The integrity of the N terminus determines αS toxicity. (A and B)Growth assay. Yeast cells with indicated plasmids were diluted, spotted ontoplates, and grown for 3 d at 30 °C. Gal, galactose; Glu, glucose. (C) Western blotof αS variants. Untagged and tagged variants of αS, αS(Δ2–10), A30P, and A30P(Δ2–10) were expressed at similar levels (induction time = 8–10 h). Cell extractswere prepared and subjected to SDS/PAGE and Western blotting. (D) GFP-A30P-mts2 localizes to the plasma membrane. Yeast cells transformed withindicated plasmids were induced for 10 h at 30 °C and then imaged by fluo-rescence microscopy. mts2, membrane-targeting sequence. (Scale bar: 5 μm.)

Fig. 2. αS disrupts the yeast cell wall integrity pathway. (A) αS disrupts theCWI pathway. Cells with high-dose plasmids (pAG426) were serially diluted,spotted onto the indicated plates, and grown for 3–5 d at 30 °C or 39 °C. Caf,caffeine (8 mM). (B) Growth assay of BY4741 transformed with indicatedhigh-dose pAG plasmids, serially diluted, and spotted onto various plates.Cer, cercosporamide (5 μg/mL). Plates were incubated at 30 °C for 2 or 3 d at30 °C. (C) A halo assay was used to test the sensitivity of cells expressinghigh-dose αS or αS(Δ2–10) to latrunculin-A. (D) Growth analysis of deletionmutants in the CWI pathway at 30 °C. Cells were serially diluted in suc-cessive 10-fold dilutions and spotted onto indicated plates, which werethen incubated 2 d at 30 °C.

2 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1206286109 Wang et al.

down compared with GST-PBD, consistent with a defect in the polobox. The kinase domain failed to pull down Tus1. Overall, αS butnot αS(Δ2–10) inhibits the binding of Tus1 to Cdc5.

In complementary experiments, we tested whether αS inhibitsthe phosphorylation of Tus(1–300)-myc by Cdc5 in cdc15-2 cells.This Tus1 variant is phosphorylated in vivo as indicated by the

Fig. 3. αS disrupts stress signaling in yeast. (A) Cdc5 phosphorylates αS at serine-129 (p-S129) in vitro. HA-Cdc5 and recombinant αS or αS(Δ2–10) (100 μM) wereincubated with Mg+2/ATP, followed by SDS/PAGE, Western blotting, and staining with indicated antibodies. (B) Phosphorylation of αS at S129 in vivo. Shown isWestern blot of cell lysates from αS or αS(Δ2–10) expressing cells using antibodies against αS. (C) αS cross-links to Cdc5. HA-Cdc5 and recombinant αS or αS(Δ2–10) (50μM) were cross-linked with 3,3′-dithiobis-[sulfosuccinimidylpropionate] and then analyzed by SDS/PAGE, Western blotting and staining with an antibody against αS.(D) Tus1-myc pull-down. Lysates from cdc15-2 cells overexpressing Tus1-myc and αS or αS(Δ2–10) were incubated with the indicated GST-Cdc5 constructs. Afterprecipitation, samples were Western blotted and the blots were stained with an anti-myc antibody. Also shown is a blot of the input and a SDS-polyacrylamide gel ofthe purified GST constructs stained with Coomassie blue. (E) αS inhibits Tus1(1–300) phosphorylation. Lysates from cdc15-2 cells expressing Tus1(1–300)-myc and αS, αS(Δ2–10), or the indicated reagents were Western blotted and blots were stained with an anti-myc antibody. PPase, calf intestinal alkaline phosphatases (10 U);inhibitors, EDTA, and Na3VO4. (F) αS decreases GTP-Rho1. cdc15-2 cells with flag-Rho1 and αS plasmids were induced for 8 h at 30 °C. Western blot of cell lysates showsthat GST-Pkc1-RBD but not GST pulled down GTP-Rho1. (G) Effect of αS on the ratio of GTP-Rho1 to Rho1 band intensity compared with control cells (pAG426). ***P <0.001 was determined for αS versus vector (n = 3 independent experiments) by one-way ANOVA, Dunnett’s post hoc analysis. (H) αS inhibits signaling to the nucleus.β-gal activity of cells transformed with high-dose αS plasmid and p1366 plasmid (PRM5::lacZ) that were induced for 10 h (last 6 h± 20 mM caffeine). ***P < 0.001 wasdetermined for αS versus vector (n = 3) by one-way ANOVA, Dunnett. (I) Cdc5 controls GTP-Rho1. β-gal assay of wild-type and cdc5-1 mutant cells grown to late logphase at 23 °C and then incubated for 40min at 37 °C (n= 4). ***P< 0.001, determined by a two-tailed Student’s t test. (J) Model for how αS disrupts the CWI pathway.

