Tubulin Polymerization-promoting Protein (TPPP/p25α) Promotes Unconventional Secretion of α-Synuclein through Exophagy by Impairing Autophagosome-Lysosome Fusion*

Background: The mechanism of unconventional secretion of α-synuclein is unknown. Results: Autophagy of α-synuclein followed by exocytosis of autophagy intermediates (exophagy) are increased by expression of TPPP/p25α. Conclusion: Exophagy of α-synuclein is increased by lysosomal dysfunction and/or altered trafficking of autophagosomes. Significance: Exophagy of α-synuclein might represent the first step in inter-neuronal spread of Lewy body disease. Aggregation of α-synuclein can be promoted by the tubulin polymerization-promoting protein/p25α, which we have used here as a tool to study the role of autophagy in the clearance of α-synuclein. In NGF-differentiated PC12 catecholaminergic nerve cells, we show that de novo expressed p25α co-localizes with α-synuclein and causes its aggregation and distribution into autophagosomes. However, p25α also lowered the mobility of autophagosomes and hindered the final maturation of autophagosomes by preventing their fusion with lysosomes for the final degradation of α-synuclein. Instead, p25α caused a 4-fold increase in the basal level of α-synuclein secreted into the medium. Secretion was strictly dependent on autophagy and could be up-regulated (trehalose and Rab1A) or down-regulated (3-methyladenine and ATG5 shRNA) by enhancers or inhibitors of autophagy or by modulating minus-end-directed (HDAC6 shRNA) or plus-end-directed (Rab8) trafficking of autophagosomes along microtubules. Finally, we show in the absence of tubulin polymerization-promoting protein/p25α that α-synuclein release was modulated by dominant mutants of Rab27A, known to regulate exocytosis of late endosomal (and amphisomal) elements, and that both lysosomal fusion block and secretion of α-synuclein could be replicated by knockdown of the p25α target, HDAC6, the predominant cytosolic deacetylase in neurons. Our data indicate that unconventional secretion of α-synuclein can be mediated through exophagy and that factors, which increase the pool of autophagosomes/amphisomes (e.g. lysosomal disturbance) or alter the polarity of vesicular transport of autophagosomes on microtubules, can result in an increased release of α-synuclein monomer and aggregates to the surroundings.

Aggregation of ␣-synuclein can be promoted by the tubulin polymerization-promoting protein/p25␣, which we have used here as a tool to study the role of autophagy in the clearance of ␣-synuclein. In NGF-differentiated PC12 catecholaminergic nerve cells, we show that de novo expressed p25␣ co-localizes with ␣-synuclein and causes its aggregation and distribution into autophagosomes. However, p25␣ also lowered the mobility of autophagosomes and hindered the final maturation of autophagosomes by preventing their fusion with lysosomes for the final degradation of ␣-synuclein. Instead, p25␣ caused a 4-fold increase in the basal level of ␣-synuclein secreted into the medium. Secretion was strictly dependent on autophagy and could be up-regulated (trehalose and Rab1A) or down-regulated (3-methyladenine and ATG5 shRNA) by enhancers or inhibitors of autophagy or by modulating minus-end-directed (HDAC6 shRNA) or plus-end-directed (Rab8) trafficking of autophagosomes along microtubules. Finally, we show in the absence of tubulin polymerization-promoting protein/p25␣ that ␣-synuclein release was modulated by dominant mutants of Rab27A, known to regulate exocytosis of late endosomal (and amphisomal) elements, and that both lysosomal fusion block and secretion of ␣-synuclein could be replicated by knockdown of the p25␣ target, HDAC6, the predominant cytosolic deacetylase in neurons. Our data indicate that unconventional secretion of ␣-synuclein can be mediated through exophagy and that fac-tors, which increase the pool of autophagosomes/amphisomes (e.g. lysosomal disturbance) or alter the polarity of vesicular transport of autophagosomes on microtubules, can result in an increased release of ␣-synuclein monomer and aggregates to the surroundings.
Parkinson disease (PD) 2 is characterized by the progressive loss of dopaminergic neurons in the substantia nigra, although other neural populations of the central nervous system (CNS) are also affected. PD develops and progress over many years and affects an increasingly larger volume of the nervous system. Braak et al. (1) have hypothesized that progression of PD correlates with a topographical spreading pattern of ␣-synuclein inclusion body disease throughout the nervous system, which is also the case of spreading ␣-synucleinopathy induced by injection of preformed ␣-synuclein fibrils into the cortex or striatum of transgenic ␣-synuclein-expressing mice (2). However, despite demonstrations that neurons are capable of secreting and internalizing ␣-synuclein (3,4), and that Lewy body pathology can be transferred from recipient to engrafted embryonic stem cells or fetal mesencephalic dopaminergic neurons in patients or experimental animals (5)(6)(7), little is known about the inter-neuronal transmission mechanisms of ␣-synuclein species.
Aggregated or modified forms of ␣-synuclein are degraded by proteasomal activity and different forms of autophagy (8,9). During quality control (QC) macroautophagy a double-layered isolation membrane, also termed the phagophore, encloses a volume of cytosol-containing damaged organelles or polyubiquitinated protein aggregates and thereby forms a vacuolar autophagosome (10). Generation of autophagosomes requires membrane lipids derived from endoplasmic reticulum, Golgi, or mitochondria (11) and is regulated by a set of conserved autophagy-related genes (ATG), including initiators (PI3K and Beclin-1) and elongators (conjugations systems ATG5-ATG12 and cytosolic light chain 3B (LC3B)) (12,13). The autophagic vacuole then quickly matures by fusion with compartments of the endosomal pathway before final fusion with lysosomes to generate an autolysosome where the luminal content of the autophagosome is degraded (14). Specifically, the fusion organelle of an autophagosome and an endosome (often a multivesicular body) is called an amphisome. In contrast to starvationinduced autophagy, which indiscriminately encloses a volume of cytosol and organelles for degradation and recycling of protein-building blocks, QC autophagy typically accepts polyubiquitinated proteins, long lived proteins, aggregates, and organelles as cargo. Selectivity is provided by ubiquitin-binding adaptor proteins p62/SQSTM1, NBR1, and HDAC6 (15,16), which link ubiquitinated cargo to LC3B, and in the case of HDAC6, additionally to dynein-dynactin motor proteins (16 -20). Neurons depend on autophagy for differentiation and survival (21), and p62/SQSTM1 and HDAC6 are required for development of inclusion bodies and aggresomes by directing minus-end transport of ubiquitinated cargo on microtubuli (15,16).
Lewy bodies invariably contain modified and aggregated ␣-synuclein as the main component along with a number of other nerve cell proteins typically highly ubiquitinated. The tendency of ␣-synuclein to form cytosolic aggregates is influenced by a number of other proteins, including synphilin-1 (22), protein interacting with NIMA 1 (PIN-1) (23), and TPPP/ p25␣ (hereafter referred to as p25␣) (24). The p25␣ protein binds to microtubules and by doing so lowers their plus-end growth rate and protects them from depolymerization (25)(26)(27). In addition, p25␣ potently stimulates aggregation of ␣-synuclein in vitro and localizes to Lewy bodies in vivo (24,28). In the CNS, p25␣ is mainly expressed in oligodendrocytes and is required for their differentiation (29), but during PD progression p25␣ becomes ectopically expressed in dopaminergic neurons. Conversely, ␣-synuclein is up-regulated in p25␣-expressing oligodendrocytes of patients with multiple system atrophy (24), and both proteins co-localize in characteristic inclusion bodies of these diseases.
In this study, we have asked the question whether ␣-synuclein species can be secreted by dopaminergic neurons following autophagy and exophagy. For this purpose, we have used the well known PC12 pheochromocytoma cell line as a model system of dopaminergic neuron-like cells to conditionally express the aggregation-prone ␣-synuclein A30P protein with or without p25␣ as a tool to increase ␣-synuclein aggregation and autophagic uptake. We show here that ␣-synuclein monomer and high molecular weight species can be secreted by exophagy and that p25␣ further augments this process by inhibiting autophagosomal fusion with lysosomes.

