Abstract
Macroautophagy is a catabolic process that coordinates with lysosomes to degrade aggregation-prone proteins and damaged organelles. Loss of macroautophagy preferentially affects neuron viability and is associated with age-related neurodegeneration. We previously found that α-synuclein (α-syn) inhibits lysosomal function by blocking ykt6, a farnesyl-regulated soluble NSF attachment protein receptor (SNARE) protein that is essential for hydrolase trafficking in midbrain neurons. Using Parkinson's disease (PD) patient iPSC-derived midbrain cultures, we find that chronic, endogenous accumulation of α-syn directly inhibits autophagosome-lysosome fusion by impairing ykt6-SNAP-29 complexes. In wild-type (WT) cultures, ykt6 depletion caused a near-complete block of autophagic flux, highlighting its critical role for autophagy in human iPSC-derived neurons. In PD, macroautophagy impairment was associated with increased farnesyltransferase (FTase) activity, and FTase inhibitors restored macroautophagic flux through promoting active forms of ykt6 in human cultures, and male and female mice. Our findings indicate that ykt6 mediates cellular clearance by coordinating autophagic-lysosomal fusion and hydrolase trafficking, and that macroautophagy impairment in PD can be rescued by FTase inhibitors.
SIGNIFICANCE STATEMENT The pathogenic mechanisms that lead to the death of neurons in Parkinson's disease (PD) and Dementia with Lewy bodies (LBD) are currently unknown. Furthermore, disease modifying treatments for these diseases do not exist. Our study indicates that a cellular clearance pathway termed autophagy is impaired in patient-derived culture models of PD and in vivo. We identified a novel druggable target, a soluble NSF attachment protein receptor (SNARE) protein called ykt6, that rescues autophagy in vitro and in vivo upon blocking its farnesylation. Our work suggests that farnesyltransferase (FTase) inhibitors may be useful therapies for PD and DLB through enhancing autophagic-lysosomal clearance of aggregated proteins.
Introduction
Parkinson's disease (PD) is the second most common age-related neurodegenerative disorder and is characterized clinically by symptoms of tremor, rigidity, bradykinesia, and postural instability (Gelb et al., 1999). In PD, Lewy bodies (LB) accumulate in the substantia nigra pars compacta (SNc), and α-synuclein (α-syn) makes up a major component of this pathology (Spillantini et al., 1997). A driving role for α-syn in disease has been reinforced by the description of a kindred experiencing autosomal dominant Parkinsonism and dementia because of multiplications in the α-syn gene, SNCA (Muenter et al., 1998; Singleton et al., 2003). In these patients, age of symptom onset and disease severity correlate with α-syn dosage, with SNCA triplication carriers demonstrating earlier onset and more severe symptoms than duplication carriers (Singleton et al., 2003, 2013). Increased abundance of wild-type (WT) α-syn is thus sufficient to cause disease, and it remains an open question how accumulation occurs in much more prevalent cases of idiopathic PD (iPD). GWAS studies of idiopathic cases have demonstrated that the strongest risk factors involve protein clearance machinery mediating the autophagy-lysosomal pathway (ALP), including genes such as GBA1, CTSB, and TMEM175 (Nalls et al., 2014; Chang et al., 2017).
Macroautophagy (or autophagy) is the branch of the ALP responsible for delivering bulk cytoplasmic aggregates and damaged organelles to lysosomes for degradation (Larsen and Sulzer, 2002). Although clearance of physiological α-syn can occur via the proteasome or chaperone-mediated autophagy (CMA), α-syn aggregates developed through transgenic overexpression have been shown to undergo degradation through autophagy (Webb et al., 2003; Cuervo et al., 2004; Vogiatzi et al., 2008; Ebrahimi-Fakhari et al., 2011; Tanik et al., 2013). Autophagy is essential for the maintenance of neuronal health, since inhibiting autophagy in mice leads to the accumulation of ubiquitinated protein aggregates in neurons and motor behavioral deficits reminiscent of PD (Hara et al., 2006; Friedman et al., 2012; Savitt et al., 2012; Sato et al., 2018). Similarly, human mutations in essential autophagy genes such as Atg5 and Atg7 result in developmental delay and severe neurologic dysfunction (Kim et al., 2016; Collier et al., 2021). Autophagy markers are increased in postmortem SNc of familial and iPD patients along with accumulation of incompletely degraded autophagosomes and phospho-ubiquitin that correlate with age and Braak stage (Dehay et al., 2010; Hou et al., 2018; Mamais et al., 2018). Autophagy is perturbed by α-syn overexpression in cell lines and animal models, but the mechanism underlying this effect is not completely understood (Winslow et al., 2010; Decressac et al., 2013; Tanik et al., 2013; Tang et al., 2021). Transgenic mice expressing human A53T mutant α-syn in dopaminergic neurons display autophagic defects preceding neurodegeneration (Chen et al., 2015), suggesting a causative role in toxicity. Collectively, such data indicate that autophagy is indispensable for neurons and may play a role in PD pathogenesis, but its relationship to Lewy body formation and neurodegeneration remains unknown. Furthermore, autophagic function in PD patient midbrain models that accumulate endogenously expressed α-syn in a chronic, age-dependent manner has not been investigated thoroughly.
Successful breakdown of autophagic substrates requires the formation of an initial membrane sac (phagophore), closure and substrate engulfment by a two-layered autophagosome membrane, trafficking of the autophagosome vesicle to a lysosome, and fusion of autophagosome and lysosome membranes to intermix their contents (autolysosome fusion; Abounit et al., 2012). In the key reaction for membrane formation, MAP1LC3B is conjugated with phosphatidylethanolamine to form LC3-II from the pool of its unconjugated form LC3-I (Tanida et al., 2008). LC3-II provides a specific and distinguishable marker for autophagosomes, and autophagic flux can be evaluated by comparing the accumulation of LC3-II following lysosomal inhibitor treatment (Klionsky et al., 2021). The discovery of LC3 binding proteins including p62 has revealed mechanisms responsible for cargo delivery of substrates into autophagic vesicles (Komatsu et al., 2007). P62 plays a critical role in the recognition and delivery of ubiquitinated protein aggregates from the cytosol into autophagosomes, and it accumulates within ubiquitinated inclusions of autophagy-deficient mice (Komatsu et al., 2007). Therefore, the accumulation of p62 is indicative of autophagic impairment. P62 inclusions have been detected in several protein aggregation diseases including PD (Zatloukal et al., 2002), and depleting p62 in a synucleinopathy mouse model enhances pathology (Tanji et al., 2015), suggesting its role in clearing aggregated α-syn. Membrane fusion events throughout autophagy require soluble NSF attachment protein receptor (SNARE) proteins (Wang et al., 2016), and recent studies have indicated an important role for synaptobrevin-homolog ykt6. Ykt6 is critical for endoplasmic reticulum (ER)-Golgi trafficking (McNew et al., 1997; Liu and Barlowe, 2002; Fukasawa et al., 2004), but also interacts with essential autophagy SNAREs to promote autophagosome assembly (Mari et al., 2010; Nair et al., 2011). Late in the autophagic process, ykt6 can mediate the final fusion step with lysosomes by interaction with SNAP-29 and syntaxin 7 (Matsui et al., 2018; Takáts et al., 2018). Studies in cell lines have shown that syntaxin 17 can independently mediate autophagosome-lysosome fusion in the absence of ykt6, suggesting functional redundancy (Matsui et al., 2018). However, it is unknown whether such redundant pathways exist in human neurons and whether ykt6 is required for autophagy.
