Sortilin acts as an endocytic receptor for α‐synuclein fibril

Cell‐to‐cell spreading of misfolded α‐synuclein (αSYN) is supposed to play a key role in the pathological progression of Parkinson's disease (PD) and other synucleinopathies. Receptor‐mediated endocytosis has been shown to contributes to the uptake of αSYN in both neuronal and glial cells. To determine the receptor involved in αSYN endocytosis on the cell surface, we performed unbiased, and comprehensive screening using a membrane protein library of the mouse whole brain combined with affinity chromatography and mass spectrometry. The candidate molecules hit in the initial screening were validated by co‐immunoprecipitation using cultured cells; sortilin, a vacuolar protein sorting 10 protein family sorting receptor, exhibited the strongest binding to αSYN fibrils. Notably, the intracellular uptake of fibrillar αSYN was slightly but significantly altered, depending on the expression level of sortilin on the cell surface, and time‐lapse image analyses revealed the concomitant internalization and endosomal sorting of αSYN fibrils and sortilin. Domain deletion in the extracellular portion of sortilin revealed that the ten conserved cysteines (10CC) segment of sortilin was involved in the binding and endocytosis of fibrillar αSYN; importantly, pretreatment with a 10CC domain‐specific antibody significantly hindered αSYN fibril uptake. The presence of sortilin in the core structure of Lewy bodies and glial cytoplasmic inclusions in the brain of synucleinopathy patients was confirmed via immunohistochemistry, and the expression level of sortilin in mesencephalic dopaminergic neurons may be altered with disease progression. These results provide compelling evidence that sortilin acts as an endocytic receptor for pathogenic form of αSYN, and yields important insight for the development of disease‐modifying targets for synucleinopathies.


| INTRODUCTION
Parkinson's disease (PD), the second most common neurodegenerative disease, is characterized pathologically by the preferential loss of dopaminergic neurons in the substantia nigra pars compacta (SNpc) and the appearance of Lewy bodies (LBs) in the remaining neurons. 1 The LBs are composed mainly of hyperphosphorylated, filamentous αsynuclein (αSYN), and point mutations and multiplication of αSYN gene (SNCA) cause familial forms of PD; thus, αSYN is considered a major culprit in the pathogenesis of PD. 2,3 Apart from αSYN pathology in neurons, αSYNpositive cytoplasmic inclusions in oligodendrocytes (glial cytoplasmic inclusions, GCI) are observed in the brain of multiple system atrophy (MSA), and these disorders are collectively termed "synucleinopathy". 4 Mounting evidence suggests that misfolded αSYN exhibits neurotoxicity through multiple cellular cascades, including endoplasmic reticulum stress, oxidative stress, mitochondrial damage, membrane disruption, and the perturbation of vesicular trafficking. [5][6][7][8][9][10][11][12] The majority of neurodegenerative diseases, including PD, have long been considered mechanistically cell-autonomous, i.e., damage within a specific population of neurons per se is sufficient to cause the disease. However, detailed neuropathological observation of a large series of postmortem brain samples has revealed that LBs appear initially in the dorsal nucleus of the vagus nerve and then spread in a caudo-rostral manner, suggesting that pathogenic αSYN species may spread among vulnerable regions to the next through anatomical connections in a prion-like manner. 13,14 This hypothetical concept is supported by autopsy results in the brains of PD patients who received transplants of fetal mesencephalic tissues more than a decade ago; in these brains, αSYNpositive LB-like inclusions were confirmed within donor-derived neurons. 15,16 Furthermore, the noncell-autonomous spread of αSYN aggregates has been verified via neuronal co-culture and animal models, which showed that misfolded αSYN released from neurons can transfer to neighboring cells and convert normal physiological αSYN to toxic misfolded species in a prion-like manner. [17][18][19][20][21][22][23] Following its validation, this "prionoid" hypothesis is now widely accepted as a common phenomenon in the pathogenesis of neurodegenerative diseases involving aberrant protein aggregates. 24 Although the precise mechanisms are not understood, the uptake, release, synthesis, and degradation machineries are known to cooperatively regulate the cellular burden and intercellular transmission of αSYN. 25 The proposed mechanism underlying the mode of αSYN aggregate uptake by cells include endocytosis, macropinocytosis, extracellular vesicles, and direct cellular contact via nanotube tunnel. 24 Several studies have highlighted the importance of the endocytic process because low temperature and the genetic and chemical ablation of dynamin, a key determinant for vesicle scission, markedly decrease the internalization of 18dm0107073; Ministry of Education, Culture, Sports, Science and Technology, Grant/Award Number: 20K07896, 23K06823, 20K07862, 19K16998 and 23K14769 receptor, exhibited the strongest binding to αSYN fibrils. Notably, the intracellular uptake of fibrillar αSYN was slightly but significantly altered, depending on the expression level of sortilin on the cell surface, and time-lapse image analyses revealed the concomitant internalization and endosomal sorting of αSYN fibrils and sortilin. Domain deletion in the extracellular portion of sortilin revealed that the ten conserved cysteines (10CC) segment of sortilin was involved in the binding and endocytosis of fibrillar αSYN; importantly, pretreatment with a 10CC domain-specific antibody significantly hindered αSYN fibril uptake. The presence of sortilin in the core structure of Lewy bodies and glial cytoplasmic inclusions in the brain of synucleinopathy patients was confirmed via immunohistochemistry, and the expression level of sortilin in mesencephalic dopaminergic neurons may be altered with disease progression. These results provide compelling evidence that sortilin acts as an endocytic receptor for pathogenic form of αSYN, and yields important insight for the development of disease-modifying targets for synucleinopathies.