MATERIALS AND METHODS
Cell Culture and Neuronal Differentiation-The rat pheochromocytoma cell line PC12 (ATCC) was seeded on collagencoated culture dishes and cultured in DMEM containing 10% horse serum, 5% fetal calf serum, and 1% penicillin and streptomycin at 37°C in 5% CO 2 . For experiments, PC12 cells were seeded at a density of 45,000 cells/cm 2 and differentiated in DMEM containing 2% horse serum, 1% penicillin and streptomycin, and 100 ng/ml nerve growth factor (NGF) (2.5S subunit, Serotec) for 2 days before transgene expression was induced for an additional 2 days with doxycycline (200 ng/ml). SH-SY5Y cells were cultured in DMEM containing 10% fetal calf serum and 1% penicillin and streptomycin at 37°C in 5% CO 2 . For experiments, SH-SY5Y cells were differentiated with 10 M alltrans-retinoic acid for 6 -8 days to obtain a neuron-like phenotype before transgene expression was induced by treatment with doxycycline (200 ng/ml) for an additional 24 h.
Lentivector Construction and Transduction-For conditional expression of transgenes, cDNAs were PCR-cloned into BamHI/SalI-restricted lentiviral vector pLOX-cPPT.TW.W expressing cDNA under the control of a tetracycline-responsive promoter (Tet-On). Already established and tested PC12-rtT〈 or SH-SY5Y-rtTA cell lines expressing the reverse tetracycline-controlled transactivator (rtTA) were superinfected with pLOX-cPPT.TW.W lentivectors expressing the cDNAs of ␤-synuclein, ␣-synuclein wt , ␣-synuclein A30P , or p25␣ or with the pHR-cPPT.CMV.W lentivector for constitutive expression of mCherry-eGFP-LC3B (kindly provided by Professor Terje Johansen, Institute of Medical Biology, University of Tromsø, Norway (20)); 3ϫHA-tagged Rab1A, Rab3A, or Rab8 (all human cDNAs from Missouri S&T cDNA Resource Center); or Rab7-GFP fusion constructs alone or in combination as specified. Dominant negative and positive Rab27A mutants Rab27A-T23N and Rab27A-Q78L, respectively, were expressed under the control of a CMV promoter in the pLenti6-V5-TOPO vector (30). For expression of polyglutamine tracts from exon 1 of human huntingtin, templates contained in plasmids pEGFP/ HD-120Q and pEGFP/HD-18Q (31) were subcloned into the plasmid PCR2.1 (Invitrogen) by PCR using primers 5Ј-GGA TCC ATT CAT TGC CCC GGT GCT G-3Ј and 5Ј-GGA TCC CCG GCT GAG GAA GCT GAG GAG-3Ј and then BamHI cloned into the pHR-cPPT.CMV.GFP.W vector to express huntingtin-115Q-GFP or -18Q-GFP. All constructs were verified by sequencing. Knockdown of HDAC6 expression was carried out by transducing PC12 cells 72 h prior to analysis with Sigma MISSION TM HDAC6 shRNA in a pLKO.1 lentivector backbone (sense sequence AGGAAAGGTTCTCGAAGCA). PC12 cell populations knocked down for ATG5 were generated by the combined constitutive expression of two ATG5 shRNAs (sense sequence, CTTGGAACATCACAGTACA and CTG-TTTACAGTCAGTCTAT) contained in the pGIPZ lentivector (Open Biosystems). At least 85% of cells carried transgenes (conditionally or constitutively) following creation of stable PC12 populations by lentiviral transduction, whereas transduction of differentiating PC12 cells yielded transduction efficiencies ranging from 40 to 60%. Vector production was carried out as described previously (32).
Western Blotting-Cells were lysed in lysis buffer (100 mM NaCl, 50 mM Tris-HCl, 1 mM EGTA, 10 mM MgCl 2 , pH 7.2) containing 1% Triton X-100, phosphatase, and protease inhibitor mixture for 5 min at room temperature and thereafter kept on ice. Cell lysates were centrifuged at 16,100 ϫ g for 5 min at 4°C, and protein concentrations of the supernatant were determined with D c protein assay (Bio-Rad), before addition of Laemmli buffer and loading of equivalent protein quantities on SDS-polyacrylamide gels for Western blotting using chemiluminescent HRP detection substrate (Millipore).
TCA Protein Precipitation-Conditioned medium was harvested and centrifuged at 800 ϫ g for 5 min, 4°C, before ice-cold TCA was added to the supernatant and incubated on ice for 10 min. The protein precipitates were pelleted by centrifugation (16,100 ϫ g, 5 min, 4°C) and washed 4 -5 times in ice-cold acetone until the pellet appeared clear white. The pellets were dried at 95°C for 10 min, dissolved in 2.5ϫ Laemmli buffer, boiled for 20 min at 95°C, and subsequently processed for Western blotting.
Exosome Preparation-Conditioned medium from 10-cm Petri dishes with either PC12-␣-synuclein A30P or ␣-synuclein A30P / p25␣ cells was pre-centrifuged at 1,200 ϫ g for 10 min followed by another round of centrifugation at 10,000 ϫ g for 30 min. The supernatants were then passed through 0.22-m filters before centrifugation at 100,000 ϫ g for 1 h in a Beckman centrifuge with SW40Ti rotor. The pellet containing exosomes was washed once in PBS, and then aliquots of pellet and supernatant were separated by gel electrophoresis and processed for Western blotting.
Homogenization and Sucrose Gradient Fractionation-PC12 cells were suspended in PBS Ϫ and pelleted by centrifugation at 800 ϫ g for 8 min, 4°C. Cells were washed once in hypotonic buffer (170 mM sucrose, 75 mM NaCl, and 10 mM HEPES, pH 7.0) containing protease and phosphatase inhibitor mixture, snap-frozen in liquid nitrogen, re-thawed in 200 l of hypotonic buffer, and then repeatedly aspirated with a 27-gauge needle. The volume was diluted to 500 l with hypotonic buffer before centrifugation at 800 ϫ g for 8 min, 4°C. Subsequently, the supernatant was centrifuged at 10,000 ϫ g for 10 min, 4°C. The post-nuclear/mitochondrial supernatant was then applied to the top of a 15-45% sucrose gradient in gradient buffer (140 mM NaCl, 25 mM HEPES, pH 7.2) and ultracentrifuged in a Beckman centrifuge with an SW40Ti rotor head at 30,000 rpm for 18 h, 4°C. Twelve 1-ml fractions were collected from the bottom with a peristaltic pump, and aliquots thereof were processed for Western blotting.
Immunofluorescence and Electron Microscopy-PC12 cells were washed once in Hanks' balanced saline solution and fixed in a phosphate buffer containing 2% paraformaldehyde, pH 7.4, for 30 min. Immunofluorescence was essentially performed as described previously (33), and images were acquired with a Zeiss LSM510 confocal laser scanning microscope with a C-Apochromat ϫ63, 1.4 NA oil immersion objective, using the argon 488-nm and the helium/neon 543-and 633-nm laser lines for excitation of Alexa 488, 568, and 633, respectively. Confocal sections of 0.8 -1.0 m were collected and saved as 512 ϫ 512-pixel or 1024 ϫ 1024-pixel images at 12-bit resolution before import to Adobe Photoshop CS2 version 9.0.2, Zeiss LSM image browser, or ImageJ for compilation and quantification. When quantifying apoptotic cells, 8 -12 images (each containing around 200 cells) were acquired for each condition with a Fluar ϫ20, 0.75 NA water immersion objective. Cells positive for cleaved caspase-3 and the total number of cells (identified by ToPro-3 iodide) were quantified by a custom-made macro in ImageJ. For ultrastructural analysis, PC12 cells were fixed in a phosphate buffer containing 2% paraformaldehyde and 0.2% glutaraldehyde, pH 7.4, and processed for either Epon embedding or cryopreservation and sectioning as described previously (33). Cryo-EM sections were incubated with monoclonal mouse anti-␣-synuclein mAb LB509 and polyclonal rabbit anti-GFP (A64655, Molecular Probes) followed by 7-and then 14-nm gold-conjugated goat anti-mouse or -rabbit antibodies, respectively. Sections were examined in a Phillips CM 100 electron microscope equipped with a digital camera.
