Lys-63-linked Ubiquitination by E3 Ubiquitin Ligase Nedd4-1 Facilitates Endosomal Sequestration of Internalized α-Synuclein*

Background: Nedd4-1 catalyzes the Lys-63-linked ubiquitination of aS. Results: The Lys-63-linked ubiquitination of aS by Nedd4-1 facilitates endosomal targeting of extracellular aS. Conclusion: Compared with C-terminal deficient mutants, wild type-aS is preferentially internalized and translocates to endosomes. The overexpression of Nedd4-1 leads to the accumulation of aS in endosomes. Significance: Nedd4-1-mediated Lys-63 ubiquitination specifies the fate of internalized aS. α-Synuclein (aS) is a major constituent of Lewy bodies, which are not only a pathological marker for Parkinson disease but also a trigger for neurodegeneration. Cumulative evidence suggests that aS spreads from cell to cell and thereby propagates neurodegeneration to neighboring cells. Recently, Nedd4-1 (neural precursor cell expressed developmentally down-regulated protein 4-1), an E3 ubiquitin ligase, was shown to catalyze the Lys-63-linked polyubiquitination of intracellular aS and thereby facilitate aS degradation by the endolysosomal pathway. Because Nedd4-1 exerts its activity in close proximity to the inner leaflet of the plasma membrane, we speculate that after the internalization of aS the membrane resident aS is preferentially ubiquitinated by Nedd4-1. To clarify the role of Nedd4-1 in aS internalization and endolysosomal sequestration, we generated aS mutants, including ΔPR1(1–119 and 129–140), ΔC(1–119), and ΔPR2(1–119 and 134–140), that lack the proline-rich sequence, a putative Nedd4-1 recognition site. We show that wild type aS, but not ΔPR1, ΔPR2, or ΔC aS, is modified by Nedd4-1 in vitro, acquiring a Lys-63-linked ubiquitin chain. Compared with the mutants lacking the proline-rich sequence, wild type-aS is preferentially internalized and translocated to endosomes. The overexpression of Nedd4-1 increased aS in endosomes, whereas RNAi-mediated silencing of Nedd4-1 decreased endosomal aS. Although aS freely passes through plasma membranes within minutes, a pulse-chase experiment revealed that the overexpression of Nedd4-1 markedly decreased the re-secretion of internalized aS. Together, these findings demonstrate that Nedd4-1-linked Lys-63 ubiquitination specifies the fate of extrinsic and de novo synthesized aS by facilitating their targeting to endosomes.

The intraneuronal aggregation of misfolded ␣-synuclein (aS), 3 known as a component of Lewy bodies (LB), is a pathological hallmark of Parkinson disease (PD). After the discovery of LB-like inclusions in the grafted neurons of PD patients who had previously received transplants of fetal mesencephalic neurons (1), increasing evidence has suggested that both monomeric and oligomeric aS can be secreted into the extracellular milieu (2), thereby affecting the physiological state of neighboring cells. Previous studies have revealed that the cellular uptake of fibrillar aS requires physiological temperatures and dynamin-1 (3,4), a master regulator of endocytic vesicle formation, suggesting the active participation of the endocytic machinery. However, aS incorporation into cells is not completely disabled by the inhibition of endocytosis (3), indicating that other pathways, such as direct penetration, macropinocytosis, pore formation, and diffusion, might be involved in aS internalization (5). This notion is supported by previous studies showing that both monomers and oligomers of aS can freely passes though plasma membranes (4).
Because the aggregated aS in the brains of PD patients is robustly ubiquitinated (6 -8), the ubiquitin modification of aS may regulate the biogenesis of LBs and could contribute to neurodegeneration in PD. Among the E3 ligases that catalyze aS ubiquitination, researchers have focused on Nedd4 (neural precursor cell expressed developmentally down-regulated protein 4) because this E3 ligase is highly expressed in neurons containing LBs and catalyzes the Lys-63-linked ubiquitination of aS (9,10). Mammalian Nedd4 exists in two isoforms, Nedd4-1 and Nedd4-2. Structurally, the Nedd4 isoforms are composed of a C2 domain, 3-4 WW domains, which recognize proline-rich motifs (PPXY or LPXY) in substrate proteins (11,12), and a catalytic domain homologous to the E6-AP C-terminal (HECT) domain at the C terminus. Nedd4-1 binds to the plasma membrane via the C2 domain in a Ca 2ϩ -dependent manner (13) and controls the ubiquitination of membrane-bound receptor proteins, such as the epithelial Na ϩ channel (11). Modification by Lys-63-linked ubiquitination promotes the sorting of receptors for endocytosis and their subsequent degradation through the endolysosomal pathway (14). Given that Nedd4-1 exerts its E3 ligase activity in close proximity to the inner leaflet of the plasma membrane, the catalytic activity of Nedd4-1 likely acts preferentially on membrane-resident aS. In this study, we explored the possible mechanism by which the ubiquitination of aS by Nedd4-1 in the juxtamembrane cytoplasm could contribute to the incorporation and endosomal targeting of aS.
