Introduction
Gradual accumulation of toxic amyloid-beta (Aβ) and tau are believed to be central to Alzheimer´s disease (AD) pathogenesis. These abnormal deposits typically appear with a hierarchical spatial distribution, which suggest that the pathological proteins can propagate between different brain areas [
7,
52,
53]. Accordingly, recent studies have demonstrated transfer of Aβ in cellular and animal models [
27,
38‐
40,
61] as well as spread of tau between neurons in the brain of transgenic mouse models [
25]. These observations are further supported by the distribution of Aß and tau PET ligands in relation to the brain connectome [
45].
Different Aβ conformational states have different properties, and intermediate products of fibril formation, such as lower molecular weight Aβ oligomers (oAβ) and protofibrils, have been suggested to be particularly neurotoxic and to act as seeds for further aggregation [
27,
61]. In addition, these soluble forms of Aß correlate better than fibrils with cognitive function in the AD brain [
39]. We have previously shown that oAβ can accumulate inside cells and subsequently spread from one cell to another [
15,
40]. These findings thus suggest that Aβ can propagate pathology in a manner similar to what is seen for prion disorders and for several other neurodegenerative diseases [
8]. However, the underlying mechanisms for the spreading of toxic oAβ remain incompletely understood [
28].
Exosomes, small extracellular vesicles (20–120 nm in diameter) [
58] developing from endosomes through multivesicular bodies, have recently emerged as key players in cellular communication and transport of molecules in both health and disease, including neuronal toxicity [
33] and neurodegenerative disorders [
59]. Exosomes can carry different cargos such as proteins, RNA and miRNA and can also contain monomeric Aß, tau and α-synuclein [
12,
17,
46,
59] and can propagate tau pathology [
4]. Thus, exosomes could potentially also carry aggregated proteins such as oAß. However, no study has so far investigated if exosomes isolated from AD brain tissue can be responsible for interneuronal protein transfer.
For the first time, we now demonstrate that intracellular oAβ is co-localized with exosomes and show that AD brain-derived exosomes can mediate neuron-to-neuron propagation of oAß. Furthermore, we show that the concentration of oAβ in exosomes is significantly increased in post mortem AD brains. In addition, exosomes carrying oAβ can be internalized in cultured neurons and spread their toxic content to nearby cells.
Methods
Brain tissues
Post mortem brain samples of temporal neocortex from healthy control (1 female, 4 male) and AD (4 female, 1 male) subjects (Table
1) were provided by the brain bank at Uppsala University, Sweden. The tissue was received as fresh-frozen or as formalin‐fixed (4% formaldehyde) paraffin-embedded blocks. The AD cases were neuropathologically diagnosed as CERAD C, Braak stages IV–VI. None of the control patients suffered from dementia or any neurodegenerative disorder. The collection and use of post mortem brain tissue was approved by the Regional Ethical Committee in Uppsala, Sweden (2005/103, 2005-06-29; 2009/089; 2009-04-22).
Table 1
Demographic and clinical characteristics of post mortem cases used in the study
AD | 61 | F | 48 | 5–6 | 5 | C | High prob | None |
AD | 64 | M | 12 | 5–6 | 5 | C | High prob | None |
AD | 85 | F | 21 | 5–6 | 3 | B | High prob | None |
AD | 90 | F | 11 | 5 | 3 | B | | None |
AD | 63 | F | 48 | 6 | 5 | C | | None |
C | 88 | M | 39 | 0 | 0 | 0 | | None |
C | 88 | F | 22 | 2 | 0 | 0 | | In S Nigra |
C | 63 | M | 30 | 0 | 0 | 0 | | None |
C | 90 | M | 30 | 0 | 1 | 0 | | None |
C | 91 | M | 27 | 3 | 4 | C | | None |
Immunohistochemistry and immunofluorescence of brain sections
Formalin-fixed, paraffin-embedded 10 μm sections of temporal neocortex from healthy control and AD brains (Table
1) were used for this study and pretreated as previously described [
6]. After 5 min blocking of endogenous peroxidase by incubation in Background Sniper (Biocare Medical), slides were washed in Tris-buffered saline (TBS) solution and incubated with rabbit polyclonal anti-flotillin-1 (Abcam, 1:200) followed by incubation with either the second primary mouse monoclonal antibody (mAb158, 1:7500, BioArctic) or mouse monoclonal antibody 82E1 (1:25, IBL International) for 30 min at room temperature (RT).