Fluorescence Recovery after Photobleaching (FRAP)-PC12 cells expressing mCherry-eGFP-LC3B were imaged in HEPES buffer, pH 7.4, 37°C. A circular region of interest (ROI) measuring 35 pixels in diameter was defined, and after three initial images mCherry fluorescence in the ROI was bleached with the 543-nm laser (30 iterations, at 100% laser strength). Subsequently, fluorescence recovery in the ROI was measured over a period of 3.5 min every 10th second. The fluorescence recovery was adjusted for general photobleaching. Eight to ten cells were measured for each experiment, and mean values were acquired from a minimum of three independent experiments.
FACS Analysis-For cell death analysis, PC12 cells were differentiated in 12-well culture plates and then were incubated with 2 g/ml propidium iodide (PI) for 15 min at RT and subsequently placed on ice. The cells were immediately analyzed in an Accuri C6 flow cytometer using the 488-nm laser for excitation and the FL3 optical filter (Ͼ670 nm) for emission detection. The fluorescence limit defining PI-positive cells was set based on control PC12 cells that did not receive PI. When analyzing the endosomal degradation pathway, PC12 cells were incubated with 40 g/ml bovine serum albumin (BSA) conjugated to the green fluorescent BODIPY FL dye (DQ-BSA) (Molecular Probes) for 6.5 h prior to flow cytometric analysis using the FITC filter. Mean fluorescence intensity of cells that did not receive DQ-BSA was subtracted from all sample values.
Statistical Analysis-Data were analyzed with one-way analysis of variance or Student's t test, and significance levels were defined as p values less than 0.05 (* or #), 0.01 (** or ##), or 0.001 (*** or ###). The largest and the smallest variances were tested for homogeneity with F-test before testing. Bar graph values are expressed as means and error bars as means Ϯ S.E.

RESULTS
Expression of p25␣ Inhibits Neurite Extension and Degradative Capacity of PC12 Cells-We used lentiviral transduction to obtain conditional (Tet-On) expression of ␣-synuclein wt or mutant ␣-synuclein A30P with or without co-expression of p25␣ in PC12 pheochromocytoma cells differentiated to catecholaminergic nerve cells with NGF. Unless otherwise specified, we adopted a differentiation protocol consisting of 2 days of pretreatment with NGF followed by 2 days of NGF and doxycycline treatment for transgene expression. At this time cells expressing ␣-synuclein, or control cells expressing the nonpathogenic but closely related ␤-synuclein, formed a fully extended neurite network (Fig. 1A), although ␣-synuclein expression, in particular ␣-synuclein A30P , caused ␣-synucleinpositive neurite blebs (Fig. 1B, arrowheads). In contrast, p25␣ expression alone or in combination with ␣-synuclein caused cell flattening and impaired neurite growth (Fig. 1, A and B). The Tet-On system provided an equal expression level of transgenes in PC12 cells with only minor leakage from the promoter (Fig. 1C). The anti-␣-synuclein antibodies used throughout this study, unless otherwise stated, react only weakly with endogenous rat ␣-synuclein in Western blotting applications. However, an alternative ␣-synuclein mAb (Abcam EP1646Y) with higher affinity to rat ␣-synuclein detects endogenous rat ␣-synuclein, and (with the limitations of cross-species antibody detection) we estimate that PC12 cells differentiated with NGF for 4 days express transgene ␣-synuclein A30P ϳ50-fold higher than the endogenous level of ␣-synuclein after 48 h of doxycycline induction. Of note, differentiated PC12 cells express considerably lower levels of ␣-synuclein than primary nerve cells from mouse (data not shown). TPPP/p25␣ expression is usually absent in cell lines, and we could not detect it in PC12 cells. All Western blot data in the following pertain to transgene ␣-synuclein, although endogenous ␣-synuclein is also recruited (see Fig. 4, C-E). Lowering the serum concentration to initiate NGF-induced differentiation causes roughly 15% of ␤-synuclein-expressing control cells to die, but within the time frame, ␣-synuclein and p25␣ co-expression only further augmented cell death through apoptosis by roughly 3% (Fig. 1, D and E). Huntingtin containing pathological polyglutamine stretches is a substrate of both the proteasomal and autophagosomal degradation pathways, which both are known to be inhibited by modified forms of ␣-synuclein (34 -36). Failure to degrade huntingtin results in the formation of large inclusion bodies called aggresomes. We therefore tested the effect of ␣-synuclein and/or p25␣ expression on the ability of PC12 cells to deal with co-expression of a polyglutamine tract (Gln-115) from huntingtin fused to GFP, or as control the nonpathogenic Gln-18-GFP. As illustrated in Fig. 1, F and G, expression of either ␣-synuclein A30P or p25␣ alone or in combination increased huntingtin inclusion body formation up to 2-fold, indicating that p25␣ and ␣-synuclein A30P expression interferes with the degradative machinery of the cells.
Expression of p25␣ Redistributes ␣-Synuclein to Cytosolic Vesicles and Causes Increased Autophagy-The majority of PC12 cells expressing ␣-synuclein wt or ␣-synuclein A30P had a compact cell body, and ␣-synuclein was distributed mainly in the cell periphery and in neurites, whereas p25␣ had a more diffuse somatic distribution (Figs. 1B and 2, A and B). In contrast, when co-expressed, p25␣ co-localized closely with ␣-synuclein A30P and induced its redistribution into cytosolic aggregates, inclusions, and/or vesicles (Fig. 2C). In a subset of PC12-␣-synuclein A30P /p25␣ cells (around 5%), large inclusion bodies with diameters up to 5 m could be observed (Fig. 2D). PC12 cells expressing ␣-synuclein alone did not contain any distinguishable inclusion bodies ( Fig. 2A). ␣-Synuclein and its mutant forms are known to be degraded by macroautophagy (henceforth "autophagy") and chaperone-mediated autophagy (8,9), and neurons are dependent on basal autophagy (21,36). We therefore analyzed conversion of the autophagosomal adaptor protein LC3B in differentiated PC12 cells. A minor significant increase was observed in ␣-synuclein-expressing PC12 cells, but p25␣ expression markedly increased conversion of LC3B-I (cytosolic; upper band in Western blots) to LC3B-II (autophagosome membrane-conjugated; lower band) (Fig. 2, E and F). Also, an ϳ3-fold increase in the LC3B-II/LC3B-I ratio was observed in differentiated SH-SY5Y cells expressing ␣-synuclein A30P with p25␣ ( Fig. 2, G and H). We next subjected the cells to ultrastructural analysis to confirm an increased autophagic flux (Fig. 2, I-N). Co-expression of p25␣ increased the number of nascent autophagosomes in ␣-synuclein-expressing PC12 cells (Fig. 2, I, M, J-L, and O, black, open arrows), which correlated with an increased number of amphisomes, clearly containing luminal vesicles derived from fusion with late endosomes (Fig. 2, M and N, white filled arrows, shows the fusion of a late endosome and an autophagosome to form an amphisome), and multilamellar bodies containing whorls of partially degraded lipids (Fig. 2M, white open arrows). Autophagosomes were evident in all parts of soma as well as neurites, and autophagosomes were observed to sequester various organelles, including endosomes, dense core vesicles, and mitochondria ( Fig. 2, J-L).