RNAi Interference-To ablate Nedd4 expression in cultured cells, siRNA specifically targeting human Nedd4-1 (sc-41079, Santa Cruz Biotechnology) or Nedd4-2 (NEDD4LHSS118599, Invitrogen) or a scrambled control siRNA (sc-36869, Santa Cruz Biotechnology) was used. To silence human CHMP2B, a target-specific siRNA (sc-72895, Santa Cruz Biotechnology) was used. For human aS silencing, a 25-nucleotide-long siRNA was used, 5Ј-GACCAAAGAGCAAGUGACAAAUGUU-3Ј (BONAC, Kurume, Japan) (15). SH-SY5Y cells in log phase growth were transfected with target-specific or control-scrambled siRNAs by electroporation. Then 24 h after gene silencing, 5 M recombinant aS was added to the culture media, and the cells were incubated for another 24 h. Subcellular Fractionation-For the subcellular fractionation of cultured cells, we adopted an established protocol (16). After being cultured for 24 h in medium containing 5 M aS, the cells (1 ϫ 10 7 ) were resuspended in 1 ml of ice-cold buffer (10 mM Tris/acetic acid, pH 7.0, and 250 mM sucrose) and homogenized using 20 strokes in a 2-ml Dounce tissue grinder. In some experiments, the cells were pretreated with 5 M chloroquine (CQ, Sigma) and/or 10 M MG132 (Millipore/Calbiochem) before exposure to aS. The cell homogenate was initially cleared by centrifugation (4000 ϫ g for 2 min) to remove debris, undestroyed cells, plasma membrane, and nuclei. The supernatant was ultracentrifuged at 100,000 ϫ g (Hitachi Koki Co., Ltd., Tokyo, Japan) for 2 min to pellet the mitochondria, endosomes,

Nedd4-1 Targets Internalized ␣-Synuclein to Endosomes
and lysosomes (fraction EL). Lysosomes were isolated from fraction EL by osmotic lysis for 10 min using a 5:1 ratio (v/v) of pellet to water. After another centrifugation step at 100,000 ϫ g for 2 min, the lysosomal proteins were recovered in the supernatant, and the mitochondria and endosomes remained in the pellet. The total protein in the culture media was extracted by trichloroacetic acid (TCA)/acetone precipitation and dissolved in 8 M urea, 5% SDS with sonication (2).
Immunocytochemistry and Confocal Laser Scanning Microscopy-Recombinant aS was labeled with Alexa Fluor 488 or Alexa Fluor 555 reactive fluorescent dyes with tetrafluorophenyl ester binding (Invitrogen). Unattached dyes were carefully removed by size exclusion chromatography using Bio-Gel P-6 (Bio-Rad). Twenty four hours before the treatment with fluorophore-labeled aS (5 M, 24 h), enhanced GFP-tagged Rab5a, Rab7, or Rab11a (markers for early, late, or recycling endosomes, respectively) was transfected into the SH-SY5Y cells. In some experiments, GFPtagged human p62 expression constructs based on the baculovirus backbone (Premo Autophagy Sensors, Invitrogen) were co-transfected 16 h prior to the microscopic observation. To visualize lysosomes or mitochondria, the cells were treated with 100 nM Lyso-Tracker Green or 100 nM MitoTracker Green for 30 min.
The cells were fixed in 4% (w/v) paraformaldehyde in phosphate-buffered saline (PBS) for 20 min, permeabilized with 0.5% Triton X-100 in PBS for 5 min, and blocked with 3% normal goat serum (Wako Pure Chemical Industries, Osaka, Japan) in PBS for 30 min. Anti-calnexin rabbit pAb (1:1000, Stress-Gen), anti-PDI rabbit pAb (1:1000, StressGen), anti-syntaxin-6 rabbit mAb (1:100, CST), anti-ubiquitin mouse mAb (1:500, Millipore), anti-Nedd4-1 rabbit mAb (1:500, Abcam), and anti-CHMP2B rabbit pAb (1:500, Abcam) were used as primary antibodies and incubated for 18 h at 4°C. Next, the cells were incubated with anti-mouse IgG Alexa Fluor 488 or 647 conjugates (1:2000, Invitrogen) for 1 h at room temperature. In some experiments, nuclei were counterstained by DRAQ7 (1:500, Biostatus, Leicestershire, UK). Fluorescent images were analyzed with the Olympus FV300 confocal laser scanning microscope system. All images were scanned by laser scanning microscope under identical conditions of 512 ϫ 512 pixels with a 12-bit/pixel resolution. Under each set of conditions, 10 -20 cells in six randomly chosen fields were analyzed to evaluate the sizes of the Rab7 vesicles that contain aS using ImageJ software (National Institutes of Health). The diameter of each vesicle was determined by comparison with a standard, and the differences between the conditions were analyzed with the Mann-Whitney U test using GraphPad Prism. The data are expressed as the means Ϯ S.E.