We previously showed that neuronal ykt6 is essential for lysosomal function and is activated upon lysosomal stress to increase trafficking of hydrolases between the ER and Golgi (Cuddy et al., 2019). α-Syn accumulation in PD patient iPSC-derived midbrain neurons (iPSn) disrupted trafficking by promoting an auto-inhibited, closed conformation of ykt6 in the cytosol (Cuddy et al., 2019). The auto-inhibited conformation requires farnesyl modification (Wen et al., 2010), and we found that inhibiting farnesylation of ykt6 promoted an open, active conformation that rescued lysosomal hydrolase activity (Cuddy et al., 2019). Here, we find that ykt6 is critical for autophagosome-lysosome fusion and is disrupted in PD iPSn. Taken together with our previous data (Cuddy et al., 2019), this indicates that ykt6 controls autophagy by simultaneous coordination of hydrolase trafficking and autophagic-lysosomal fusion.
Materials and Methods
Experimental design and statistical analysis
Each assay was performed with at least 2–3 technical replicates and confirmed with 2–4 individual passages or differentiation sets. The study design, including the number of replicates and passage numbers, was based on preliminary data and our previous studies on the lysosomal system (Cuddy et al., 2019; Stojkovska et al., 2022). These studies provided estimations of effect size and sample variability for each assay and model system. Statistical analyses were performed using Prism software. Statistical outliers were confirmed and removed using ROUT test (Q = 5%). A Student's t test was used when comparing two groups, and one-way ANOVA with Dunnett's multiple comparisons or Tukey's post hoc test was used when comparing more than two, as indicated in the figure legends. For ease of representation and comparison across modalities, graphs most often show fold change in the relevant value compared with each experiment's control. Throughout, *p < 0.05, **p < 0.01, ***p < 0.001. Extended Data Table 1-1 lists the sources of cell lines and catalog numbers of reagents.
Raw, unprocessed full length western blot data is available at https://data.mendeley.com/datasets/374r67jc2h/draft?a=69ffcc35-f399-42d1-af5b-8092ebd90358.
Extended Data Table 1-1
Resources and reagents. Download Table 1-1, DOCX file.
Culture models
iPS cells and neuronal differentiation
iPSC maintenance and dopaminergic differentiation has been described previously (Mazzulli et al., 2011, 2016). iPSCs were maintained on Cultrex coated plates in mTeSR+ media (Stemcell Technologies). Dopaminergic neuron differentiation followed a well-established dual SMAD inhibition protocol (Kriks et al., 2011). Cells were treated with growth factor cocktail for 40–50 d followed by maintenance in Neurobasal media supplemented with NeuroCult SM1 (Stemcell Technologies), 1% pen/strep, and 1% L-glutamine (Thermo Fisher Scientific). Neurons were cultured for 60–120 d before use. Except as stated in results and figure legends, experiments were performed between days 88 and 95. SNCA triplication iPSCs, healthy control line (2131), and their differentiated neurons have been extensively characterized in two of our previous papers (Mazzulli et al., 2016; Stojkovska et al., 2022). Standard authentication as described (Mazzulli et al., 2011) includes genotyping for PD-associated mutations, karyotyping, pluripotency analysis, ensuring efficient expression of neuronal dopaminergic markers following maturation, and mycoplasma testing. All lines were previously found to differentiate with similar efficiencies (Mazzulli et al., 2016; Stojkovska et al., 2022). Efficient differentiation of each set was ensured by monitoring cell body and neurite morphology throughout maturation and measuring βIII-tubulin/GAPDH ratio by Western blotting and α-syn in cell lysates after harvesting. From day 40 through day 180, between 80% and 90% of surviving cells are typically positive for FOXA2, TH, and/or βIII-tubulin (Mazzulli et al., 2016; Stojkovska et al., 2022). Neuron survival measured by neurofilament staining was previously found to be unchanged in SNCA triplication lines at day 90 but decreases to ∼40% by day 330 (Mazzulli et al., 2016; Stojkovska et al., 2022).
H4 cells
Inducible H4 neuroglioma cells that overexpress WT α-syn under a tetracycline-responsive promoter (tet-off) have been described and authenticated previously (Mazzulli et al., 2011, 2016). Culture media is OptiMem (Invitrogen) supplemented with 5% heat-inactivated fetal bovine serum (HI-FBS), 1% penicillin/streptomycin (pen/strep), and 200 mg/ml Geneticin (G418) and Hygromycin (both Thermo Fisher Scientific). For controls, α-syn expression was suppressed by 72-h 1 µg/ml doxycycline (+dox) treatment (Sigma).
SH-SY5Y
Generation of pCDNA3.1 (vector), pCDNA3.1-wt-α-syn, pEGFP, and pEGFP-ykt6 stably expressing lines was described previously (Cuddy et al., 2019; from naive ATCC #CRL-2266; female). The same approach was taken to generate lines expressing pCDNA3.1-wt-α-syn with pLKO1 or pLKO1-ykt6-shRNA. Briefly, Lipofectamine 2000 transfected cells were exposed to incrementally higher concentrations of G418 over two weeks, then remaining colonies were expanded and screened for α-syn and ykt6 expression. Stable lines were maintained in DMEM (Invitrogen) with 10% HI-FBS, 1% pen/strep, and 200 mg/ml G418. Cells were differentiated for experimental use by addition of 10 μm all trans retinoic acid for 5 d (Sigma).
FTI treatment in culture
SH-SY5Y and iPSn were treated for 7 d with 10 nm LNK-754 (Link Medicine; Cuddy et al., 2019) in DMSO or DMSO alone. H4 were treated for 3 d. Cell lines had media with compound refreshed daily while iPSn were refreshed every other day. For ykt6 knock-down studies, treatment began 3 d postinfection in iPSn and 1 d post-transfection in H4.
Mouse model
Transgenic α-syn mice
DASYN53 mice expressing human α-syn under the dopamine transporter promoter have been described (Chen et al., 2015) and were graciously provided by the producing lab. Care and genotyping was identical to our previous paper (Cuddy et al., 2019). Housing, breeding, care, and use followed guidelines of Northwestern University's Institutional Animal Care and Use Committee and the US National Institutes of Health Guide for the Care and Use of Laboratory Animals. Standard rodent chow and water were provided ad libitum. Mice were numbered by ear tags and tail samples were collected for genotyping service by Transnetyx at weaning or ∼21 d old. Equal numbers of male and females were used for experiments at six to nine months old. Individuals carrying the A53T transgene were compared with nontransgenic littermates as controls. Animal use was approved by Northwestern IACUC under protocol number IS00011551.
FTI treatment of mice
Mice were injected with LNK-754 (Link Medicine) for 28 d as described previously (Cuddy et al., 2019). They received daily intraperitoneal injections of 0.9 mg/kg LNK-754 in 0.5% sodium carboxymethylcellulose (Sigma) or this vehicle alone. After treatment, mice were perfused with PBS, brains rapidly dissected, and brainstem and other regions frozen for analysis. Tissue was lysed in 1% Triton X-100 buffer and used for Western blotting as below. Detection of LC3-II was aided by use of LC3B (Sigma L7543 rabbit 1:500) for 48 h with CQ-treated mouse cell line N2A used as a positive control.
Biochemistry and cell biology
Plasmids and lentiviral production
Generation of ykt6-CS, ykt6-GFP, WT α-syn, and ykt6-shRNA plasmids was described previously (Mazzulli et al., 2016; Cuddy et al., 2019). SNAP29 in pLenti6.3/V5-DEST (HsCD00942173) was obtained from DNASU. FLAG-SNAP29 was a gift from Noboru Mizushima (Addgene plasmid #45915; http://n2t.net/addgene:45915; RRID:Addgene_45915; Itakura et al., 2012).
Lentiviral plasmids were packaged as described (Mazzulli et al., 2016) and titer determined using Zeptometrix HIV1-P24 ELISA kit. SH-SY5Y and iPSn were infected at MOI3. Ykt6-CS rescue studies were harvested 7 d postinfection. Ykt6 knock-down, SNAP-29-V5, and α-syn overexpression were harvested 5 d postinfection. Empty vector or pLKO1-scramble were used as controls.