␣-Synuclein A30P Co-localizes with Markers of Autophagosomes-To further pursue the involvement of autophagosomal protein clearance, we chose to focus on the aggregation prone mutant ␣-synuclein A30P , which is a more likely substrate for QC autophagy than ␣-synuclein wt . By indirect immunofluorescence of PC12 cells expressing ␣-synuclein A30P , we only observed few vesicular structures with overlapping fluorescence between ␣-synuclein A30P and the autophagosomal markers LC3B (Fig. 3A, arrow). This co-localization was increased considerably by co-expression of p25␣, which also to a lesser degree caused co-localization of ␣-synuclein A30P and the autophagosomal adaptor protein p62/SQSTM1 (Fig. 3, B and C, arrows), indicating that p25␣ substantially increases the suitability of ␣-synuclein as an autophagosomal substrate. We verified these light microscopic observations by cryo-immunogold labeling for EM, and for this purpose we made use of PC12 cells constitutively expressing an mCherry-eGFP-LC3B fusion construct (20). ␣-Synuclein detected by the monoclonal antibody LB509 could be co-localized with LC3B (detected with anti-GFP antibody) in autophagosomes in both PC12-␣- synuclein A30P (Fig. 3, D-F) and to a higher extent in ␣-synuclein A30P /p25␣-expressing cells (Fig. 3, G-I). In PC12-␣-synuclein A30P cells, ␣-synuclein reactivity could also be detected in unidentified compartments without LC3B labeling (Fig. 3F, black filled arrow). In addition, ␣-synuclein and LC3B co-localized, in particular in PC12-␣-synuclein A30P /p25␣ cells, in matured autophagosomal elements, likely representing amphisomes (Fig. 3, E, G, and H; white open arrows). For reasons discussed below, we also consider (LC3B-positive) multilamellar bodies (Fig. 3G, white open arrow) as amphisomes, although in the absence of functional antibodies for cryo-EM labeling of rat late endosomal markers we cannot conclusively say so. Cells expressing ␤-synuclein yielded only very sparse nonspecific labeling (data not shown). Collectively, the data demonstrate that in the presence of p25␣ expression, overexpressed ␣-synuclein A30P becomes a suitable substrate for autophagy.
Expression of p25␣ Induces Selective Autophagy of ␣-Synuclein A30P and Also Affects Endogenously Expressed ␣-Synuclein-The p25␣-mediated inclusion of ␣-synuclein into autophagosomes was selective, inasmuch as pathogenic Htt-115Q co-expressed in the same cells segregated entirely to p62/SQSTM1-positive aggresomes (Fig. 4, A and B). To rule out that the observed aggregation and autophagy of ␣-synuclein in the presence of p25␣ was due to overexpression of ␣-synuclein A30P , we also studied the effects on endogenous rat ␣-synuclein in PC12 cells expressing only p25␣ by immunofluorescence. As shown in Fig. 4, C and D, p25␣ caused a marked redistribution of endogenous ␣-synuclein from the cell periphery into ubiquitin-positive structures, which in themselves increased in numbers compared with control cells expressing ␤-synuclein. These structures were also immunoreactive for p25␣, as well as late endosome/amphisome markers KAI1 and mannose phosphate receptor (Fig. 4E).
␣-Synuclein A30P Is Excluded from Lysosomes in p25␣expressing Cells-By immunofluorescence we observed considerable co-localization between ␣-synuclein and lysosome-associated membrane protein 1 (LAMP1) in ␣synuclein A30P -expressing PC12 cells after treatment with protease inhibitors leupeptin and pepstatin A (Fig. 5, arrows), indicating that vesicular ␣-synuclein observed by immunofluorescence and cryo-immunogold labeling eventually is degraded in lysosomes. However, in PC12 cells co-expressing p25␣, ␣-synuclein A30P -positive vesicles and inclusions were conspicuously devoid of LAMP1 immunoreactivity suggesting impaired fusion of amphisomes with lysosomes (Fig. 5, B and C). Although some LAMP1-positive lysosomes (Fig. 5B, red arrows) were distributed in close proximity to ␣-synuclein-positive vesicles (Fig. 5B, green arrows), there was no evident overlap, and this was also the case for larger inclusion bodies (Fig.  5C, white arrows). To substantiate this further, we first fractionated homogenates from ␣-synuclein A30P -and ␣-synuclein A30P / p25␣-expressing PC12 cells on sucrose gradients by velocity centrifugation. Most ␣-synuclein A30P was contained within low density fractions (10 -12) of the gradient corresponding to cytosolic proteins, including LC3B-I, and low density membrane (Fig. 5D). In PC12 cells expressing ␣-synuclein A30P , a small accumulation of ␣-synuclein could be detected in the high density fractions (1-4) of the gradient corresponding to where lysosomes sediment (we were not able to identify a functional anti-rat LAMP antibody for Western blotting). These fractions contained only the LC3B-II form, which is incorporated into the membrane of autophagosomes. However, the high density fraction content of ␣-synuclein was diminished in PC12 cells co-expressing p25␣ (Fig. 5D). In addition, we performed a functional assay to investigate endosomal trafficking to lysosomes by incubating PC12 cells with fluorophoreconjugated BSA (DQ-BSA), which is internalized by endocytosis and becomes fluorescent only upon lysosomal proteolysis. Although PC12 cells expressing ␤-synuclein or ␣synuclein A30P showed a similar proficiency in DQ-BSA degradation, expression of p25␣ alone or in combination with ␣-synuclein A30P decreased the level of DQ-BSA fluorescence (Fig. 5E). More dramatic decreases, which were largely cell lineindependent, were obtained by either blocking lysosomal proteolysis with leupeptin/pepstatin A or by blocking fusion of late endosomes/amphisomes with lysosomes using the V-ATPase inhibitor bafilomycin A1, which inhibits endosomal and lysosomal acidification. Collectively, the results indicate that p25␣ impairs the delivery of ␣-synuclein to lysosomes by interfering with the fusion of autophagosome maturation intermediates including amphisomes with lysosomes.
Co-expression of p25␣ Impairs Autophagosome Maturation-We next analyzed the autophagosomal flux in greater detail by superinfecting PC12 cell lines with lentivector expressing the mCherry-eGFP-LC3B tandem construct (20). The construct is incorporated into the outer and inner leaflet of the isolation membrane of forming autophagosomes via LC3B. Although eGFP fluorescence is pH-sensitive and decreases upon fusion with low pH lysosomes, the mCherry fluorophore is pH-insensitive, which enables distinction of recently formed autophagosomes (co-localizing green and red dots) and autophagosomes, which have matured by fusion with lysosomes to form autolysosomes (only red dots). In PC12 cells expressing ␤-synuclein, ϳ80% of vesicular LC3B was associated with exclusively red dots indicating a continuous flux of autophagosomes that fuse with lysosomes (Fig. 6, A, F, and G). However, the ratio between mCherry-and mCherry/eGFP-positive vesicular compartments was decreased by both ␣-synuclein A30P and in particular p25␣, indicating an impairment of the autophagosomal flux, which could be mimicked in PC12 ␤-synuclein cells by treatment with bafilomycin A1 (Fig. 6, B, C, and E-G). PC12 cells expressing p25␣ alone or together with ␣-synuclein A30P furthermore displayed large autophagosomes/amphisomes (Ͼ1 m), and a subset of cells (2%) displayed structures with diameters ranging up to 5 m (Fig. 6, D and H). These structures correlate in size with the ␣-synuclein A30P -positive aggregates shown in Fig. 2D and could as above be instigated by bafilomycin A1 treatment of control cells alone (Fig. 6, E and H).
Co-expression of p25␣ Causes Unconventional Secretion of ␣-Synuclein by Exophagy-Despite the ability of p25␣ to impair autophagosome maturation and autolysosomal degradation, p25␣ expression surprisingly correlated with an augmented clearance of ␣-synuclein (Fig. 7A). In addition, clearance of ␣-synuclein in PC12 cells without p25␣ was not completely suppressed by addition of leupeptin/pepstatin A, which inhibits endosomal and lysosomal degradation. We therefore tested if ␣-synuclein was secreted to the surroundings by the PC12 cells. The conditioned medium after 2 days of transgene expression was precipitated with TCA and then analyzed for presence of ␣-synuclein by Western blotting. When expressed alone, both ␣-synuclein wt and to a higher extent ␣-synuclein A30P was secreted into the medium; however, secretion of ␣-synuclein was greatly up-regulated by p25␣ co-expression (Fig. 7B). TPPP/p25␣ was also secreted (Fig. 7B), it but had a longer cellular half-life than ␣-synuclein (data not shown).