After rinsing in TBS, the slides were incubated for 30 min with MACH 2 Double Stain (Biocare Medical). Following this step, a 30 min incubation was performed with an AP linked chromogen IP Warp red/HRP linked chromogen Vina Green cocktail (Biocare Medical). After rinsing in deionized water, the sections were counterstained with Mayers haematoxylin (Histolab Products AB) and mounted with Pertex mounting media (Histolab Products AB) and micrographs were obtained using 100 × oil immersion objective (Nikon Eclipse 80i, Digital Sight DS-Fi1).
After blocking and incubating with primary antibodies as described above, the slides were rinsed in TBS and incubated with a Fluorescence Enhancement Probe Mouse/Fluorescence Enhancement Probe Rabbit cocktail (Biocare Medical) for 20 min subsequent to a 40 min incubation with a Goat-anti-Mouse DyLight 549/Goat Anti-Rabbit DyLight 488 cocktail (Biocare Medical) diluted 1:200, respectively. After rinsing in deionized water, the sections were nuclear counterstained with DAPI and mounted with Fluoro Care Anti-Fade Mountant (Biocare Medical) and analysed with a Zeiss LSM 700 confocal microscope. The differential interface contrast (DIC) mode and 405, 488, 555 and 639 lasers were used to acquire the images with 63x/1.40 oil immersion plan-apochromatic DICII objectives. The micrographs were processed using Huygens (Scientific Volume Imaging) and ZEN lite (blue edition) software.
Cell lines and differentiation
Two different cultured cell types were used in the study: the human-induced pluripotent stem cells, AF22 (from a control subject) and the human neuroblastoma cell line, SH-SY5Y (ECACC: Sigma-Aldrich). The neuroepithelial stem cell line, AF22, derived from human-induced pluripotent stem cells from human skin fibroblasts was provided by Dr. Anna Falk, Karolinska Institute, Sweden. The process of reprogramming human cells was approved by the Ethical Committee at Karolinska Institute, Sweden (dnr 2012/208-31/3 with addendum 2012/856-32). All samples were given with informed consent. The AF22 cell line has previously been shown to have a stable neuronal differentiation competence and the capacity to generate functionally mature human neurons [
20], denoted hiPSC. The hiPSCs were cultured on 0.01% poly-
l-ornithine and laminin (10 µg/mL, Sigma-Aldrich) coated cell culture flasks (Corning) in DMEM/F12 media (Gibco by Life Technologies), supplemented with EGF (10 ng/mL, PeproTech), FGF2 (10 ng/mL, PeproTech), N2 (5 µl/ml, Life Technologies) and B27 (1 ml/L, Life Technologies) and further differentiated for 40 days to functionally mature human neurons in 1:1 DMEM, neurobasal media containing B27 (10 µl/ml), Laminin (1 µl/ml) and N2 (5 µl/ml) as previously described [
20].
Neuronal differentiation of the human neuroblastoma cell line SH-SY5Y was performed as previously described [
1]. In brief, SH-SY5Y cells were cultured and pre-differentiated for 7 days using 10 μM retinoic acid (RA; Sigma-Aldrich; denoted as raSH-SY5Y). Pre-differentiated raSH-SY5Y cells were seeded on 6-, 12- or 24-well glass plates coated with 20% extracellular matrix (ECM) gel (BD Bioscience) and further differentiated for 10 days with serum-free MEM (Gibco by Life Technologies) supplemented with brain-derived neurotrophic factor (BDNF, 50 ng/ml, PeproTech), neuregulin β1 (NRGβ1,10 ng/ml, R&D Systems), nerve growth factor (NGF, 10 ng/ml, R&D Systems) and vitamin D3 (VitD3, 24 nM, Sigma-Aldrich). These fully differentiated cells are denoted dSH-SY5Y.
In addition, SH-SY5Y cells expressing a CD63-EGFP fusion protein were generated using AddGene plasmid #62964.