Immunohistochemistry-Formalin-fixed, paraffin-embedded sections of the substantia nigra from five human patients with PD were subjected to immunohistochemical investigations using the avidin-biotin-peroxidase complex method with a Vectastain ABC kit (Vector Laboratories, Burlingame, CA). Polyclonal antibodies against Nedd4-1 (Sigma, 1:50; Millipore, 1:50) and Rab7A (Sigma, 1:50) were used as the primary antibodies. The sections were pretreated by heating for 15 min at 121°C. Diaminobenzidine was used as the chromogen. The sections were counterstained with hematoxylin.

RESULTS
aS Is Internalized and Accumulates in the Endosomes of Neuronal Cells-First, we assessed the molecular weight of recombinant aS by BN-PAGE. The total recombinant aS migrated at ϳ40 kDa and showed two smaller peaks over 700 kDa (Fig. 1a).
To eliminate the HMW aS species, recombinant aS was further separated using a 100-kDa pore-size filter, and LMW aS migrating at ϳ40 kDa was collected (designated as Ͻ100-kDa aS in Fig. 1a). To visualize the internalized aS, we prepared less than 100-kDa aS covalently bound to a fluorescent compound (Alexa-aS). The labeling efficiency was estimated to be ϳ2.29 (mol of dye)/(mol of protein) by fluorometric analysis (Alexa-aS in Fig. 1b). In BN-PAGE, Alexa-aS has two peaks, and the conformational changes might be caused during the process of fluorescence labeling or size exclusion chromatography (Fig. 1c). After treatment with 5 M Alexa-aS for 24 h, acceptable amounts of internalized aS were detected in SH-SY5Y cells (Fig. 1d). Subcellular fractionation analysis revealed that internalized aS appeared in the cytosol in the early phase (8 h) and thereafter gradually increased to 24 h. However, the increased aS in the endosomes and lysosomes was inversely proportional to the cytosolic aS, indicating the translocation of cytosolic aS to endo/lysosomal compartments (Fig. 1e). The kinetics of aS re-secretion into the medium was similar to that of the cytosolic aS. It should be noted that the endosomal fraction isolated by our method contains both early and late endosomes because this fraction is positive for Rab5 and Rab7 (Fig. 1e). Next, we performed immunocytochemical analysis using SH-SY5Y cells expressing enhanced GFP-tagged Rab5a, Rab7, and Rab11a together with the acidic organelle marker LysoTracker or the mitochondrial marker MitoTracker. Most of the internalized aS co-localized with Rab7-positive late endosomes and Lyso-Tracker-positive structures and co-localized to a lesser degree with Rab5a-positive early endosomes and Rab11a-positive recycling endosomes. Note that none of the Alexa-aS corre-

Nedd4-1 Targets Internalized ␣-Synuclein to Endosomes
sponds with the fluorescence of MitoTracker (Fig. 1f). Another intriguing finding is that large Alexa-aS inclusions were occasionally surrounded by Rab7-positive vesicular structures. The level of internalized aS was in proportion to the amount of aS added to the culture medium (Fig. 1g). We also observed re-secreted aS in the medium 1 h after replacing it with fresh medium. The endosomal fraction used in this study also contains mitochondria, Golgi, autophagosomes, and endoplasmic reticulum because this fraction was positive for mitochondrial Tom20, the Golgi marker, syntaxin-6, the autophagosome marker p62, and the endoplasmic reticulum marker BiP (Fig.  1g). To exclude the possibility that internalized aS was located in these organelles, SH-SY5Y cells treated with Alexa-aS were fixed and subjected to double immunostaining. As shown in Fig. 1h, none of the Alexa-aS-positive red dots were co-localized with endoplasmic reticulum proteins (calnexin and PDI) or syntaxin-6. In addition, we performed p62 immunostaining using the aS-exposed, Nedd4-1-expressing cells in the presence of the autophagy inhibitor CQ (Fig. 1i). The treatment with CQ caused numerous p62-positive fluorescent puncta, indicative of autophagosomes; however, the Alexa-aS-positive puncta were scarcely co-localized with p62.