In yeast, exocytosis of autophagosomes constitutes an unconventional secretion pathway (37). Consistent with this and our observations above, addition of the autophagy inhibitor 3-MA, which inhibits class III PI3K activity required for forma-tion of the isolation membrane, significantly reduced secretion of ␣-synuclein in PC12 cells co-expressing ␣-synuclein A30P and p25␣ (Fig. 7, B and F). Autophagy was effectively inhibited at the concentrations of 3-MA used, as the drug caused p62/ SQSTM1 accumulation (Fig. 7B) (15,20). In cells only expressing ␣-synuclein A30P , 3-MA on the contrary induced an increase in secretion of ␣-synuclein (Fig. 7, B and F). This is likely due to the persistent inhibitory effect of 3-MA on class I PI3K (which normally activates the mTOR pathway, and thereby opposes autophagy) and a transient inhibition of class III PI3K (required in both mTOR-dependent and -independent autophagy) as reported previously (38).
For space considerations, Western blots of conditioned medium have generally been cropped to show only monomer ␣-synuclein, but Fig. 7C shows that SDS-insoluble oligomeric and high molecular weight ␣-synuclein immunoreactive species were secreted alongside monomers and correlated in intensity with each other. By densitometry of Western blots, we estimate that between 6 and 12% of the total cellular ␣synuclein A30P was present in the medium after 2 days of culture. This makes it difficult to see changes in cellular levels of ␣-synuclein by Western blotting, but p25␣ expression markedly lowered the levels of endogenous ␣-synuclein in cell lysates, and this correlated with secretion of ␣-synuclein into the medium (Fig. 7D). The low amounts of secreted endogenous ␣-synuclein in the medium could only be detected by immunoprecipitation and not by TCA precipitation.
TPPP/p25␣-induced secretion of ␣-synuclein was not a consequence of the inability of p25␣-expressing cells to form an extensive neurite network (see Fig. 1, A and B). If PC12 cells were fully differentiated with NGF for 4 -6 days before induction of p25␣ expression with doxycycline, there was no subsequent effect on the neuritic network, and the cells maintained an augmented secretion of ␣-synuclein 4.9 Ϯ 1.3 (n ϭ 3)-fold higher than control cells without p25␣ (data not shown). Secretion was however a function of NGF differentiation, as the amount of ␣-synuclein secreted from undifferentiated PC12 cells (with or without p25␣ expression) was less than 25% that of differentiated cells (data not shown).
Chemical or Genetic Modification of the Autophagosomal-Lysosomal Pathway Affects Secretion of ␣-Synuclein-We further tested the effects of chemical modulation of the autophagosomal-lysosomal pathway on the secretion of ␣-synuclein by using a panel of inhibitors or enhancers of different forms of autophagy. As seen in Fig. 7, E and F, inhibition of lysosomal function by either leupeptin/pepstatin A or even more promi- nently with bafilomycin A1, which mimics the p25␣-induced block of autophagosomal fusion with lysosomes, caused a clear up-regulation of ␣-synuclein secretion. This was also the case when PC12 cells with or without p25␣ expression were treated with trehalose, an mTOR-independent enhancer of autophagy (39), which correlated with marked increases in LC3B-II levels (Fig. 7E). The increase in LC3B-II levels caused by bafilomycin and leupeptin/pepstatin is directly due to suppression of degradation of conjugated LC3B-II in lysosomes, whereas the increase mediated by trehalose is due mainly to increased conversion of LC3B. In contrast, rapamycin, an enhancer of mTOR-dependent autophagy (starvation-induced autophagy), did induce transient conversion of LC3B (see below), but only increased secretion moderately in PC12-␣-synuclein A30P cells and not at all in p25␣-expressing cells. Trichostatin, which inhibits HDAC6 and thereby deacetylation of cortactin required for fusion of QC autophagosomes with lysosomes (40), caused a modest up-regulation of ␣-synuclein secretion in PC12 cells with or without p25␣ expression.
The involvement of autophagosomes in secretion was verified by shRNA knockdown of ATG5, which is required for elongation and formation of the autophagosomal isolation membrane (12). Using a combination of two different shRNA constructs, ATG5 protein expression in PC12-␣synuclein A30P /p25␣ cells was reduced to 30%, which correlated with a decreased conversion of LC3B-I to LC3B-II and a 60% reduction in the amount of ␣-synuclein secreted (Fig. 7, G-I).
Collectively, the data demonstrate that p25␣ mediates unconventional secretion of ␣-synuclein through a pathway, which is dependent on autophagosome formation. In contrast, the levels of Htt-115Q-GFP detected in the conditioned medium of PC12 cell lines co-expressing huntingtin constructs were low and did not change significantly upon p25␣ expression (Fig. 7, J and K), likely reflecting the aggresome inclusion of Htt-115Q-GFP noted previously (Fig. 4, A and B).
Trehalose and Rapamycin-induced Autophagosomes Are Differentially Affected by p25␣ Expression-The different effects of trehalose and rapamycin on LC3B conversion and ␣-synuclein secretion after 48 h made us examine more carefully the effect of these autophagy enhancers. Fig. 8A shows the typical distribution and fluorescence of the mCherry-eGFP-LC3B construct in untreated PC12-␣-synuclein/p25␣ cells. As expected, 3-MA caused the number of autophagosomes to decrease, and apart from a few large vacuoles LC3B was mainly cytosolic. Trehalose induced an increased number of LC3B-positive structures, but as in the control situation, the autophagosomes were inhibited in their maturation (both red and green fluorescence). In contrast, the effect of rapamycin on LC3B conversion was transient (Ͻ24 h) and did not differ significantly from untreated cells after 48 h (Figs.   7, E and F, and 8, E and F). However, maturation of rapamycininduced autophagosomes in terms of acidification progressed unimpeded by p25␣ expression as assessed by the mCherry-eGFP-LC3B construct, which displayed an increased number of vesicular profiles with only red fluorescence (Fig. 8D). Collectively, the data therefore indicate that p25␣ expression specifically affects selective mTOR-independent autophagy.
Secretion of ␣-Synuclein Depends on Exocytosis of a Compartment with Late Endosomal/Amphisomal Characteristics-Exocytosis of late endosomal elements, including amphisomes, which arise from fusion with late endosomes, causes insertion of phosphatidylserine into the exoplasmic leaflet of the plasma membrane, and this can be detected by annexin-V staining (41). As binding of annexin-V to the cell surface is also used for detection of apoptosis, we co-stained the cells with anticaspase-3 antibodies to separate the two phenomena. As shown in Fig. 9, A and B, binding of Alexa-conjugated annexin-V to the cell surface was ϳ3-fold increased up to 45% by the expression of p25␣, while at this time point (2 days of transgene expression) apoptosis detected by immunoreactivity for cleaved caspase-3 did not exceed 4% (Fig. 9A, see also Fig. 1E) indicating exocytosis of late endosomes and amphisomes. Exocytosis of late endosomes causes the release of small intraluminal vesicles, termed exosomes, to the extracellular medium (42), and it has recently been forwarded that ␣-synuclein can be taken up directly into late endosomes and secreted with exosomes (43). As shown in Fig. 9C, PC12-␣-synuclein A30P /p25␣ cells expectedly secreted a higher amount of exosomes (identified by flotillin-1 expression) into the medium than ␣-synuclein A30Pexpressing PC12 cells. However, almost all the secreted monomeric or high molecular weight ␣-synuclein was soluble and did not associate with the exosome fraction, which we estimate contained less than 3% of the total secreted ␣-synuclein. It has recently been shown that the small GTPase Rab27A is essential for exocytosis of late endosomes by promoting the fusion of late endosomes with the plasma membrane (44). When we either silenced Rab27A by shRNA or expressed Rab27A mutants in PC12-␣-synuclein A30P /p25␣ cells, we encountered an increased mortality, which precluded analysis of active secretion of ␣-synuclein. However, in PC12 cells expressing ␣-synuclein A30P alone, expression of dominant active (Q78L) or negative (T23N) mutants of Rab27A increased or decreased secretion of ␣-synuclein, respectively, despite a quite low efficiency of transduction with this pLenti vector (ϳ40% transduced cells for Rab27A-Q78L and 25% for Rab27A-T23N) (Fig. 9, D-F).