Wang et al. PNAS Early Edition | 3 of 6

BIOCH

EMISTR

Y

large electrophoretic mobility shift that occurs upon phosphatasetreatment (14) (Fig. 3E, Left). Whether αS alters the mobility ofthe hyperphosphorylated Tus1(1–300)-myc was tested. Lysates ofcdc15-2 cells that expressed Tus1(1–300)-myc and coexpressedαS or αS(Δ2–10) were probed with an anti-myc antibody. Thehigh molecular mass band at ∼75 kDa, which we attribute tohyperphosphorylated Tus1(1–300)-myc, occurred in cellsexpressing αS(Δ2–10) and in control cells, but was absent fromcells expressing αS (Fig. 3E, Right); instead, a lower molecularmass band was observed. The results are consistent with αS butnot αS(Δ2–10) partially inhibiting the phosphorylation of Tus1(1–300) by Cdc5.Third, to determine whether αS alters the total cellular level of

GTP-Rho1, a pull-down assay with GST-Pkc1-PBD beads (14)was used. PBD is a Pkc1 domain that specifically binds a GTP-Rho1 molecule. Lysates from the cdc15-2 strain were tested. αSdecreased the total cellular level of GTP-Rho1 in cdc15-2 cellscompared with vector control cells; whereas, the total cellularlevel of GTP-Rho1 in αS(Δ2–10)-expressing cells was in-distinguishable from vector control cells (Fig. 3 F and G).Fourth, to determine whether the αS disrupts signaling to the

nucleus upon activation of the CWI pathway, a reporter plasmidwas used in which the bacterial lacZ gene was under the controlof the Rlm1-regulated promoter of PRM5 (PRM5::lacZ) (22).The effect of αS-induced transcriptional activation of this reporterwas quantified by measuring β-galactosidase (β-gal) activity.Strikingly, for wild-type cells treated without caffeine and withcaffeine, αS inhibited β-gal activity by 48% and 75%, respectively,

compared with cells expressing αS(Δ2–10) or vector control cells(Fig. 3H); this data indicates that membrane-bound but not cy-tosolic αS inactivates transcription of this reporter gene. A parallelexperiment showed that inactivating Cdc5 by using a tempera-ture-sensitive mutant (cdc5-1) decreased β-gal activity comparedwith wild-type cells (Fig. 3I), which is consistent with Cdc5controlling the CWI pathway (Fig. 3J). The experiments in Fig. 3show that αS but not αS(Δ2–10) specifically blocks transcriptionof a gene in the CWI pathway by its ability to inhibit Cdc5. αSalso disrupts Cdc5-dependent cell cycle functions; see SI Textand Figs. S4 and S5.

αS Disrupts Stress Signaling in SH-SY5Y Cells. αS has been reportedto inhibit the phosphorylation of the p38 MAPK and acceleratethe death of mouse neuroblastoma cells in culture (23). Smallmolecular inhibitors of p38 also trigger cell death (24), includingin SH-SY5Y cells (Fig. S6A). Our interpretation is that αS inhibitsthe phosphorylation of p38 and accelerates cell death because itperturbs upstream interactions between Plk2 with Rho/Ras GEFsand/or Plk2 with Rho/Ras GAPs, which lowers the level of activeRho/Ras and, thus, shuts down this pathway. Because detectingchanges in active Rho/Ras levels in human cells, which have∼56 of these proteins (25), is quite challenging, we probed thephosphorylation of various MAPKs in human neuroblastomaSH-SY5Y cell lines that stably overexpress αS or αS(Δ2–10). Itwas first confirmed that GST-Plk2 phosphorylates αS more effi-ciently than αS(Δ2–10) in vitro (Fig. 4A), and nearly the sameresults were found in vivo (Fig. 4 B and C). The phosphorylation