Lys-63 Linkage-specific Ubiquitination Enhances the Incorporation and Endosomal Targeting of Extracellular aS-After crossing the plasma membrane, some of the internalized aS exits the membrane, and some remains in the cytoplasm for minutes. To confirm this possibility, we performed pulse-chase experiments to determine how long internalized aS remains in the cytosolic fraction. After exposure to 5 M recombinant aS for 24 h, the cells were extensively washed with PBS and further cultured for the indicated periods (0 -60 min) in fresh medium in the absence of aS. The cytosolic fraction was isolated and subjected to Western blot analysis (Fig. 3a). After the removal of aS from the culture medium, the aS in the cytosolic fraction decreased by half within 5 min, demonstrating that the free cytosolic aS can easily cross the plasma membrane and immediately disappear from the cytosol. Based on these observations, it is tempting to speculate that specific ubiquitin modifications by Nedd4-1 might prevent aS secretion by leading to the juxtamembrane localization of aS to endosomal compartments. To confirm this possibility, we investigated whether Nedd4-1 silencing affects the re-secretion of internalized aS into the culture medium. As expected, after removing aS from the medium, Nedd4-1 silencing increased the aS re-secreted in the medium, whereas the elimination kinetics of cytosolic aS seem to be unaltered (Fig. 3, a and b). In addition to this re-secretion to the extracellular space, there are several possibilities that could explain the decrease of cytosolic aS as follows: proteolysis by the ubiquitin-proteasome system and

Nedd4-1 Targets Internalized ␣-Synuclein to Endosomes
sequestration to autophagic degradation. To investigate this, we examined the rate of aS disappearance in the presence of CQ and proteasomal inhibitors (MG132). Of note, treatment with CQ and/or MG132 did not affect the aS turnover up to 60 min (Fig. 3, a and b). These results suggest that most of the aS that disappeared from the cytosol is not removed by lysosomal and/or proteasomal degradation. To investigate the effect of Nedd4-1 on the stability of extracellularly derived aS, we then investigated the time-dependent change of internalized aS in cells in the presence of the protein synthesis inhibitor cycloheximide. In the presence of 50 g/ml cycloheximide, the basal amount of endogenous aS in Nedd4-1expressing cells was significantly lower compared with control cells, but the decay kinetics of intracellular aS seem to be Western blotting using P4D1 antibody, which recognizes both mono-and polyubiquitin, shows an HMW smear in the samples incubated with UbcH5a, UbcH5b, UbcH5c, UbcH6, UbcH7, and UbcH8 (lower panel). Mono-ubiquitin (Ub) bands are weakly apparent in each lane. E2 enzymes with one or two ubiquitins were occasionally detected (double asterisk). Note that the HMW aS smear (aS-Ub n ) visualized by anti-aS Ab (Syn-1) was solely detected in samples incubated with UbcH5a, UbcH5b, UbcH5c, and UbcH6. c, types of ubiquitin linkages preferentially generated by Nedd4-1. Recombinant wild type aS and Nedd4-1 were incubated together with ubiquitin that has no active lysine residue (Lys-0), or ubiquitin that has only one active lysine (K) residue (Lys-6, Lys-11, Lys-27, Lys-29, Lys-33, Lys-48, and Lys-63). Western blot using P4D1 antibody shows that the polyubiquitination smear is exclusively detected in the samples with wild type and Lys-63 (lower panel). Polyubiquitinated, HMW aS species were detected in the samples incubated with wild type, Lys-11, Lys-33, and Lys-63 ubiquitin (upper panel). Several bands at ϳ20 -40 kDa are believed to be unspecific because these bands were also detected in the reaction with Lys-0 ubiquitin (asterisk). Note that the autophagy/lysosome inhibitor CQ and/or proteasome inhibitor MG132 pretreatment did not alter the kinetics of aS elimination from the cytosol (middle panel). The results of densitometric analyses are presented in the lower panel. Statistical analyses were performed by Dunnett's multiple comparisons test. IB, immunoblot. b, to evaluate the aS re-secreted into the medium, the cells were incubated for 1 h with fresh medium after the aS exposure (5 M for 24 h). Note that aS was detected in the medium treated with recombinant aS (Rec aS). Silencing of Nedd4-1 substantially increased the level of re-secreted aS (upper panel). CQ and/or MG132 (CQ and CϩM) treatments did not affect the aS re-secretion. The results of densitometric analyses are shown in the lower panel. Asterisk, p Ͻ 0.05 by Dunnett's multiple comparisons test. c, Nedd4-1 expression affects the stability of endogenous aS. In the presence of the protein synthesis inhibitor cycloheximide (CHX), the basal amount of aS in cells overexpressing Nedd4-1 was significantly lower than that in the control cells (asterisk, p Ͻ 0.01 by Sidak's multiple comparisons test). Densitometry from three independent experiments is presented (right panel).
unchanged between control and Nedd4-1-expressing cells (Fig. 3c). We observed that Nedd4-1 did not influence the turnover of de novo synthesized HA-aS (data not shown).
Nedd4-1 Facilitates the Endosomal Targeting of aS-Intracellular aS is divided into two types based on its derivation as follows: aS internalized from the extracellular space and as de novo synthesized aS. Thus, the intracellular aS of different origins may have distinct fates during the process of intracellular vesicular transport. To determine whether Nedd4 specifies the intracellular trafficking of aS of different origins, we co-expressed wild type, C867A, or ⌬WW Nedd4-1 in SH-SY5Y cells stably expressing HA-tagged aS. Following 48 h of Nedd4-1 expression, the amount of aS increased in the endosomal fraction (Fig. 4a). This finding is consistent with earlier research showing that Nedd4-1 promotes the endolysosomal sorting of aS for degradation (9). After washing off the medium, we did not detect re-secreted aS within 1 h, but a detectable amount of extracellular aS was observed up to 24 h (Fig. 4a). In an inverse correlation with endosomal aS, the level of re-secreted aS was slightly decreased in wild type Nedd4-1-overexpressing cells.
