GABA transmission via ATP-dependent K + channels regulates a -synuclein secretion in mouse striatum

a -Synuclein is readily released in human and mouse brain parenchyma, even though the normal function of the secreted protein has not been yet elucidated. Under pathological conditions, such as in Parkinson’s disease, pathologically relevant species of a - synuclein have been shown to propagate between neurons in a prion-like manner, although the mechanism by which a -synuclein transfer induces degeneration remains to be identiﬁed. Due to this evidence extracellular a -synuclein is now considered a critical target to hinder disease progression in Parkinson’s disease. Given the importance of extracellular a -synuclein levels, we have now investigated the molecular pathway of a -synuclein secretion in mouse brain. To this end, we have identiﬁed a novel synaptic network that regulates a -synuclein release in mouse striatum. In this brain area, the majority of a -synuclein is localized in corticostriatal glutamatergic terminals. Absence of a -synuclein from the lumen of brain-isolated synaptic vesicles suggested that they are unlikely to mediate its release. To dissect the mechanism of a -synuclein release, we have used reverse microdialysis to locally administer reagents that locally target speciﬁc cellular pathways. Using this approach, we show that a -synuclein secretion in vivo is a calcium-regulated process that depends on the activation of sulfonylurea receptor 1-sensitive ATP-regulated potassium channels. Sulfonylurea receptor 1 is distributed in the cytoplasm of GABAergic neurons from where the ATP-dependent channel regulates GABA release. Using a combination of speciﬁc agonists and antagonists, we were able to show that, in the striatum, modulation of GABA release through the sulfonylurea receptor 1-regulated ATP-dependent potassium channels located on GABAergic neurons controls a -synuclein release from the glutamatergic terminals through activation of the presynaptic GABA B receptors. Considering that sulfonylurea receptors can be selectively targeted, our study highlights the potential use of the key molecules in the a -synuclein secretory pathway to aid the discovery of novel therapeutic interventions for Parkinson’s disease.


Introduction
Under conditions of low dopamine levels, as is the case in Parkinson's disease, the firing pattern of striatal neurons is dramatically altered thereby affecting GABAergic microcircuits.Thus, deregulation of GABAergic transmission in the striatum can greatly contribute to basal ganglia dysfunction (Gittis and Kreitzer, 2012).In the brain, adenosine triphosphate (ATP)-regulated potassium (K ATP ) channels are thought to modulate GABA release (Amoroso et al., 1989;Maneuf et al., 1996).K ATP channels are regulated by their sulfonylurea receptor [SUR1 (encoded by ABCC8) or SUR2 (encoded by ABBC9)] regulatory subunit.They function to couple metabolic state to cellular excitability and therefore are considered to be the 'metabolic sensors' of neuronal cells (Zeng et al., 2007).Of importance, K ATP channels have been implicated in the selective vulnerability of dopaminergic substantia nigra pars compacta (SNpc) neurons in Parkinson's disease (Liss et al., 2005).
-Synuclein is a presynaptic neuronal protein that is biochemically and genetically linked to Parkinson's disease (Vekrellis et al., 2011).Beyond the established involvement of -synuclein in the pathology of neurological diseases, collectively known as synucleinopathies, the normal function(s) of this protein still remains unknown (Bendor et al., 2013).In mice, during early postnatal development,synuclein is predominantly localized at presynaptic terminals (Iwai et al., 1995) where it is thought to play a role in active synaptogenesis including synapse formation, maintenance and pruning (Jakowec et al., 2001).In presynaptic boutons, the protein is also associated with synaptic vesicles (Fortin et al., 2005;Burre et al., 2010).
-Synuclein attached to the synaptic vesicle membrane is found to bind to the v-SNARE protein, synaptobrevin-2, thereby promoting SNARE-complex assembly in vivo and in vitro (Burre et al., 2010(Burre et al., , 2014)).Interestingly, changes in presynaptic mobilization of the readily releasable pool of vesicles that reduces the capacity of the reserve pool have been reported in -synuclein-deficient mice not only for dopamine but also for glutamate (Abeliovich et al., 2000;Cabin et al., 2002;Yavich et al., 2004;Gureviciene et al., 2007).
There is now considerable experimental in vivo evidence demonstrating that -synuclein can be taken up by neighbouring neurons leading to the seeding of the endogenous protein in recipient neurons (Desplats et al., 2009;Hansen et al., 2011;Angot et al., 2012;Luk et al., 2012).This cellto-cell transmission of -synuclein could be the mechanism of the pathology spreading observed in the Parkinson's disease brain (Kordower et al., 2008;Li et al., 2008).It becomes obvious that regulation of extracellular -synuclein levels can be of great importance for the initiation, progression and/or spreading of disease pathology.
-Synuclein is normally secreted from neuronal cells via a non-classical mechanism that is regulated by intracellular calcium and partially involves exosomes (Emmanouilidou et al., 2010).In this cellular system, the small GTPase Rab11 and the AAA (ATPases associated with diverse cellular activities)-ATPase, VPS4 are considered to play a role in the mechanism of -synuclein externalization in vitro (Haseqawa et al., 2011;Chutna et al., 2014).However, the mechanism of -synuclein secretion in vivo has not been elucidated.
Given the valid hypothesis that deregulation of extracellular -synuclein levels could be critical for the initiation and/ or the progression of Parkinson's disease, we sought to investigate the sites of -synuclein secretion as well as the mechanism(s) that regulate this release in vivo using wildtype and A53T -synuclein overexpressing animals.We have found that in mouse striatum the majority of -synuclein is localized in corticostriatal glutamatergic nerve terminals with lower amounts being present in thalamostriatal and nigrostriatal afferents.Using in vivo reverse microdialysis we show that the levels of -synuclein in mouse brain parenchyma are tightly controlled by specific cellular networks.Importantly, we demonstrate that -synuclein secretion is mediated by GABA B receptors present on the glutamatergic endings and is regulated by GABA through the modulation of SUR1-K ATP channels located on neighbouring GABAergic neurons.We believe that identification of the central players in -synuclein secretory pathway will provide us with novel targets to modulate extracellular -synuclein levels and may thus represent potential new routes of therapeutic intervention for Parkinson's disease.

Animals
Male homozygous transgenic C57BI/C3H mice expressing human A53T SNCA under the control of the prion promoter (Jackson Laboratory) and wild-type littermates were used at 7-10 months of age.The generation and phenotype of these mice has been previously described (Giasson et al., 2002).C57BL6/ JOlaHsd synuclein null mice (Harlan Laboratories) of the same sex and age were used as controls to verify antibody specificity.Animals were housed in the animal facility of the Biomedical Research Foundation of the Academy of Athens (BRFAA) in a room with a controlled light-dark cycle (12 h light-12 h dark) and free access to food and water.

Primary cortical and striatal neurons
Cultures of mouse (embryonic Day 16) cortical or striatal neurons were prepared as previously described (Vogiatzi et al., 2008).Dissociated cells were plated onto poly-D-lysinecoated dishes or glass coverslips at a density of $250 000/ cm 2 .Cells were maintained in Neurobasal Õ medium (Gibco, Invitrogen), with B27 serum-free supplements (Gibco, Invitrogen), 0.5 mM L-glutamine, penicillin (100 units/ml) and streptomycin (100 mg/ml).Cells were treated with compounds as indicated for 3 h at 37 C. Following treatment, both conditioned medium and neurons were collected forsynuclein measurement by enzyme-linked immunosorbent assay (ELISA).Prior to analysis, conditioned medium was centrifuged first at 500 g for 5 min and then at 4000 g for 10 min to remove cell debris.Neurons were harvested, washed twice with ice-cold phosphate-buffered saline (PBS) and lysed with STET lysis buffer [50 mM Tris (pH 7.6), 150 mM NaCl, 1% Triton TM X-100, 2 mM EDTA] on ice for 15 min.Protein content was estimated using the Bradford method (Bio-Rad).