Fig. 4. αS disrupts stress signaling in human cells. (A) Plk2 phosphorylates αS at S129 in vitro. GST-Plk2 and recombinant αS or αS(Δ2–10) (50 μM) were in-cubated with Mg+2/ATP, followed by SDS/PAGE, Western blotting, and staining with antibodies against αS and p-S129-αS. (B) αS is phosphorylated at S129 inSH-SY5Y cells. Cell lysates were Western blotted and stained with indicated antibodies (n = 2). (C) Plot shows the p-S129 to αS ratio. (D) αS inhibits MAPKphosphorylation. SH-SY5Y cells overexpressing αS or αS(Δ2–10) were incubated at 37 °C or 45 °C for 30 min; lysates were Western blotted with the indicatedantibodies. Bar graph is fold change of the phospho-protein to protein band intensity relative to control cells (pcDNA). Values are means ± SD (n = 3). In eachset of three samples αS is compared with vector and to αS(Δ2–10). **P < 0.01 and ***P < 0.001 were determined by using a one-way ANOVA with a Tukey posthoc test. (E) Cell viability after heat shock. Cells were heat shocked at 51 °C for 30 min, returned to a 37 °C incubator, and the viability was measured 24 h laterby using the MTT assay. Values are means ± SD (n = 3). P value range (**, 0.001 < P < 0.01) of indicated comparisons was determined by using one-wayANOVA with Tukey post hoc test. (F) Model of αS inhibiting Plk interactions with Rho GEFs/GAPs. Inhibition can occur in solution or when αS and Plksconcentrate at the plasma membrane.

4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1206286109 Wang et al.

of p38, c-Jun N-terminal kinase (JNK), and extracellular-signal-regulated kinases 1 and 2 (ERK1/2), which are activated by var-ious stresses (26), and the transcription factor c-Jun was assessedby Western blotting with phospho-specific antibodies after in-cubating SH-SY5Y cells for 0.5 h at 37 °C or 45 °C. Upon heatstress, αS partially inhibited the phosphorylation of p38, JNK,and c-Jun compared with control cells, whereas αS(Δ2–10) hadno effect (Fig. 4D and Fig. S6B). We also examined cell viability24 h after the brief heat shock by using several different SH-SY5Y clones that overexpress αS (clone 2, 12, 14) or αS(Δ2–10)(clone 9, 12). αS but not αS(Δ2–10) accelerated cell death com-pared with vector control cells as evidenced by the 59 ± 6%decrease in viability of SH-SY5Y clones overexpressing αS com-pared with vector control cells and SH-SY5Y clones overexpressingαS(Δ2–10) (Fig. 4E). The inhibition of the phosphorylation ofp38 by αS but not by αS(Δ2–10) in human cells is similar to theinhibition of the CWI pathway by αS in yeast.

DiscussionThis study has provided evidence for the following mechanism ofαS pathology: high expression levels of αS prevent Plks fromactivating Rho regulatory proteins (GEFs/Gaps), which, in turn,leads to a decrease in the total cellular level of GTP-Rho. αS istoxic is because it inhibits Plks and disrupts signaling (Fig. 4F),not because it is phosphorylated by Plks at Ser-129. Although αS-Ser-129-Pi may aggregate and forms inclusions, whether it istoxic to cells is debatable.Deleting the first nine N-terminal residues of αS causes a loss