K E Y W O R D S
endocytosis, fibril, glial cytoplasmic inclusion, Lewy body, multiple system atrophy, Parkinson's disease, receptor, sortilin, synucleinopathy, αsynuclein | 3 of 19 ISHIYAMA et al. αSYN in cell cultures and animal models. 17,20,26 Studies have sought to determine whether the endocytic uptake of αSYN is receptor-mediated, because the existence of specific receptors for misfolded αSYN on the cell surface might explain the region-and cell-type-specific neurodegeneration in synucleinopathies. 24,27 Several molecules on the plasma membrane (PM) have been proposed as putative binding partners for αSYN fibrils 24,27 ; however, some of these membrane proteins are preferentially expressed in non-neuronal cells, and it is possible that unknown receptors for misfolded αSYN in neuronal and glial cells remain to be discovered. In the present study, we performed a comprehensive screening for the cell surface receptors of αSYN fibrils using affinity chromatography, mass spectrometry (MS), and a uniquely engineered brain-derived membrane protein library (MPL). 28

| Membrane protein library
Preparation of the MPL from the mouse whole brain was conducted according to method reported previously. 28 Briefly, mouse whole brain tissue (1.5 g) was homogenized in 10 mL of 10 mM sodium phosphate buffer (pH 7.4) containing DNase I (70 U/μL; Takara, Shiga, Japan) and 1 mM dithiothreitol (DTT) using a Potter-Elvehjem grinder (DWK Life Sciences, Millville, NJ, USA). After centrifugation of the homogenate at 2500 rpm for 5 min, the supernatant was collected. The supernatant was then embedded in a 40% (w/v) sucrose solution and ultracentrifuged at 95 000 × g for 1 h (Optima Max™; Beckman Coulter, Brea, CA, USA). The membrane fraction in the middle layer was collected and stored at −80°C. To obtain artificial liposomes, 1.0 g of egg yolk lecithin and 0.2 g of cholesterol were dissolved in chloroform, and vacuumdried at room temperature, and 10 mL of 10 mM Tris-HCl (pH 7.4) was added and stirred vigorously to prepare an artificial liposome suspension (stored at 4°C). To produce a 50 mg equivalent cocktail of the membrane fraction and liposome suspension, 5 mL of 10 mM Tris-HCl (pH 7.4) was added and sonicated at 100 W for 15 min (Insonator™ 201 M sonicator; Kubota, Tokyo, Japan). After three freeze-thaw cycles, the membrane was sonicated again and passed through a polycarbonate membrane with a pore size of 200 nm to yield the MPL.

| Affinity chromatography
Recombinant mouse αSYN (250 μg) monomer and preformed fibrils (PFF) were exchanged with 0.2 M NaHCO 3 /0.5 M NaCl (pH 8.3) after desalting using a gel filtration column (PD MidiTrap™ G-25; Cytiva, Tokyo, Japan). After transferring 0.5 mL of the carrier (NHSactivated Sepharose™ 4 Fast Flow; Cytiva) to an empty column and replacing the solvent with 1 mM HCl, the αSYN monomer and PFF were added, respectively, and the ligands were fixed at 4°C on a carousel overnight. Subsequently, 0.1 M Tris-HCl (pH 8.5) was added to the columns along with a carrier-only control, after which the reaction was blocked for 2 h. On the completion of this reaction, the carriers were washed with 2.5 mL of 0.1 M Tris-HCl (pH 8.5) and 2.5 mL of 0.5 M NaCl/0.1 M acetic acid (pH 4.0) to obtain αSYNimmobilized and control carriers. After centrifugation at 3000 × g for 1 min, the carriers were washed and the membrane proteins bound to the carriers were eluted with 15 μL of elution buffer. Five bottles were prepared for each carrier, yielding 75 μL of eluate in total. The proteins in eluted samples were separated using sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and quantified by silver staining (Silver Stain MS kit, FUJIFILM Wako Chemicals, Osaka, Japan).

| Mass spectrometry
The samples eluted via affinity chromatography were separated using SDS-PAGE, reductively alkylated with DTT and iodoacetamide, and digested with lysyl-endopeptidase and trypsin according to the conventional method. Digests were desalted on solid-phase extraction columns and subjected to liquid chromatography (LC)-MS/MS using an Easy-nLC™ 1200 system (Thermo Fisher Scientific, Waltham, MA, USA) to perform ultrahigh-performance LC with an Orbitrap Fusion Lumos Tribrid™ mass spectrometer (Thermo Fisher Scientific). Each sample was run in triplicate, and proteins that differed from those in the control were quantified and identified using labelfree quantification via the analysis program Proteome Discoverer version 2.2 (RRID:SCR_014477; Thermo Fisher Scientific). In addition, the Sequest HT™ search engine (Thermo Fisher Scientific) was used to search the database with the following parameters: (a) trypsin as an enzyme with up to two missed cleavages; (b) a precursor mass tolerance of 10 ppm; (c) a fragment mass tolerance of 0.6 Da; (d) the carbamidomethylation of cysteine as a static modification; (e) the oxidation of methionine and deamidation of asparagine and glutamine as dynamic modifications; (f) glutamine → pyro-glutamic acid, glutamic acid → pyro-glutamic acid and the carbamylation of the N-terminus as dynamic modifications; and (g) the acetylation of the N-terminus as a dynamic modification. Precursor ion intensities were used for label-free quantitation, and the total peptide amount mode was selected as the normalization mode.

| Primary cortical neuron culture
Primary cultures of rat cortical neurons were prepared according to a previous method, with slight modifications. 30 In brief, The dissociated cortical neurons were plated at a density of 1 × 10 6 cells on a poly-L-lysine-coated glass bottom dish (Cat# D11141H; Matsunami Glass Co., Ltd., Osaka, Japan) and cultured in Neurobasal™ A (Thermo Fisher Scientific) medium supplemented with 2% B27 (Thermo Fisher Scientific), 25 mM glutamate, 18 mM glucose and 0.5 mM L-glutamine. Half of the culture medium was replaced with fresh medium minus glutamate every 3 days. On day 10 after the initiation of culture, cells were used for further experiments.