Compartmentalized cultures of mouse cortical neurons
Cultures of primary cortical neurons from C57BL/6 mice (embryonic Day 16) were prepared as above.Dissociated cells (3.5 Â 10 5 cells / device) were plated in microfluidic compartmentalization devices (SND450, Xona Microfluidics), which effectively separate neuronal soma from axons, and were used after $8-10 days in culture.Conditioned medium from both somal and axonal sides was collected, centrifuged and analysed for the presence of -synuclein by ELISA.

Immunocytochemistry
Primary cortical or striatal neurons were plated in glass coverslips and fixed with 3.7% paraformaldehyde in PBS for 15 min.After washing and blocking with 5% normal goat serum in PBS, cells were incubated with primary antibodies (Table 1) for 16 h at 4 C. Cells were washed and incubated with Cy2-and Cy3-labelled secondary antibodies (Jackson Labs) for 1 h at room temperature.The fluorescent marker, Hoechst 33258 (1 mM; Sigma) was used to detect cell nuclei.Finally, cells were analysed by confocal microscopy.The purity of the neuronal cultures was examined by measuring the cells that were positive for the neuronal marker, NeuN, the astrocytic marker, GFAP and the microglial marker, Iba-1.

Surgery and reverse microdialysis
CMA-12 custom-made guide cannulas were stereotaxically implanted in the right striatum under isoflurane anaesthesia as previously described (Emmanouilidou et al., 2011).Four days after surgery, mice were removed to the microdialysis cage.CMA-12 custom made probes (2 mm, 100 kDa cut-off) were inserted and connected to a CMA 402 pump with a constant flow rate of 0.6 ml/min.Perfusion was performed in artificial CSF (147 mM NaCl, 2.7 mM KCl, 1.2 mM CaCl 2 , 0.85 mM MgCl 2 ) containing 0.15% bovine serum albumin.Prior to sample collection, the probe was allowed to equilibrate for 6 h with the same flow rate.Samples were collected hourly using a CMA 470 refrigerated fraction collector and stored at À80 C until analysed by ELISA.The average concentration of -synuclein in the first samples (at least three samples) was considered as the baseline concentration.Compounds were infused through the dialysis membrane by changing the perfusion buffer to artificial CSF containing the compound.In the case of elevated K + perfusion fluid, KCl was increased to 60 mM and NaCl was decreased to 88 mM to maintain osmolarity.At the end of the experiment, the brain was excised, fixed in 4% paraformaldehyde and analysed for probe placement by 2% Evans blue staining.
Throughout the study, the basal concentration of interstitial fluid -synuclein was found to be 0.61 AE 0.03 ng/ml (mean AE SEM, n = 54 mice).This variation in the baseline concentration values, although not significant, was probably due to inconsistent experimental parameters such as differences in membrane permeability and protein recovery among the microdialysis probes used or to variability in striatalsynuclein levels between the transgenic animals (Supplementary Fig. 1A and B).
All efforts were made to minimize animal suffering and to reduce the number of the animals used, according to the European Communities Council Directive (86/609/EEC) guidelines for the care and use of laboratory animals.All animal experiments were approved by the Institutional Animal Care and Use Committee of BRFAA (permit number A.05.1/ 6/02-07).

Pharmacological compounds
A23187, diazoxide, glyburide and d-amphetamine were purchased from Sigma; cromakalim, baclofen and muscimol from Santa Cruz; CNQX, gabazine and CGP55845 from Tocris; thapsigargin was obtained from Calbiochem.All compounds used in reverse microdialysis experiments were administered locally into the striatum through the microdialysis probe after dissolving them in artificial CSF.The concentrations of the infused compounds are indicated in the 'Results' section.

Ultra-sensitive ELISA for a-synuclein
The development and established protocol of the ELISA assay for -synuclein measurement has been described in detail elsewhere (Emmanouilidou et al., 2011;Kapaki et al., 2013).

Immunohistochemistry
Mice were anaesthetized by isoflurane and transcardially perfused with PBS followed by ice-cold 4% paraformaldehyde in PBS.Brains were excised and fixed in 4% paraformaldehyde in PBS for 16 h.For tissue dehydration, brains were transferred to 15% sucrose in PBS for 24 h and then to 30% sucrose in PBS for 24 h.Finally, brains were frozen at À45 C and stored at À80 C until sectioning.Free-floating coronal sections (30 mm) were collected using a Bright cryostat at À25 C at the levels of striatum (AP, 0.2 mm from bregma).After washing and blocking with 2% normal goat serum, free-floating sections were incubated with primary antibodies for 48 h at 4 C (Supplementary Table 1).Sections were washed and incubated with Cy2-and Cy3-labelled secondary antibodies (Jackson Labs) for 1 h at room temperature.DAPI (1:2000, Molecular Probes) or TO-PRO Õ -3 iodide (1 mM, Molecular Probes) staining was used to detect cell nuclei.Sections were mounted on slides and analysed by confocal microscopy.

Confocal microscopy
Fluorescent images were obtained in a Leica SP5-II confocal microscope and processed by the LAS software (Leica).For colocalization analysis, confocal images were captured using a 100 Â oil immersion objective by sequentially scanning each channel with attention to eliminate the crosstalk of fluorophores and avoid signal saturation.Images were acquired with a screen resolution of 1024 Â 1024 pixels.Co-localization was quantified from the reconstructed 3D image using the Volocity software (Perkin Elmer) according to the guidelines of Costes et al. (2004).Pearson's correlation coefficients (r) were estimated by the same software as means of quantification of the overlapping pixels from each channel (Manders et al., 1993;Barlow et al., 2010) with values 40.5 indicating the strongest correlation.

Isolation of synaptic vesicles
Synaptic vesicles were isolated from freshly prepared whole brain of 2-month-old wild-type mice according to the protocol of Ahmed et al. (2013).The integrity of the isolated vesicles was verified by electron microscopy.To remove non-integral membrane proteins, synaptic vesicles (150 mg) were treated with 100 mM Na 2 CO 3 (pH 11) for 30 min.Vesicle membrane fragments were recovered by centrifugation at 100 000 g for 1 h.Supernatant and pellet fractions were subsequently analysed by western blotting.Synaptic vesicles (50 mg) were also treated with 0.2% Triton TM X-100 for 30 min.Triton TM X-100 soluble and insoluble fractions were separated by centrifugation at 50 000 g for 1 h and subsequently analysed by ELISA.All treatments were carried out at 4 C.

Isolation of synaptosomes
The isolation of the crude synaptosomal fraction was performed as described by Postupna et al. (2014) with slight modifications.Briefly, striatal tissue was dissected from three adult mice and immediately homogenized in 10 volumes of icecold homogenization buffer (0.32 M sucrose, 10 mM Tris, pH 7.4) supplemented with protease inhibitors (Roche) using a Teflon-glass homogenizer.The homogenate was centrifuged at 1000 g for 10 min at 4 C to remove cell debris and nuclei.The supernatant was collected and centrifuged again at 10 000 g for 20 min at 4 C to obtain the crude synaptosome pellet (P2).The P2 pellet, containing mainly synaptosomes, mitochondria, myelin and membranes, was washed once in 2 ml PBS.Finally, the P2 pellet was collected by centrifugation at 5000 g for 4 min at 4 C, resuspended in Krebs-Ringer buffer (118 mM NaCl, 5 mM KCl, 4 mM MgSO 4 , 1 mM CaCl 2 , 1 mM KH 2 PO 4 , 16mM Na 2 HPO 4 , pH 7.4) containing 10 mM glucose and stored at À80 C until further analysis.

Synaptic vesicles
Negative staining was performed as described by Jahn and Maycox (1988) with some modifications (Ohi et al., 2004).

Synaptosomes
The synaptosome pellet was fixed with 4% formaldehyde/1% glutaraldehyde in phosphate buffer for 1 h at 4 C.After washing with phosphate buffer the pellet was dissected into 1-mm cubes and embedded in epoxy resins.Ultrathin sections were cut with a Diatome diamond knife at a thickness of 65-70 nm on Leica EM UC7 ultramicrotome (Leica Microsystems) and mounted onto 200 mesh nickel grids for immunogold labelling.