General Western blotting
A total of 40 µg of protein lysate was run on a tris-glycine SDS-PAGE gel (time and acrylamide percentage specified in individual assay sections titled “LC3-II autophagic flux assay,” “Immunoprecipitation,” “Size exclusion chromatography,” and “HDJ-2 shift assay”) and transferred onto Millipore immobilon-FL PVDF membranes at 30 V for 1 h. Membranes were postfixed in 0.4% paraformaldehyde (PFA) for 30 min, washed three times in pure water, and blocked for 1 h in 1:1 Odyssey blocking buffer (OBB) and tris-buffered saline (TBS). Primary antibodies were incubated overnight rotating at 4° in 1:1 blocking buffer and TBS with 1% Tween (TBS-T). Membranes were then washed three times in TBS-T and secondary antibody in OBB:TBS-T was incubated at room temperature for 1–2 h (Alexa 680 1:10,000; Alexa 790 and IRDye800 1:5000). Membranes were washed as before and scanned on an Odyssey or Sapphire infrared imaging system. Manual band analysis was performed in Odyssey Image Studio software. Lanes were always normalized to a loading control: total protein using Coomassie stain on gel for most assays (CBB), neuron specific enolase (NSE) for SEC, and GFP for immunoprecipitation (IP). GAPDH and tubulin were also evaluated as additional loading and quality controls.
LC3-II autophagic flux assay and triton soluble protein extraction
A total of 40 μm chloroquine (CQ) in water or an equivalent volume in fresh media was added to SH-SY5Y or iPSn, and cells were collected in ice cold PBS 72 h later. This treatment was found to be nontoxic but caused a significant increase in LC3-II (Extended Data Fig. 1-1A). After spinning down at 400 × g, 5 min, 4°C and discarding supernatant, cell pellet was lysed in 1% Triton X-100 buffer [1% Triton X-100, 20 mm HEPES pH 7.4, 150 mm NaCl, 10% glycerol, 1 mm EDTA, 1.5 mm MgCl2, 1 mm phenylmethanesulfonyl fluoride (PMSF), 50 mm NaF, 2 mm Na orthovanadate, and a protease inhibitor cocktail]. Neuron lysates were homogenized ∼15 times in a conical glass vessel using a motor driven Teflon homogenizer at 4000 RPM while SH-SY5Y were broken up by vigorous pipetting. Lysates were incubated in ice slurry for 30 min then cleared by centrifugation at 100,000 × g, 4°C for 30 min; 40 µg of protein determined by BCA assay was run on 15% tris-glycine SDS-PAGE gel for ∼1.5 h, and Western blotting proceeded as described above (LC3 Sigma L8918 rabbit 1:500).
Signal was normalized to CBB. To calculate flux, LC3-II signal in CQ-treated lanes was divided by the average signal in the condition's untreated lanes on the same blot. Fold change in flux compared with control is shown.
qRT-PCR for mRNA measurement
Neuron cell pellet was divided so that one half could be used for mRNA and one half used for protein. PureLink RNA Mini kit (Thermofisher Scientific) was used to extract total RNA, and RevertAid First Strand cDNA Synthesis kit (Thermofisher Scientific) was used to synthesize cDNA. Applied Biosytems 7500 Fast system was used for qRT-PCR with the following predesigned Taqman-primer probes: ACTB (Hs01060665_g1), MAP1LC3 (Hs01076567_g1), SQSTM1 (Hs00177654_m1), and FNTA (Hs00357739_m1). Target mRNA expression was normalized to ACTB and evaluated using δ-δ-CT method (two to three technical replicates; three to five biological replicates).
Immunoprecipitation
Co-IP of GFP-ykt6 from SH-SY5Y was performed using GFP_Trap (Chromotek) beads as described (Cuddy et al., 2019). Briefly, cells from a 10-cm plate were lysed in Co-IP buffer (10 mm Tris-HCl pH 7.5, 150 mm NaCl, 0.5 m EDTA and 0.5% NP-40 containing protease inhibitor cocktail, PMSF, and NaF), and 800–1000 µg total protein was incubated overnight with 20 µL GFP_trap beads rotating at 4°C. Complexes were washed then eluted by boiling in 50 µl 2× Laemmeli sample buffer, half of which was loaded on the gel. The gels were 12% acrylamide and run for ∼1 h.
Size exclusion chromatography
Neurons were lysed in 1% Triton X-100 buffer as above. A total of 800 µg in 250 µl was injected on a Superdex 200 HR 10/300 gel filtration column using Agilent HPLC 1200 series pumps, injector, UV detector, and fraction collector. PBS pH 7.4 was used as mobile phase: flow rate 0.5 ml/min, collecting 0.5-ml fractions, with an injection volume of 250 µl. Fraction size identity was determined by standard calibration of the column before use. Fractions were combined and concentrated using Millipore 10,000 MWCO filters and analyzed by Western blot as above (SNAP29 Abcam rabbit 1:500; ykt6 Santa Cruz mouse 1:500; see previous description by Cuddy et al., 2019). Complexes were measured as the percentage of SNAP-29 signal that appeared at 120 Å over total SNAP-29 summed from all fractions.
HDJ-2 shift assay
DMSO-treated neuron Triton X-100 lysates were run as above on an 8% gel for ∼1 h to separate bands (HDJ-2/Hsp40 Santa Cruz mouse 1:500). Along with molecular weight, farnesylated and nonfarnesylated bands were ensured by comparison to control FTI treated lane (7-d 10 nm LNK-754).
LDH toxicity assay
A total of 40 μm CQ was pipetted into iPSn media 72–24 h before harvest. Media was collected and cell protein extracted for later analysis. Cytotoxicity was measured using the CyQUANT LDH Cytotoxicity Assay (Invitrogen) according to manufacturer's instructions. Samples included three technical replicates and three biological replicates. Maximum LDH activity was used as a positive control, and reaction alone was used as a negative control. LC3-II was confirmed to be increased by Western blotting.
H4 immunofluorescence
Images were acquired using a Leica TCS SPE CTR4000/DMI4000B microscope. At 63×, a 5-µm Z-section centered on DAPI focus was imaged using 0.5-µm slices and sequential scan that started from long wavelength excitation to shortest wavelength (DAPI). Approximately 15 fields of view were acquired from two to four cover glass per condition. Images were analyzed using Fiji software. Slices were stacked in a maximum projection, and individual cell bodies excluding the nucleus were manually selected as regions of interest (ROIs) for measurement. Secondary antibody incubation alone was used as a negative control for autofluorescence and background.
Transfection
For experiments requiring transfection, 80 ng of plasmid was transfected per 50,000 cells using X-treme Gene HP transfection reagent (Roche). Nontransfected wells, empty vector, or GFP were used as controls. Cells were fixed 3 d post-transfection.
LC3 flux
Different fixation conditions were optimized for each assay. For LC3 flux assay, cells on cover glass were treated for 2 h with 200 nm Baf or DMSO then fixed in 100% HPLC grade methanol for 20 min at −20°C. Following three PBS washes, cells were blocked for 1 h in Triton X-100 blocking buffer (0.2% Triton X-100 made in PBS with 5% normal goat serum and 2% bovine serum albumin). Primary incubation with 1:200 LC3 (Sigma L8918 rabbit) was conducted in this same buffer at 4°C for 24 h. LC3 nuclear stain appeared under these conditions but not others, and therefore assumed to be a fixation artifact.
The following steps were identical for each assay. After washing three times in PBS with 0.05% Tween, secondary incubation was performed in Triton X-100 blocking buffer using 1:500 Alexa Fluor 568 and/or Alexa Fluor 488 for 2 h at room temperature. After washing, cover glass were mounted using DAPI fluoromount G (Southern Biotech), and slides were dried for at least 24 h before imaging.