| Recombinant protein preparation
Mouse and human recombinant αSYN proteins were prepared according to the method published previously. 31 In brief, αSYN cDNA was subcloned into the bacterial expression vector pET23a(+) (RRID:Addgene_13387; Novagen/ Merck, Darmstadt, Germany) and expressed in E. coli BLR(DE3) (Cat# 69450-M; Novagen/Merck KGaA). Cells were suspended in purification buffer [50 mM Tris-HCl (pH 7.5), containing 1 mM ethylenediaminetetraacetic acid (EDTA), 0.1 mM DTT, and 0.1 mM phenylmethylsulfonyl fluoride (PMSF)], disrupted via sonication, and centrifuged at 14000 rpm for 30 min. Streptomycin sulfate (2.5%; FUJIFILM Wako Chemicals) was added to the supernatant to remove nucleic acids. After the removal of nucleic acids via centrifugation, the supernatant was heated to 90°C for 15 min and then centrifuged again. The supernatant was then precipitated by adding solid ammonium sulfate to 70% saturation, after which it was centrifuged, dialyzed overnight, and then applied to a Resource-Q™ column (GE Healthcare, Little Chalfont, UK) with 50 mM Tris-HCl buffer (pH 7.5) containing 0.1 mM DTT and 0.1 mM PMSF as the running buffer. Samples were eluted using a linear gradient of 0-1 M NaCl. The protein concentration of αSYN was estimated using the absorbance at 280 nm with a molar extinction coefficient of 5960 M −1 cm −1 . 32 Fibrillation was achieved by incubating 300 μL of αSYN monomer for 7-10 days with constant stirring at 37°C (1 μg of the original fibrils was added in subsequent fibrillations). αSYN PFF were sonicated into small pieces using an ultrasonic homogenizer (Smurt NR-50 M; Microtec, Urayasu, Japan) prior to use. The concentration of αSYN PFF was estimated after treatment with 6 M guanidine hydrochloride at room temperature for 24 h. Prepared αSYN proteins were evaluated by Coomassie brilliant blue staining and Western blot (WB) analysis. The structure of synthesized αSYN was observed using transmission electron microscopy (TEM). For live cell imaging and immunocytochemistry, αSYN monomer and PFF were fluorescently labeled using an Alexa Fluor-555/647 Microscale Protein Labeling Kit (Cat# A30007 and A30009; Thermo Fisher Scientific).

| Transmission electron microscopy
Recombinant protein (10 μM) was suspended in distilled water with gentle vortex mixing. The suspensions were applied to a 200-mesh formvar-coated copper grid (Nissin EM, Tokyo, Japan) and allowed to stand for 5 min before being negatively stained with 2% uranyl acetate. The ultrastructure of the samples was observed using a TEM system (model H-7650; Hitachi High-Tech, Tokyo, Japan).

| RNA interference
To suppress the expression of endogenous sortilin in cultured cells, small interfering RNAs (siRNAs) specifically targeting sortilin were used (Silencer Select™ pre-designed siRNAs, Cat# s12404, s12406 and s224558 for human sortilin and s136061 for rat sortilin; Thermo Fisher Scientific). Cells were transfected with target-specific or scrambled control (Cat# sc-37 007; Santa Cruz Biotechnology, Dallas, TX, USA) siRNAs (150 pmol for 2 × 10 5 cells) using RNAi max Transfection Reagent (Thermo Fisher Scientific). After 48 h of silencing, the cells were harvested and subjected to further experiments.

| Co-immunoprecipitation
Co-immunoprecipitation (co-IP) was performed according to the standard protocol. Briefly, HEK293 cells stably expressing candidate proteins were exposed to 0.5 μM αSYN monomer or PFF for 30 min. After washing with ice-cold phosphate-buffered saline (PBS) once, the cells were lysed with ice-cold lysis buffer containing 100 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM EDTA, 1% Triton X, and protease inhibitors. Lysates containing 500 μg of protein were incubated for 1 h at 4°C on a carousel with 0.8 μL of anti-αSYN antibody (Cat# ab138507; Abcam, Cambridge, MA, USA) followed by an additional incubation with 40 μL of Pierce protein A magnetic beads (Thermo Fisher Scientific) at 4°C for 1 h. After being washed three times with 0.1% Triton-X in PBS, the protein complexes were eluted with 2× Laemmli buffer (4% SDS, 20% glycerol, 10% 2-mercaptoethanol, 0.004% bromophenol blue and 0.125 M Tris HCl, pH 6.8) at 70°C for 10 min and subjected to WB analysis.

| Subcellular fractionation
For the subcellular fractionation of cultured cells, we used an established protocol with slight modifications. 35 All steps of the fractionation scheme were performed at 0-4°C with ice-cold reagents. Cells (1 × 10 6 ) were resuspended in 2 mL of ice-cold fractionation buffer [10 mM Tris/acetic acid (pH 7.0) and 250 mM sucrose] and homogenized using 20 strokes in a 2-mL Dounce tissue grinder (GPE, Bedfordshire, UK). The cell homogenate was initially cleared using three successive centrifugation steps (2000 × g for 2 min, twice) to remove debris and undestroyed cells. The supernatant was transferred to a new tube and centrifuged at 4000 × g for 2 min to pellet the PM and nuclei. The supernatant was ultracentrifuged at 100000 × g (HIMAC™ CP70MX ultracentrifuge with P50S2 swing rotor; Hitachi Koki Co., Ltd., Tokyo, Japan) for 2 min to pellet the mitochondria, endosomes, and lysosomes (fraction EL). Lysosomes were isolated from the fraction EL via 10 min of osmotic lysis using distilled water at 5 × the pellet volume. After another centrifugation step at 100000 × g for 2 min, late endosome and lysosomes remained in the supernatant, whereas mitochondria and early and late endosomes were in the pellet. The amount of sample loading for WB was adjusted so that the loading controls for each fraction were aligned.