level of lysosome fusion (40). It has recently been demonstrated that p25␣ inhibits the deacetylase activity of HDAC6 (27). In accordance with this, the level of acetylated tubulin, a substrate of HDAC6, was increased by p25␣ expression in PC12 cells (Fig.  10A), which could critically influence transport of mature organelles or autophagosome formation. As shown above, chemical inhibition of HDAC6 with trichostatin A increased secretion of ␣-synuclein (Fig. 7, E and F), so we started out by substantiating this correlation further by silencing HDAC6 with shRNA in PC12 cells only expressing ␣-synuclein A30P to mimic the p25␣-induced inhibition of HDAC6. Fig. 10, B and C, shows that expression of HDAC6-specific shRNA increased the level of acetylated tubulin more than 2-fold in PC12-␣synuclein A30P cells without altering the total pool of ␣-tubulin. Although this was a modest increase in acetylated tubulin levels compared with ␣-synuclein A30P /p25␣-expressing PC12, it translated into a disproportionally large increase in secreted ␣-synuclein (Fig. 10, B and D), probably reflecting that HDAC6 alters transport as well as fusion of autophagosomes with lysosomes. We further tested the effect of HDAC6 inhibition on autophagosomal maturation by flow cytometric analysis of the pH-sensitive eGFP moiety of the mCherry-eGFP-LC3B construct, which was expressed at equal levels in the PC12 cell populations (Fig. 10E). As expected, expression of p25␣ increased eGFP fluorescence in the cells due to inhibited mat-uration and acidification, and this effect was replicated in PC12 cells either expressing ␤-synuclein or ␣-synuclein A30P by treatment with trichostatin (Fig. 10F).
During live cell imaging of the mCherry-eGFP-LC3B construct, we noticed that the mobility of autophagosomes FIGURE 7. TPPP/p25␣ induces ␣-synuclein secretion, which can be modified by regulators of the autophagosomal degradation pathway. A, PC12 cells expressing ␣-synuclein A30P alone (␣-SNC A30P ) or with p25␣ (␣-SNC A30P /p25␣) were predifferentiated with NGF for 2 days and transgene expression induced for additionally 2 days. Doxycycline was then withdrawn, and the cells chased for 6 days with or without leupeptin (50 g/ml) and pepstatin A (67 g/ml). Representative Western blots using anti-␣-SNC (BD Transduction Laboratories) shows clearance of ␣-SNC from cell lysates acquired from day 0, 2, 4, and 6 after doxycycline withdrawal. B, representative Western blots using anti-␣-SNC (BD Transduction Laboratories), p62/SQSTM1 (SQSTM1), and ␤-actin of cell lysates and TCA-precipitated conditioned media obtained from PC12 cell lines expressing ␤-SNC, ␣-SNC wild type (␣-SNC wt ) , or ␣-SNC A30P with or without p25␣ in the presence or absence of 3-MA (10 mM) (n ϭ 3). C, representative Western blot of TCA-precipitated medium obtained from PC12 cells expressing ␣-SNC A30P alone or together with p25␣ using mouse anti-␣-SNC (BD Transduction Laboratories). Notice the presence of both monomeric (17 kDa) and high molecular weight (High Mw) forms of secreted ␣-SNC. D, PC12 cells expressing ␤-SNC (as control) or p25␣ were treated or not with leupeptin/pepstatin A (50 and 67 g/ml) for the last 24 h of culture before analysis of endogenous ␣-SNC in cell lysates and immunoprecipitates from conditioned medium (using BD Transduction Laboratories and LB509 mAbs) by Western blotting (using anti-␣-SNC rabbit mAb EP1646Y) as indicated. Data are representative of three independent experiments. E, PC12 cells expressing ␣-SNC A30P or ␣-SNC A30P /p25␣ were treated with bafilomycin A1 (15 nM), leupeptin (50 g/ml)/pepstatin A (67 g/ml), trichostatin A (20 M), rapamycin (0.5 M), trehalose (100 mM), or left untreated (control) for 48 h concurrently with doxycycline induction before Western blot analysis of TCA-precipitated conditioned media and cell lysates using antibodies as indicated. F, bar graph shows fold increase in integrated optical density (IOD) of TCA Western blot bands of secreted ␣-SNC (BD Transduction Laboratories) obtained from B and E relative to untreated PC12-␣-SNC A30P cells. Mean Ϯ S.E. of three independent experiments are shown. G, TCA-precipitated conditioned medium or cell lysates from PC12-␣-SNC A30P /p25␣ cells with or without stable co-expression of either control (pGIPZ) or ATG5 shRNA were analyzed by Western blotting using anti-␣-SNC (BD Transduction Laboratories), ATG5, or LC3B antibodies as indicated. H, bar graph shows IOD of ATG5 western bands normalized to PC12 cells expressing control shRNA (pGIPZ) and represents mean Ϯ S.E. of three independent experiments. I, bar graph shows IOD of ␣-SNC western bands (TCA samples) normalized to PC12 cells expressing control shRNA (pGIPZ) and represents mean Ϯ S.E. of three independent experiments. J and K, PC12 cell populations as indicated were transduced with either Htt-18Q-GFP or Htt-115Q-GFP for 2 days before transgene induction for a further 2 days. Conditioned medium was then TCA-precipitated and analyzed by Western blotting with polyclonal rabbit anti-GFP antibodies. J, a ϳ50-kDa band (upper arrow) immunoreactive with anti-GFP antibodies is seen exclusively in the conditioned medium from HTT-115Q-GFP-expressing cells but not from control HTT-18Q-GFP cells. K, similar secretion of HTT-115Q-GFP was observed in PC12 cells expressing ␤-SNC, p25␣, or ␣-SNC A30P with or without p25␣ expression. The increased reactivity to monomer GFP to the right on the blot is caused by overflow from an adjacent (data not shown) cell lysate lane. Upper and lower arrows indicate HTT-115Q-GFP fusion protein and monomeric GFP, respectively. appeared severely reduced in p25␣-expressing cells, and to quantitate this, we applied FRAP on PC12 cells constitutively expressing the mCherry-eGFP-LC3B construct. The percentage of recovered mCherry fluorescence after bleaching in small regions of interest is shown in Fig. 10, G-I, where it is evident that p25␣ expression suppressed recovery with 80%. Although co-expression of ␣-synuclein depressed recovery further, ␣-synuclein alone had no effect. Inhibition of cytosolic deacetylase activity with either trichostatin or HDAC6 shRNA in ␣synuclein A30P -expressing control cells replicated the effect of p25␣ expression (Fig. 10I). The less efficient reduction in autophagosome mobility after HDAC6 knockdown when compared with trichostatin treatment could be due to inefficient transduction as the knocked down cells could not be identified before selecting cells for FRAP analysis. See the supplemental movie of autophagosome mobility in the ␤-synucleinand ␣-synuclein A30P /p25␣-expressing PC12 cells from Fig. 10, G and H.