of toxicity in yeast and human cells compared with the full-lengthprotein, and that the lack of toxicity of αS(Δ2–10) is a conse-quence of its failure to interact with Plks (Figs. 3 A–I and 4 A–E),not to its failure to interact with the plasma membrane. For in-stance, for the two proteins forced into the yeast plasma mem-brane, A30P-mts and A30P(Δ2–10)-mts, the former is extremelytoxic, whereas the latter is not (Fig. 1 A–C). Strikingly, in bothyeast and human SH-SY5Y cells αS but not αS(Δ2–10) blocksa MAPK pathway and accelerates death due to heat shock (Figs.2 A–D and 4 D and E and Fig. S2 B–E). A recent report showedthat, in contrast to yeast, deleting the first 10 N-terminal residueshas no effect on αS membrane binding, aggregation, and cellviability in human SH-SY5Y cells (27). Specifically, αS and αS(Δ2–11) each bind to the plasma membrane and each exhibitidentical toxicity when stably expressing SH-SY5Y cells are trea-ted with a proteasome inhibitor. Although αS and the N-terminaldeletion mutant are equally toxic in the proteasomal inhibitionassay, the two proteins have unequal toxicities in the heat shockassay; thus, whether αS and the N-terminal deletion mutant havethe same or different toxicities depends on the cellular pathway.The efficiency by which Plks phosphorylate αS at Ser-129 was

recently shown to depend on the N terminus of αS. Specifically,αS(103–140) fails to be phosphorylated by all four Plks (Plk1–4)tested, whereas the intact protein is efficiently phosphorylated(28). Our results show that deletion of only 9 N-terminal residuesabolishes the ability of Cdc5/Plk2 to phosphorylate αS in vitroand in vivo (Figs. 3 A and B and 4 A–C). Our interpretation ofour findings is that membrane-bound αS/A30P are extremelytoxic because they inhibit membrane-bound Cdc5, which, in turn,decreases the level of GTP-Rho1, and the membrane-bound de-letion mutants are not toxic because they cannot interact withCdc5. Whether the N terminus of αS/A30P directly binds to thepolo-box of Cdc5/Plk2 or whether a unique conformation, whichbinds to the polo-box, exists only in full-length protein cannot bedetermined at this time. Whether membrane-associated or soluble,Plks that interact with Rho GEFs and Rho GAPs should be sub-ject to inhibition and consequent disruption of cell signaling by αS.We propose that αS does not inhibit Plk2 under normal cel-

lular conditions because its concentration is not high enough.Were the concentration of αS to increase with age, however, then

its inhibition of Plk2-GEF/GAP interactions would proportion-ally increase, and cells would become more and more sensitive tostress (Fig. 4F). An increase in the concentration of neuronal αScan come from an age-dependent decline in the function of thelysosome or the proteasome (3). In our model, decreasing theconcentration of αS, increasing the concentration of Plk2 (9), orblocking the binding of αS to Plk2 should rescue αS toxicity.A therapeutic strategy for PD is to block the binding of αS toPlks without interfering with the binding of other substrates. Ourmechanism explains how soluble, monomeric αS can becometoxic to cells with age.

Materials and MethodsYeast Strains and Media. For additional methods, see SI Materials andMethods. Yeast strains and plasmids used in this study are given in Tables S1and S2, respectively. Synthetic complete drop-out media were prepared asdescribed in ref. 29. Raffinose and galactose were used in the noninducingand inducing media, respectively. Drop-out media were purchased fromSigma-Aldrich and United States Biological, and, unless otherwise noted, allchemicals were purchased from Sigma-Aldrich. Yeast cells were grown withshaking at 30 °C, or, for heat stress experiments, at 39 °C.

Growth Assay. Yeast cells transformed with various plasmids were grownovernight in liquid raffinose medium and then diluted (to OD600 = 0.2) intoliquid galactose medium and incubated for 6–8 h. Cultures were normalizedto the same OD, serially diluted in 10-fold increments, spotted (10 μL) ontosolid plates, and incubated for 2–6 d at 30 °C. Assays using latrunculin A andβ-galactosidase are described in SI Materials and Methods.

