α-Synuclein binds to the ER–mitochondria tethering protein VAPB to disrupt Ca²⁺ homeostasis and mitochondrial ATP production

Mammalian expression vectors for human myc-tagged VAPB and haemagglutinin (HA)-tagged PTPIP51 in pCIneo, pCIneo control vector expressing Escherichia coli chloramphenicol acetyltransferase (CAT), wild-type α-synuclein, α-synucleinA53T and α-synucleinA30P in pcDNA3.1(−) and enhanced green fluorescent protein (EGFP) tagged versions in pEGFPC1, and AT1.03 cytosolic ATeam FRET based ATP reporter was all as described and [20, 34, 42, 71]. For the production of stable cell lines, wild-type and mutant untagged α-synuclein cDNA were cloned as NheI–HindIII fragments in pcDNA6.V5-His with a stop codon prior to the His tag. GST-VAPB and pRK172 α-synuclein E. coli expression vectors were as described [43, 78]. Control and human α-synuclein siRNAs were from Santa Cruz Biotechnology (sc-37007 and sc-29619, respectively).

Rabbit and rat antibodies to VAPB and PTPIP51 were as described [20]. Mouse anti-glyceraldehyde 3-phosphate dehydrogenase (GAPDH), mouse anti-hemagglutinin (HA), mouse anti-myc 9B11 and rabbit anti-glycogen synthase kinase-3β (GSK3β) phosphorylated on serine-9 (inactive GSK3β) were from Cell Signalling. Rabbit anti-mitofusin-2, rabbit anti-HA, mouse anti-heat shock protein-60 (HSP60) and mouse anti-neurofilament heavy-chain (NFH) (N52) were from Sigma. Rabbit (sc-7011) and mouse (211) anti-α-synuclein, rabbit anti-translocase of the outer mitochondrial membrane protein-20 (TOM20), rabbit anti-fatty acid coenzyme A ligase long-chain 4 (FACL4) and mouse anti-Sigma-1 receptor were from Santa Cruz Biotechnology. Mouse anti-α-synuclein and mouse anti-calnexin were from BD Biosciences. Rabbit anti-EGFP and mouse anti-β-actin were from Abcam. Rabbit anti-PTPIP51 and chicken anti-MAP2 were from Genetex. Mouse anti-protein disulphide isomerase (PDI) was from Affinity Bioreagents, rabbit anti-myc was from Upstate and rabbit anti-GSK3β was from StressGen.

SH-SY5Y and HEK293 cells were purchased from the European Collection of Cell Cultures. Cells and were maintained in Dulbecco’s modified Eagle’s medium (DMEM) containing 4.5 g/l glucose (HEK293 cells) or DMEM/F-12 (1:1) containing 3.15 g/glucose (SH-SY5Y cells) supplemented with 10% fetal bovine serum (Sera Laboratories), 2 mM l-glutamine, 1 mM sodium pyruvate, 100 IU/ml penicillin and 100 μg/ml streptomycin (Invitrogen). Cells were transfected with plasmids and siRNAs using Lipofectamine 2000 according to the manufacturer’s instructions (Invitrogen). For production of stable cell lines, cells were selected with media containing either 15 μg/ml blasticidin (for vector pcDNA6.V5-His) or 500 μg/ml geneticin sulphate (G418) (for vector pEGFPC1) for 4 weeks (Santa Cruz Biotechnology). Transiently transfected cells were analysed 16–24 h post-transfection and siRNA treated cells 72 h post-transfection. Rat cortical neurons were prepared and transfected with Lipofectamine 2000 as previously described [81].

Induced pluripotent stem (iPS) cells from a familial Parkinson’s disease patient carrying gene triplication of SNCA encoding α-synuclein (α-synuclein triplication; AST cells) and an unaffected first-degree relative control (normal α-synuclein; NAS cells) were maintained and differentiated into dopaminergic cells as described [21, 89]. 54–60% of the cells were neuronal based upon immunostaining for markers. Two different disease AST clones and two different control NAS clones were used in the studies and pooled data shown. For analyses, iPS cell-derived neurons were grown on 35 mm IBIDI dishes (BD Biosciences) as described [21].