| Live cell imaging
Cells cultured in a glass bottom chamber were incubated with conditioned medium containing Alexa 488-conjugated HA antibody (1:1000, Cat# 2350; MBL, Tokyo, Japan) at 37°C for 30 min. After washing with prewarmed fresh medium, Alexa 555-tagged human αSYN PFF (1 μM) were added to the medium and further incubated at 37°C with 5% CO 2 . Time-lapse images were obtained from 0 to 60 min using a FLUOVIEW™ FV300 confocal laser scanning microscope (Olympus, Tokyo, Japan) equipped with a heat-stable CO 2 chamber system (MMW-T-001; Matsunami, Tokyo, Japan). Captured images were analyzed by image analyzing software Fiji.

| Immunocytochemistry
Cells cultured in a glass bottom chamber were exposed to Alexa 555-tagged αSYN PFF (0.25 μM) for 30 min. The cells were fixed with 4% (w/v) PFA for 20 min, permeabilized with 0.5% Triton X-100 in PBS for 10 min and blocked with 3% normal goat serum (FUJIFILM/Wako Chemicals) in PBS for 30 min. After the blocking step, cells were incubated with primary antibodies (1:2000, anti-sortilin, Cat# 12369-1-AP; Proteintech) for 2 h at room temperature, followed by the incubation with Alexa 488-conjugated anti-rabbit IgG (Cat# A-11008; Thermo Fisher Scientific). Images were captured with a FLUOVIEW FV300 confocal laser scanning microscope (Olympus). Thirteen images were taken at 0.5 μm intervals in the Z-axis direction to create Z-stack images. Using image analysis software Fiji, side-view images obtained by reslicing the specified portion of the Z-stack image were produced, and the signal intensity for each fluorescence was quantified.

| Human subjects and immunohistochemistry
Tissue samples were obtained from the Department of Neuropathology at the Institute of Brain Science, Hirosaki University Graduate School of Medicine, Hirosaki, Japan.

| RNA-seq analysis
RNA-seq analysis was performed using publicly available datasets from previously published paper. 36 In brief, SNpc neurons in the midbrain were collected by laser capture microdissection (LCM) from the postmortem brains of 12 patients with sporadic PD (Braak stage 1-4, aged 69-92) and 17 patients with incidental LB disease (ILBD, Braak stage 1-4, aged 53-94 years). Bulk RNA sequencing (RNAseq) datasets from the LCM samples were mapped onto hg19 human genome using STAR method. 36,37 The Reads Per Kilobase of exon per Million mapped reads (RPKM) and matrix data were obtained from the Gene Expression Omnibus database repository (RRID:SCR_005012, accession number: GSE182622). The log fold RPKM of sortilin gene (SORT1) expression was visualized using ggplot2 (RRID:SCR_014601). The code used for this study is available via GitHub (https://github.com/naosu ge/sortilin).

| Statistical analysis
Statistical analyses were conducted using Graph Pad Prism 9 software (RRID:SCR_002798; GraphPad Software, San Diego, CA, USA). Data were presented as means ± standard errors. Statistical difference was considered significant if p-value was less than .05.

| Screening of fibrillar αSYN receptor from mouse brain
The procedure for MPL preparation using the mouse whole brain is illustrated in Figure 1A. First, membrane proteins extracted from brain homogenate were mixed with liposome suspension from egg-yolk lecithin to induce membrane fusion, yielding artificial multilayer proteoliposomes in which brain-derived membrane proteins were rearranged. The MPL composed of monolayer liposomes was then prepared using repeated freeze-thaw cycles, sonication, and size-exclusion chromatography, and the membrane proteins in the MPL bound to αSYN monomer or PFF were detected using affinity chromatography and MS. The biochemical and structural nature of recombinant mouse αSYN used in the present study are shown in Figure 1B. To eliminate the high molecular weight (HMW) αSYN species, recombinant αSYN was further separated using a 100 kDa pore-size filter, and low molecular weight αSYN migrated at approximately 17-18 kDa on SDS-PAGE gel was collected (designated as "monomer" in Figure 1B). In contrast, the αSYN PFF formed by agitation showed SDS-insoluble HMW smear of up to 210 kDa. Representable TEM images of recombinant mouse αSYN revealed the presence of the filamentous structure of PFF ( Figure 1C). The morphology of recombinant human αSYN was similar to that of mouse αSYN (data not shown). Notably, after shearing via sonication, αSYN PFF showed short fragments of the original molecular architecture of the fibril. After separating the brain-derived membrane proteins bound to αSYN monomer or sonicated PFF (hereafter designated as "PFF") on 1-D SDS-PAGE gel ( Figure 1D, silver staining), digesting the proteins in-gel, extracting the peptides, and analyzing these using LC-MS/MS, we identified 3481 distinct proteins bound to αSYN (Table S1). After narrowing down the list of candidate molecules based on the number of binding peptides (>5), the affinity ratio of PFF versus monomer αSYN (>1.2), and localization in the PM, the following five cell surface proteins remained: sortilin, CNTNAP2, NTM, LSAMP, and DIP2A ( Figure 1E).