Western Blot Analysis. For yeast experiments, lysate preparation and Westernblotting were carried out as described (30). Fifty micrograms of total proteinwas loaded per well. For mammalian cell culture experiments, cells werecultured, harvested, and lysed in RIPA buffer [50 mM Tris at pH 7.2, 150 mMNaCl, 1% sodium deoxycholate, 0.1% SDS, 1% Triton X-100, 10 mM NaF,1 mM Na3VO4, and protease inhibitor mixture (1:1,000; Sigma-Aldrich)]. Allsteps were carried out at 4 °C. Lysates were sonicated for 10 s and centrifugedat 14,000–15,000 × g for 10 min. Protein concentration was determined bybicinchoninic acid assay (Pierce). Equivalent amounts of protein were sepa-rated on a 7.5–12% (wt/vol) or 4–20% (wt/vol) SDS–polyacrylamide gel (Bio-Rad) and transferred to polyvinylidene difluoride membrane. Membraneswere blocked for nonspecific binding and then incubated with primaryantibodies at the recommended dilutions followed by the appropriate sec-ondary antibody conjugated to horseradish peroxidase. Immunoreactivebands were visualized by using enhanced chemiluminescence solution(Pierce). Membranes were stripped and blotted for tubulin as a loadingcontrol. Bands were quantified by densitometry using the Photoshop(Adobe) histogram function. Antibodies and methods for analyzing the ef-fect of αS on the phosphorylation of Tus1(1–300) are given in SI Materialsand Methods.

GTP-Rho1 Pull Down. The 3Flag-Rho1 labeled strain transformed with pAG426,pAG426-αS, or pAG426-αS(Δ2–10) was pregrown in noninducing medium andthen induced in galactose medium for 8 h. Lysates were prepared and in-cubated with either Pkc1-RBD (David Pellman, Harvard University, Boston)bound to GST agarose beads or GST agarose beads as a control, as described inref. 14. After several washes with PBS, the beads were analyzed for retentionof GTP-Rho1 by SDS/PAGE followed by Western blotting with an anti-Flagantibody. Total Rho1 was used as loading control.

GST-Cdc5 Pull Down. The cdc15-2 strain with integrated TUS1-13-myc trans-formed with pAG426, pAG426-αS, or pAG426-αS(Δ2–10) was pregrown insucrose medium at 23 °C and then diluted (OD600 = 0.2) into galactosemedium and induced overnight at 23 °C. The cultures were shifted to 37 °Cto arrest the cell cycle at anaphase for 4 h and then released by returning to23 °C. After a 1-h recovery, the cells were harvested and lysed. A sample ofsupernatant (3 mg of total protein) was mixed with an aliquot of GST-PBD,GST-Pincer, or GST-KD bound to agarose beads. After overnight incubation at4 °C, the beads were washed, boiled for 5 min at 95 °C, and centrifuged. Sampleswere subjected to SDS/PAGE followed byWestern blotting. Amousemonoclonalanti-myc (9E10) (gift from Kelly Tatchell, Shreveport, LA) was used as theprimary antibody.

Wang et al. PNAS Early Edition | 5 of 6

BIOCH

EMISTR

Y

Kinase and Cross-Linking Assays. HA-Cdc5 was immunoprecipitated from ayeast cell lysate as follows. BY4741 cells transformedwith HA-Cdc5 expressingplasmid p384 (Angela Amon, Massachusetts Institute of Technology, Cam-bridge, MA) or empty vector were grown, harvested, and lysed with glassbeads. The supernatant was incubated with anti-HA antibody (2367P; CellSignaling) for 10 h, then protein-A-agarose beads were added, and the so-lution was incubated for another 2 h. The kinase assay consisted of 200-μLsamples of 0.1 mM αS or αS(Δ2–10), 10 mM MgCl2, 1 mM ATP, and 20 μL ofHA-Cdc5 bound agarose beads (or controls beads with bound antibody butno Cdc5) in 20 mM Tris·Cl (pH 7.4). After 24 h at 30 °C, the reaction mixturewas centrifuged and the supernatant was subjected to SDS/PAGE followedby Western blotting with an antibody specific for phospho-Ser-129 αS(Epitomics). The cross-linking assay consisted of 200-μL samples containing0.5 mM αS or αS(Δ2–10), 0.5 mM DTSSP (3,3′-dithiobis-[sulfosuccinimidyl-propionate]) (Pierce), and 20 μL of the agarose beads prepared as above in50 mM Hepes (pH 7.4). After 1-h incubation at 4 °C, the reaction was stoppedby adding 50 mM Tris. After several washes, the beads were subjected toSDS/PAGE followed by Western blot analysis with an antibody against αS.