Cells were harvested for SDS-PAGE and immunoblotting by scraping into SDS-PAGE sample buffer containing 2% SDS, 100 mM dithiothreitol, 10% glycerol, 0.1% bromophenol blue and protease inhibitors (Complete Roche) in 50 mM Tris–HCl pH 6.8 and heating to 100 °C for 5 min. Other samples were prepared by addition of SDS-PAGE sample buffer and heating to 100 °C for 5 min. Samples were separated on 8–15% (w/v) acrylamide gels and transferred to Protran nitrocellulose membranes (Schleicher and Schuell) using a Mini-PROTEAN 3 gel electrophoresis system and Transblot system (BioRad). The immunoblots were then blocked by incubation in 5% (w/v) dried milk/0.1% (w/v) Tween-20 in Tris buffered saline (TBS) pH 7.5 for 1 h and then probed with primary antibodies diluted in blocking solution for 16 h at 4 °C. Following washing in blocking solution, they were then incubated with horseradish peroxidase-conjugated goat anti-mouse, anti-rabbit or anti-rat Igs (GE Healthcare). Immunoblots were developed using an enhanced chemiluminescence Luminata Forte Western HRP substrate system according to the manufacturer’s instructions (Millipore). Signals on immunoblots were quantified using ImageJ after scanning with an Epson Precision V700 Photo scanner essentially as described by us in previous studies [78].

Recombinant GST, GST-VAPB fragments and α-synuclein were produced in E. coli BL21 (DE3). GST tagged proteins were purified on glutathione-Sepharose beads according to the manufacturer’s instructions (GE Healthcare) and as described by us previously [78]. α-Synuclein was purified as described [73]. For pull-down assays, GST or GST-VAPB proteins were incubated overnight with 50 μg purified recombinant α-synuclein in ice-cold lysis buffer containing PBS with 1% Triton X-100 and Complete protease and PhosStop inhibitor cocktail tablets (Roche). GST-fusion protein complexes were pulled down via the glutathione-Sepharose beads and washed three times with ice-cold lysis buffer prior to analyses by SDS-PAGE and immunoblotting. For cellular GST pull-down assays, baits were incubated with cell lysates and captured essentially as described above. HEK293 cells were used for immunoprecipitation and GST pull down assays since they transfect with high efficiency and so produce robust readouts that are particularly suitable for such biochemical studies.

ER, MAM and mitochondria were prepared from adult Sprague–Dawley rat brains essentially as described [87]. Briefly, fresh rat brains were washed in ice-cold isolation buffer containing 225 mM mannitol, 75 mM sucrose, 0.5% BSA and 0.5 mM EGTA in 30 mM Tris–HCl buffer pH 7.4. All procedures were then performed on ice. Three brains were chopped into fine pieces and washed three times in isolation buffer and then prepared as a 10% w/v mixture in isolation buffer. Tissues were homogenized using a Dounce homogenizer (20 strokes by hand using the tightest pestle) and the homogenate centrifuged twice at 740×g for 5 min to remove nuclei and unbroken cells. The crude mitochondrial fraction was then pelleted by centrifugation at 9000×g for 10 min and the remaining supernatants were centrifuged at 100,000×g for 30 min to pellet ER/microsomes. To isolate MAM and pure mitochondria, the crude mitochondrial pellet was resuspended in isolation buffer and layered on top of a self-forming 30% Percoll gradient (225 mM mannitol, 1 mM EGTA, 0.05% BSA, 30% Percoll, 25 mM Na-HEPES pH 7.4). After centrifugation at 95,000×g for 30 min the mitochondria were recovered at the bottom of the gradient and MAM retrieved above. To remove residual Percoll, the mitochondrial fraction was diluted in mitochondria resuspending buffer (5 mM HEPES pH 7.4 containing 250 mM mannitol and 0.5 mM EGTA) and the mitochondria washed twice by centrifugation at 6300×g for 10 min. The MAM band was diluted with isolation buffer and centrifuged once at 6300×g for 10 min to remove any residual contaminating mitochondria. MAM was then pelleted from the resulting supernatant by centrifugation at 100,000×g for 1 h. All final organelle pellets were resuspended and lysed in in RIPA buffer containing 50 mM Tris–HCl pH 6.8 containing 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS with Complete protease inhibitor (Roche). Protein concentrations were determined using a Bradford assay (Bio-Rad Laboratories).