To monitor the trafficking of extracellular aS in more detail, endogenous aS of SH-SY5Y cells was silenced prior to rec-aS exposure (Fig. 4, b and c). Compared with the results using HA-aS expressing cells, extracellularly derived aS tended to accumulate in the endosomal compartments (Fig. 4c). Furthermore, the ectopic expression of wild type Nedd4-1 substantially augmented the endosomal sorting of internalized aS. In contrast to the up-regulation of endosomal aS, we found that aS levels in the lysosomal fraction decreased slightly in Nedd4-1expressing cells. The excessive amount of Nedd4-1 may have  (Fig. 4a), extracellularly derived aS tended to accumulate in the endosomal compartments. Furthermore, the ectopic expression of wild type Nedd4-1 considerably increased the amount of endosomal aS (double asterisk, p Ͻ 0.01). In contrast, the amount of aS in the medium and the lysosomal fraction slightly, but significantly, decreased in wild type Nedd4-1expressing cells (asterisk and double dagger, respectively, p Ͻ 0.05). The densitometric values of monomeric aS in each fraction are presented in the right panel. d, Nedd4-1 expression significantly increased the size of aS-positive inclusions surrounded by Rab7-positive endosomal structures (white arrows) in SH-SY5Y cells exposed to Alexa-aS. Scale bar, 10 m. Asterisk, p Ͻ 0.05 by two-tailed Mann-Whitney U test (right panel). e, number of Rab7-positive or Lysotracker-positive vesicles having Alexa-aS was quantitatively analyzed. Asterisk, p Ͻ 0.01 Sidak's multiple comparisons test. f, cells overexpressing Rab7-EGFP and Nedd4-1 were treated with Alexa-aS and then subjected to immunostaining. Scale bar, 10 m. g, de novo synthesized HA-aS in the endosome was 20% versus the cytosolic aS ( §). By contrast, the internalized recombinant aS was more extensively distributed in the endosome (50% compared with the cytosolic aS). (dagger, p Ͻ 0.01; double dagger, p Ͻ 0.05; pilcrow, p Ͻ 0.05; double pilcrow, p Ͻ 0.05; asterisk, p Ͻ 0.05 by Dunnett's multiple comparisons test). The densitometric values of mono-aS are shown in the right panel. In these cells, the level of endosomal aS was decreased (double asterisk, p Ͻ 0.05), whereas re-secreted aS was significantly increased (asterisk, p Ͻ 0.001). The densitometric values of monomeric aS in each fraction is presented in the right panel.
hampered extracellularly derived aS transport from the endosome to the lysosome because Nedd4-1 expression significantly increased both the occurrence and size of aS-positive inclusions surrounded by Rab7-positive endosomal structures (Fig. 4, d  and e). Theoretically, the cargo's ubiquitin modification needs to be detached before entry into the endosomal luminal space (22). Thus, it is feasible that only 3% of Alexa-labeled recombinant aS in Rab7-positive vesicles was ubiquitinated in this cel-lular model (Fig. 4f). The importance of ubiquitin-dependent sorting machinery is strengthened by the fact that the silencing of CHMP2B, a component of the endosomal sorting complex required for transport (ESCRT), resulted in a marked reduction of endosomal targeting of recombinant aS (Fig. 5a). As far as the subcellular distribution is concerned, de novo synthesized HA-aS in the endosome was 20% versus the cytosolic aS. By contrast, the internalized recombinant aS was more extensively

Nedd4-1 Targets Internalized ␣-Synuclein to Endosomes
distributed in the endosome (50% compared with the cytosolic aS) (Fig. 4g). To further establish the role of Nedd4 in the endosomal targeting of aS, we knocked down endogenous Nedd4-1 or Nedd4-2 prior to aS exposure. As shown in Fig. 4h, a substantial decrease of endosomal aS and an elevation of extracellular aS were observed in Nedd4-1-deficient cells, although this effect was unremarkable in Nedd4-2-silenced cells. This find-ing may indicate that Nedd4-1 and Nedd4-2 are functionally distinct and that Nedd4-1 is the main human Nedd4 isoform that affects the fate of internalized aS in human neuronal cells.
Nedd4-1 Is a Component of aS-positive Inclusions in Cellular Model and the Brain Lesion of PD-To verify the functional role of Nedd4-1 in the formation of aS-positive inclusions, SH-SY5Y cells transfected with Nedd4-1 were treated with Alexa-aS for in Fig. 5b, aS-positive inclusions were partly positive for Nedd4-1 and a component of ESCRT machinery, CHMP2B. This outcome indicates that the ESCRT pathway is closely involved in the process of aS inclusion formation. Furthermore, we found that the core structure of LBs, a pathological hallmark of PD, showed strong immunoreactivity with Rab7A (Fig. 5c) and Nedd4-1 (Fig. 5, d and e).