Labelling and oligomerization of Aβ1-42
Recombinant Aβ1-42 peptides (Innovagen,) was dissolved in 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP, Sigma-Aldrich) and vacuum dried overnight. Aβ1-42 (1.054 mM final concentration) was resuspended in Na2CO3 (0.1 M pH 8.5) and incubated with the fluorophore Alexa Fluor 700 (AF700) succinimidyl ester (1.58 mM final concentration, Life Technologies) for 40 min at 4 °C or the fluorophore 6-carboxytetramethylrhodamine succinimidyl ester (TMR, Invitrogen), in a molar ratio of 2:, incubated overnight at 4 °C. Labelled Aβ1-42 was diluted to a final concentration of 100 µM in HEPES 20 mM pH 7.4, vortexed, sonicated for 10 min and incubated at 4 °C overnight. After the overnight incubation, Aβ1-42 was separated from free dye with size exclusion chromatography (SEC). A Sephadex 75 10/300 GL column coupled to a liquid chromatography system (ÄKTA pure, GE Healthcare) was equilibrated with NH4HCO3 50 mM pH 8.5 and 500 µl of sample was injected into the column. To estimate the molecular weight of the Aβ species, LMW gel filtration calibration kits (GE Healthcare) were used. Oligomeric and monomeric Aβ species were eluted at a flow rate of 0.5 ml/min, collected and lyophilized. Then, Aβ species (oAβ-AF700) were resuspended in phosphate-buffered saline (PBS) solution and quantified spectrophotometrically at 215 nm by using the Aβ1-42 extinction coefficient (Aβ1-42 ɛ214 nm = 75,887 M−1 cm−1) according to Lambert–Beer’s law. Protein aliquots were stored at − 80 °C.
Exosome purification, characterization and labelling
Isolation of brain exosomes from extracellular space of freshly frozen human brain tissues (250 mg) was performed as previously described [
42]. Tissue was dissociated with papain (20 units/ml, 15 min at 37 °C, Sigma-Aldrich) followed by filtration through 40 µm mesh filter (BD Biosciences) and a 0.2 µm syringe filter (Thermo Scientific) to separate extracellular matrix from cells. The crude exosomes were then isolated by differential centrifugation method and subsequently purified by sucrose density gradient as previously described and resuspended in PBS, lysis buffer or diluent C (Sigma-Aldrich) for further experiments [
42].
Exosomes from conditioned media of raSH-SY5Y cells were isolated by differential ultracentrifugation. In brief, 50–80 million raSH-SY5Y cells were incubated with oAβ-AF700for 3 h at 37 °C. After PBS washing, cells were kept for 48 h in MEM supplemented with exosome-free serum (System Biosciences). Culture supernatants were collected and spun at 1000 × g for 10 min for removal of cellular debris. The supernatants were then filtered through 0.22 µM filter and sequentially centrifuged at 5000×g, 10,000×g, and 100,000×g. The final pellet was then resuspended in PBS, lysis buffer or diluent C for further analysis.
Exosomes were labelled with PKH67 or PKH26 dye (Sigma-Aldrich), according to the manufacturer’s protocol. Briefly, 4 μL PKH67 dye was mixed with exosome suspension in diluent C and incubated for 10 min at 37 °C. The labelling reaction was stopped by adding 20 ml chilled PBS. Labelled exosomes were ultra-centrifuged at 100,000×g for 70 min, washed with PBS, ultra-centrifuged again at 100,000×g and the pellet was resuspended in PBS.
Cellular uptake of exosomes
dSH-SY5Y or hiPSCs cells were plated on coverslips in the respective serum-free growth medium. Before the uptake assay, exosomes were isolated from brain tissue or conditioned media of raSH-SY5Y cells and labelled with PKH67 or PKH26 as described above. In order to use equal amounts of exosomes in the cell cultures, exosomal protein content was quantitated by using BCA (Bio-Rad) or QuantIT (Invitrogen). Brain exosome abundance was quantified according to the AChE activity (EXOCET Exosome Quantification kit; System Biosciences) according to the manufacturer’s protocol. The uptake was performed by incubating cell cultures with 100 µl of exosome solutions (corresponding to an exosomal protein content of 0.62 ± 0.28 μg/μl from brain or 0.71 ± 0.33 μg/μl from conditioned media; equal to 1.4e10 exosome abundance from brain) in a humid chamber for 3 h (37 °C, 5% CO2). For inhibition experiments, cultured cells were pre-incubated for 30 min with the endocytosis inhibitors, dynasore (dynamin inhibitor, 80 µM), phenylarsine oxide (clathrin inhibitor 20 µM), genistein (caveolae inhibitor, 200 µM) all from Sigma-Aldrich. Isolated exosomes in PBS were added to cells for 3 h as above and flow cytometry was performed.