The Plk2 kinase assay was conducted in 15 μL of 20 mM Tris·Cl buffer (pH7.4) containing 0.05 mM purified αS or αS(Δ2–10), 10 mM MgCl2, 1 mM ATP,and 0.13 μg/mL human GST-Plk2 (PV4204; Invitrogen). After 3-h incubationat 30 °C, the reaction mixture was halted by adding 15 μL of 2× SDS/PAGEloading buffer and boiled for 5 min. The supernatant was subjected to SDS/PAGE followed by Western blotting with the phospho-specific αS antibody.Mammalian cell culture. SH-SY5Y human neuroblastoma cells (ATCC) weremaintained in a 1:1 mixture of EMEM (ATCC) and Ham’s F-12 medium

(Invitrogen) supplemented with 10% FBS (Biowest), 1% penicillin-streptomycinsolution in a humidified incubator, 5% CO2 at 37 °C. Stably transfected celllines (SI Materials and Methods) were kept in the same medium supple-mented with 200 μg/mL G418.Fluorescence microscopy. Fluorescent images of yeast cells were acquired withan Olympus AX70 microscope equipped with an Olympus UPlanFl 100×/1.35N.A. objective and a CoolSNAP HQ CCD camera (Roper Scientific). For GFPdetection, a Chroma 41001 filter was used (excitation 480/40 nm, emission535/50 nm; Chroma Technology). Image analysis, filter wheels, shutters, andz axis stepping motor were under the control of imaging software Slide-book 4.0 (Intelligent Imaging Innovations). Images were acquired at roomtemperature.Statistical analysis. P values were determined by an unpaired, two-tailedStudent’s t test when comparing two samples, by one-way ANOVA with aDunnett post hoc test when comparing more than two samples to a control,or by one-way ANOVA with a Tukey post hoc test when comparing multiplesamples. Experimental values are means ± SD of typically three to fiveindependent experiments. Microsoft Excel and KaleidaGraph were usedfor the statistical tests.

ACKNOWLEDGMENTS. We thank Angelika Amon, David Levin, SusanLindquist, David Pellman, and Kelly Tatchell for strains and plasmids. Thiswork was supported by National Institutes of Neurological Disorders andStroke Grant NS057656 and a grant by the Parkinson’s Disease Resource ofNorthwest Louisiana (to S.N.W.).

1. Dawson TM, Dawson VL (2003) Molecular pathways of neurodegeneration in Par-kinson’s disease. Science 302:819–822.

2. Spillantini MG, et al. (1997) Alpha-synuclein in Lewy bodies. Nature 388:839–840.3. Mazzulli JR, et al. (2011) Gaucher disease glucocerebrosidase and α-synuclein form

a bidirectional pathogenic loop in synucleinopathies. Cell 146:37–52.4. Polymeropoulos MH, et al. (1997) Mutation in the alpha-synuclein gene identified in

families with Parkinson’s disease. Science 276:2045–2047.5. Krüger R, et al. (1998) Ala30Pro mutation in the gene encoding alpha-synuclein in

Parkinson’s disease. Nat Genet 18:106–108.6. Zarranz JJ, et al. (2004) The new mutation, E46K, of alpha-synuclein causes Parkinson

and Lewy body dementia. Ann Neurol 55:164–173.7. Singleton AB, et al. (2003) alpha-Synuclein locus triplication causes Parkinson’s dis-

ease. Science 302:841.8. Anderson JP, et al. (2006) Phosphorylation of Ser-129 is the dominant pathological

modification of alpha-synuclein in familial and sporadic Lewy body disease. J BiolChem 281:29739–29752.