C-terminal Residues of aS Are Required for Nedd4-1-mediated Endosomal Targeting-Nedd4-1 recognizes PR regions of target proteins via the WW domain, thereby exerting its E3 ligase activity. Although aS does not have a canonical PPXY motif, it contains a relatively proline-rich domain (PVDPDNEAYEMPSE-EGYQDYEPEA) at its C terminus (9). To determine the functional importance of the PR sequence in Nedd4-1-mediated ubiquitination, three deletion mutants of aS, designated ⌬PR1(1-119 and 129 -140), ⌬PR2 (1-119 and 134 -140), and ⌬C(1-119), were generated and expressed in E. coli (Fig. 6, a and b). To discriminate among these mutants by Western blotting, the following Abs were used: Syn-1, Syn211, and 2628. The Syn-1 Ab, which recognizes aa 91-99, detected all mutant proteins, whereas the Syn211 Ab, which recognizes aa 121-125, did not detect all mutant proteins. The 2628 antibody, the exact specificity of which is unknown, detected wild type and ⌬PR1 aS but not ⌬PR2 and ⌬C aS (Fig. 6b). Not only wild type aS but also mutant aS migrated to ϳ30 -40 kDa (Fig. 6c). After the in vitro ubiquitination assay, all recombinant proteins were subjected to Western blotting using the Syn-1 Ab. We found that wild type aS produced high molecular weight bands in the presence of UbcH5b. In contrast, high molecular weight bands appeared less noticeable in samples containing these mutants (Fig. 6d). These results provide evidence that the PR sequence is required for aS ubiquitination by Nedd4-1. To further substantiate and extend these observations, SH-SY5Y cells were exposed to 5 M wild type aS or to mutant aS lacking the PR sequence for 24 h. Intriguingly, we found that the endolysosomal as well as the cytosolic targeting of all mutants was greatly inhibited compared with that of wild type aS, whereas the PR mutation substantially increased the amount of re-secreted aS in the culture media (Fig. 6e). These results suggest that mutant aS forms that show disturbance in Nedd4-mediated polyubiquitination cannot be sorted into endolysosomal compartments. Furthermore, at least the nine amino acids (PR sequence 120 PDNEAYEMP 128 ) in the C terminus of aS may be required for endosomal targeting of aS by Nedd4-1.
Pro-120 and Pro-128 in the PR Sequence Are Essential for Nedd4-1-mediated Endosomal Targeting of aS-To further elucidate which proline residue(s) within the PR sequence are essential for the Nedd4-1-mediated aS ubiquitination, we generated recombinant aS in which proline 120 or proline 128 was replaced with alanine (P120A and P128A, respectively, in Fig. 7,  a and b). Because HECT-type ubiquitin ligase sometimes attaches to phosphorylated serine/threonine residues (23), serine 129 (Ser-129), which is a major phosphorylation site in aS, was also substituted with alanine (S129A, in Fig. 7, a and b). After size exclusion filtration using a 100-kDa Amicon filter, all mutant aS and wild type aS migrated to ϳ40 kDa on BN-PAGE (Fig. 7c). In vitro ubiquitination assays using the UbcH5b E2 enzyme revealed that P120A and P128A aS mutants were less prone to being polyubiquitinated by Nedd4-1 compared with wild type aS, whereas the ubiquitination of S129A aS by Nedd4-1 was comparable with that of wild type-aS (Fig. 7d), suggesting that Pro-120 and Pro-128 are key residues for the Nedd4-1-mediated ubiquitination of aS. Intriguingly, we found that the level of intracellular aS was slightly decreased in P120A mutant aS, whereas P120A substitution substantially increased the amount of re-secreted aS in the culture media (Fig. 7e). Furthermore, subcellular fractionation analysis showed that the cytosolic and endosomal targeting of P120A and P128A mutants was disturbed compared with wild type and S129A aS. Cumulatively, these findings confirm the functional importance of the Pro-120 and Pro-128 residues for Nedd4-1-mediated endosomal targeting of aS.