Fluorescence was measured as integrated density in each channel. LC3 flux was analyzed by dividing each baf treated cell's intensity by the average of untreated cell intensities.
Colocalization
For lamp1/LC3 colocalization, cells were fixed in 4% PFA for 15 min at room temperature, permeabilized in 0.1% Triton X-100 for 30 min, and blocked in 0.1% Triton X-100 blocking buffer for 30 min. Incubation continued for 48 h in 1:50 lamp1a (Santa Cruz mouse) and 1:200 LC3 (Sigma L8918 rabbit).
For ykt6/SNAP-29 colocalization, cells were first permeabilized with 50 µg/ml digitonin for 2 min, washed 3× in PBS, then fixed in 4% PFA. Permeabilization and blocking was performed in 0.2% Triton X-100 blocking buffer as above. Endogenous ykt6 was stained with 1:100 ykt6 (Santa Cruz polyclonal) for 48 h.
Secondary incubation and imaging as above. “EzColocalization” plugin was used for colocalization analyses (Stauffer et al., 2018), and Pearson's R was compared.
To induce autophagosome formation in Atg9 movement study, media was replaced with Optimem alone 2 h before permeabilization in digitonin and fixation in 4% PFA.
Neuron immunofluorescence
Images were acquired as for H4 cells above with the following change: 4.8-µm Z-stacks were imaged in 0.4-µm slices at 12-bit depth resolution.
LC3 flux
Cells were fixed in methanol for 15 min at −20°, permeabilized with 0.2% Triton X-100 for 30 min at room temperature, and blocked overnight at 4° with blocking buffer described above. Incubation with 1:100 LC3A (Novus Biologicals) was performed overnight in blocking buffer at 4°. Fiji software was used to measure LC3 integrated density in cell bodies of a single slice excluding the nucleus. Autophagic flux was calculated as above.
Colocalization
Cells were permeabilized with 50 µg/ml digitonin for 2 min, then fixed in 4% PFA for 15 min at room temperature. Permeabilization with 0.2% Triton X-100 and blocking overnight was performed as for LC3 flux. Ykt6 (Santa Cruz monoclonal) and SNAP-29 (Abcam) were both incubated at 1:100 for 48 h at 4°. Analysis was performed using Fiji software as above.
Electron microscopy (EM) analysis
Full details of cell preparation can be found previously (Stojkovska et al., 2022). Day 90 neurons were fixed in 2.5% glutaraldehyde, postfixed in 1% osmium tetroxide (OsO4), and stained with 1% uranyl acetate (Electron Microscopy Sciences). They were finally embedded in LX112 (Ladd Research Industries) resin mix and thin sectioned at ∼70-nm width. Imaging was performed on a FEI Tecnai Spirit G2 TEM. For qualitative analysis, blind images were searched for structures with features of immature to mature autophagosomes (double-membraned, roughly circular or crescentiform, medium radiodensity, may include cargo; Jung et al., 2019; Klionsky et al., 2021) which were labeled for review. A second pass removed questionable structures, images were unblinded, and similarities within conditions assessed with discussion (6–12 cells per line).
Postmortem synucleinopathy brain tissue analysis
Postmortem frontal cortex from the Northwestern University Alzheimer's disease pathology core were sequentially extracted for protein as described in detail previously (Stojkovska et al., 2022). Full details regarding control and DLB patient characteristics can be found there as well. Briefly, five steps of increasingly harsh detergents were used: high-salt buffer (50 mm Tris-HCl pH 7.4, 750 mm NaCl, 10 mm NaF, 5 mm EDTA), buffer with 1% Triton X-100, 1% Triton X-100 with 30% sucrose, 1% Sarkosyl, and remaining Sarkosyl-insoluble. LC3 (Sigma L8918 rabbit 1:500) was probed in the 1% Sarkosyl fraction, and FT-α (Santa Cruz mouse 1:500) was probed in the high-salt fraction. Samples were de-identified and blinded until analysis.
Results
Accumulation of wild-type α-syn reduces autophagy flux through impeding autolysosome fusion
To assess autophagy phenotypes in PD patient iPSC-derived midbrain neurons (iPSn), we employed iPSC lines from three different members of the Iowa kindred carrying an SNCA triplication and their corresponding isogenic CRISPR-corrected controls as previously described (Stojkovska et al., 2022). These lines differentiate with appropriate midbrain and maturity markers by day 60 and exhibit accumulation of pathogenic insoluble synuclein by day 90 (Mazzulli et al., 2016; Stojkovska et al., 2022). Autophagic flux was assessed by quantifying LC3-II protein levels in response to an autolysosomal inhibitor chloroquine (CQ) with treatment conditions determined to be nontoxic (Extended Data Fig. 1-1A). While analysis of day 60 cultures showed no difference (Fig. 1A), day 90 iPSn from each triplication carrier showed a ∼50% reduction in autophagic flux compared with their isogenic corrected controls and a healthy control line previously published as “2131” (Fig. 1B; Mazzulli et al., 2016). These results were replicated in differentiated SH-SY5Y cells that overexpress α-syn (Extended Data Fig. 1-1B; Cuddy et al., 2019). P62/SQSTM1 is a linker between LC3 and ubiquitinated MA substrates that is constantly degraded by MA (Abounit et al., 2012). Although both soluble and insoluble p62 levels increased on CQ treatment in isogenic control iPSn as expected, SNCA triplication lines PD iPSn showed no response to CQ (Fig. 1C; Extended Data Fig. 1-1C). RNA expression of LC3 was increased and p62 unchanged in PD iPSn, suggesting that perturbations in autophagic flux do not occur from decreased transcriptional response (Fig. 1D). This is consistent with elevated protein levels of both LC3-II and p62 in untreated (CQ–) PD iPSn (Fig. 1E), suggesting a late-stage blockade in autophagy. To assess the effect of α-syn accumulation on autophagy in vivo, we measured LC3-II protein in postmortem cortical samples from patients diagnosed with Dementia with Lewy bodies (DLB). We found that LC3-II was elevated along with α-syn, similar to our observations in PD iPSn (Fig. 1F). These data suggest that α-syn inhibits autophagic flux through post-transcriptional mechanisms at a stage beyond LC3 or autophagosome synthesis.
Macroautophagy flux is decreased in aged SNCA triplication neurons and DLB brain. A, Western blot analysis of day 60 iPSn from patient line #1 (Syn. 3×) and isogenic controls (Corr.) that were treated for 72 h with 40 μm chloroquine (CQ). Coomassie Brilliant Blue (CBB) was used as a loading control. Flux is measured as the ratio of normalized LC3-II levels in inhibitor treated lanes to the average of untreated lanes. Graphs represent fold change compared with controls on the same blot (n = 6). B, Western blot analysis of day 90 iPSn from three patient lines (Syn. 3×) and isogenic controls (Corr.) evaluated as in A (Healthy control, patient 1, and patient 2: n = 5–8; patient 4: n = 2). The time course of CQ effects is shown in Extended Data Figure 1-1A. A similar phenotype in α-syn overexpressing SH-SY5Y cells is shown in Extended Data Figure 1-1B. C, Western blot analysis of day 90 iPSn as in B showing Triton X-100-soluble p62 with and without CQ treatment. Graph shows the increase in p62 with CQ treatment compared with the average of untreated lanes, similar to flux (n = 2–8). Triton X-100 insoluble fractions are shown in Extended Data Figure 1-1C. D, RT-PCR analysis of LC3 and P62 mRNA from patient 1 normalized to actin represented as fold change relative to corrected control (n = 3–5). E, Baseline LC3-II and p62 protein levels measured in untreated (CQ–) lanes of each patient normalized to corresponding controls. Patients are shown as different shaded points (n = 12 corrected line; n = 19 Syn. 3×). F, Representative Western blotting of LC3 extracted from cortex of DLB patients and controls (n = 4 controls; n = 9 DLB). For all quantifications, values are the mean ± SEM; *p < 0.05, **p < 0.01, ***p < 0.001. Student's t test or ANOVA with Tukey's post hoc test was used for panels B, C. Student's t test was used for panels A, D–F. Please see Extended Data Figure 1-1 for more details.