Immunoblotting
Freshly excised brain tissue was homogenized with a teflonglass homogenizer in D-PBS [1.5 mM KH 2 PO 4 , 8mM Na 2 HPO 4 (pH 7.3), 3 mM KCl, 137 mM NaCl] containing 1% Triton TM X-100 and phosphatase inhibitors.To isolate mouse striatum, wild-type and transgenic mice were decapitated at 8 months of age; the brains were harvested and dissected on ice to obtain the regions encompassing the striatum, the midbrain and the cortex.The homogenate was transferred to Eppendorf tubes and incubated on ice for 20 min.Soluble proteins were recovered in the supernatant fraction following centrifugation at 14 000 g for 5 min at 4 C. Protein concentration was estimated by the Bradford assay.Denaturing gel electrophoresis was performed on 12% sodium dodecyl sulphatepolyacrylamide gel electrophoresis (SDS-PAGE) gels in Trisglycine buffer.Immunoblotting was performed using the following antibodies: anti--synuclein (C20, rabbit polyclonal from Santa Cruz; Syn1, mouse monoclonal from BD Transductions; 4B12, mouse monoclonal from GeneTex), anti-tyrosine hydroxylase (AB152 rabbit polyclonal and MAB318 mouse monoclonal, both from Millipore), antisynaptophysin, anti-SNAP25, anti-synaptobrevin-2 (all mouse monoclonal antibodies from Synaptic Systems), anti-flotillin 1 (mouse monoclonal from Santa Cruz), anti-g-tubulin (mouse monoclonal from Sigma), anti-actin (mouse monoclonal from Sigma), anti-SUR1 (rabbit polyclonal from Santa Cruz), anti-SUR1 (rabbit polyclonal, custom made, kind gift of Dr. Mark Simard).The intensity of the immunoreactive bands was estimated by densitometric quantification using the Gel Analyzer v1.0 software.

Toxicity assay
The toxicity was assessed by using the LIVE/DEAD Õ viability/ cytotoxicity kit (Invitrogen).Briefly, cells were stained by 0.2 mM Ethidium Homodimer (EthD-1), which labels dead cells, for 20 min.Hoechst 33258 dye was used at the same time for total nuclei staining.After staining, cells were washed once with Neurobasal Õ medium and immediately visualized on a Leica DMIRB inverted fluorescence microscope.Images were captured under 20 Â magnification with a Leica DFC-350FX digital cooled CCD camera using the LAS AF software.Toxicity was defined in each image as the percentage of dead cells versus the total number of cells, counting at least 150 cells per image.

Statistical analysis
Data analysis was carried out using the GraphPad Prism 4 software.As indicated, statistics were performed by the unpaired Student's t-test or the one-way ANOVA test followed by Tukey's Multiple Comparison test.P-values 5 0.05 were considered significant.