DISCUSSION
One of the most exciting themes emerging from recent neurodegenerative research is the transcellular spread of pathogenic protein aggregates in affected brain lesions. To understand how aggregated proteins, such as aS, travel from cell to cell, the underlying mechanism responsible for the uptake and secretion of aggregate-prone proteins must be elucidated. The internalization of aS by cells is thought to be initiated by aS attachment to the outer leaflet of the plasma membrane via its amphipathic N-terminal domain (24,25), which induces membrane curvature, tubulation, and breaking (26). Prior evidence has suggested that endocytic processes play a role in aS internalization in both neuronal and glial cells (3,4,27); however, aS internalization was not found to be completely blocked by the disruption of the endocytic machinery. These findings indicate that mechanisms other than endocytosis may contribute to aS internalization (3,4,27). Indeed, there is evidence showing that fibrillar and nonfibrillar oligomeric aS species are incorporated via the endocytic machinery and that monomeric aS directly passes through the plasma membrane (4). Unfortunately, how aS crosses the plasma membrane remains to be determined.  119), were generated to determine the functional importance of the PR motif in Nedd4-1-mediated ubiquitination. b, characterization of the aS deletion mutants by SDS-PAGE. Syn-1 antibody, which recognizes amino acids (aa) 91-99, detected all mutant proteins, whereas Syn211 antibody, which recognizes aa 121-125, did not detect all mutant proteins. The 2628 antibody, the exact specificity of which is unknown, detected wild type and ⌬PR1 aS, but not ⌬PR2 and ⌬C aS. c, characterization of the aS deletion mutants by BN-PAGE. Both wild type aS and mutant aS migrated to ϳ30 -40 kDa. d, in vitro ubiquitination assays using UbcH5b reveal that all of the PR mutants had a reduced ability to form Nedd4-1-mediated polyubiquitinated aS (aS-Ub n ). The densitometric values of aS-Ub n smear on long exposure image were normalized by the values of mono-aS on short exposure images (p Ͻ 0.01; each mutant is compared with wild type by Dunnett's multiple comparisons test). e, SH-SY5Y cells were exposed to 5 M wild type and mutant aS lacking the PR sequence for 24 h and subjected to Western blot analysis. Note that the endolysosomal as well as the cytosolic targeting of all of the PR mutants was greatly inhibited compared with that of wild type aS, whereas the PR mutation substantially increased the amount of re-secreted aS in the medium (asterisk, p Ͻ 0.01; each mutant is compared with wild type by Dunnett's multiple comparisons test). The right panel shows the amount of monomeric aS normalized by each fraction marker. IB, immunoblot.

Nedd4-1 Targets Internalized ␣-Synuclein to Endosomes
Several possibilities have been postulated, including direct penetration (25), the formation of annular pore-like structures (28), and macropinocytosis (5,29). Although this notion is provocative, it is supported by analogous studies in other neurodegen-erative diseases, such as polyglutamine disease, in which polyQ aggregates can rapidly enter the cytosolic compartment of mammalian cells and nucleate the aggregation of soluble proteins with these polyQ tracts (30). Regardless of the mecha-FIGURE 7. Pro-120 might be a core frame of the aS PR sequence for Nedd4-1-mediated aS endosomal targeting. a, schematic presentation of PR sequence ( 120 PDNEAYEMP 128 ) in human aS. The PR sequence contains two proline residues (Pro-120 and Pro-128), and Ser-129 flanks the PR sequence. b, characterization of the aS proline-substituted mutants by SDS-PAGE. All mutants appear as monomeric aS (15-20 kDa) under denaturing conditions. c, mutants and wild type aS (after 100-kDa size exclusion) showed LMW aS by BN-PAGE. d, in vitro ubiquitination assays using UbcH5b reveal that P120A and P128A aS mutants were less prone to polyubiquitination (aS-Ub n ) by Nedd4-1 compared with wild type aS, whereas the ubiquitination of S129A aS by Nedd4-1 was comparable with that of wild type-aS. The values for the amount of aS-Ub n from long exposure are divided by the values of mono-aS from short exposure (asterisk, p Ͻ 0.01; double asterisk, p Ͻ 0.05 by Dunnett's multiple comparisons test). e, SH-SY5Y cells were exposed to 5 M wild type-aS or mutant, prolinesubstituted aS for 24 h and subjected to Western blot analysis. The levels of intracellular and cytosolic aS were slightly decreased in the P120A mutant aS (dagger, p Ͻ 0.001; double asterisk, p Ͻ 0.05), although this mutant substantially increased the amount of re-secreted aS in the culture media (asterisk, p Ͻ 0.05). Note that the cytosolic and endosomal targeting of P120A and P128A mutants was disturbed compared with wild type and S129A aS (double dagger, p Ͻ 0.01; pilcrow, p Ͻ 0.05 by Dunnett's multiple comparisons test). The right panel shows the amount of monomeric aS normalized by each fraction marker. IB, immunoblot.

Nedd4-1 Targets Internalized ␣-Synuclein to Endosomes
nisms involved in aS internalization, some extrinsic aS species can likely enter neuronal and/or glial cells directly, where they gain access to the cytosolic compartment and are subjected to further processing, modification, and transport.
In this study, we found that Nedd4-linked Lys-63 ubiquitination specified the fate of extrinsic and de novo synthesized aS by facilitating aS targeting to endosomal compartments. It appears that immediately after passing through the plasma membrane, the majority of the internalized aS is located just beneath the plasma membrane. Because Nedd4 localizes to the cytosolic space by associating with the inner plasma membrane leaflet via its C2 domain, Nedd4-1 likely preferentially catalyzes juxtamembrane aS localization. This notion is supported by previous studies showing that Nedd4-1-mediated ubiquitination is closely linked to the turnover and trafficking of cell-surface receptors (31,32). Although aS does not contain the canonical PPXY motif known to interact with the WW domain of Nedd4-1, the WW domain can recognize several motifs other than PPXY with varying affinities (33). For example, the WW domains of Nedd4 families recognize PPLP, PR motifs, and phospho-(Ser/Thr) residues, as well as PPXY (23). Our obser-vation using deletion mutants provides evidence that the PR sequences in the C terminus of aS, particularly residues Pro-120 and Pro-128, are important for proper recognition by Nedd4-1. It is uncertain why Nedd4 silencing did not show a similar effect on the cytosolic aS accumulation as the ⌬PR and P120A mutations. However, this could be attributed to the insufficient silencing efficacy of Nedd4 in cultured cells. The fact that a major phosphorylation site, Ser-129, occurs in the region flanking the aS PR sequence (34,35) raises the possibility that phosphorylation at Ser-129 might affect Nedd4-1-mediated ubiquitination. However, this effect is not likely, as we found no difference in aS ubiquitination regardless of the presence of Ser-129.