Cells were fixed with 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer pH 7.2 for 3 h at 20 °C and then harvested by a gentle scraping with a plastic scaper. The cells were pelleted by centrifugation at 800 g (av) for 10 min, washed in 0.1 M sodium cacodylate buffer and post-fixed for 1 h in 2% osmium tetroxide and 1.5% ferricyanide in 0.1 M sodium cacodylate buffer. The cells were then stained for 1 h with 1% uranyl acetate in water before dehydration and embedding in epoxy resin (TAAB). 150 nm semi-fine sections were cut on a Reichert Ultra cut E ultramicrotome and stained for 6 min in 0.16% lead citrate in 0.1 M NaOH followed by 3 washes in distilled water. Samples were viewed on a Tecnai 12 electron microscope at 4800× magnification. Digital images were acquired and the circumference of each mitochondria and the proportions of the mitochondrial surface closely associated (<30 nm) with ER were calculated. Cells were randomly selected for analyses without prior knowledge of transfected plasmid or siRNA. All clearly identified mitochondria in the samples were scored and the numbers of mitochondria and ER profiles were quantified from lower magnification images. Image analyses were performed using ImageJ.

For super resolution structured illumination microscopy, cells plated on glass coverslips were fixed with 3% paraformaldehyde and 0.1% glutaraldehyde in PBS for 10 min at 20 °C. After washing with PBS, samples were quenched by incubation with 50 mM NaBH₄ in PBS for 7 min, washed in PBS and then permeabilized and blocked for 30 min in PBS containing 0.2% Triton X-100 and 3% bovine serum albumin (BSA). Samples were then incubated with primary antibodies diluted in blocking solution, washed with PBS and incubated with goat anti-rabbit or anti-mouse secondary Igs conjugated to AlexaFluor 546 or AlexaFluor 633 (Invitrogen). Following final washings in PBS, the samples were mounted in Mowiol-DABCO mounting medium containing 10% (w/v) Mowiol 4-88 (Calbiochem), 25% (w/v) glycerol and 2.5% (w/v) DABCO (1,4-diazobicyclo[2.2.2]octane) in 100 mM Tris–HCl pH 8.5. Samples were analysed on a Nikon Eclipse Ti-E inverted microscope equipped with a Nikon N-SIM Super Resolution System. Images were captured using a CFI Plan Apo IR SR 60× water immersion objective and then reconstructed using Nikon Imaging Software Elements AR with N-SIM module. ER–mitochondria interactions were quantified by Mander’s coefficient using ImageJ software. Co-localized signals were displayed using the ImageJ 1.44p RG2B co-localization plugin to determine co-localized pixels.

For confocal and wide-field microscopy, cells were fixed in 4% paraformaldehyde in PBS, immunostained essentially as described [28] and analysed using a Leica TCS-SP5 confocal with x63HCX PL APO lambda blue CS 1.4 objective or a Leica DM5000 fluorescence microscope equipped with a 40×/0.75NA HCX-PL-FLUOTAR lens. Mitochondria and ER morphologies and cellular distributions were determined using ImageJ with the Analyse Particle function and Mitochondrial Morphology [17] and AnalysedSkeleton [3] Plugins.