9. Gitler AD, et al. (2009) Alpha-synuclein is part of a diverse and highly conserved in-teraction network that includes PARK9 and manganese toxicity. Nat Genet 41:308–315.

10. Inglis KJ, et al. (2009) Polo-like kinase 2 (PLK2) phosphorylates alpha-synuclein atserine 129 in central nervous system. J Biol Chem 284:2598–2602.

11. Archambault V, Glover DM (2009) Polo-like kinases: Conservation and divergence intheir functions and regulation. Nat Rev Mol Cell Biol 10:265–275.

12. Pak DT, Sheng M (2003) Targeted protein degradation and synapse remodeling by aninducible protein kinase. Science 302:1368–1373.

13. Lee KJ, et al. (2011) Requirement for Plk2 in orchestrated ras and rap signaling,homeostatic structural plasticity, and memory. Neuron 69:957–973.

14. Yoshida S, et al. (2006) Polo-like kinase Cdc5 controls the local activation of Rho1 topromote cytokinesis. Science 313:108–111.

15. Levin DE (2005) Cell wall integrity signaling in Saccharomyces cerevisiae. MicrobiolMol Biol Rev 69:262–291.

16. Srinivasa SP, Bernstein LS, Blumer KJ, Linder ME (1998) Plasma membrane localizationis required for RGS4 function in Saccharomyces cerevisiae. Proc Natl Acad Sci USA 95:5584–5589.

17. Outeiro TF, Lindquist S (2003) Yeast cells provide insight into alpha-synuclein biologyand pathobiology. Science 302:1772–1775.

18. Sussman A, et al. (2004) Discovery of cercosporamide, a known antifungal naturalproduct, as a selective Pkc1 kinase inhibitor through high-throughput screening.Eukaryot Cell 3:932–943.

19. Yarmola EG, Somasundaram T, Boring TA, Spector I, Bubb MR (2000) Actin-latrunculinA structure and function. Differential modulation of actin-binding protein functionby latrunculin A. J Biol Chem 275:28120–28127.

20. Yoshida S, Bartolini S, Pellman D (2009) Mechanisms for concentrating Rho1 duringcytokinesis. Genes Dev 23:810–823.

21. Sakchaisri K, et al. (2004) Coupling morphogenesis to mitotic entry. Proc Natl Acad SciUSA 101:4124–4129.

22. Jung US, Sobering AK, Romeo MJ, Levin DE (2002) Regulation of the yeast Rlm1transcription factor by the Mpk1 cell wall integrity MAP kinase. Mol Microbiol 46:781–789.

23. Iwata A, Maruyama M, Kanazawa I, Nukina N (2001) alpha-Synuclein affects theMAPK pathway and accelerates cell death. J Biol Chem 276:45320–45329.

24. Nemoto S, Xiang J, Huang S, Lin A (1998) Induction of apoptosis by SB202190through inhibition of p38beta mitogen-activated protein kinase. J Biol Chem 273:16415–16420.

25. Wennerberg K, Rossman KL, Der CJ (2005) The Ras superfamily at a glance. J Cell Sci118:843–846.

26. Martindale JL, Holbrook NJ (2002) Cellular response to oxidative stress: Signaling forsuicide and survival. J Cell Physiol 192:1–15.

27. Vamvaca K, Lansbury PT, Jr., Stefanis L (2011) N-terminal deletion does not affectα-synuclein membrane binding, self-association and toxicity in human neuroblastomacells, unlike yeast. J Neurochem 119:389–397.

28. Mbefo MK, et al. (2010) Phosphorylation of synucleins by members of the Polo-likekinase family. J Biol Chem 285:2807–2822.

29. Burke D, Dawson D, Stearns T (2000) Methods in Yeast Genetics (Cold Spring HarborLab Press, Cold Spring Harbor, NY), pp 103–174.

30. Lee YJ, Wang S, Slone SR, Yacoubian TA, Witt SN (2011) Defects in very long chainfatty acid synthesis enhance alpha-synuclein toxicity in a yeast model of Parkinson’sdisease. PLoS ONE 6:e15946.

6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1206286109 Wang et al.