| Sortilin affects the surface binding and uptake of αSYN fibrils
For the second screening, HEK293 cells stably expressing the human homologs of each candidate protein identified via LC-MS/MS were cultured in conditioned medium containing either human αSYN monomer or PFF, and then subjected to co-IP analysis. Among the five candidates, sortilin was most efficiently co-immunoprecipitated by exogenously added αSYN with stronger binding to αSYN PFF than the monomer form ( Figure 2A). Structurally, sortilin is a single-pass, type I membrane protein, and the extracellular portion of sortilin (sort-ecto; usually called as the VPS10P domain) consists of a ten-bladed βpropeller region and 10CC domain ( Figure 2B). 38 Based on this structure, we attempted to verify the binding site using clones lacking each extracellular domain. Extracellular αSYN PFF bound robustly with 10CC regions, but not with the βpropeller domain ( Figure 2C). We also examined how the expression of sortilin on the cell surface affects the binding and uptake of αSYN using HEK293 cells exposed to αSYN monomer or PFF (0.5 μM for 1 h). The semi-quantitative densitometric analysis of WB data revealed that the overexpression of the sort-ecto domain on the surface of cells significantly increased the cellular levels of αSYN (both the monomer and insoluble HMW smear) in cells treated with αSYN PFF ( Figure 3A). Although there was no significant difference in the level of αSYN monomer, a slight increasing trend was observed in sort-ecto-overexpressing cells. The immunocytochemistry revealed that significantly more sortilin/αSYN double-positive puncta were observed in HA-tagged sortecto-expressing cells exposed to αSYN PFF than in those exposed to monomer ( Figure 3B). On the other hand, the silencing of endogenous sortilin slightly but significantly reduced the uptake of αSYN PFF ( Figure 3C). The effect of sortilin silencing on the uptake of αSYN monomer was repeatedly examined, but no significant changes were obtained (data not shown). We further examined the reproducibility of the above results using neuronal cells. In rat dopaminergic PC12 cells, a mild but significant decrease in αSYN PFF uptake was observed under sortilin silencing in both WB and immunostaining ( Figure S1A,B). Since WB could not detect endogenous αSYN in either scrambled or sortilin siRNA-transfected cells without αSYN PFF exposure, it is unlikely that silencing of sortilin itself affected endogenous αSYN expression in αSYN PFF-treated cells. Similarly, in primary rat cortical neurons, silencing of endogenous sortilin slightly but significantly reduced the uptake of αSYN PFF ( Figure S2). Furthermore, we attempted to determine which part of the extracellular domain of sortilin contributes most to the surface binding and uptake of αSYN. As expected, cells overexpressing the 10CC domain of sortilin, a putative binding site for extracellular αSYN, exhibited significantly increased uptake of αSYN PFF ( Figure 3D). Overexpression of the 10CC domain slightly increased the uptake of αSYN monomers, but repeated experiments failed to show a significant difference. The overexpression of the βpropeller domain had no significant effect on αSYN uptake ( Figure 3E).
Together, these results suggest that the surface expression of sortilin, specifically its 10CC domain, contributes to the surface binding and uptake of αSYN PFF.
Given that the extracellular 10CC domain of sortilin contributes to the binding and internalization of fibrillar αSYN, we postulated that the preoccupation of the binding site with a target specific antibody could mitigate the binding and subsequent uptake of αSYN fibrils. To test this hypothesis, we treated HEK293 cells expressing the sort-ecto domain on the cell surface with rabbit monoclonal antibody (termed "10CC Ab") that specifically bound to the amino-terminal portion of the 10CC domain of human sortilin (GVNPVREVKDLKKK: residues 726-739) ( Figure 3F). Compared with cells mock-treated with normal rabbit immunoglobulin, cells pretreated with 10CC Ab (100 μg/mL, 30 min) exhibited a significantly reduced amount of incorporated αSYN PFF. Repeated experiments showed that the 10CC Ab tended to slightly lower the uptake of monomeric αSYN, but did not show a significant difference ( Figure 3G).

| Sortilin regulates the endocytic uptake of αSYN fibrils
As the name implies, sortilin is involved in the endocytosis and intracellular sorting of various ligands. 38 Given the notable effect of sortilin expression on the surface binding and cellular levels of αSYN PFF, sortilin is likely to modulate the endocytosis and subsequent intracellular sorting of αSYN fibrils. To confirm this, the uptake and subcellular trafficking process of αSYN and sortilin were chronologically monitored under a confocal microscope using HEK 293 cells, which were used because neurons are less efficient for transfection and their cytoplasm is too small to observe intracellular vesicular trafficking. Specifically, cells expressing Alexa 488-labeled HA-sortilin (i.e., sort-ecto domain) were exposed to conditioned medium containing Alexa 555-labeled αSYN PFF and observed under a confocal microscope. As shown in Figure 4A and Video S1, αSYN PFF and sortilin were concomitantly internalized from the cell surface to the intracellular space in a timedependent manner. To evaluate the subcellular trafficking of αSYN PFF and sortilin in detail, we prepared HEK293 cells stably expressing enhanced green fluorescent protein (EGFP)-tagged Rab5A (early endosome marker) or Rab7A (late endosome and lysosome marker) together with HAtagged sortilin. Because the intracellular cytosolic tail of sortilin influences its subcellular trafficking, 38 full-length sortilin was used in this experiment. HA-sortilin was labeled with Alexa 555-conjugated HA antibody for 30 min. Subsequently, cells were exposed to Alexa 647-labeled αSYN PFF and subjected to time-lapse imaging. After a few minutes of observation, some puncta of the internalized αSYN PFF and sortilin were co-localized with Rab5A and Rab7A ( Figure 4B and Videos S2 and S3). In agreement with the results of microscopic imaging, subcellular fractionation revealed that the overexpression of sortilin upregulated the level of insoluble HMW αSYN, mainly in the endolysosomal fractions ( Figure 5).

| Exogenous αSYN fibril is found to co-localize with endogenous αSYNpositive aggregates along with sortilin
In the paradigm of the prion-like hypothesis, exogenous misfolded protein are supposed to convert endogenous normal protein to abnormal forms by seed-templating mechanism. Given that cell surface sortilin functions as an endocytic receptor for misfolded αSYN, one could imagine that sortilin might be involved in the aggregation process of endogenous αSYN using exogenously incorporated pathogenic αSYN as a template. To confirm this, HEK293 cells stably expressing mCherry-tagged αSYN were transiently transfected with plasmid expressing HA-sortilin, and sortilin on the cell surface was fluorescently labeled with Alexa488-conjugated anti-HA antibody. Then, the cells were exposed to Alexa647labeled αSYN PFF for 2 h, fixed, and observed under a confocal laser microscope ( Figure S3). To clearly show the co-localization of each molecule in three dimensions, sideview images were obtained by reslicing the Z-stack images. Intriguingly, some of the extracellularly incorporated αSYN PFF were detected as dot-like structures together with endogenous mCherry-tagged αSYN and sortilin. This result indicates that some of the internalized αSYN PFF together with sortilin might engulf endogenous mCherry αSYN to form aggregates.