Proximity ligation assays to quantify VAPB-PTPIP51 and α-synuclein-VAPB interactions were performed essentially as described previously using Duolink reagents (Sigma-Aldrich) [20, 28, 78]. Cells were fixed in 4% paraformaldehyde in PBS and probed with primary antibodies. Signals were developed using the Duolink In Situ Orange kit. Cells were counterstained with 4′,6-diamidino-2-phenylindole (DAPI) to show nuclei and/or further immunostained for NFH or MAP2. Proximity ligation assay signals were quantified using the Particle Analysis function of ImageJ.

Ca²⁺ measurements were performed essentially as described previously [20, 78, 79]. SH-SY5Y cells were loaded with 2 μM Fluo4-AM and/or Rhod2-AM dye (Invitrogen) in external solution (145 mM NaCl, 2 mM KCl, 5 mM NaHCO₃, 1 mM MgCl₂, 2.5 mM CaCl₂, 10 mM glucose, 10 mM Na-HEPES, pH 7.25) containing 0.02% Pluronic-F27 (Invitrogen) for 15 min at 37 °C, followed by washing in external solution for 15 min at 37 °C. Fluo4 and Rhod2 fluorescence were timelapse recorded (1 s intervals) at 37 °C using either an Axiovert S100 microscope (Zeiss) driven by MetaMorph (Molecular Dynamics) and equipped with GFP (Fluo4) and DsRed (Rhod2) filtersets (Chroma Technology), a 40× Plan-Neofluar 1.3NA objective (Zeiss), and a Photometrics Cascade-II 512B36 EMCCD camera or a Nikon Ti-E microscope using a CFI Plan Apo VC 20× objective and Nikon Andor Neo sCMOD high-resolution camera and appropriate filter sets. The cells were kept under constant perfusion with external solution (0.5 ml/min). Inositol 1,4,5-trisphosphate (IP3) receptor-mediated Ca²⁺ release from ER stores was triggered by application of 100 μM oxotremorine-M for 2 min. Ca²⁺ levels were calculated as relative Rhod2 or Fluo4 fluorescence compared to baseline fluorescence (F/F ₀) at the start of the measurement. Oxotremorine-M was from Santa Cruz Biotechnology and was dissolved in water.

ATP levels in cultured cells were measured using a ViaLight ATP kit (Lonza) according to the manufacturer’s instructions; luminescence signals were obtained with a FluoSTAR luminometer (BMG Labtech). To determine ATP levels generated by oxidative phosphorylation in mitochondria, cells were first treated with 100 μM iodoacetate (Sigma) for 2 h to inhibit ATP produced by glycolysis. ATP levels were also determined using a FRET based plasmid reporter (Adenosine 5′-Triphosphate indicator based on Epsilon subunit for Analytical Measurements; ATeam reporter) [42]. To do so, cells were transfected with AT1.03 cytosolic ATeam reporter and then imaged in Hanks Balanced Salt Solution (HBSS) without phenol red at 37 °C by timelapse microscopy (12 s intervals) on a Zeiss Axiovert S100 microscope equipped with a 40×/1.3NA Plan-Neofluar objective and a Photometrics Cascade-II 512B EMCCD driven by MetaMorph (Molecular Dynamics). FRET filtersets (ECFP excitation filter ET 430/24×; ECFP emission filter 470/24 m; EYFP emission filter ET545/40m) were from Chroma Technology. 1 mM KCN in HBSS was applied using a peristaltic pump (0.5 ml/min). YFP/CFP ratios prior to and after KCN treatment were measured as described and used to calculate relative ATP levels in the different samples which were displayed as bar charts [42].

All experiments were repeated at least three times. Statistical analyses were performed with Prism 5.0 (GraphPad Software).