Extended Data Figure 1-1
Limited chloroquine treatment is nontoxic and macroautophagy flux is decreased by WT α-syn in SH-SY5Y. A, Triplication iPSn were treated for 24–72 h with 40 μm CQ and media was analyzed for lactate dehydrogenase (LDH) as an indication of cell death (n = 3). Neg. ctrl = media alone sample without cells. Right, CQ efficacy was confirmed by LC3-II increase by Western blot analysis. CBB (Coomassie Brilliant Blue) was used as a loading control from the corresponding gel. B, SH-SY5Y stably overexpressing WT α-syn were treated for 2 h with 200 nm Bafilomycin A1 (Baf). Autophagic flux was measured by Western blot analysis as described in Figure 1. Vector indicates a matched control stable line (n = 12). CBB was used as a loading control. C, Western blot analysis of triton insoluble (SDS soluble) protein fractions from triplication iPSn CQ flux experiments. Right graph shows fold increase in p62 in CQ-treated lanes relative to untreated average, similar to flux calculation. Differently colored points signify different patient lines. (Values are the mean ± SEM; *p < 0.05, **p < 0.01, ***p < 0.001.) ANOVA with Dunnett's test was used for panel A where no CQ treatment was considered the control. Student's t test was used for panels B, C. Download Figure 1-1, TIF file.
To directly visualize autophagosome puncta by immunofluorescence, we used a previously characterized human neuroglioma H4 model overexpressing wild-type (WT) α-syn under a tet-off system that exhibits lysosomal dysfunction (Mazzulli et al., 2011). We previously established that the degree of α-syn overexpression in this model is similar to SNCA triplication neurons (Mazzulli et al., 2016). α-Syn overexpression increased LC3 puncta at baseline, and reduced LC3 response on inhibitor treatment (Fig. 2A). The same pattern was observed for p62 puncta (Fig. 2B), suggesting decreased autophagosome degradation. Colocalization between LC3 and the lysosome marker LAMP1 was decreased by α-syn accumulation, consistent with reduced autolysosomal fusion (Fig. 2C). Phagophore formation requires movement of Atg9-containing vesicles from the Golgi to provide lipid and scaffolding components (Ohashi and Munro, 2010; Sawa-Makarska et al., 2020). Atg9 movement from the Golgi was assessed in H4 cells by measuring colocalization with the resident Golgi SNARE syntaxin 5 (STX5) following starvation to induce autophagy. We found similar changes in Atg9/STX5 colocalization after starvation in both high and low expressing α-syn cells (Extended Data Fig. 2-1), suggesting that autophagic formation is not dramatically perturbed by α-syn. We confirmed immunofluorescence results in SNCA triplication iPSn, which showed increased autophagosome puncta within cell bodies and neural extensions of fixed SNCA triplication iPSn compared with corrected controls (Fig. 2D). As with Western blotting, LC3 puncta did not increase proportionally with chloroquine treatment in SNCA triplication iPSn (Fig. 2D), indicating decreased flux through the autophagic pathway.
Immunofluorescence and ultrastructural analysis of autophagy in fixed PD culture models. A, Representative Z-stack projected confocal images of fixed H4 cells overexpressing α-syn (−dox) treated for 2 h with 200 nm inhibitor Bafilomycin (Baf), fixed, and immunostained for LC3. Each point represents intensity measured in a manually selected cell body. Any detectable nuclear signal was excluded. DAPI indicates nuclei and was used for normalization. Upper, Graph of flux quantification analyzed by Student's t test. Lower, Graph showing individual LC3 puncta intensity analyzed by ANOVA with Tukey's post hoc test (n = 50–80 cells; 15 fields of view; scale bar = 12 µm, N.S., not significant). B, Immunofluorescence analyses of H4 cells as in A showing p62 staining. Graph shows p62 intensity in individual cells analyzed by ANOVA with Tukey's post hoc test (n = 29–35 cells; 15 fields of view; scale bar = 12 µm; N.S., not significant). C, Representative confocal images of H4 cells stained for LC3 and LAMP1. Pearson's R calculated in manually selected cell bodies. Each point is one cell (n = 65–70 cells; 15 fields of view; scale bar = 12 µm). Arrows and white spots illustrate colocalized points. Extended Data Figure 2-1 evaluates colocalization of other markers to address autophagosome formation. D, Immunofluorescence images showing LC3 puncta in fixed day 90 PD iPSn and controls. A 1.6-µm projection is shown. Chloroquine (CQ) was used as before to evaluate flux. Left graph compares LC3 intensity in manually selected cell bodies in untreated wells. Right graph compares LC3 intensity as a response to chloroquine as in Figure 1 and A (n = 20–33 cells; scale bar = 12 µm) Student's t test used for panels C, D (values are the mean ± SEM; *p < 0.05, **p < 0.01, ***p < 0.001). E, Representative electron microscopy images of day 90 SNCA triplication and healthy control iPSC-neurons showing autophagic structures. Black arrows label multilamellar vesicles while white arrows label vesicles with traditional autophagosome features. Upper row, Scale bar = 2 µm. Lower row, Zoomed scale bar = 500 nm (6–12 cells per line). Please see Extended Data Figure 2-1 for more details.
Extended Data Figure 2-1
Atg9 localization response to starvation is not perturbed by α-syn. H4 were starved for 2 h in serum reduced media to induce autophagosome formation, then fixed and immunostained for Golgi marker Stx5 and autophagic protein Atg9. Colocalization was quantified by Pearson's R (n = 40–50 cells; scale bar = 12 µm). Values are the mean ± SEM; ***p < 0.001, **p < 0.01, Student's t test. Scale bars = 12 µm. Download Figure 2-1, TIF file.
We next sought to directly examine autophagosomes by electron microscopy analysis of fixed PD iPSn. Qualitative ultrastructural analysis of SNCA triplication iPSn revealed unusual vesicles with autophagosome features: double-membraned and completely closed, but abnormally wrapped in multiple layers of lamina (Fig. 2E; Jung et al., 2019; Klionsky et al., 2021). Some vesicles appeared to contain electron dense, undegraded material (Fig. 2E, black arrows). This supports the notion that PD iPSn are capable of autophagosome formation and cargo engulfment, but are compromised in their ability to clear engulfed substrates. Altogether, these data indicate that α-syn disrupts autophagic flux at a stage after autophagosome formation.
α-Syn decreases the association between ykt6 and autolysosome fusion SNARE SNAP-29
Since previous studies in cell lines showed that ykt6 plays a role in autolysosomal fusion (Matsui et al., 2018), we sought to determine whether ykt6 played a similar role in human midbrain cultures. Ykt6 was knocked down (KD) in healthy control iPSn using previously established shRNA constructs (Cuddy et al., 2019), followed by analysis of autophagy by LC3 Western blot. Although we only achieved a ∼50% KD of ykt6, this was sufficient to cause a dramatic sevenfold accumulation of LC3-II (Fig. 3A). Ykt6 KD also abolished the LC3-II response to CQ, indicating a severe impairment in autophagic flux that is similar to the effect induced by α-syn accumulation (Fig. 3B). These data suggest that ykt6 is critical for late-stage autophagic flux in midbrain iPSn.