| Sortilin is colocalized with phospho-αSYN in LB and GCI
αSYNpositive LBs are known to colocalize with numerous endosome-resident proteins including Rab7A, VPS4, and charged multivesicular body protein 2B. 31,39,40 Furthermore, a pale body, thought to be the precursor of LB, also contains the endolysosomes and vacuolar structures, 4 suggesting that the sortilin-mediated endocytic process might be involved in the biogenesis of LBs in vivo. To confirm this, we performed immunohistochemistry using brain samples obtained in the autopsies of patients with PD and DLB. Intriguingly, both brainstem and cortical LBs exhibited strong immunoreactivity toward sortilin ( Figure 6A). Confocal microscopic images of double immunofluorescent staining showed similar distribution of phosphorylated αSYN and sortilin in cortical LB, whereas the former was distributed in halo and the latter was more centrally localized in brainstem LB. Since sortilin is highly expressed in oligodendrocytes in human brain (www.prote inatl as.org), we also examined whether sortilin is detected in GCI in MSA brain ( Figure 6B). As expected, sortilin was coimmunostained with phospho-αSYN in GCI, although staining was weaker than in LB.

| Sortilin expression in midbrain dopaminergic neurons is altered by Braak stage
According to the Braak's pathological staging, LB pathology in PD is thought to progress from the lower brainstem toward the rostral region. 13 Intriguingly, transcriptome analyses of postmortem brain tissue samples obtained by LCM have revealed spatiotemporal changes in the expression levels of various genes associated with PD. 36,41 Furthermore, recent bioinformatics analysis using publicly available Gene Expression Omnibus database, comparing differentially expressed genes in PD patients and normal controls, revealed that sortilin is upregulated in the substantia nigra of PD and serves as a hub molecule in the protein-protein interaction network. 42 Given this situation, we used a publicly available RNA-seq dataset of neuromelanin-containing midbrain dopaminergic neurons obtained from the brain samples of PD and ILBD to evaluate the expression profile of sortilin during disease progression ( Figure 6C,D). 36 Interestingly, the expression level of sortilin in the midbrain SN tended to be higher in Braak stage 3, when LB pathology reached the midbrain, than in earlier (stage 1 and 2) and later (stage 4) stages ( Figure 6E).