To determine the effects of α-synuclein on ER–mitochondria associations, we quantified ER–mitochondria contacts in polyclonal populations of SH-SY5Y cells stably expressing either enhanced green fluorescent protein (EGFP) control vector, EGFP-α-synuclein or familial Parkinson’s disease mutant EGFP-α-synucleinA53T or EGFP-α-synucleinA30P. Numerous studies have utilized α-synuclein tagged with EGFP in this way (e.g. [35, 61, 64, 72, 77]). Probing of immunoblots with an EGFP antibody revealed that they expressed similar levels of exogenous protein and that expression of α-synuclein did not affect expression of the ER–mitochondria tethering proteins VAPB and PTPIP51, of mitofusin-2 which has been proposed as a further ER–mitochondria tether, or of the Sigma-1 receptor which has also been linked to ER–mitochondria tethering [5, 19, 29, 86] (Fig. 1a). Likewise, no changes in the levels of TOM20 or PDI were detected (Supplemental Fig. 1a). The absence of changes in the levels of PTPIP51, TOM20 and mitofusin-2 (mitochondrial proteins) or of VAPB and PDI (ER proteins) suggest that there are no overall changes in mitochondrial or ER masses in the different transfected cells. We also quantified the numbers of mitochondria and ER profiles, and mitochondrial circumferences in the EM and detected no differences between the different transfected cells (data obtained from 30 to 35 cells; analysed by one-way ANOVA). However, confocal and structured illumination microscopy (SIM) revealed that mitochondria in EGFP-α-synuclein and EGFP-α-synucleinA53T cells had increased circularity (rounding up) (EGFP-α-synucleinA30P displayed a trend for increased circularity) and that mitochondria in wild-type and mutant EGFP-α-synuclein cells had reduced cytosolic distributions (increased clustering) (Supplemental Fig. 1b, c). We detected no changes in ER cytosolic distributions, the numbers of ER branch points or branch length (i.e. overall ER morphology) in wild-type or mutant EGFP-α-synuclein cells compared to control (Supplemental Fig. 1d, e, f).

ER–mitochondria associations were, therefore, quantified by determining the proportion of the mitochondrial surface that was closely apposed (less than 30 nm) to ER following analyses by EM. This and similar EM approaches have been used previously [15, 16, 23, 24, 26, 52, 78, 85]. Compared to control cells, expression of wild-type and mutant α-synuclein all led to significant reductions in ER–mitochondria associations (Fig. 1b).

We also enquired whether loss of α-synuclein influenced ER–mitochondria associations. To do so, we quantified ER–mitochondria associations by EM in untreated SH-SY5Y cells and in cells treated with control or α-synuclein siRNAs. Probing of samples on immunoblots revealed that the α-synuclein siRNAs produced an approximate 90% knockdown of α-synuclein but this did not affect expression of VAPB, PTPIP51, mitofusin-2 or the Sigma-1 receptor (Fig. 1c). However, compared to untreated or control siRNA treated cells, the EM studies revealed no change in ER–mitochondria associations in the α-synuclein knockdown cells (Fig. 1d).

To complement the EM studies, we monitored the effects of α-synuclein expression on ER–mitochondria associations using super resolution structured illumination microscopy (SIM) [39]. For these studies, SH-SY5Y cells were transfected with EGFP control vector, EGFP-α-synuclein, EGFP-α-synucleinA53T or EGFP-α-synucleinA30P and then immunostained for PDI and TOM20 to label ER and mitochondria, respectively. Co-localisation analyses of PDI and TOM20 revealed that wild-type and mutant α-synuclein again disrupted ER–mitochondria associations (Fig. 2). Further analyses of these images revealed that a proportion of α-synuclein localized with the co-localized PDI/TOM20 signals (i.e. some α-synuclein is present at ER–mitochondria contact sites) (Supplemental Fig. 2).