Results
The majority of a-synuclein is localized in glutamatergic synaptic endings in the adult mouse striatum Several neuronal types are present in adult rodent striatum; GABAergic neurons (medium spiny neurons and interneurons) account for more than the 97% of the cells in striatum with the rest being mostly large cholinergic interneurons.Projection input to the striatum includes dopaminergic (from substantia nigra) and glutamatergic (from cortex and thalamus) terminals (Tepper et al., 2004).To elucidate the neuronal network involved in the secretion of -synuclein we first investigated the localization of the protein in the adult mouse striatum of wild-type and A53T transgenic mice, a Parkinson's disease model of moderate -synuclein overexpression.Even though -synuclein is overexpressed to a similar degree in all brain regions (Supplementary Fig. 1A and Giasson et al., 2002), transgenic mice display an age-dependent progressive decrease in the TH-positive terminals in the spinal cord but show no pathology in the nigrostriatal system (Sotiriou et al., 2010).In our hands, these mice also showed no signs of terminal loss in the striatum and midbrain as monitored by the levels of TH in wild-type versus transgenic animals (Supplementary Fig. 1B and C).To assess -synuclein localization, coronal sections of wild-type and transgenic mouse brain were labelled with specific neuronal markers and visualized by confocal microscopy.Our analysis demonstrated that, in the striatum, -synuclein (either wild-type or overexpressed) exhibits exclusively a neuropil staining consistent with localization solely in neuronal terminals and not cell bodies (Fig. 1A and B).The observation that -synuclein does not display cytoplasmic distribution in the striatum was verified by co-staining with the neuronal cytoskeletal markers, b3 tubulin (TUJ1) and microtubule-associated protein 2 (MAP2), which localize in both cytoplasm and terminals (Fig. 1A and Supplementary Video 1).Previous data in mice have also shown striatal neuropil staining of -synuclein during adulthood whereas cell body staining was detected only during the first postnatal week (Jakowec et al., 2001).Such neuropil staining was also evident, even though not exclusive, in other areas of the brain such as the cortex, the midbrain and the hippocampus and was demonstrated using three different -synuclein antibodies; Syn1 and C20, which recognize both mouse and human -synuclein, and Syn211, which binds only human -synuclein (Fig. 1 and Supplementary Fig. 2A and B).Diffuse cytosolic localization of -synuclein was also detected in dopaminergic neurons in the SNpc as shown by labelling with the dopaminergic markers, TH and dopamine transporter (DAT), and consistent with the literature (Galvin et al., 2001;Totterdell et al., 2004;Chu and Kordower, 2007;Boassa et al., 2013;Garcia-Reitboeck et al., 2013;Zharikov et al., 2015) (Supplementary Fig. 2A  and B).Further analysis using our experimental setup demonstrated that, except for expression at the terminals, $50% of dopaminergic neurons in SNpc (as identified by TH or DAT labelling) also showed cytoplasmic expression of -synuclein (Supplementary Fig. 3).As expected,synuclein was co-localized with the synaptic markers, synaptophysin, synaptobrevin-2 (Syb2) and SNAP25, confirming the synaptic topology of the protein (Fig. 1B and  E).This co-localization was further quantified using Volocity 3D Image Analysis software, which estimates a thresholded Pearson's correlation coefficient to describe the distance between green and red-coloured pixels based on the reconstructed two-coloured 3D image (Barlow et al., 2010).The Pearson's correlation coefficients obtained throughout this study are summarized in Supplementary Tables 2 and 3.Such unbiased quantitative analysis revealed highly efficient co-localization of -synuclein with all the synaptic markers in both wild-type and transgenic mice, and especially with Syb2 (Fig. 1B and 2E, Supplementary Fig. 5A and Supplementary Table 2).In addition, no significant differences were observed in the total levels of the synaptic markers in the striatum of wild-type and transgenic mice (Supplementary Fig. 4B).These initial experiments suggested that -synuclein in both wild-type and transgenic mice share the same spatial pattern and synaptic localization.
To gain further insight as from which neuronal terminals -synuclein is secreted, striatal sections of wild-type and transgenic mouse brain were co-stained with antibodies against human and mouse -synuclein and specific depicting the striatum of wild-type (Wt) and transgenic mouse following immunohistochemistry of coronal brain sections using three different antibodies against -synuclein.All three antibodies revealed a similar pattern of punctate neuropil expression.Images on the right represent magnification of the boxed area.Scale bar = 20 mm.(A) Representative images from sections of wild-type and transgenic mouse following costaining of -synuclein with the neuronal cytoskeletal marker, TUJ1, which is present in both cytoplasm and terminals.-Synuclein was detected using either the Syn1 antibody that recognizes both mouse and human -synuclein or, in the case of the transgenic mouse, using the Syn211 antibody that only binds to the human protein.(B) Representative sections of transgenic striatum showing the colocalization of -synuclein with the synaptic markers, synaptophysin, SNAP25 and synaptobrevin2 (Syb2).-Synuclein was detected using the C20 antibody that recognizes both the mouse and the human protein.
neuronal-type markers.Confocal images were obtained and co-localization -synuclein with each of the marker protein was assessed using the Volocity software.Such analysis demonstrated that the majority of -synuclein was localized in glutamatergic neuronal terminals, predominately in vGLUT1-positive corticostriatal terminals and to a lesser extent in vGLUT2-positive thalamostriatal terminals (Fig. 2A, Supplementary Table 2 and Supplementary  Video 2).On the contrary, -synuclein was not present in GABAergic or cholinergic neurons as demonstrated by the lack of co-localization with vesicular GABA transporter (vGAT) and choline acetyltransferase (CHAT), respectively (Fig. 2B and C).The absence of -synuclein from GABAergic neurons, which are highly abundant in the striatum, was further verified by co-staining with glutamic acid decarboxylase (GAD65), the rate-limiting enzyme in GABA synthesis (Fig. 2B).In agreement with previous studies (Kaneko and Fujiyama, 2002), immunoreactivity of   vGAT, GAD65, vGLUT1 and vGLUT2 in the striatum was punctate whereas CHAT labelled the large cholinergic cell bodies as well as their terminals (Perez et al., 2007).
Finally, we investigated the presence of -synuclein in the dopaminergic terminals in the striatum, which are the main projections of the dopaminergic neurons in SNpc.Initial immunofluorescence analysis revealed that TH does not exhibit a typical synaptic localization as shown by the lack of co-localization with synaptophysin.On the contrary, vGLUT1 and DAT showed synaptic topology as they were significantly co-localized with synaptophysin (Supplementary Fig. 5A and B and Supplementary Table 3).For this reason, the presence of -synuclein in dopaminergic neurons was assessed using DAT as a marker protein.Similar immunohistological analysis and quantification of co-localization using Volocity software revealed that a clearly detectable amount of -synuclein is indeed present in dopaminergic terminals in the striatum of wild-type and transgenic mice (Fig. 2D and E, Supplementary Table 2 and Supplementary Video 3).We hypothesize that this amount of -synuclein corresponds to the protein which originates from the SNpc dopaminergic neurons and localizes at the synaptic end of their afferents in the striatum.We cannot exclude the possibility that a different fraction of -synuclein (i.e.-synuclein modified by dopamine or multimerized conformations of -synuclein) is present in TH-positive terminals but cannot yet be picked up by our quantitative immunofluorescence approach due to interference with antibody binding.Such conformations have already been described to be present in the mouse brain (Conway et al., 2001;Martinez-Vicente et al., 2008;Dettmer et al., 2015).Comparison of the Pearson's coefficients obtained for all protein markers showed similar co-localization pattern for -synuclein in the striatum of wild-type and transgenic mice (Fig. 2e).
Taking this immunofluorescence approach, our data indicated that the majority of -synuclein expressed in the striatum is primarily localized in the corticostriatal glutamatergic afferents and to a lesser extent to the nigrostriatal dopaminergic and thalamostriatal glutamatergic terminals.Moreover, over-expression of human A53Tsynuclein as in our transgenic animal model did not alter the localization pattern observed suggesting that the expression pattern of human A53T -synuclein mimics the wildtype -synuclein.To further support these findings, we isolated crude synaptosomes from wild-type and transgenic mouse striatal tissue and analysed the presence ofsynuclein in these preparations using immuno-electron microscopy.A representative image showing the integrity of functional synaptic terminals present in such preparations is illustrated in Fig. 2F.To assess the localization of -synuclein, the synaptosomes were labelled with antibodies against -synuclein (C20), TH and vGLUT1.In both wild-type and transgenic animals, -synuclein was present primarily in the presynaptic terminals often colocalizing with synaptic vesicles.Thorough observation of the -synuclein-positive terminals suggested that $60% of these terminals were also vGLUT1-positive whereas $20% were TH-positive indicating that even though -synuclein was present in dopaminergic and glutamatergic terminals, the precedence in vGLUT1-positive endings was much more frequent (Fig. 2G).
To ensure the validity of the above observations, the following control experiments were performed.First, the specificity of the antibodies against -synuclein (Syn1 and C20), which were mostly used in the study, was tested by immunoblotting (Supplementary Fig. 1B and C) and immunohistochemistry (Supplementary Fig. 6A) using brain tissue from -synuclein knock-out mice.As expected, none of the -synuclein antibodies demonstrated a positive reaction with knock-out tissue.The specificity of these antibodies had already been assessed previously by ELISA measurements in brain tissue from knock-out, wild-type and transgenic mice (Kapaki et al., 2013).Second, as the wildtype littermates used had been generated in a C3H/C57BL6 mixed background, we verified our findings in striatal tissue from control animals with different genetic background.Costaining of striatal sections from C57BL6 and BALBc mice with antibodies against TH, vGlut1 and -synuclein (C20) further supported the observation that -synuclein is preferentially detected in glutamatergic terminals (Supplementary Fig. 6B).Finally, to exclude any regional localization variability, we assessed the staining pattern of -synuclein by immunohistochemistry including every fourth section throughout the whole striatum.Co-staining of serial striatal sections with TH and -synuclein exhibited similar distribution pattern (Supplementary Fig. 7).a-Synuclein is not secreted through synaptic vesicle fusion -Synuclein has been previously found to localize on the surface of synaptic vesicles where it interacts with the v-SNARE protein, synaptobrevin-2 (Burre et al., 2010).Our data imply that -synuclein export could be facilitated by the fusion of synaptic vesicles during neuronal firing for glutamate release.To investigate this, we isolated synaptic vesicles from wild-type mouse brain (Ahmed et al., 2013).The integrity of the isolated vesicles was confirmed by electron microscopy (Fig. 3A).The diameter of the synaptic vesicles in our preparation was estimated in the range of 27-45 nm.Western blotting analysis demonstrated the presence of -synuclein in these synaptic vesicles (Fig. 3B).However, analysis of the synaptic vesicle-free fractions showed that brain synaptic vesicles contain, but are not enriched in -synuclein (Fig. 3B and C).As expected, the integral synaptic vesicle membrane protein, synaptophysin, and the lipid raft protein, flotillin-1, were found to be enriched in the synaptic vesicle fraction (Fig. 3C).-Synuclein was also detected by our ELISA in intact brain synaptic vesicles which indicated that the protein is present on the surface of the vesicles with its C-terminus (at least amino acids 90-140) exposed to the extravesicular space.Estimation of -synuclein concentration by ELISA in brain homogenate and brain-derived synaptic vesicles revealed that $1ø of total brain -synuclein is associated with synaptic vesicles (-synuclein concentration: 29.34 AE 0.20 mg/ml and 22.85 AE 5.71 ng/ml, for brain homogenate and synaptic vesicles, respectively, n = 4) (Fig. 3D).Interestingly,synuclein was found to be embedded in the synaptic vesicle membrane since it was still recovered in the membrane fraction following treatment of the vesicles with Na 2 CO 3 but not in the lumen fraction (Fig. 3E).The association of -synuclein with the vesicle membrane was verified by solubilization of the synaptic vesicle membranes with 0.2% Triton TM X-100.Such treatment resulted in the recovery, almost exclusively, of -synuclein in the Triton-soluble fraction suggesting that it is an integral vesicle membrane protein like synaptophysin (Fig. 3F and G).These results suggested that -synuclein is an integral membrane protein of the synaptic vesicle membrane but is not present in the vesicle lumen and, therefore, it is unlikely to be secreted via synaptic vesicles.