| DISCUSSION
The isolation and characterization of membrane proteins is always challenging. Indeed, the extraction and solubilization of highly hydrophobic membrane proteins require detergents, but the poor compatibility between detergents and MS analysis is a major obstacle to achieve success. Importantly, the MPL used in the present study does not require detergent treatment and can maintain the structural characteristics and ligand-binding ability of the membrane proteins; thus it enables a more efficient and comprehensive search of putative αSYN receptors. 28 Using protein library screening and subsequent validation experiments, we found that sortilin, a membrane protein highly enriched in the nervous tissue, serves as the endocytic receptor for αSYN fibrils. The incorporation of αSYN PFF was significantly dependent on the surface expression of sortilin, especially the membraneproximal 10CC domain. In support of this finding, αSYN PFF bound to the 10CC domain of sortilin and were transported to endosomal compartments. Given that some parts of internalized αSYN PFF did not colocalize with sortilin, and that sortilin silencing had a significant but minor effect of reducing αSYN uptake in PC12 cells and primary cultured neurons, it is likely that sortilinindependent mechanisms also contribute to the uptake of αSYN fibrils. Furthermore, since sortilin also binds to monomeric αSYN, although not to the same extent as fibrillated αSYN, and sortilin overexpression and silencing slightly affect the uptake of monomeric αSYN, we cannot exclude the possibility that αSYN monomer is also recognized as a ligand for sortilin. Nevertheless, abovementioned results together with the slight but significant reduction in αSYN PFF uptake in the presence of 10CC domain-specific antibody implies that sortilin does influence the internalization of pathogenic form of αSYN. Given that SNCA duplication, i.e., only a 1.5-fold elevation of αSYN expression, is sufficient to cause full-blown | 11 of 19 ISHIYAMA et al. PD, 43 even small changes in αSYN expression may contribute to disease progression over several decades. It is also interesting to note that sortilin was detected in the LB and GCI of the autopsied brain of the patients with synucleinopathy, and the RNA-seq analysis revealed that the expression level of sortilin in midbrain dopaminergic neurons might be altered by consecutive Braak's pathological stages. F I G U R E 2 Sortilin strongly binds to αSYN fibrils via its 10CC domain. (A) HEK293 cells stably expressing each candidate were exposed to either αSYN monomer or PFF for 30 min, and the interactions were verified via co-IP. Among the five candidates, sortilin (ectodomain) is most efficiently co-immunoprecipitated by exogenously added αSYN with stronger binding to αSYN PFF than to the monomer. (B) Schematic illustration of the deletion clones lacking each extracellular domain of sortilin. (C) Results of co-IP analyses using clones lacking each extracellular domain of sortilin (other panels). Note that extracellular αSYN PFF binds strongly with the 10CC region, but not with the βpropeller domain. IP, immunoprecipitation; αSYN, αsynuclein; PFF, preformed fibrils; CNTNAP2, contactin associated protein 2; NTM, neurotrimin; LSAMP, limbic system associated membrane protein; DIP2A, disco-interacting protein 2 homolog A; VPS10P, vacuolar protein sorting 10 protein; 10CC, ten conserved cysteines.
Various mechanisms have been proposed for the cellular uptake of αSYN, e.g., dynamin-mediated endocytosis, macropinocytosis, extracellular vesicles including exosomes, membrane penetration, and tunneling nanotubes. 20,31,40,[44][45][46] The αSYN burden in the central nervous system (CNS) is thought to be balanced by this secretion, uptake, and intracellular transport machinery. 25 However, if the equilibrium state is disrupted, the intracellular accumulation and intercellular propagation of abnormal proteins may be triggered. Another important question pertain to whether the endocytic uptake of αSYN occurs in a receptor-dependent manner. 24,27 Several cell surfaceresident proteins, e.g., lymphocyte-activation gene 3, Na + / K + -ATPase, prion protein, heparan sulfate, Fcγ receptor IIb, toll-like receptor 2/4, connexin-32, CD36 scavenger receptor, and low-density lipoprotein receptor-related protein 1, have been reported as putative receptors for αSYN fibrils [47][48][49][50][51][52][53][54][55][56][57] ; however, some of these proteins are preferentially expressed in non-neuronal cells and to a lesser extent in neurons. 58 In contrast, immunohistochemistry in human postmortem brains has revealed that sortilin is highly expressed in neurons but rarely found in astrocytes and microglial cells. 59 Notably, the expression of sortilin is particularly high in dopaminergic neurons in the SNpc, cholinergic neurons in the nucleus basalis Meynert, and noradrenergic neurons in the locus coeruleus, in which LB formation and neuronal loss are prominent in PD brain. 2,59 In addition to the abundant expression of sortilin in the vulnerable regions in PD, the presence of sortilin in LB and GCI, and the altered expression of sortilin in the midbrain dopaminergic neurons during disease progression collectively suggest that sortilin-mediated cargo trafficking along the endocytic pathway may be involved in the pathological process of synucleinopathy in vivo.
Our findings that sortilin functions as a novel endocytic receptor for αSYN fibrils and influences cellular αSYN dynamics are worthy of discussion. In the CNS, sortilin acts as a receptor of neurotrophic factors and neuropeptides, but also as a co-receptor to cytokine receptors, tyrosine receptor kinases, and G-protein coupled receptors. 38 On the other hand, sortilin, particularly its ectodomain, also serves as the endocytic receptor of several neurodegeneration-related proteins and regulates their subcellular trafficking. 38 For example, progranulin, a key molecule in frontotemporal dementia (FTD), binds to the βpropeller region of sortilin, which is followed by endocytosis and translocation to lysosomes. 60 Progranulin plays crucial roles in lysosomal function, neuronal repair, stress response and innate immunity, 61 and its dysfunction has been implicated in a variety of neurodegenerative diseases, including amyotrophic lateral sclerosis (ALS), Alzheimer's disease and PD, in addition to FTD. 62 Sortilin antibody has been shown to increase progranulin levels by inhibiting its lysosomal degradation in cultured cells, animal models and humans, 63,64 and clinical trials are currently underway to test the efficacy of the antibody in patients with FTD due to progranulin gene (GRN) mutation and ALS with C9orf72 mutation. 62 While still speculative, sortilin antibody therapy may have a synergistic effect by increasing progranulin and inhibiting pathological αSYN uptake, thereby mitigating the progression of PD and other synucleinopathy. In addition, both normal and infectious scrapie prion proteins interact with the 10CC domain of sortilin and are transported to lysosomes for F I G U R E 3 Sortilin expression on the cell surface influences the uptake of αSYN fibrils. (A) Effect of the surface expression of sortilin on the binding and uptake of αSYN. Semi-quantitative densitometry reveals that the overexpression of the sort-ecto domain significantly increases the levels of αSYN PFF (both the monomer and insoluble HMW smear) in HEK293 cells treated with αSYN PFF. Although there is no significant difference in the level of αSYN monomer, a slight increasing trend is observed in sort-ecto-overexpressing cells. (B) Immunocytochemistry shows that more sortilin/αSYN double-positive puncta (i.e., colocalized pixels visualized by white dots) are found in cells exposed to αSYN PFF than in those exposed to monomer. Scale bar: 10 μm. (C) Silencing of endogenous sortilin significantly increases the levels of αSYN PFF (both the monomer and insoluble HMW smear) in cells treated with αSYN PFF. The effect of sortilin silencing on the uptake of αSYN monomer was repeatedly examined, but no