We also used in situ proximity ligation assays [75] to monitor the effects of α-synuclein on ER–mitochondria associations. Proximity ligation assays have a resolution similar to that detected by resonance energy transfer between fluorophores (i.e. approximately 10 nm) [75] and so are suitable for quantifying ER–mitochondria associations. For these assays, we used VAPB and PTPIP51 primary antibodies as markers for ER and mitochondria, respectively, since these proteins interact directly to tether ER with mitochondria [20, 78, 79]. Indeed, proximity ligation assays have already been used to quantify ER–mitochondria associations and the VAPB-PTPIP51 interaction [5, 20, 36]. We have previously demonstrated the specificity of these VAPB-PTPIP51 proximity ligation assays via control experiments involving the absence of each or both primary antibodies [20, 79]. Transfection of SH-SY5Y cells with EGFP-α-synuclein, EGFP-α-synucleinA53T or EGFP-α-synucleinA30P all produced significant reductions in ER mitochondria associations in these assays (Fig. 3a).

Since we detected no α-synuclein-induced changes in expression of VAPB or PTPIP51 (Fig. 1a, c), these proximity ligation assays not only confirm that α-synuclein reduces ER–mitochondria associations, but also show that this reduction involves breaking of the VAPB-PTPIP51 tethers. To test this further, we performed immunoprecipitation assays to monitor the effect of wild-type and mutant α-synuclein on binding of VAPB to PTPIP51. To do so, we co-transfected cells with hemagglutinin (HA)-tagged PTPIP51 and either control vector, α-synuclein, α-synucleinA53T or α-synucleinA30P and monitored the amounts of VAPB bound to immunoprecipitated PTPIP51-HA by immunoblotting of the samples. Consistent with the proximity ligation assays, both wild-type and mutant α-synuclein decreased the amounts of endogenous VAPB bound to immunoprecipitated PTPIP51-HA in these assays (Fig. 3b).

To determine whether α-synuclein also disrupts the VAPB–PTPIP51 interaction in Parkinson’s disease patient material, we utilized proximity ligation assays to probe ER–mitochondria associations and the VAPB-PTPIP51 interaction in iPS cell-derived dopaminergic neurons from a SNCA triplication patient (α-synuclein triplication; AST) and a non-disease first degree relative control (normal α-synuclein; NAS) [21]. SNCA triplication causes autosomal dominant familial Parkinson's disease and neurons carrying this mutation have higher levels of α-synuclein compared to controls [21, 41, 74]. Compared to controls, the VAPB-PTPIP51 proximity ligation assay signals were significantly reduced in the SNCA triplication neurons (Fig. 4). Thus, expression of both wild-type and mutant α-synuclein reduces ER–mitochondria associations and the VAPB–PTPIP51 interaction in a number of different assays and this includes in dopaminergic neurons derived from familial Parkinson’s disease patients carrying SNCA gene triplication.