Nedd4-1 catalyzes both the mono-ubiquitination and Lys-63-linked polyubiquitination of target proteins (36). Lys-63linked ubiquitination is implicated in various cellular activities, including protein trafficking, DNA repair, stress responses, and signal transduction (37). Although mono-ubiquitination appears to be involved in endocytic trafficking, additional Lys-63-linked polyubiquitination is known to accelerate this trafficking process (38). More specifically, Lys-63-linked polyubiq-FIGURE 8. Nedd4-1 determines the fate of internalized aS in neuronal cells. Some extracellular aS is incorporated into neuronal cells by a dynamin-dependent mechanism and is transported into early endosomes (A). In contrast, a substantial amount of aS directly enters the cytosolic space, presumably via penetration of the plasma membrane (B). After crossing the plasma membrane, some of the internalized aS will exit the cells (C) and some will remain in the cytosol for minutes. Plasma membrane-resident Nedd4-1 binds to the C terminus of internalized aS through the WW domain and attaches a Lys-63-linked polyubiquitin chain to aS, thereby facilitating endosomal targeting (D). Most likely, the ESCRT complex recognizes the Lys-63-ubiquitinated aS and transports the aS into the late endosome through invagination of the endosomal membrane (E), which may promote aS degradation in lysosomes. The aS mutant that lacked the PR motif failed to sort into the late endosomes, most likely because it cannot be recognized by Nedd4-1 (F).

Nedd4-1 Targets Internalized ␣-Synuclein to Endosomes
uitination serves as a signal for protein sorting into the ESCRTdriven multivesicular body pathway by inward membrane invagination of endosomes (39). Previous studies have shown that LB showed immunoreactivity against ESCRT components such as CHMP2B and VPS4 (2,40,41). These findings are interesting when considering the biogenesis of LB because the pale body, a possible precursor of LB, often contains lysosomes, vacuolar structures, and ubiquitinated proteins (42). Moreover, our observation that Nedd4-1 is a component of LB also strengthens the hypothesis that the Nedd4-regulated endo/lysosomal sorting machinery might be involved in the buildup of aS-positive aggregates in affected brain lesions.
The mechanism by which cytosolic aS moves into the endosomal vesicle is poorly understood; however, our result showing that the silencing of CHMP2B, a component of ESCRT-III, can disrupt the endosomal accumulation of aS indicates the functional relevance of ESCRT machinery in the endolysosomal targeting of aS. Mechanistically, the endosomal targeting of ubiquitinated cargo and the formation of multivesicular bodies are mediated by the upstream ESCRT complexes (ESCRT-0, -I, and -II) on the surface of the endosomal membrane. ESCRT-III then recruits de-ubiquitinating enzymes to remove ubiquitin from the cargo before incorporating them into the intraluminal vesicles of multivesicular bodies (43,44). This could be a reason why we failed to detect strong ubiquitination in aS-positive inclusions surrounded by Rab7-positive late endosomes. Another important finding of this study is that the Nedd4-1mediated endosomal targeting of aS was accompanied by a prominent enlargement of the late endosome. Intriguingly, a marked enlargement of endosomal vesicles has also been shown in the affected brain and in a cellular model of Alzheimer disease (45,46). Why the overexpression of Nedd4-1 downregulated the lysosomal accumulation of aS in this study is uncertain. One possible explanation is that the excessive accumulation of aS might prevent early-to-late endosome transition. Indeed, previous studies have shown that aS itself is closely involved in vesicular trafficking events, such as Rab-mediated endoplasmic reticulum-Golgi transport and endosomal trafficking (47)(48)(49). An alternative possibility is that aS aggregates may decrease the lysosomal burden by inducing lysosomal rupture (50).
In summary, we found that Nedd4-1 markedly facilitated aS internalization, which was linked to Lys-63 linkage-specific polyubiquitination. Our results demonstrate how Lys-63linked ubiquitination contributes to the endosomal targeting and the endosomal accumulation of aS and therefore may be involved in the propagation and formation/clearance of Lewy pathology in PD (Fig. 8). Although the concept of the cell-tocell transmission of aberrant proteins has been recognized as a common phenomenon in many neurodegenerative diseases, the molecular mechanisms underlying the spread of protein misfolding likely differ depending on the biochemical nature of the protein aggregate, the level of cellular stress, and the cell type. Further studies are needed to gain insight into the cellular mechanisms of disease progression and to identify molecular targets for therapeutic intervention in PD and other neurodegenerative diseases.