Co-culture model
Co-culture of donor-recipient cells was performed by using two different methods namely the coverslip system (where physical contact of synapses is possible) or the transwell system (where physical contact of synapses is not possible). In both cases, donor cells (raSH-SY5Y 12,500 cells/cm2, or hiPSCs 25,000 cells/cm2) were seeded on glass coverslips coated with 0.1 mg/ml poly-l-ornithine and 10 µg/ml laminin and cultured as described above for 3 h at 37 °C with either 1 µM of oAβ-AF700 or labelled exosomes from brain tissue or conditioned media, and thereafter washed twice with PBS.
In the transwell system, donor cells were seeded on a polycarbonate membrane filter with a 0.4 µm pore size (Falcon, Corning), placed on top of recipient cells (dSH-SY5Y) and subsequently co-cultured for 24 h. At the end of incubation, the membrane filter was removed and recipient cells were washed with PBS and analysed with flow cytometry or fixed with 4% PFA for immunofluorescent labelling.
In the coverslip system, the donor cells were seeded on glass coverslips (VWR International) and placed upside down on top of recipient cells, predifferentiated as described above (resulting in donor cells facing recipient cells) and subsequently co-cultured for 24 or 48 h at 37 °C. For gel cultured cells this results in a 3D environment. Thereafter, the coverslips with donor cells were removed and recipient cells were washed with PBS and either analysed with flow cytometry or fixed with 4% PFA for immunofluorescent staining. Additionally, to control for donor cell contamination in the recipient cell samples, donor cells were transfected with BacMam 2.0 early endosomes Rab5a-RFP (Life Technologies) at a final concentration of 30 particles per cell before co-culture and RFP fluorescence was monitored in recipient cells by flow cytometry.
Immunocytochemistry
Co-localization of oAβ and flotillin-1 and TSG101 were visualized with immunostaining using 1: 5000 solution of mouse anti-mAb158, 1: 200 solution of mouse anti-flotillin-1 and a 1: 200 solution of mouse anti-TSG101. The secondary antibodies were Cy3 conjugated (Jackson Immuno Research, 1:1000) and Alexa Fluor 488-conjugated (Invitrogen, 1:400) goat anti-mouse IgG. GFP was detected using 1:200 rabbit anti-GFP (Life Technologies) and 1:400 goat anti rabbit Alexa flour 647 (Life Technologies).
Cell microscopy
Images of fixed cells were acquired with a Zeiss LSM 700 confocal microscope. The differential interface contrast (DIC) mode and 405, 488, 555 and 639 lasers were used to acquire the images with 63x/1.40 oil immersion plan-apochromatic DICII objectives. Live cell imaging were done using a Zeiss Primo Vert microscope. The micrographs were processed using Huygens (Scientific Volume Imaging) and ZEN lite (blue edition) software.
Flow cytometry
To detect PKH67 labelled exosomes or oAβ-AF700, cells were released from the ECM gel using Corning Recovery Solution (Corning) according to manufacturer’s instructions, filtered through CellTrics 30 µm filters (Sysmex), re-suspended in PBS, and subsequently analysed on a BD FACSAria™ (BD Biosciences) flow cytometer.
After inhibiting the gene expression of exosome markers TSG101 and VPS4A in raSH-SY5Y cells, using RNA interference, the number of secreted exosomes were analysed using the Exo-Flow™ kit (System Biosciences, USA), targeting CD9, CD63 and CD81, as per manufacture’s instruction.
Enzyme-linked immunosorbent assay (ELISA) analysis
Altogether, 96-well EIA/RIA plates (Corning Inc.) were coated at 4 °C overnight with 200 ng/well of mAb158 in PBS. Plates were blocked with 1% bovine serum albumin (BSA) in TBS. Exosome samples prepared in RIPA buffer (150 mM sodium chloride, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate and 50 mM Tris, pH 8.0) were added to the plates in duplicates and incubated for 2 h at 37 °C. A total of 1 mg/mL of biotinylated mAb158 was added and incubated for 1 h at 22 °C, followed by 1 h at RT incubation of streptavidin-coupled poly-HRP (Mabtech). K-blue enhanced (ANL product, Sweden) was used as HRP substrate and plates were read in a spectrophotometer at 450 nm, using Spectra MAX 190 and then analysed with SOFT Max Pro. Wells were washed three times in TBST between each step. The 82E1 sandwich ELISA was performed same as above using 0.25 µg/ml for capture and detection antibody [
56]. The amount of oAβ within exosomes was quantified with respect to a standard curve created with serial dilution of synthetic Aβ oligomers and expressed as picomolar/mg of protein.