A primary function of ER–mitochondria associations is to facilitate delivery of Ca²⁺ to mitochondria from ER stores. This delivery is mediated via IP3 receptors located in MAM [16, 19, 46, 48, 63, 70, 82]. Thus, disruption to the VAPB–PTPIP51 interaction via siRNA loss of VAPB or PTPIP51 to loosen ER–mitochondria associations perturbs this delivery [20, 78, 79]. Since expression of wild-type and mutant α-synuclein reduces both ER–mitochondria associations and the VAPB–PTPIP51 interaction, we, therefore, monitored the effect of α-synuclein on mitochondrial Ca²⁺ uptake following its release from ER stores. For these experiments, we used SH-SY5Y cells stably expressing control empty vector, α-synuclein, α-synucleinA53T or α-synucleinA30P and triggered physiological IP3 receptor-mediated Ca²⁺ release from ER by application of the M3 muscarinic acetylcholine receptor agonist oxotremorine-M. This approach has been utilized previously [20, 78, 79]. In agreement with previous studies [20, 78, 79], oxotremorine-M induced a time-dependent increase in mitochondrial Ca²⁺ levels (Fig. 5a). However, compared to controls, the peak values were lower in cells expressing wild-type or mutant α-synuclein (Fig. 5a). This finding suggests that α-synuclein perturbs Ca²⁺ delivery to mitochondria. To test this further, we monitored cytosolic and mitochondrial Ca²⁺ levels following oxotremorine-M mediated Ca²⁺ release from ER. Compared to controls, the time-lags between cytosolic and mitochondrial peak values were longer in cells expressing wild-type and mutant α-synuclein (Fig. 5b). Finally, we enquired whether the reduced Ca²⁺ uptake by mitochondria in α-synuclein expressing cells following IP3 receptor-mediated release could be rescued by transfection of VAPB to restore ER–mitochondria associations; VAPB overexpression increases ER–mitochondria contacts [78]. In agreement with previous studies [28], transfection of VAPB increased mitochondrial Ca²⁺ uptake (Fig. 5c). Moreover, VAPB overexpression completely abrogated the effect of α-synuclein on mitochondrial Ca²⁺ uptake (Fig. 5c). Thus, α-synuclein induced breaking of the VAPB-PTPIP51 tethers and loosening of ER–mitochondria associations are accompanied by disruption to delivery of Ca²⁺ to mitochondria from ER stores.

Ca²⁺ is required by mitochondria for generating ATP via the tricarboxylic acid cycle [30] and so the reduced mitochondrial Ca²⁺ levels seen in both wild-type and mutant α-synuclein expressing cells predict that α-synuclein impairs mitochondrial ATP production. We, therefore, monitored mitochondrial ATP production in the stably expressing α-synuclein SH-SY5Y cells. To do so we used a bioluminescent assay that measures total cellular ATP levels (generated by both glycolysis and oxidative phosphorylation) and inhibited glycolytic ATP production with iodoacetic acid. This method has been used by others to measure ATP generated by oxidative phosphorylation in mitochondria [4]. To first confirm the effect of iodoacetic acid, we measured ATP production in control SH-SY5Y cells treated with either vehicle or iodoacetic acid; as predicted, iodoacetic acid markedly reduced ATP levels (Fig. 6a). We then measured ATP levels in the different transfected cells following treatment with iodoacetic acid. Compared to control cells, the levels of ATP generated by oxidative phosphorylation were significantly reduced in cells expressing α-synuclein, α-synucleinA53T and α-synucleinA30P (Fig. 6b).

To complement these studies, we also used a FRET reporter system that permits ATP quantification in single living transfected cells [42]. Cellular ATP is generated by a combination of oxidative phosphorylation and glycolysis and so we assayed ATP production in cells treated with potassium cyanide (KCN) which inhibits cytochrome C oxidase to block oxidative phosphorylation. Monitoring ATP levels in the individually transfected cells prior to and after KCN treatment thus permits calculation of the levels of mitochondrial ATP production [42, 79]. SH-SY5Y cells were, therefore, co-transfected with the ATP FRET reporter and either control vector, wild-type or mutant α-synuclein and the relative ATP levels generated by oxidative phosphorylation determined. These studies revealed that compared to control transfected cells, both wild-type and mutant α-synuclein reduced mitochondrial ATP production (Fig. 6c). We also used the ATP FRET reporter to quantify mitochondrial ATP production in rat cortical neurons co-transfected with control vector, wild-type or mutant α-synuclein. These experiments revealed that α-synuclein likewise reduced mitochondrial ATP production in the neurons (Fig. 7). Thus, the α-synuclein-induced disruptions to IP3 receptor-mediated Ca²⁺ delivery to mitochondria are accompanied by reductions in mitochondrial ATP production.