a-Synuclein secretion is a calcium-regulated process
First, we aimed to identify whether the -synuclein secretory pathway in vivo involved a membrane fusion event or a direct translocation across the plasma membrane similar to that described for the unconventional secretion of FGF2 (Zehe et al., 2006).Because intracellular calcium (Ca 2 + ) is the most important trigger of regulated exocytosis, we have used reverse microdialysis to pharmacologically manipulate -synuclein release in the striatum of transgenic mice by locally applying reagents that elevate intracellular Ca 2 + concentration.
Although it was considered the best choice for our study, we could not use wild-type animals to perform the reverse microdialysis experiments due to technical reasons related to the low interstitial fluid -synuclein concentration.Therefore, we decided to perform all the reverse microdialysis experiments in transgenic animals since their increased interstitial fluid -synuclein levels would be sufficient to allow determination of interstitial fluid -synuclein concentration before and after drug administration.Previous work using in vivo microdialysis demonstrated that the excess of -synuclein in the transgenic mouse resulted in increased interstitial fluid -synuclein levels in a manner that reflects the expression levels of the protein (Emmanouilidou et al., 2011).Moreover, the expression data provided here show that the localization pattern of -synuclein in the transgenic striatum is similar with the endogenous wild-typesynuclein.
Intracellular Ca 2 + concentration was increased through K + -evoked depolarization.Infusion of KCl (60 mM) induced a sharp elevation in -synuclein levels in the interstitial fluid (interstitial fluid -synuclein concentration, 0.75 AE 0.03 and 0.93 AE 0.04 ng/ml for basal and KCl-post infusion levels, respectively, mean AE SEM, n = 6) (Fig. 4A).This result was confirmed by increasing intracellular Ca 2 + concentration through two independent pathways; calcium mobilization by the Ca 2 + -ionophore, A23187 (1 mM) or release of endoplasmic reticulum (ER)stored calcium by the SERCA pump inhibitor, thapsigargin (2 mM), which targets smooth ER in both cell bodies and terminals (McGraw et al., 1980).Infusion of both compounds increased interstitial fluid -synuclein levels to the same extent as KCl (Fig. 4B and C).
During reverse microdialysis, drugs are administered by diffusion through the microdialysis probe.Therefore, the drug concentration actually applied to the striatum cannot be precisely estimated but is surely much less than the initial concentration used.Using the initial concentrations administered in vivo, we applied the Ca 2 + modulators in primary cortical neurons to assess their impact on cell viability.None of the compounds had a toxic effect on the recipient neurons suggesting that the observed increase of interstitial fluid -synuclein concentration observed after local administration of the Ca 2 + modulators was not due to cell death (Supplementary Fig. 8).Collectively, these data indicated that -synuclein secretion in the mouse striatum is a Ca 2 + -regulated process.
a-Synuclein secretion is modulated by the SUR1-regulated K ATP channel Sulfonylurea-sensitive ATP-regulated K + (K ATP ) channels are widely expressed in the brain, especially in the SNpr and the striatum where they play a critical role in neurosecretion at the nerve terminals (Zini et al., 1993).Potent blockage of these channels by sulfonylureas leads to depolarization, activation of the L-type Ca 2 + -channels, Ca 2 + entry and finally neurotransmitter release (Avshalumov and Rice, 2003).In the striatum, GABA is the major inhibitory neurotransmitter which, via feedforward or feedback inhibition, regulates local and distal neuronal networks (Tepper et al., 2004).GABA release is thought to be regulated by the operation of the K ATP channel, which in turn is controlled by its regulatory subunits, SUR1 or SUR2 (Amoroso et al., 1989;Maneuf et al., 1996).Indeed, analysis of striatal sections from wild-type or transgenic mice, revealed that medium spiny neurons that express the dopamine D2 receptor (D2DR, encoded by DRD2) and the paralvumin (PRV)-positive interneurons, were found to contain SUR1 (Fig. 5A).To investigate whether GABA levels affect -synuclein secretion, we used reverse microdialysis to locally administer in the striatum of transgenic mice pharmacological agents that modulate the operation of SUR1, the most abundant receptor type in this brain region.Interestingly, we found that administration of the specific SUR1 blocker, glyburide (10 mM), resulted in a significant decrease of -synuclein concentration in the interstitial fluid (0.51 AE 0.03 and 0.33 AE 0.03 ng/ml for basal and glyburide-post infusion levels, respectively, mean AE SEM, n = 5) (Fig. 5B).Administration of the specific SUR1 opener, diazoxide (100 mM), resulted in a robust increase of interstitial fluid -synuclein levels (0.39 AE 0.02 and 0.50 AE 0.02 ng/ml for basal and diazoxide-post infusion levels, respectively, mean AE SEM, n = 5) (Fig. 5C).The observed effect was specific to SUR1 since administration of the specific SUR2 activator, cromakalim (100 mM), did not affect interstitial fluid -synuclein concentration (Fig. 5D).Toxicity of SUR1 modulators was assessed on primary cortical neurons as above.None of the reagents used reduced cell viability at the concentrations used (Supplementary Fig. 9A and B).Glyburide even had a protective effect reducing the basal cytotoxicity of these cultures (Supplementary Fig. 9B), a result consistent with previous findings (Simard et al., 2012).These data suggested that -synuclein secretion follows a SUR1-sensitive pathway in which the SUR1-K ATP possibly regulates secretion of the protein in an opposite manner to GABA release.
In the striatum of both wild-type and transgenic mice, SUR1 exhibited a punctate cytoplasmic distribution as verified by co-staining with TUJ1 (Fig. 5E).SNpr, another SUR1-rich brain region, clearly showed a similar pattern of SUR1 localization (Fig. 5F).Moreover, wild-type and transgenic mice possess similar levels of SUR1 as shown by immunoprecipitation and immunoblotting with the anti-SUR1 antibody (Supplementary Fig. 10A and B).Importantly, SUR1 was not found to co-localize withsynuclein in the mouse striatum (Fig. 5G, Supplementary Table 2 and Supplementary Video 4).Taking into account our previous findings where -synuclein was mostly found to be localized in glutamatergic terminals in the striatum, these data suggest that GABA release via the SUR1-K ATP channels could indirectly regulate -synuclein secretion through the crosstalk between the GABAergic neurons and the glutamatergic terminals.