Rab GTPases Modulate ␣-Synuclein Secretion and Toxicity-The Rab family of small GTPases are important regulators of vesicular transport and fusion events along the biosynthetic and endosomal pathway. Recent genetic screens in yeast and further experimentation in animal models showed that Rab1A and Rab8 alleviate ␣-synuclein-induced cytotoxicity (45)(46)(47). We therefore overexpressed Rab1A and Rab8 in PC12 cells to observe the effect on ␣-synuclein secretion (Fig. 11A). Additionally, we expressed a Rab7-GFP fusion protein, as this GTPase is associated with maturation of late endosomal compartments by regulating fusion with lysosomes (48,49). Rab3A is known to regulate exocytosis of chromaffin-containing vesicles in PC12 cells and was used as a control. Although expression of Rab1A increased secretion of ␣-synuclein A30P 2-fold in PC12 cells only expressing ␣-synuclein A30P , it had no significant effect in PC12 cells co-expressing p25␣ (Fig. 11, B and C). However, most markedly, Rab8 increased ␣-synuclein A30P secretion severalfold in both cell lines without altering LC3B conversion significantly. To our surprise, expression of Rab7-GFP, which we expected to promote autophagosome maturation and lessen secretion, also modestly increased secretion. Interestingly, while promoting secretion, expression of Rab7 or Rab8 both reduced cytotoxicity of ␣-synuclein and p25␣ expression (Fig. 11, H and I), underscoring that within the time frame of analysis used throughout the study, release of ␣-synuclein is an active process and not correlated with cell death. When we analyzed the effect of Rab expression on the distribution of ␣-synuclein by confocal microscopy, we observed that neither Rab3 (data not shown) nor Rab1A affected the distribution of ␣-synuclein A30P (compare Figs. 11D with 2C).
In contrast, Rab8, which was mainly localized to subcortical areas of the cytosol, caused a more peripheral distribution of ␣-synuclein A30P (Fig. 11E). Rab8 could be seen to coat peripheral membrane elements also positive for LC3B as shown in Fig.  11F, identifying them as autophagosomes/amphisomes. Interestingly, Rab8 consistently and specifically (compared with Rab3 as control) caused accumulation of particular material on the cell surface or substratum, which was immunoreactive with anti-KAI1 antibodies. A fraction of this material was also positive for ␣-synuclein, and it likely represents the secreted and aggregated luminal contents of amphisomes and late endosomes (Fig. 11G). FIGURE 9. Secretion of ␣-synuclein is mediated by compartments with late endosomal/amphisomal characteristics. A, p25␣-expressing PC12 cells were incubated with Alexa 568-conjugated annexin-V on ice before fixation and indirect immunofluorescence with antibodies against cleaved caspase-3. Note widespread annexin-V surface staining in the absence of caspase-3 immunoreactivity. Bar, 10 m. B, PC12 cell lines labeled with Alexa 488-conjugated annexin-V on ice were analyzed by flow cytometry. The bar graph shows percentage annexin-V positive cells of the whole cell population and represents mean Ϯ S.E. of three independent experiments. C, conditioned medium (CM) from PC12 cells expressing ␣-synuclein A30P (␣-SNC A30P ) with or without p25␣ was centrifuged at 100,000 ϫ g to obtain a pellet (containing exosomes) and a supernatant (sup). Aliquots were then Western blotted with anti-␣-SNC (BD Transduction Laboratories) or anti-flotillin-1 (exosome marker) antibodies. The exosome fraction was applied on the gel at 6-fold the relative load of supernatant and medium. The blot is representative of two independent experiments. D-F, PC12-␣-SNC A30P cells were differentiated for 2 days and transduced at two different multiplicities of infection with lentivectors constitutively expressing FLAG-tagged Rab27A-Q78L (dominant positive) or Rab27A-T23N (dominant negative). Control wells received a GFP-expressing vector (pLenti). D, after 2 days of transgene induction cell lysates were prepared to show expression of Rab27A using anti-Rab27A mAb 4B12 (arrow points to transgene) or anti-FLAG antibodies. E, conditioned medium was TCA-precipitated and Western blotted with anti-␣-SNC antibodies (BD Transduction Laboratories). F, bar graph shows mean Ϯ S.E. of the integrated optical density (IOD) of ␣-SNC Western blot bands normalized to control-transduced cells (n ϭ 3). * and # denotes a statistic significant increase or decrease, respectively, in ␣-SNC secretion when compared with control-transduced PC12 cells.

DISCUSSION
The aim of this study was to evaluate the role of autophagy in degradation and unconventional secretion of ␣-synuclein species from dopaminergic-like neurons in culture.
TPPP/p25␣ Impairs the Autophagosomal Degradation Pathway-Aggregated or modified forms of ␣-synuclein are substrates of all major degradative systems in vivo and in vitro, including the ubiquitin proteasome system, chaperone-medi-ated autophagy, and macroautophagy (8,9,50,51). In this study we have used p25␣ expression as a tool to increase aggregation and autophagosomal uptake of ␣-synuclein. However, p25␣ is also in its own right an interesting protein in the context of ␣-synucleinopathy, because p25␣ is ectopically expressed in dopaminergic neurons in PD, and it is a component of Lewy bodies purified from post-mortem brain of patients with PD (24). Upon prolonged culture p25␣-expressing cells eventually Exophagy of ␣-Synuclein succumb to apoptotic cell death. However, within the time frame of experimentation used here, ␣-synuclein or p25␣ expression only raised mortality associated with neuronal differentiation by a few percent (from 15% in NGF-differentiated control cells to 18% in ␣-synuclein/p25␣-expressing cells).
It has recently been shown that p25␣ inhibits the deacetylase activity of HDAC6 (27), which is the predominant deacetylase in the cytosol of neurons, and in accordance with this, p25␣ expression in PC12 cells increased tubulin acetylation about 6-fold. Expression of HDAC6 shRNA in the absence of p25␣ afforded a 2-fold increase in levels of acetylated tubulin, but this still correlated with a slowed maturation of autophagosomes and a 3-fold increase in levels of secreted ␣-synuclein, suggesting that these effects of p25␣ expression are a consequence of HDAC6 inhibition. HDAC6 binds both polyubiquitinated cargo and the dynein-dynactin motor complex, and it is required for the retrograde transport of ubiquitinated cargo and formation of aggresomes (18,52). However, under conditions of proteasomal inhibition HDAC6 is also an important mediator of the autophagosomal-lysosomal pathway of protein degradation in neurons, and when overexpressed HDAC6 can rescue neurodegeneration in a fly model (16). Expression of p25␣ could potentially influence autophagosomal processing due to HDAC6 inhibition on several levels (see Fig. 12 for a schematic representation). First, p25␣ could interfere with the interactions between HDAC6 and dynactin, thereby uncoupling retrograde transport of polyubiquitinated cargo (18) and potentially autophagosomes (16). As estimated by FRAP, p25␣ expression severely reduced the mobility of autophagosomes, and direct inhibition of HDAC6 with trichostatin or shRNA knockdown similarly decreased mobility. We note that p62/ SQSTM1-mediated retrograde transport of Htt-115Q-GFP to form conspicuous aggresomes was not disturbed by p25␣ expression, suggesting that hyper-acetylation of tubulin does not by itself affect retrograde transport. Second, HDAC6 has recently been pinpointed to an important role further downstream in the autophagic pathway as it deacetylates cortactin, which in turn is specifically required for fusion of QC autophagosome intermediates with lysosomes (40). This fusion event was severely affected by p25␣ expression, as evidenced by the following: (i) lack of co-localization between ␣-synuclein and LAMP1; (ii) decreased amount of ␣-synuclein in heavy frac-tions following sucrose gradient fractionation; (iii) decreased mCherry to eGFP fluorescence ratio of the LC3B fusion protein, and (iv) decreased DQ-BSA degradation, which all testify to an inhibition of autophagosomal/amphisomal fusion with lysosomes. Therefore, by several mechanisms p25␣ expression caused the build up of late autophagosomal elements. These likely include the multilamellar bodies observed in abundance by EM in ␣-synuclein A30P /p25␣-expressing cells as late endosomes and thereby amphisomes receive lysosomal hydrolases from the biosynthetic pathway to commence degradation of luminal contents.