α-Syn decreases the association between ykt6 and autolysosome fusion SNARE SNAP-29. A, Ykt6 was knocked down in day 90 iPSn from controls for 5 d by lentiviral infection of shRNA (ykt6 KD) or a scrambled vector, and LC3 and flux measured by Western blotting (n = 2–3). B, Control uninfected or ykt6 KD iPSn were treated with CQ and analyzed by LC3 Western blot analysis (n = 3). C, Representative zoomed confocal images of H4 cotransfected with SNAP-29-Flag and ykt6-GFP. Pearson's R calculated in cytosol excluding nucleus as before (n = 10–20 cells; 15 fields of view; scale bar = 12 µm). GFP transfection was used as a negative control. Graph on the right compares relative intensities for each marker across conditions. Each point represents an individual cell. D, Representative immunofluorescence projections of fixed day 90 PD iPSn showing colocalization between endogenous SNAP-29 and ykt6 (n = 19–22 cells; 18 fields of view; scale bar = 12 µm). Graph compares Pearson's R in manually selected cell bodies as in B. E, Immunoprecipitation of GFP from SH-SY5Y stably expressing ykt6-GFP infected to express SNAP-29-V5 with α-syn (+) or empty vector (−). V5 signal normalized to ykt6-GFP pulldown (n = 5). F, SNAP-29 and ykt6 levels from day 90 triplication and corrected iPSn (n = 7). G, Triton X-100 soluble day 90 iPSn lysates separated by size exclusion chromatography (SEC) followed by Western blot analysis of collected fractions. NSE was used as control (n = 4). Input for column injection is shown to the right. Graph shows the percentage of SNAP-29 signal in the first fraction (120 Å) out of the sum of SNAP-29 in all fractions. For all quantifications, values are the mean ± SEM; *p < 0.05, **p < 0.01, ***p < 0.001. Student's t test was used in panels A, B, D–G; ANOVA with Dunnett's post hoc test was used for panel C using +dox as the control condition.
We then examined whether the interaction between ykt6 and its primary SNARE binding partner required for autolysosome fusion, SNAP-29 (Matsui et al., 2018; Takáts et al., 2018) is disrupted in cells that accumulate α-syn. H4 cells were transfected to express Flag-tagged SNAP-29 and GFP-tagged ykt6 to facilitate detection by immunofluorescence analysis. Although the total levels of GFP-ykt6 and Flag-SNAP-29 did not change, colocalization was decreased in α-syn overexpressing cells compared with controls (Fig. 3C), consistent with a decline in autophagic-lysosomal fusion (Fig. 2). We verified this finding using antibodies that detect endogenous ykt6 and SNAP-29 in patient iPSn (Fig. 3D). We stained endogenous ykt6 and SNAP-29 in fixed SNCA triplication iPSn and found that colocalization was reduced in patient cells (Fig. 3D). To measure SNARE complexes, we immunoprecipitated GFP-ykt6 from stably expressing SH-SY5Y (Cuddy et al., 2019) and probed for V5-tagged SNAP-29 following expression of WT α-syn or empty vector. In this model, the amount of V5-SNAP-29 that pulled down with ykt6 was decreased by α-syn accumulation (Fig. 3E). To confirm these findings in PD patient iPSn with a distinct method, we used size exclusion chromatography (SEC) to measure endogenously expressed SNARE complexes in lysates from day 90 SNCA triplication cultures and isogenic controls. Quantification of the input injected on the SEC column indicated that the total amount of SNAP-29 in SNCA triplication iPSn lysates was not different compared with isogenic controls (Fig. 3F). SEC analysis indicated that ykt6-SNAP-29 complexes eluted as a 120-Å-sized particle, and that the percentage of SNAP-29 co-eluting with ykt6 was dramatically reduced in SNCA triplication iPSn compared with corrected controls (Fig. 3G). Together, these data indicate that reduced association of SNAP-29 and ykt6 occurs in cells that accumulate α-syn, providing a mechanistic explanation for the inhibition of autophagic flux.
Farnesyltransferase protein is increased in PD iPSn and DLB brain
We next wanted to gain mechanistic insight into the cause of reduced ykt6-SNAP-29 complexes and the connection to autophagic dysfunction. Ykt6 is autoinhibited by farnesylation, which promotes a closed, inactive conformation that is incapable of forming SNARE complexes (Tochio et al., 2001; Wen et al., 2010). Our previous study indicated that inactive, cytosolic ykt6 is elevated and interacts with α-syn in PD patient iPSn (Cuddy et al., 2019). To assess the role of farnesylation in SNCA triplication iPSn, we first measured the levels of farnesyltransferase (FTase). Western blot analysis showed increased levels of FTase subunit α (FT-α) in day 90 SNCA triplication iPSn compared with corrected and healthy controls (Fig. 4A). This change occurred post-transcriptionally, since mRNA levels were unchanged (Fig. 4B). To assess FTase activity, we quantified the levels of farnesylated and nonfarnesylated HDJ-2, an established marker of FTase activity (Adjei et al., 2000; Capell et al., 2008). The proportion of farnesylated HDJ-2 was increased in PD iPSn compared with corrected controls (Fig. 4C), suggesting increased FTase activity. To determine FTase levels in vivo, we analyzed frontal cortical lysates from DLB brain and found FT-α was increased in DLB patients compared with age-matched healthy controls (Fig. 4D). These data indicate that FTase is increased in both iPSC-derived PD patient neurons and DLB brain, and its increased activity may contribute to autophagic dysfunction in PD.
Farnesyltransferase protein is increased in PD iPSn and DLB brain. A, Day 90 iPSn lysates from triplication (3×), corrected (corr), and healthy control (ctrl) showing FT-α (n = 5–7). CBB and GAPDH were used as loading controls. B, Measurement of mRNA by RT-PCR from matched cultures to quantify FNTA expression, normalized to actin (n = 4–5). C, HDJ-2 from corrected (corr) and triplication (3×) iPSn was separated into farnesylated (lower band: f) and nonfarnesylated (upper band: non-f) forms. Fold change ratio of farnesylated to nonfarnesylated is shown (n = 5). 3× iPSn were treated for 7 d with 10 nm LNK-754 (Ctrl + FTI) and used as a control. D, Representative blot of FT-α extracted from cortex of DLB patients and controls (n = 4 controls; 9 DLB). For all quantifications, values are the mean ± SEM; *p < 0.05, **p < 0.01, ***p < 0.001 (Student's t test).
Inhibition of farnesyltransferase improves macroautophagy through ykt6
To determine whether increased FTase activity contributes to autophagic failure in PD cultures, we next tested whether treatment with a specific farnesyltransferase inhibitor (FTI), LNK-754, could rescue autophagy in PD iPSn. We had previously found that LNK-754 reduces farnesylation of ykt6, promotes its movement from cytosol to membrane, and reduces pathologic α-syn levels in A53T mutant iPSn (Cuddy et al., 2019). Here, we find that FTI treatment rescued autophagic flux, as indicated by immunofluorescence analysis of fixed H4 α-syn cells that showed increased LC3 response to lysosomal inhibition (Fig. 5A). Western blot analysis of LC3-II corroborated these findings, demonstrating that FTI treatment increased autophagic flux in α-syn overexpressing SH-SYSY (Fig. 5B). We verified that reducing FTase activity improves autophagic flux in PD patient cultures since FTI treatment of 2 distinct patient lines completely rescued LC3-II flux to that of isogenic corrected lines (Fig. 5C). FTI treatment had no effect on the mRNA of LC3 or p62, suggesting the effect is post-transcriptional (Fig. 5D).