GABA regulates a-synuclein secretion through GABA B receptors
The actions of GABA in the basal ganglia are mediated via the fast-acting ionotropic GABA A receptors and the slower-acting metabotropic GABA B receptors.Specifically in the striatum, glutamatergic synapses express predominantly functional presynaptic GABA B receptors (Lacey et al., 2005).Activation of the presynaptic GABA B receptors leads to G-protein-mediated inhibition of presynaptic Ca 2 + channels which decreases Ca 2 + influx and thereby inhibits transmitter release (Emson, 2007).As the majority ofsynuclein was found in the glutamatergic terminals, we wanted to test the hypothesis that GABA mobilization affects -synuclein secretion through activation of the GABA B receptors.
Using reverse microdialysis, we applied in the striatum of A53T transgenic mice specific agonists and antagonists known to modulate GABA levels.Direct administration of the specific GABA A agonist, muscimol (50 mM) and the specific GABA B agonist, baclofen (50 mM), did not affect -synuclein levels detected in the interstitial fluid (Fig. 6A  and B) (Tanaka et al., 2003;Darbin et al., 2008).However, infusion of the potent GABA A antagonist, gabazine (5 mM), resulted in a significant reduction in interstitial fluid -synuclein concentration (0.86 AE 0.02 and 0.73 AE 0.03 ng/ml for basal and gabazine-post infusion -synuclein levels, mean AE SEM, n = 4) (Fig. 6C).Similar results were obtained when 10 mM of gabazine was applied.Higher doses (25 mM and 50 mM) induced the compulsive ipsilateral turning behaviour that has been reported previously (Marks et al., 2008) and were thus not used further.Gabazine has been reported to induce intense firing of the GABAergic neurons that blocks striatal and cortical activity in monkeys (Darbin et al., 2008).The effect of gabazine on -synuclein interstitial fluid levels was abolished when the GABA B antagonist, CGP55845 (100 mM), was co-infused with gabazine, indicating that GABA B receptors mediate the modulation of -synuclein secretion (Fig. 6D).Infusion of CGP55845 alone reduced ($15%) the interstitial fluid -synuclein levels although this reduction did not reach statistical significance (0.71 AE 0.02 and 0.60 AE 0.05 ng/ml for basal and CGP55845-post infusion levels, mean AE SEM, n = 4) (Fig. 6E).In addition, infusion of the AMPA antagonist, CNQX (50 mM), caused a significant increase in interstitial fluid -synuclein (0.89 AE 0.03 and 1.07 AE 0.07 ng/ml for basal and CNQX-post infusion levels, mean AE SEM, n = 4), probably by reducing the excitatory input to the GABAergic neurons (Fig. 6F).Finally, mobilization of dopamine by infusion of d-amphetamine (10 mM) failed to alter interstitial fluid -synuclein levels (Fig. 6G) (Abercrombie and DeBoer, 1997).This result was probably due to the lower levels of -synuclein present in the dopaminergic terminals surrounding the microdialysis probe or to the presence of modified/oligomeric species that cannot enter the probe.
Collectively, these data suggest that GABA release can regulate -synuclein secretion possibly via activation of the presynaptic GABA B receptors.To obtain direct evidence for this hypothesis, we infused the SUR1 activator, diazoxide (100 mL), in the presence of the GABA B agonist, baclofen (50 mM).Consistent with our previous observations, diazoxide infusion produced a robust increase in interstitial fluid -synuclein levels.Perfusion of baclofen for 60 min before and during diazoxide stimulation completely suppressed the diazoxide-induced increase of interstitial fluid -synuclein concentration (Fig. 6H) indicating that GABA B receptor activation is required for -synuclein secretion.
a-Synuclein can be secreted from neuronal cell bodies and axons -Synuclein can be secreted from neurons in culture (Emmanouilidou et al., 2010).As our results point towards terminal localization of -synuclein in mouse brain, we wanted to investigate the neuronal sites of -synuclein secretion.To this end, we cultured cortical primary neurons from wild-type mice in microfluidic compartmentalization chambers that enable effective separation of neuronal axons from cell bodies (Supplementary Fig. 11A).Cortical neurons were cultured for $8-10 days during which a robust, viable neurite network was formed in the axonal compartment (Supplementary Fig. 11B).Microfluidic devices were examined for synuclein leakage from one compartment to the other by spiking recombinant -synuclein (1 mg/ml) in culture medium at 37 C for 7 days.Nosynuclein was detected by ELISA in the axonal compartment indicating that there is no protein transfer from the somal to the axonal compartment (Supplementary Fig. 11C).To assess -synuclein secretion from axons, neurons were cultured for 7 days and conditioned medium was collected from both the axonal and somal compartments ($100 ml).Analysis of the conditioned medium by our inhouse ELISA showed that -synuclein was present in the conditioned medium of both compartments (somal compartment: 0.22 AE 0.05 ng/ml, n = 7 chambers, axonal compartment: 0.11 AE 0.03 ng/ml, mean AE SD, n = 4 chambers) (Supplementary Fig. 11D).These results strongly indicate that axons (and somata) from primary mouse cortical neurons can readily release -synuclein.

SUR1-K ATP -mediated a-synuclein secretion is dependent on the integrity of the in vivo neuronal network
To determine whether the suggested GABA-mediated mechanism of -synuclein release depends on the local inhibitory networks of GABAergic neurons in the striatum, gabazine and the SUR1 modulators, glyburide and diazoxide, were directly applied in the growth medium of mouse striatal primary neurons in culture.This isolated neuronal cell system is enriched in a neuronal NeuN-positive population (83% of the total number of cells) and also contains a small percentage of astrocytic GFAP-positive cells but no Iba-1-positive microglial cells (Supplementary Fig. 12A).Cultured neurons express the neuronal markers TUJ1 and synaptophysin and are GABAergic in nature as they express GAD65 and vGAT (Fig. 7A).Importantly, these neurons also express SUR1 and -synuclein (Fig. 7B).In this embryonic stage, both proteins are distributed in the cytoplasm of neuronal cells.In agreement with our results from the adult mouse striatum, SUR1 showed similar punctate distribution in the cytoplasm of the striatal embryonic neurons whereas -synuclein showed a distinct pattern of diffuse distribution in both neuronal cell bodies and axons (Fig. 7B).Expression of -synuclein was restricted to neurons and it was completely absent from astrocytes (Supplementary Fig. 12B).Analysis of the conditioned medium of mouse striatal neurons revealed that these neurons readily secrete -synuclein (Fig. 7C).Application of KCl (50 mM) and thapsigargin (2 mM) indicated that -synuclein was secreted  in a Ca 2 + -dependent fashion in accordance with our in vivo data.However, neither diazoxide (100 mM), glyburide (10 mM) or gabazine (10 mM) treatment affected -synuclein secretion in this isolated neuronal cell-system (Fig. 7c) suggesting that the -synuclein secretory mechanism greatly depends on the integrity of the neuronal network in the striatum, which allows the continuous crosstalk between different types of neurons.Supplementation of GABA (50 mM) into the medium of striatal neurons also failed to mobilize -synuclein release (Fig. 7C) suggesting that GABA B receptor activation alone could not modulate -synuclein secretion from striatal neurons.
We also tried to challenge the proposed mechanism in primary cortical cultures which were also enriched in neurons (71% of the total number of cells) but contained a small fraction of astrocytic cells (11% of the total number of cells) and no microglial cells (Supplementary Fig. 12C).Cortical neurons were positive for TUJ1, vGLUT1 and vGLUT2 but negative for vGAT indicating that they are mostly glutamatergic neurons (Fig. 7A).KCl treatment stimulated -synuclein secretion consistent with a Ca 2 +dependent pathway of release (Fig. 7D).These neurons, which show a similar expression pattern of SUR1 andsynuclein as striatal neurons (Fig. 7B), displayed no change in -synuclein release following application of diazoxide or glyburide.Primary mouse cortical neurons readily express GABA B receptors (Kurokawa et al., 2012).To challenge the suggested mechanism for -synuclein secretion, we exogenously applied GABA in the conditioned medium of these neurons.GABA addition only moderately increased the secreted levels of -synuclein, possibly via in vitro stimulation of GABA B receptors in these neurons (Fig. 7D).These results indicate that GABA B receptor activation can, at least in part, induce -synuclein release from cortical, but not striatal, neurons in vitro.However, these data strongly suggest that the proposed GABAmediated mechanism for -synuclein secretion in vivo that involves SUR1-K ATP activation is synergistic between neurons and cannot be recapitulated in isolated neuronal cultures.