Recently, GSK-3β has been shown to regulate ER–mitochondria associations and binding of VAPB to PTPIP51 [78, 79]. Inhibition of GSK-3β promotes whereas activation inhibits the VAPB-PTPIP51 interaction and this leads to complementary changes in ER–mitochondria associations [78, 79]. The α-synuclein-induced loosening of ER–mitochondria associations and the VAPB–PTPIP51 interaction may, therefore, involve activation of GSK-3β. A major route for regulating GSK-3β activity involves inhibitory phosphorylation of serine-9 [44] and so we monitored GSK-3β serine-9 phosphorylation in control, α-synuclein, α-synucleinA53T and α-synucleinA30P expressing SH-SY5Y cells by immunoblotting. However, we detected no differences in GSK-3β serine-9 phosphorylation between these different cells (Supplemental Fig. 3).

An alternative mechanism whereby α-synuclein could disrupt the VAPB-PTPIP51 tethers might involve its binding to either VAPB or PTPIP51 so as to sterically interfere with their interaction. In support of this notion, a proportion of α-synuclein localizes to MAM [31, 67]. To test this possibility further, we first sought to confirm the biochemical localization of α-synuclein to MAM and so prepared MAM, mitochondria and ER fractions from rat brain, and probed these for the presence of α-synuclein on immunoblots. In agreement with earlier studies [31], we detected a significant proportion of α-synuclein in MAM (Fig. 8a). We, therefore, monitored binding of wild-type and mutant α-synuclein to PTPIP51 and VAPB using immunoprecipitation assays from PTPIP51 + α-synuclein and VAPB + α-synuclein co-transfected cells. Although we detected no interaction between α-synuclein and PTPIP51 in these assays, we obtained robust signals for binding of both wild-type and mutant α-synuclein to VAPB (Fig. 8b, c). Moreover, we obtained stronger signals for binding of VAPB to α-synucleinA53T and α-synucleinA30P than to wild-type α-synuclein (Fig. 8c). We also used immunoprecipitation assays from rat brain to determine whether endogenous α-synuclein and VAPB interact. Again, α-synuclein bound to VAPB in these assays (Fig. 8d, e).

To confirm the binding of VAPB to α-synuclein using other methods, we first performed proximity ligation assays in non-transfected SH-SY5Y cells using VAPB and α-synuclein antibodies. Omission of primary VAPB and/or α-synuclein antibodies produced very few signals but inclusion of both antibodies generated robust signals consistent with a direct interaction of the two endogenous proteins in situ (Fig. 9a). We also performed these assays in the control (NAS) and α-synuclein triplication (AST) iPS cell neurons and these likewise produced signals consistent with a direct interaction (Fig. 9b). Moreover, we obtained greater numbers of signals in the AST neurons that express higher levels of α-synuclein [21] (Fig. 9b).

To test this direct binding further, we prepared recombinant VAPB and α-synuclein in E. coli and monitored their binding in vitro. For these experiments, we generated different domains of VAPB as GST fusion proteins and used the GST moiety to isolate the VAPB “baits” along with any bound α-synuclein. VAPB is anchored in the ER via a C-terminal membrane-spanning domain with its N-terminus projecting into the cytoplasm through which it interacts with PTPIP51 [20, 78]. The VAPB cytoplasmic region contains an N-terminal major sperm protein (MSP) domain and a centrally located coiled-coil domain (Fig. 10a). We, therefore, prepared GST-VAPB “baits” comprising the entire cytoplasmic domain, the MSP domain, the coiled-coil domain and sequences encompassing the C-terminal cytosolic region of VAPB, and tested their abilities to bind to α-synuclein in pull-down assays. Both the entire VAPB cytoplasmic domain and the MSP domain, but not the coiled-coil or C-terminal domains bound to α-synuclein in these assays (Fig. 10a). Finally, we used the GST-VAPB “baits” to pull down α-synuclein from transfected cell lysates. Again, only the entire VAPB cytoplasmic domain and the MSP domain bound to VAPB in these cellular assays. Thus, VAPB and α-synuclein interact in a variety of assays including in vitro assays in the absence of other proteins and this interaction involves the VAPB MSP domain.