Immunoblot analysis
Exosomes were prepared as described above. Brain lysates were prepared from homogenised brain tissue followed by addition of lysis buffer (150 mM NaCl, 0.5% deoxycholate, 1% Triton X-100, 50 mM tris-HCL pH 7.5, 20 μl/ml phosSTOP (Roche), 10 μl/ml Halt Protease inhibitor cocktail (Thermo Fisher Scientific)), clarified by centrifugation at 10,000 x g for 5 min and sonicated using an ultrasonic probe. Cell lysates were prepared from cells collected in lysis buffer followed by homogenisation and sonication. Samples were mixed with 4x Laemlli loading buffer and separated on a ClearPAGE SDS Gel 4–12% or 10% (C.B.S. Scientific), and transferred onto a nitrocellulose membrane (Invitrogen). The gel was subsequently stained using InstantBlue protein stain (Expedeon). Additionally, exosomes, isolated from Control and AD brains or conditioned media of oAβ-AF700-treated raH-SY5Y cells were lysed by freeze-thawing and subsequently run by SEC using the conditions described above. The eluted proteins were collected in fractions of 1 ml, lyophilized, resuspended in 15 µl PBS and spotted on 0.2 µm nitrocellulose membrane. Membranes were then blocked by 3% BSA followed by primary antibody incubation. The following antibodies were used: anti-flotillin-1 (1: 500, BD Transduction Laboratories); anti-alix (1: 1000, EMD Millipore); anti-TSG101 (1:1000, Thermo Fisher Scientific); 1:1000, anti-VPS4A (Abcam), anti-calnexin (1:1000, Abcam), anti-synaptophysin (1:1000, Synaptic Systems), mAb158 (1: 5000/10000, BioArctic) and anti-glyceraldehyde 3-phosphate dehydrogenase (GAPDH, 1:40000, Synaptic Systems,). Anti-rabbit IgG, horseradish peroxidase (HRP)-linked antibody (1:3000, Dako) and anti-mouse IgG, HRP-linked antibody (1:3000, Dako) were used as secondary antibodies. The blots were visualized using Amersham™ ECL™ (GE Health Care) or SuperSignal® (Thermo Scientific) detection systems and analysed by ImageJ software.
Negative staining and transmission electron microscopy of exosomes
Exosome suspensions were fixed in 4% paraformaldehyde (at 1:1 dilution, for a final paraformaldehyde concentration of 2%) overnight at 4 °C and stored at − 20 °C until use. Thawed exosome suspensions were vortexed briefly and centrifuged in a microcentrifuge for 30 s. Exosomes were adsorbed on Formvar-coated Ni mesh grids by placing the grids on 5 µl drops of exosome suspension for 20 min in a dry chamber. Negative staining was performed by gently dripping 100 µl 2% aqueous uranyl acetate onto the grid, followed by removal of excess uranyl acetate solution using a lens paper. The grids were examined in a JEOL JEM-1230 electron microscope at 100 kV accelerating voltage. Electron micrographs were obtained at 150,000–200,000 × magnification, for a final image scale of 3.1–4.2 pixels/nm.
Tunable resistive pulse sensing by qNano
Exosome size and particle number were analysed by TRPS analysis using a qNano instrument (IZON Science, UK) as described previously [
37]. First, isolated exosomes from brain tissue or conditioned media of dSH-SY5Y cells were diluted and passed through a 0.2 μm filter (Millipore). Subsequently, particle numbers were counted for a maximum of 5 min or until 500 particles had been counted, using 8 mbar pressure and the NP150 or NP100 nanopore membranes with a stretch between 45 and 47 mm. Voltage was set to 0.1 and 0.25 mV to achieve a stable current. Particle size histograms were recorded when root mean square noise was below 13 pA and particle rate in time was linear. Calibration was performed using known concentration of beads CPC70D (mode diameter 70 nm) or CPC100B (mode diameter: 110 nm) (all from IZON) diluted in 1:500 0.2 μm filtered PBS.