Discussion
Under pathological conditions, such as in Parkinson's disease and other synucleinopathies, extracellular -synuclein is thought to propagate from a diseased to a healthy neuron through a prion-like mechanism (Desplats et al., 2009;Hansen et al., 2011).To this end, identification of the molecular pathway of -synuclein secretion will provide valuable information about its physiological role and its contribution to the initiation and progression of Parkinson's disease pathology.
We demonstrate here that in mouse striatum -synuclein is secreted by a calcium-dependent mechanism that is tightly regulated by GABA transmission.In this area of the brain, the majority of -synuclein is found to be localized in synaptic endings of corticostriatal glutamatergic terminals.Using reverse microdialysis to locally administer several compounds as modulators of specific functions (Fig. 8B), we have shown that modulation of GABA release through the SUR1-regulated K ATP channels located on GABAergic neurons controls -synuclein release from glutamatergic terminals.Importantly, this secretion is mediated by activation of the presynaptic GABA B receptors present on these terminals.The cartoon in Fig. 8A schematically illustrates this proposed secretory pathway for -synuclein in mouse striatum.
Our study is based on data obtained from both wild-type and transgenic mouse models.Our results indicate that in terms of protein localization the human A53T -synuclein, which is moderately overexpressed in this transgenic model, exhibits similar characteristics with the wild-type mouse protein.However, the reverse microdialysis study was carried out only in transgenic animals.It should be noted that, although it was considered the best choice for our study, we could not use wild-type animals to perform the reverse microdialysis experiments due to technical reasons related to the low interstitial fluid -synuclein concentration in these mice.
Our results clearly suggest a functional role of SUR1dependent K ATP channels in the regulation of -synuclein secretion in the striatum.K ATP channels primarily regulate potassium flux as a response to alterations in the ATP/ADP ratio, thus coupling metabolic state to membrane excitability.Evidence from Parkinson's disease mouse models suggested that selective activation of the SUR1-K ATP channels in dopaminergic substantia nigra neurons leads to loss of electrophysiological activity and may thus contribute to the differential vulnerability and degeneration of these neurons in Parkinson's disease (Liss et al., 2005).Importantly, K ATP channels are also involved in the regulation of secretory events.Recent in vivo data indicate that in medial substantia nigra dopaminergic neurons SUR1-K ATP channels can act as cell-type specific gates to promote burst firing as in the case of pancreatic b-cells (Schiemann et al., 2012).In addition, SUR1-K ATP channels have been shown to modulate neurotransmitter release such as glutamate (Tanaka et al., 2003), GABA (Maneuf et al., 1996) or dopamine (Avshalumov and Rice, 2003).The evidence presented here supports the idea that, at least in mouse striatum, SUR1-K ATP channels operate to modulate GABA release, which in turn activates GABA B receptors resulting in -synuclein secretion.In this sense, -synuclein in the extracellular space is not secreted through, but is tightly regulated by, synaptic transmission.
The activity of metabotropic GABA B receptors seems to be a prerequisite for normal physiological processes (Kornau, 2006).Activation of postsynaptic GABA B receptors results in hyperpolarization due to activation of K + channels whereas activation of presynaptic GABA B receptors suppresses neurotransmitter release by inhibiting Ca 2 + channels.GABA B receptors are widely expressed at presynaptic and postsynaptic sites in rat striatum.GABAergic modulation of GABA B receptors has previously been reported to regulate glutamate release in rat hippocampus in vivo (Tanaka et al., 2003).In this study, the effects of GABA B receptor activation were due to the inhibition of presynaptic Ca 2 + channels regulated by the receptor (Darbin et al., 2008).It is therefore plausible that a similar mechanism applies for -synuclein release.
Indeed, the data presented here support the idea thatsynuclein secretion is a Ca 2 + -regulated process.In the adult mouse brain, we found -synuclein to be located specifically on the nerve terminals close to the synaptic zone where it is partially associated with synaptic vesicles.However, synaptic vesicles do not contain -synuclein in their lumen and therefore it is unlikely that they mediate -synuclein secretion.The synaptic vesicle-bound -synuclein is suggested to have a distinct function as a chaperone to promote SNARE-complex assembly (Burre et al., 2010(Burre et al., , 2014) ) or as a regulator of synaptic vesicle endocytosis (Vargas et al., 2014).In agreement with this, we found that -synuclein is actually embedded in the synaptic vesicle membrane having a large part of its C terminus sequence (i.e. at least amino acids 90-140) exposed to the extravesicular space.The finding that -synuclein is not secreted via synaptic vesicles was not unexpected.Due to their size, synaptic vesicles are restricted to carry small molecules and not proteins or other macromolecules (Sudhof, 2004).Based on our findings, we cannot speculate on how -synuclein is actually transported out of the nerve terminal.Release via secretory granules still remains an attractive possibility, even though -synuclein lacks a signal peptide that would direct the protein to the regulated secretory pathway.Such a mechanism of export has previously been described in the adult rodent striatum for the release of BDNF from extrinsic afferents (Altar, 1997).An alternative strong candidate mechanism for -synuclein export, at least in part, still remains release via exosomes (Emmanouilidou et al., 2010;Alvarez-Erviti et al., 2011).
Our study focused in the area of striatum where the majority of -synuclein was found to be localized in presynaptic corticostriatal glutamatergic nerve endings and to a lesser extent in thalamostriatal and nigrostriatal afferents.In the SNpc, -synuclein is distributed in the cytoplasm and terminals of dopaminergic neurons.As expected, a fraction of SNpc-related -synuclein can be detected in DAT-positive terminals in the striatum.Consistent with the synaptic topology of the protein, -synuclein was not co-localized with TH, a protein not found close to the area of the synapse.This finding is in agreement with a previous study which showed that only a small fraction of SNpc neurons contains -synuclein with the majority of the protein been present in the TH-negative terminals in the SNpr (Jakowec et al., 2001).The terminal localization of -synuclein is of particular importance since in conditions where the integrity of the dopaminergic system becomes compromised, such as in pharmacologically-induced Parkinson's disease or during ageing, the localization of the protein is reported to shift to a more cytoplasmic pattern (Vila et al., 2000;Chu and Kordower, 2007).
In the context of Parkinson's disease pathology, dopamine depletion due to the loss of dopaminergic neurons in SNpc leads to dramatic alterations in the firing pattern of the striatal medium projection neurons.GABA levels in the striatum are disturbed resulting in a deregulation ofsynuclein secretion possibly via over-activation of GABA B receptors.The changing levels of extracellular -synuclein could be another parameter that disrupts the fragile neuronal communication under these conditions.
To this end, this work highlights the importance of understanding the secretory pathway involved in the regulation of extracellular -synuclein levels in vivo.Our data are consistent with the idea that not only the release but also the maintenance of extracellular -synuclein levels in the brain parenchyma are critical for neuronal homeostasis.This is supported by the reverse microdialysis experiments where a number of reagents were locally administered in vivo in mouse striatum and their effects in -synuclein levels were monitored in real time.Importantly, on removal of the agents that either stimulate or reduce secretion, compensatory mechanisms seemed to be activated to bringsynuclein to basal levels.
There is now substantial evidence emphasizing the critical role of extracellular -synuclein on the initiation and progression of Parkinson's disease pathology.In this context, targeting the molecules that regulate -synuclein secretion could be a promising therapeutic approach for Parkinson's disease.Pharmacological modulation of GABA B receptors has already been proposed as an anti-parkinsonian tool (Galvan et al., 2011).However, efforts using GABA B agonists and antagonists have been hindered by the widespread distribution of GABA B receptors in the CNS and by the fact that these reagents cannot distinguish between pre-and postsynaptic receptors.The finding that the secretion of -synuclein is regulated by SUR1 receptors, which can be specifically targeted, highlights the potential use of this molecule as a novel therapeutic intervention for Parkinson's disease.Further supporting a role of SUR1 in Parkinson's disease, post-mortem analysis of brain tissue from patients with Parkinson's disease and control subjects revealed significant differences in SUR1 expression levels in substantia nigra neurons, whereas ABBC9 (SUR2) and KCNJ11 (Kir6.2) mRNA levels remained unchanged (Schiemann et al., 2012).In conclusion, our study suggests that the secretion of -synuclein in vivo is tightly regulated by GABA transmission via activation of the SUR1-K ATP channels and reinforces the idea that the regulatory elements of this secretory pathway could aid the development of future neuroprotective strategies.
Secretariat of Research and Technology of Greece (GSRT) to KV and an ARISTEIA II GSRT grant to KV.