Inhibition of farnesyltransferase improves macroautophagy in PD cultures and in vivo. A, Representative images showing LC3 in −dox (high α-syn) H4 cells treated for 3 d with 10 nm FTI and 2-h 200 nm Baf (n = 30–40; scale bar = 12 µm). Nuclear LC3 signal was excluded from quantification. Each point represents a single cell. B, Western blot analysis of stably expressing α-syn SH-SY5Y cells treated for 7 d with 10 nm FTI and 72 h CQ (n = 6). C, Western blot analysis of day 90 iPSn to quantify the LC3-II response to CQ following 7 d 10 nm FTI treatment (n = 5–11). D, RT-PCR analysis of LC3 and P62 mRNA from FTI treated iPSn, normalized to actin (n = 4–5). Differently colored points show variation among passages. E, Brainstem lysates of six-month-old DASYN53 mice and littermate controls (WT) receiving 28-d intraperitoneal injections of 0.9 mg/kg FTI or vehicle (n = 4–5). N2A cells treated for 24 h with 40 μm CQ were included as a positive control to identify LC3-II. Syn211 is a human specific synuclein antibody. For all quantifications, values are the mean ± SEM; *p < 0.05, **p < 0.01, ***p < 0.001. ANOVA with Dunnett's test was used in panels A–C, E (middle graph), using +α-syn/+DMSO as the control condition. Student's t test was used for panels D, E (right graph).
Finally, we wanted to determine whether FTI treatment can alter autophagy in vivo. We used a previously characterized transgenic mouse model that accumulates human A53T α-syn within dopamine neurons, called DASYN53 mice (Chen et al., 2015). These mice were injected for 28 d with LNK-754 or a vehicle, and brainstem lysates were analyzed by LC3 Western blot. This treatment not only normalized LC3-II levels, but also significantly reduced α-syn (Fig. 5E). This is consistent with our previous findings that showed reduced α-syn pathology and improvement of motor behavior in DASYN53 mice (Cuddy et al., 2019). Collectively, these data indicate that increased FTase activity contributes to autophagic flux inhibition in PD, and that inhibiting FTase in PD patient iPSn or animal models can rescue autophagy and reduce α-syn.
Since our data indicated that ykt6-SNAP-29 complexes are reduced in PD iPSn leading to reduced autophagic flux, we next sought to determine whether reducing FTase activity could restore ykt6-SNAP-29 complexes. FTI treatment in SNCA triplication iPSn increased the percentage of ykt6-SNAP-29 complexes that co-eluted as a 120-Å-sized SEC particle, suggesting increased complex formation (Fig. 6A). In fixed α-syn overexpressing H4 cells transfected to express ykt6-GFP and SNAP-29-flag, reducing FTase activity improved the colocalization of ykt6 and SNAP-29, supporting our findings from SEC analysis (Fig. 6B). Quantification of immunostained cells for endogenous ykt6/SNAP-29-flag confirmed increased colocalization with FTI treatment (Extended Data Fig. 6-1A). Finally, we expressed a ykt6 mutant that resists farnesylation by mutating the farnesyl-modified cysteine residue at position 195 (called ykt6-C195S or CS). Ykt6-CS increased the colocalization with SNAP-29, consistent with increased complex formation (Extended Data Fig. 6-1B). Ykt6-CS mediated rescue was not further increased by FTI treatment, suggesting that the FTI enhances complex assembly by reducing ykt6 farnesylation rather than reducing the farnesylation of other proteins (Extended Data Fig. 6-1B). These data indicate that reducing FTase activity promotes ykt6-SNAP-29 complex assembly and improves autophagic flux in cells that accumulate α-syn.
FTI rescue of macroautophagy depends on ykt6. A, Triplication iPSn cultures were treated for 7 d with 10 nm FTI, and lysates were analyzed by SEC/Western blot of collected fractions. Percentage SNAP-29 eluting as a complex at 120 Å was quantified on the right (n = 2). B, Representative immunofluorescence images of α-syn-overexpressing H4 transfected with SNAP-29-flag and ykt6-GFP treated for 3 d with 10 nm FTI. Quantification of colocalization is shown to the right (n = 15–18; scale bar = 12 µm). Points represent individual cells. Colocalization rescue of endogenous ykt6 and constitutively active ykt6-CS is shown in Extended Data Figure 6-1A,B. C, Stable SH-SY5Y expressing α-syn or empty vector were infected to overexpress constitutively active ykt6-CS for 7 d. Cultures were analyzed by Western blot to measure the LC3-II response to CQ. D, Triplication iPSn were infected and analyzed as in panel C. Differently colored points illustrate individual passages (n = 5–6). E, H4 cells cotransfected with ykt6 shRNA and GFP, treated with FTI and Baf as in Fig. 3. LC3 intensity was measured in manually selected GFP-positive cells (n = 10–12 cells; scale bar = 12 µm). F, Triplication iPSn infected with ykt6 shRNA (ykt6 KD) or scramble vector (scrb) for 3 d followed by 7 d FTI treatment (n = 3). Cultures were extracted and analyzed by Western blot to measure LC3-II response. NS, not significant. The same effect is quantified in SH-SY5Y cells in Extended Data Figure 6-1C. For all quantifications, values are the mean ± SEM; *p < 0.05, **p < 0.01, ***p < 0.001. ANOVA with Dunnett's post hoc test was used for panels B–D where the +α-syn/+DMSO or vector condition was considered the control. ANOVA with Tukey's post hoc test was used for panel F. Student's t test was used for panels A, E. Please see Extended Data Figure 6-1 for more details.
Extended Data Figure 6-1
Reducing the farnesylation of ykt6 increases ykt6/SNAP-29 colocalization and improves autophagic flux. A, H4 α-syn cells were transfected with SNAP-29-flag and treated with 10 nm FTI for 3 d. Cultures were fixed and immunostained to detect endogenous ykt6 and flag, and analyzed for colocalization as in Figure 6. B, H4 α-syn cells were infected with WT ykt6 or ykt6-CS, with or without FTI. Ykt6/SNAP-29 colocalization was determined in immunostained cells. C, Stable lines of SH-SY5Y cells overexpressing α-syn and either scrambled control (Ctrl) or ykt6 knock-down (KD) were treated with FTI as in Figure 6. Autophagic flux was determined by measuring the LC3 response to lysosomal inhibitor treatment by Western blotting (n = 5–6). Values are the mean ± SEM; *p < 0.05, **p < 0.01, ***p < 0.001. ANOVA with Dunnett's post hoc test where DMSO (−) dox was considered as the control used in panels A, B. Student's t test used in panel C. Download Figure 6-1, TIF file.
We further assessed the role of reducing farnesylation on autophagic activity by LC3 Western blot analysis. This showed that ykt6-CS rescued autophagic flux in α-syn SH-SY5Y and PD patient iPSn, since the LC3-II response to CQ was more prominent in ykt6-CS expressing cells (Fig. 6C,D). We also found that the FTI had no effect on autophagic flux in fixed H4 cells that were transfected with Ykt6 shRNA constructs and immunostained for LC3 (Fig. 6E). Western blot analysis of LC3 in both cell lines and PD iPSn models confirmed that ykt6 was required to rescue autophagic flux upon FTI treatment (Fig. 6F; Extended Data Fig. 6-1C). These findings indicate that autophagic flux can be rescued by FTIs through ykt6.