significant changes were obtained (data not shown). (D) Cells overexpressing the 10CC domain of sortilin, a binding site for αSYN, exhibits the significant upregulation of cellular αSYN level in cells treated with αSYN PFF. Overexpression of the 10CC domain slightly increases the uptake of αSYN monomers, but does not show a significant difference. (E) Overexpression of the βpropeller domain does not affect αSYN uptake. (F) An antibody targeting the 10CC domain of sortilin reduces the uptake of αSYN fibrils. Schematic illustration of the epitope (GVNPVREVKDLKKK: residues 726-739 in human sortilin) for the 10CC domain-specific, rabbit monoclonal antibody (10CC Ab). In HEK293 cells expressing the sort-ecto domain on the cell surface, pretreatment with 10CC Ab significantly reduces the level of incorporated αSYN PFF. Repeated experiments show that the 10CC Ab tends to slightly lower the uptake of monomeric αSYN, but does not show a significant difference. *p < .05 and **p < .01 (n = 4-8) according to a Mann-Whitney U test (A, D, E and G) or one-way ANOVA (C). **p < .01 (n = 12) according to a Spearman's rank correlation and Mann-Whitney U test for the quantitative assessment of the colocalization of sortilin and αSYN (confocal image in A). PFF, preformed fibrils; αSYN, αsynuclein; siRNA, small interfering RNA; 10CC, ten conserved cysteines; SP, signal peptides; 10CC, ten conserved cysteines; TM, transmembrane domain; CT, cytoplasmic tail; aa, amino acids; Ab, antibody degradation. 65 Because the 10CC domain contributes to the stabilization of the βpropeller region, 66 it is possible that the binding of prion and αSYN to the 10CC domain modulates the higher-order structure of sortilin, which in turn influences the binding ability of other ligands. In addition to its extracellular domain, the intracellular region of sortilin may be involved in the neurodegenerative process. For instance, the cytoplasmic tail of sortilin binds to amyloid precursor protein (APP) and regulate the lysosomal degradation of APP. 67 Furthermore, the retrograde trafficking of βsite APP-cleaving enzyme 1, a rate-limiting enzyme in the production of Amyloid β, is uniquely regulated by the cytoplasmic domain of sortilin. 68 Moreover, the C-terminal fragments of sortilin can be deposited as senile plaque-like lesions. 69 Hypothetically, sortilin may therefore be involved in the intracellular trafficking and regulatory expression of neurodegenerative-related proteins, serving as a common receptor for these proteins.
We acknowledge that the present study has several limitations. First, the αSYN PFF used in this study are rather artificial, and it remains uncertain whether sortilin would also affect the uptake of more physiological/pathological αSYN species, such as those secreted from neurons or pathological fibrils derived from autopsied brains of synucleinopathy patients. 70,71 In addition, recent studies using cryo-electron microscopy have shown that distinct molecular conformers of assembled αSYN exist in samples from patients with different types of synucleinopathy. [72][73][74] Therefore, it is possible that each fibrillar αSYN strain has a different affinity for cell surface receptor. These points warrant further investigation. Second, post-endocytic fate of sortilin and αSYN has not been fully elucidated. Sortilin has a complex cellular trafficking itinerary, in which it functions as a receptor in the TGN, endosomes, secretory vesicles, multivesicular bodies, and the cell surface. 38 Indeed, sortilin is thought to negatively regulate the aggregation of prion via lysosomal degradation, 65 and decrease the extracellular secretion of progranulin. 60 Thus, the possible role of sortilin in αSYN degradation and secretion at later time points is worthy of exploration in future study. Finally, the majority of our results were obtained using in vitro cellular models; thus, it will be necessary to conduct in vivo experiment in animal models to validate the present findings.
In summary, this study revealed the novel role played by sortilin in the endocytic uptake and subcellular transport F I G U R E 5 Sortilin overexpression markedly upregulated the level of insoluble HMW αSYN mainly in the endolysosomal fractions in αSYN PFF exposed cells: Subcellular fractionation using αSYN PFF exposed HEK293 cells indicated that the overexpression of sortilin markedly upregulated the level of insoluble HMW αSYN mainly in the endolysosomal fractions. αSYN, αsynuclein; PFF, preformed fibrils; EGFP, enhanced green fluorescent protein; HSP90, heat shock protein 90. F I G U R E 4 Sortilin regulates the endocytic uptake of αSYN fibrils. (A) Time-lapse images showing the uptake of αSYN PFF and sortilin under a confocal microscope. Both Alexa 555-labeled αSYN PFF (red) and Alexa 488-labeled HA-sortilin (i.e., sort-ecto domain; green) are concomitantly internalized from the cell surface to the intracellular space in a time-dependent manner (white arrowheads, see also Video S1). The right upper panel shows semi-quantitative measurements of red and green signal intensities in the specified section (between a and b) after 28 min of observation. Scale bar: 5 μm. (B) To visualize the subcellular trafficking of αSYN PFF and sortilin in detail, HEK293 cells stably expressing EGFP (pseudocolored yellow)-tagged Rab5A (EE marker) or Rab7A (late endosome and lysosome marker) together with the Alexa 555-labeled HA-sortilin (pseudocolored magenta) were used. These cells were exposed to Alexa 647-labeled αSYN PFF (pseudocolored cyan) and were subjected to time-lapse imaging. Notably, after a few minutes of observation, some puncta of the internalized αSYN PFF and sortilin are co-localized with Rab5A and Rab7A positive structures (white arrowheads, also see Videos S2 and S3, respectively). Below panels are the semi-quantitative values of yellow, magenta and cyan signal intensity in the specified section (between a and b) at 17.75 min (Rab5A) and 20.75 min (Rab7A) after the observation. Scale bars: 2 μm.
of αSYN fibrils. Although the detailed mechanism underlying cell-type specific vulnerability in synucleinopathy remains to be elucidated, the present findings deepen our understanding of the biological processes that lead to the topographic progression of αSYN pathology in the patient brain and provide clues for novel disease-modifying therapy. Nevertheless, further studies are needed to fully elucidate the underlying pathomechanisms and to accurately identify molecular targets for therapeutic intervention in PD and other synucleinopathy. F I G U R E 6 Sortilin is colocalized with phosphorylated αSYN in LB and GCI: (A) DAB-stained immunohistochemical image using autopsied brain samples from patients with PD and DLB. Core structures of the brainstem and cortical LBs show strong immunoreactivity toward a sortilin antibody (black arrow). Confocal microscopic images show similar distribution of Ser129-phosphorylated αSYN and sortilin in cortical LB, whereas phospho-αSYN is distributed in halo and sortilin is more centrally localized in brainstem LB (white arrow). Scale bars: 10 μm. (B) Sortilin is coimmunostained with phospho-αSYN in GCI. DAB-immunohistochemical image (black arrowhead) and confocal immunostaining images are shown (white arrowhead). Scale bars: 10 μm. Sortilin expression in midbrain dopaminergic neurons is altered by Braak stage: (C) Schematic illustration of experimental design. Midbrain sections from PD and ILBD were used to isolate pigmented dopaminergic neurons by LCM and subjected to RNA-seq analysis. (D) List of patients who provided autopsied brain samples in this study. (E) Changes in expression levels of sortilin in midbrain SN at different stages of disease. Note that the expression level of sortilin tends to be higher in Braak stage 3, when LB pathology reaches the midbrain, than in earlier (stage 1 and 2) and later (stage 4) stages. *p < .05 according to One-way ANOVA with post-hoc Tukey's multiple comparison test. DAB, diaminobenzidine; LB, Lewy body; GCI, glial cytoplasmic inclusion; LCM, laser capture microdissection.