Proteinase K digestion
To examine whether exosome-associated Aβ is luminal or bound to the exterior exosome surface, exosomes were isolated from conditioned media (oAβAF700-treated) of dSH-SY5Y cells and incubated with proteinase K (Sigma-Aldrich,1 mg/ml) for 30 min at 37 °C. 4-(2-aminoethyl)-benzene-sulfonyl fluoride (Sigma-Aldrich, 0.5 mM) was subsequently added to the vesicle fraction to inactivate the enzyme prior to two rounds of 100,000 × g centrifugation. The final pellet was resuspended in PBS and AF700 fluorescence was measured in Tecan Safire2 microplate reader at Ex/Em 696/719 nm.
Cytotoxicity assay
To investigate the toxic effect of exosomes on neurons, equal amounts of exosomes (based on exosomal protein estimation by BCA) from brains or cells were added to dSH-SY5Y cells and hiPSCs in our co-culture model for 48 h, as described above. At the end of incubation, donor cells were removed and cell medium was collected to assess the release of lactate dehydrogenase (LDH) in the medium. Collected medium was centrifuged 2000×g for 5 min at 4 °C and LDH assay (Pierce) was performed according to manufacturer’s instructions. The absorbance was measured in a microplate reader (SpectraMAX 190) at 490 nm with subsequent blank at 680 nm. Furthermore, XTT (2,3-bis [2-methoxy-4-nitro-5-sulfophenyl]-5-[(phenylamino) carbonyl]-2H-tetrazolium hydroxide) assay using the Cell Proliferation Kit II (Roche Diagnostics GmbH) was performed on acceptor cells according to the manufacturer’s instructions. The reduced XTT product produced by mitochondrial enzymes in viable cells, formazan (bright orange in colour) was measured after 8 h of incubation at 450 and 750 nm using a Victor 3 V 1420 multilabel plate reader (PerkinElmer). Both LDH and XTT values were presented as percentage of untreated control.
RNA interference
Cells were seeded at a density of 12 500 cells/cm2 in a 6-well plate and transfected 24 h later with TSG101 or VPS4 mRNA-targeting siRNA or a non-targeting siRNA with no homology to any known human gene (All Stars Negative Control siRNA) with the HiPerFect transfection reagent (all from Qiagen) according to manufacturer’s protocol. TSG#6 (CAGTTTATCATTCAAGTGTAA), TSG#3 (ACTGTCAATGTTATTACTCTA), VPS#7 (AAGCTGAAGGATTATTTACGA) and VPS#5 (CTCAAAGACCGAGTGACATAA) siRNAs were used for this study. The final siRNA concentrations in the culture medium ranged from 10 to 20 nmol/l. Twenty-four hours after transfection, knockdown was verified by quantitative real-time PCR and Western Blot analyses, and a decrease in the mRNA level of 70% or greater was considered sufficient downregulation.
Gene expression analysis
Total RNA was extracted with the RNeasy Mini Kit (Qiagen), and cDNA was obtained with the High Capacity RNA-to-cDNA Kit (Applied Biosystems). The expression levels of TSG101 or VPS4 mRNA were analysed with a 7500 Fast Real-Time PCR system and FAM/MGB probes (Applied Biosystems) to confirm downregulation after siRNA treatment. All reactions were performed according to the manufacturer’s instructions. GAPDH was amplified as an internal standard. The data were calculated according to the comparative Ct method to present the data as fold differences in the expression levels relative to the control sample.
Statistics
All statistical analyses were performed using GraphPad Prism Software. Data were expressed as the mean ± SEM, and statistical comparisons were made using two-tailed unpaired Student’s t tests with Welch’s correction or one-way ANOVA with Tukey's correction. Every batch of cell cultures was treated as one independent experiment (n = 1). P values less than or equal to 0.05 were deemed statistically significant.
Study approval
The collection and use of post mortem brain tissue was approved by the Regional Ethical Committee in Uppsala, Sweden (2005/103, 2005-06-29; 2009/089; 2009-04-22). The process of reprogramming human cells was approved by the Ethical Committee at Karolinska Institute, Sweden (dnr 2012/208-31/3 with addendum 2012/856-32). A written informed consent was received from all donors.