␣-Synuclein Is Secreted by Exophagy-Even though p25␣ imposed a partial inhibition of the degradative capacity of the cell, the turnover of ␣-synuclein was increased by p25␣ expression. We found that this phenomenon could be explained by a markedly increased secretion of ␣-synuclein into the external environment, and our data indicate that exocytosis of autophagosomes and amphisomes containing ␣-synuclein is responsible for this. In particular, amphisomes may be involved in the release of ␣-synuclein, as they constitute the compartment immediately preceding the p25␣-mediated maturation block. Such a role also correlates well with the known physiological exocytosis of late endosomes (42), as well as the recently described unconventional secretion of the cytosolic Acb1 protein through exophagy of autophagosome intermediates in yeast (37,53). In partial agreement with a recent report (43), we find that a small fraction (Ͻ3% of total secreted ␣-synuclein) of monomer ␣-synuclein is associated with exosomes. The significance of this is at present unclear, but secretion of exosomes represents a physiological mechanism for transferring cytosolic proteins in between cells that should be explored further in the context of disease transmission (54 -56). Exosomes, however, are clearly not required to start the pathological cascade of inter-neuronal transmission of ␣-synuclein-derived prionoids, as ␣-synuclein aggregates made in vitro from recombinant ␣-synuclein is taken up by neurons in culture, and suffices to induce propagating inclusion disease in ␣-synuclein transgenic mice (2,5).
Formation of autophagosomes was a prerequisite for secretion of ␣-synuclein. First of all, by cryo-immunogold labeling almost all vesicular ␣-synuclein was present in autophagosomes, amphisomes, or autolysosomes. Second, expression of p25␣ correlated with increased levels of phosphatidylserine on the plasma membrane in the absence of apoptosis, indicative of exocytosis of late endosome/amphisome elements. Finally, secretion of ␣-synuclein could be controlled by chemical or genetic modulation of autophagy initiation (trehalose, 3-MA, and Rab1), elongation (ATG5 shRNA), or autophagosome maturation (bafilomycin). Knockdown or introduction of mutants of Rab27A, which regulates fusion of late endosomes with the Exophagy of ␣-Synuclein plasma membrane (44), in PC12 ␣-synuclein A30P /p25␣ cells was problematic in terms of cell death, but expression of Rab27A dominant positive and negative mutants in PC12-␣synuclein A30P cells modulated the release of ␣-synuclein. In addition, the fact that ␣-synuclein A30P was also present in autophagosomes in the absence of p25␣ expression and that the basal release of ␣-synuclein in PC12-␣-synuclein A30P cells was up-regulated by trehalose (and Rab1A, Rab8, or HDAC6 FIGURE 11. Expression of Rab8 enhances ␣-synuclein secretion while lowering cell death. A, representative Western blots showing expression levels of HA-tagged Rab1A, Rab3A, or Rab8 and GFP-tagged Rab7 in PC12 cell lines expressing ␣-synuclein A30P (␣-SNC A30P ) or ␣-SNC A30P and p25␣ using anti-HA or -GFP antibodies. B, representative Western blots of TCA-precipitated media and cell lysates obtained from PC12 cells expressing ␣-SNC A30P or ␣-SNC A30P /p25␣ with or without co-expression of Rab proteins as indicated using antibodies against ␣-SNC (BD Transduction Laboratories) and LC3B. C, quantified integrated optical density (IOD) of ␣-SNC Western blot bands obtained in B are presented as fold increase relative to ␣-SNC A30P -expressing PC12 cells. Mean Ϯ S.E. of five independent experiments are shown. * denotes a statistic significant increase in IOD when compared with control cells within the respective cell line. D-F, indirect immunofluorescence of PC12-␣-SNC A30P /p25␣ cells co-expressing either HA-Rab1A (D) or HA-Rab8 (E and F). Rab8 was predominantly distributed toward the cell surface and caused a more peripheral distribution of ␣-SNC A30P (BD Transduction Laboratories) (E) and was co-localized partly with LC3Bpositive autophagosomes/amphisomes (F). Bars, 10 m. G, PC12-␣-SNC A30P /p25␣ cells co-expressing HA-tagged Rab8 were fixed after 2 days of transgene induction and then processed for immunofluorescence to visualize KAI1 (green), ␣-SNC (red), or F-actin detected with Alexa 633-conjugated phalloidin (blue). Note that Rab8 expression results in accumulation of material on the cell surface, which is immunoreactive with anti-␣-SNC and KAI1 antibodies (arrows). The  Proposed mechanism for p25␣ effects and exophagy of ␣-synuclein. The diagram illustrates how p25␣ may alter trafficking pathways (marked by numbers) and autophagosome dynamics. Expression of p25␣ causes aggregation and autophagosomal uptake of ␣-synuclein (␣-SNC), involving QC autophagosome adaptors p62/SQSTM1 and possibly HDAC6 (1). The autophagosome (AU) then fuses (2) with a late endosome (LE) to generate an amphisome (AM), which can travel retrogradely (3) to fuse with a lysosome (Lyso) thereby forming an autolysosome, which degrades ␣-SNC. Retrograde transport involves HDAC6-mediated interactions between LC3B and the minus-end-directed dynein-dynactin motor complex, which may be inhibited by p25␣. Fusion of amphisomes with lysosomes is promoted by HDAC6 deacetylase activity, which is inhibited by p25␣. Under conditions where pathway 3 is blocked (compromised HDAC6 activity, lysosomal dysfunction, and/or altered ratio of minus-to plus-end-directed trafficking of amphisomes), anterograde transport of amphisomes toward the cell surface takes place (4), where a fraction of competent amphisomes can undergo exocytosis regulated by Rab27A (5), to release ␣-SNC in monomer and aggregated/modified forms to the extracellular environment. The extent to which autophagosomes may directly contribute to exocytosis is unclear (6).
shRNA), indicates that exophagy of ␣-synuclein is a mechanism that is likely operational also in the absence of p25␣ expression. Exophagy was specific for ␣-synuclein inasmuch as co-expressed Htt-115Q-GFP was not taken up into autophagosomes and was only to a minor degree secreted into the medium. Both early and late endosomes can fuse with newly formed autophagosomes to form amphisomes (57). Recently, it was reported that interference with VPS4 function, required for formation of late endosomes, caused secretion of ␣-synuclein potentially through elements of the early endosomal pathway (58), but whether this involved exocytosis of early amphisomal elements is unknown.
The Rab family of small GTPases constitutes important regulators of biosynthetic and endosomal vesicular trafficking. Rab1A or Rab8 have been shown to rescue ␣-synuclein-induced cytotoxicity in small organism models, possibly by alleviating endoplasmic reticulum-to-Golgi transport deficits induced by ␣-synuclein expression (45,46). However, Rab8 also facilitates polarized vesicular transport toward the plasma membrane (59 -61). When overexpressed in PC12 cells, Rab8 caused a pronounced redistribution of ␣-synuclein A30P elements to the plasma membrane, which was accompanied by a significantly increased secretion of ␣-synuclein with or without p25␣ expression. Rab1A also increased ␣-synuclein secretion in PC12 cells expressing only ␣-synuclein A30P likely reflecting the reported role of Rab1A in autophagy initiation (36). Rab7 regulates dynamics and fusion events of late endosomal and autophagosomal elements (48,49,62). However, Rab7 overexpression did not circumvent the p25␣-imposed fusion block and instead increased the secretion of ␣-synuclein, potentially as a consequence of Rab7-promoted plus-end-directed microtubular transport of amphisomes through interactions with the LC3B-binding protein FYCO (63). Secretion of cytotoxic ␣-synuclein might be beneficial for the individual neuron as both Rab7 and Rab8, which promoted secretion, also reduced mortality significantly. However, secretion of ␣-synuclein may be a pathological mechanism of PD pathology transmission to nearby neurons (2,3,7), and in addition it can activate resident microglia, which mediate neuroinflammation and thereby exacerbate disease progression (64 -66).
In conclusion, we propose that exophagy, with or without ectopic expression of p25␣, may be an important physiological mechanism for unconventional secretion of ␣-synuclein from dopaminergic neurons, and we predict that lysosomal dysfunction observed in neurodegenerative diseases including PD (50) will aggravate such a release.