Discussion
Our data indicate that α-syn disrupts late-stage autophagic flux and support a crucial role for ykt6 in autophagosome-lysosome fusion in iPSC-derived human midbrain neurons. Previous studies have shown that ykt6-SNAP29 complexes are similarly important for late-stage fusion in HeLa cell lines, although autophagic flux can still occur in the absence of ykt6 through an independent STX17-SNAP-29 complex (Matsui et al., 2018). Our data indicate that midbrain neurons are highly dependent on ykt6 for autophagosome-lysosome fusion, since partial ykt6 KD resulted in a near-complete inhibition of autophagic flux (Fig. 3A,B). Furthermore, expression of ykt6-CS in PD iPSn completely restored autophagic flux to control levels (Fig. 6). Ykt6 is a unique SNARE protein since its activity is regulatable by post-translational lipidation and phosphorylation (Wen et al., 2010; Karuna et al., 2020; McGrath et al., 2021). Similarly, autophagy is an inducible clearance pathway which requires regulated machinery to rapidly sense and respond to various types of cellular stress. Our previous work showed that ykt6 coordinates with TFEB, the master transcriptional regulator of lysosomal biogenesis and autophagy (Sardiello et al., 2009; Settembre et al., 2011), during periods of lysosomal stress to enhance hydrolase trafficking between the ER and Golgi (Cuddy et al., 2019). Consistent with previous studies, we show here that ykt6 also regulates autophagy in iPSn models, indicating its critical role in integrating two distinct trafficking pathways to enhance cellular clearance (Fig. 7).
Summary of the role of ykt6 in the autophagic-lysosomal pathway and its disruption in synucleinopathies. Under physiological conditions, ykt6 controls autophagic-lysosomal activity by coordinating both hydrolase trafficking between the ER and Golgi (step 1), and autophagic cargo delivery into lysosomes (step 2) in human iPSC-derived midbrain neurons. α-Synuclein accumulation perturbs the autophagic-lysosomal system through blocking ykt6-SNAP-29 complexes to prevent auto-lysosomal fusion and simultaneous blockade of lysosomal hydrolase trafficking (demonstrated by Cuddy et al., 2019). Enhancing ykt6 activity can be achieved by preventing its farnesylation, which promotes an open, active conformation. Subsequent palmitoylation in either Golgi or autophagic membranes promotes SNARE complex formation, membrane fusion, and enhanced proteolysis, resulting in the reduction of protein aggregates.
Multiple studies have shown that α-syn can bind to SNARE proteins under physiological or pathologic conditions, including synaptobrevin-2, which shares sequence homology with ykt6 (Burré et al., 2010; Choi et al., 2013). We previously found that α-syn impedes ykt6 SNARE complex assembly between the ER and Golgi by interacting with a closed, inactive form of ykt6 in the cytosol (Cuddy et al., 2019). Our data here show that α-syn inhibits autophagic flux through a similar mechanism, by impeding ykt6-SNAP-29 complexes during autophagosome-lysosome fusion. Other work in cell lines with viral α-syn overexpression showed that α-syn may inhibit autophagic flux at a similar stage by SNAP-29 depletion (Tang et al., 2021). Although we did not document SNAP-29 depletion in PD patient iPSn, we find a reduction in ykt6-SNAP-29 complexes (Fig. 3B–E). Ultrastructural analysis suggests that autophagosomes are synthesized in PD iPSn and capable of engulfing substrates, which is consistent with the accumulation of LC3-II. The complete multi-lamellar vesicles we observed further suggest that autophagosomes are formed but not processed, perhaps with multiple attempts at engulfment (Fig. 2C). Human dopaminergic iPSn with CRISPR-corrected isogenic controls were selected as a primary model for this investigation, to account for relevant differences in autophagic processes of human neurons as compared with other species and cell types that rapidly divide (Larsen and Sulzer, 2002; Maday and Holzbaur, 2016; Benito-Cuesta et al., 2017). Just as age is a primary risk factor for parkinsonism, reduced autophagy only appeared when iPSn were aged to 90 d in culture; no difference was observed at day 60 (Fig. 1). Collectively, this indicates that a similar mechanism of chronic endogenously accumulated α-syn simultaneously impedes multiple trafficking pathways required for autophagy. This, in turn, disables essential proteostasis pathways required to remove aggregates, promoting the growth of α-syn aggregates in a self-propagating pathogenic cycle.
We found that increased FTase levels inhibit autophagic flux in PD patient iPSn and DLB brain (Fig. 4). Other groups have shown that elevated FTase occurs in neurodegenerative disease (Jeong et al., 2021), and that inhibiting FTase can activate proteolytic pathways and reduce protein aggregates (Pan et al., 2008; Hernandez et al., 2019; Cuddy et al., 2022). FTase is the only mammalian enzyme that mediates farnesyl modification of ykt6 and other proteins that harbor the CaaX motif (Casey and Seabra, 1996), and we found it to be increased in disease (Fig. 4). The mechanisms leading to increased FTase activity are not fully understood, however elevated FTase proteins levels in the absence of mRNA changes suggest that increased FTase protein stability may play a role. Farnesylation of ykt6 is a well-established regulatory mechanism that promotes a closed, inactive conformation in the cytosol (Wen et al., 2010; Cuddy et al., 2019). Our previous work demonstrated that inactive ykt6 accumulates in the cytosolic fraction of PD patient iPSn and that FTase inhibition promotes its active conformation and rescues protein trafficking (Cuddy et al., 2019). Similarly, here we find that inhibiting FTase promotes ykt6-SNAP-29 complexes and restores autophagic flux, suggesting that aberrant FTase activity plays a mechanistic role in autophagy impairment in PD. The dysfunction we describe is attractive for intervention because ykt6 is amenable to pharmacological activation through brain penetrant small molecules, as we demonstrate here with the FTI, LNK-754. Clinically-validated FTIs are already available for other diseases including progeria (Moulder et al., 2004; Gordon et al., 2018), and it is possible that these agents may be repurposed for the treatment of neurodegenerative disease as ykt6-ALP activators. Knocking down ykt6 prevented FTI-mediated rescue of autophagy, and overexpressing ykt6-CS replicated the effect of FTIs (Fig. 6), together suggesting that LNK-754 improves autophagy through ykt6. Independently, FTase activity was found to be increased in postmortem Alzheimer's brain, and its suppression in mouse models reduced β-amyloid and neuron loss (Jeong et al., 2021; Cuddy et al., 2022). However, while we found increased FTase in DLB patient cortex, another group reports no difference in this region in PD and decreased FTase in the substantia nigra (Jo et al., 2021). The discrepancy could be because of pathologic distinctions between DLB and PD or other technical differences, but developing more robust measures of FTase activity will likely provide clarification beyond protein levels. Increasing farnesylation can also improve phenotypes of PD models induced by loss of Parkin activity through rescuing transcriptional dysregulation (Jo et al., 2021). Therefore, FTI treatment may not be suitable for certain PD subtypes caused by loss of Parkin. Nonetheless, activating ykt6 to stimulate protein clearance by FTIs or other methods may be a viable therapeutic strategy to reduce protein aggregates in multiple types of neurodegenerative diseases including sporadic synucleinopathies. These therapies may be especially effective in combination with treatments that synergistically target other elements of the proteostasis pathway (Stojkovska et al., 2022).
While medical and surgical treatments are available to manage PD symptoms, there are currently no disease-modifying therapies capable of reversing its underlying neuropathology. Development of such treatments will likely require a detailed understanding of how basic cellular processes malfunction in response to α-syn accumulation, and one of the most relevant common pathways revealed by genetic studies is the ALP (Nalls et al., 2014; Chang et al., 2017). Our study highlights the importance of the ALP in PD pathogenesis, and provide a novel therapeutic strategy that simultaneously rescues multiple steps in the autophagic pathway through manipulation of a single SNARE protein.
Footnotes
Research reported in this publication was supported by the National Institute of Neurological Disorders and Stroke of the National Institutes of Health under Award Number R01NS092823. J.R.M. is an inventor on patents that cover methods to activate cellular clearance, United States Provisional Patent Application Ser. No. 63/264,224. (Pending), and United States Patent Application 17/073,603 (Filed April 22, 2021). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
The authors declare no competing financial interests.
- Correspondence should be addressed to Joseph R. Mazzulli at jmazzulli{at}northwestern.edu