Discussion
Recent evidence suggests that toxic Aβ aggregates can spread pathology in the Alzheimer brain. The nature of the propagating species has not been established, although several studies have indicated a particularly pathogenic role of soluble oAβ on synaptic and cellular functions and structure [
13,
32,
50,
60]. It has been speculated that exosomes might transfer neurodegenerative proteins in the affected brain [
28]. Accordingly, exosomes from blood [
22], CSF [
17,
49] and cell cultures [
46] have been shown to contain monomeric Aβ and tau, but so far, no study has addressed the presence of oAβ in exosomes from human AD brains. In this study, we could show that AD brain exosomes contain an increased amount of oAβ compared to non-neurological control brains and found evidence that exosomes can be responsible for the neuron-to-neuron transfer of toxic oAβ. These findings suggest that exosomes might be the main mediator of the pathogenic progression in AD as was recently suggested for dementia with Lewy bodies [
41]. In addition, we found a co-localization between oAβ and exosomes inside neurons, which might indicate that exosomes play a role in Aβ sorting and oligomerization [
16]. The AD brain exosomes were further shown to effectively transfer oAβ from one neuron to another, with subsequent toxic effects on the recipient cells. Interestingly, at least a part of the exosomes seems to be transferred intact to further cells, consistent with a recent study showing that a substantial fraction of exosomes internalized in one cell were subsequently passed on to a second cell [
44]. Importantly, there is increasing evidence of correlations between intra-neuronal oAβ and cell death [
43]. We have previously demonstrated that transfer of oAβ causes neurotoxicity [
40] which has also been shown with Aß containing exosomes isolated from AD CSF [
33]. Accordingly, we now found signs of neurotoxicity both morphologically and with the LDH and the XTT assays after transfer of exosomes carrying oAβ. This observation not only reinforces the role of intracellular oAβ in AD pathogenesis but also establishes the disease relevance associated with the exosomal neuron-to-neuron transfer of intercellular oAβ.
The molecular content of exosomes is a fingerprint of the releasing cell type and, because of their small size, neuronal exosomes are released into accessible body fluids such as blood and CSF [
48]. Since neuronal exosomes display unique neuron-specific surface markers [
22,
26] they may be a valuable biomedical marker for early diagnosis and treatment in AD. Indeed, exosomes have recently been highlighted as diagnostic biomarkers in various disease conditions, including AD [
29,
36]. In concordance, the finding of increased levels of oAβ in brain exosomes opens the possibility that similar features could be detected also in easily accessible body fluids, such as plasma and CSF. Hence, measurement of increased oAβ in exosomes from such patient samples could potentially serve as a diagnostic tool.
The intercellular propagation of oAβ and its ensuing toxicity could also serve as a potential treatment target by inhibiting either formation, secretion, or cellular uptake of exosomes. Indeed, downregulation of TSG101 and VPS4A, proteins necessary for exosome formation and secretion, was found to result in decreased release of exosomes and a reduced subsequent transfer of oAβ, thus supporting the possibility of modulating this mechanism. Moreover, these observations are in line with recent studies showing that interfering with exosome release can impact the release of specific proteins [
9,
14]. An alternative therapeutic target could be the dynamin-dependent uptake of exosomes [
23,
24] as the dynamin inhibitor dynasore decreased exosome propagation, spread of oAβ and the associated neuronal toxicity, leading to rescued cell viability. Dynasore itself would not be a feasible therapeutic substance, but phenothiazine-derived antipsychotic drugs have been suggested to inhibit dynamin dependent endocytosis [
11] and could thus be suitable for further drug development.
In conclusion, our results point to a role for exosomes in the spreading of toxic oAβ and the associated disease progression in the AD brain (summarized in Supplementary Fig. S6). It has been suggested that exosomal release may provide an alternative disposal mechanism to lysosomal degradation of oAβ [
5] or other proteins that are resistant to degradation [
55]. We speculate that this alternative mechanism of clearance, which initially could be beneficial for the cells, over time becomes a liability with increased propagation of pathological proteins throughout the brain. The possibility of inhibiting exosome transfer and the related spread and toxicity of oAβ may lead to the identification of new pharmaceutical targets for AD.