Figure 1
Figure 1 a-Synuclein is localized in the synaptic nerve endings in the adult mouse striatum.Confocal images (200Â magnification)

Figure 2
Figure 2 The majority of striatal a-synuclein is localized corticostriatal glutamatergic neuronal terminals.(A-D) Confocal images (200Â magnification) depicting the striatum of wild-type (Wt) and transgenic mice following immunohistochemistry of coronal brain sections.Images show co-staining of -synuclein with (A) vGLUT1 and vGLUT2 as markers of glutamatergic corticostriatal and thalamostriatal terminals, respectively; (B) vGAT and GAD65 as markers for GABArgic terminals; (C) choline acetyltransferase (ChAT) as a marker for cholinergic neurons; (D) TH and DAT as markers for dopaminergic terminals.For the detection of -synuclein the monoclonal Syn1 antibody and the polyclonal C20 antibody were used.Lack of -synuclein from TH-positive terminals was verified using two different -synuclein antibodies, D37A6 that detects only mouse -synuclein, and Syn211 that detects only human -synuclein.In the case of the transgenic mouse, TH was also detected using a different TH antibody, AB152.Right images represent magnification of the boxed area.White arrows indicate yellow puncta indicative of co-localization.Scale bar = 20 mm.(E) Graphical representation showing the Pearson's coefficients obtained from the co-localization analysis of -synuclein with synaptic and glutamatergic neuronal markers in wild-type and transgenic mice (see also Supplementary Table2).Statistics by Student's t-test comparing wild-type versus transgenic for each neuronal marker (data presented as mean AE SEM, ns = not significant).(F) Representative electron micrograph of synaptosomes isolated from striatal tissue, post-fixed with osmium tetroxide, showing preserved synapses (asterisks).Synaptic vesicle load (arrow) is visible in the presynaptic site.Scale bar = 0.2 mm.(G) Representative micrographs of striatal synaptosomes from wild-type and transgenic animals following double immunolabelling.-Synuclein is depicted with the small gold particles (10 nm, asterisks) and vGLUT1 and TH are represented by large gold particles (15 nm, arrowheads).-Synuclein was detected in both vGLUT1and TH-positive terminals.Scale bar = 0.2 mm. 2 ). Statistics by Student's t-test comparing wild-type versus transgenic for each neuronal marker (data presented as mean AE SEM, ns = not significant).(F) Representative electron micrograph of synaptosomes isolated from striatal tissue, post-fixed with osmium tetroxide, showing preserved synapses (asterisks).Synaptic vesicle load (arrow) is visible in the presynaptic site.Scale bar = 0.2 mm.(G) Representative micrographs of striatal synaptosomes from wild-type and transgenic animals following double immunolabelling.-Synuclein is depicted with the small gold particles (10 nm, asterisks) and vGLUT1 and TH are represented by large gold particles (15 nm, arrowheads).-Synuclein was detected in both vGLUT1and TH-positive terminals.Scale bar = 0.2 mm.

Figure 3
Figure 3 a-Synuclein is found on the membrane but not in the lumen of synaptic vesicles.(A) Representative negative-stain electron micrographs showing synaptic vesicles isolated from whole wild-type mouse brain.Left image obtained with Â 9100 original magnification (Scale bar = 150 nm).Right image is magnified from a micrograph obtained with Â 13 900 original magnification (Scale bar = 50 nm).Arrowheads indicate individual synaptic vesicles.(B) Synaptic vesicles were analysed by immunoblotting for the presence of -synuclein using the C20 polyclonal antibody.The integral membrane protein synaptophysin was used as a marker for the synaptic vesicle fraction (H = whole brain homogenate; P1 = first pellet of the isolation procedure containing large cell fragments and nuclei; SV = synaptic vesicles; F1 and F5 = synaptic vesicle-negative fractions 1 and 5 obtained after the purification of synaptic vesicles on a sucrose cushion).(C) Western blotting analysis of all fractions obtained following purification through the sucrose cushion using antibodies against -synuclein (C20), synaptophysin and flotillin-1.Equal volume of each fraction was loaded on the gel.In contrast to synaptophysin and flotillin-1, -synuclein is not enriched in synaptic vesicle fractions (H = whole brain homogenate; SV = pooled synaptic vesicle-positive fractions; 1-10 = synaptic vesicle-negative fractions).(D) Estimation of -synuclein concentration by ELISA in whole homogenate and synaptic vesicles (SVs) from wild-type mouse brains (n = 4 mice).(E) Immunoblotting analysis of synaptic vesicles following treatment with Na 2 CO 3 using -synuclein (C20) and synaptophysin antibodies (H = brain homogenate; SV = synaptic vesicles; P = Na 2 CO 3 -insoluble fraction; S = Na 2 CO 3 -soluble fraction).(F) Western blotting analysis of synaptic vesicles following treatment with Triton TM X-100 (H = brain homogenate; SV = synaptic vesicles; P = Triton TM X-100-insoluble fraction; S = Triton TM X-100-soluble fraction).-Synuclein in these fractions was also quantified using ELISA (G).

Figure 5
Figure 5 a-Synuclein secretion is modulated by the SUR1-regulated K ATP channel.(A) SUR1 is present in GABAergic neurons in mouse striatum.Confocal images from wild-type (Wt) and transgenic (Tg) mouse striatum showing co-staining of SUR1 and the GABAergic markers, D2DR, which labels a subset of spiny projection neurons, and paralvumin (PRV), which stains a subcategory of interneurons.For each marker, a magnification of the boxed area from the merged image is shown on the right.Scale bar = 10 mM.(B) Intrastriatal infusion of glyburide (GLY, 10 mM) by reverse microdialysis caused a significant decrease in interstitial fluid -synuclein concentration.Top: One representative experiment is shown.Arrow indicates the time of glyburide application.Bottom: Comparison of the basal versus the post-infusion -synuclein levels in the interstitial fluid.Data are presented as mean AE SEM, n = 5 mice and statistics were performed by Student's t-test, *P 5 0.05.(C) Striatal administration of diazoxide (DZ, 100 mM) through the microdialysis probe induces a significant increase in interstitial fluid -synuclein levels.Top: One representative experiment is shown.Time of infusion is indicated by the arrow.Bottom: Comparison of the basal versus the postinfusion interstitial fluid -synuclein levels (mean AE SEM, n = 5 mice, Student's t-test, **P 5 0.01).(D) Infusion of cromakalim (CRO, 100 mM) did not affect interstitial fluid -synuclein concentration.Top: One representative experiment is shown.Time of infusion is indicated by the arrow.Bottom: Comparison of the basal versus the post-infusion interstitial fluid -synuclein levels (mean AE SEM, n = 4 mice, Student's t-test, P = 0.5393).(E and F) Representative confocal images of striatum (E, Â 200 magnification; Scale bar = 10 mm) and SNpr (F, Â 189 magnification; Scale bar = 15 mm) showing the cytoplasmic distribution of SUR1 in neuronal cells as indicated by co-staining of SUR1 (red) with the cytoskeletal marker, TUJ1 (green).(G) SUR1 is not co-localized with -synuclein.Confocal images of wild-type and transgenic striatum co-stained with antibodies against SUR1 (red) and -synuclein (Syn1, green).Images were captured under Â 200 magnification.Scale bar = 10 mm.

Figure 8
Figure8GABA transmission via SUR1-K ATP channels regulates a-synuclein secretion in the mouse striatum.(A) The in vivo secretory pathway of -synuclein proposed by this study is based on the crosstalk of GABAergic neurons with glutamatergic terminals.Left: Activation of the SUR1-K ATP channels located on the membrane of GABAergic neurons causes membrane hyperpolarization and a decrease in GABA release.Locally lower GABA levels reduce the activation of GABA B receptors on neighbouring glutamatergic nerve endings.As a result, the inhibition of surrounding Ca 2 + channels is compromised and the raise in intracellular Ca 2 + triggers -synuclein release.Right: Closing of the SUR1-K ATP channels leads to membrane depolarization and increase in GABA release.Increased GABA levels cause higher activation of GABA B receptors in glutamatergic terminals, inhibiting surrounding Ca 2 + channels and thus reducing -synuclein release.(B) Schematic representation of the action and primary target(s) of the compounds used in vivo in this study.Amph = amphetamine, acts on dopaminergic endings to mobilize dopamine; Mus = muscimol, agonist of GABA A receptors; GZ = gabazine, antagonist of GABA A receptors; Bac = baclofen, agonist of GABA B receptors; CGP = CGP55845, antagonist of GABA B receptors; CNQX = antagonist of AMPA receptors; DZ = diazoxide, opener of SUR1-K ATP channels; GLY = glyburide, blocker of SUR1-K ATP channels.