Extracellular Vesicles as Mediators of Cellular Information in the Central Nervous System

The classical concept of interneuronal communication, requiring direct cell-to-cell contact or exocytosis of vesicular contents, as in synaptic transmission, has been challenged by the realization that neurons release extracellular vesicles (EVs) that contain and transmit information through uptake by recipient cells in the immediate environs and also at a distance.

Extracellular vesicles can transfer proteins and lipids (Subra et al. 2010), as well as DNA and RNA (Valadi et al. 2007). Thus, genetic information can be carried over from one cell to another and it has, for example, been demonstrated that vesicular mRNA can be translated in the recipient cells (Valadi et al. 2007; Skog et al. 2008; Lai et al. 2015). Moreover, it has been shown that transfer of non-coding RNAs, such as microRNAs (miRNAs) and other non-coding RNAs, can mediate gene regulation in recipient cells (Montecalvo et al. 2012).

In contrast to synaptic vesicles, which fuse with the pre-synaptic terminal to release classical neurotransmitters, EVs are released intact from neurons (Faure et al. 2006) and can either be shed from multivesicular bodies (MVBs), which are derived from endosomes, or bud directly out from the plasma membrane (reviewed in Cocucci and Meldolesi 2015). Once released into the extracellular space, they are referred to as exosomes or ectosomes (or microvesicles), depending on whether they have formed from inner or outer cell membranes, and are taken up by other cells through endocytosis or fusion.

In the following, we will either use the more general term “extracellular vesicle” or apply the terms from the studies that are referred to (for a more thorough characterization of the different vesicles and their functions, please see (Lotvall et al. 2014) or the initial chapter of this Special Issue. We will provide recent examples from the literature, suggesting that EVs may be centrally involved in both the generation and spreading of toxic α-synuclein. Moreover, we will describe the ongoing efforts to measure EV-associated α-synuclein as a novel disease biomarker and discuss some of the methodological aspects in the preparation of such samples.

α-Synuclein and Its Role in Neurodegeneration

The protein pathology in most neurodegenerative disorders describes a hierarchical anatomical pattern. For amyloid plaques and neurofibrillary tangles, pathological features of the Alzheimer’s disease (AD) brain, the entorhinal cortex is typically the first affected location prior to involvement of other areas, such as hippocampus and neocortex. In PD, certain nuclei of the brain stem as well as the substantia nigra feature Lewy body and Lewy neurites at early disease stages before the pathology becomes more widespread as the disease progresses. At an advanced disease stage, a large part of the cerebral cortex may display Lewy bodies and Lewy neurites (Fig. 1). Other disorders with Lewy body brain pathology include dementia with Lewy bodies, multiple system atrophy, and the Lewy body variant of AD (reviewed in Puschmann et al. 2012).

Fig. 1
figure 1

Neuropathology of Parkinson’s disease. Lewy bodies (black arrow) and Lewy neurites (red arrow) are found in pigmented dopaminergic neurons in the substantia nigra from a PD patient, stained with a polyclonal antibody directed against α-synuclein amino acids 126–135 (×20 magnification). Photo: Leire Almandoz Gil, Uppsala University (Color figure online)

Full size image

The main constituent of Lewy bodies and Lewy neurites is the 140 amino acid long protein α-synuclein. The normal physiological function of α-synuclein is unclear, but it seems to involve neurotransmitter release (Nemani et al. 2010) and interaction with the synaptic SNARE (soluble N-ethylmaleimide-sensitive factor attachment protein receptors) complex (Burre et al. 2010). In its pathological state, α-synuclein has lost its native structure and is prone to aggregate into β-sheet-rich fibrils, which are formed via a number of intermediate steps (Uversky et al. 2001). Conformationally altered α-synuclein monomers might initiate the formation of dimers and trimers, which continue to form soluble oligomers and protofibrils that eventually deposit as fibrils and mature aggregates (Fig. 2).

Fig. 2
figure 2

Theory on how α-synuclein aggregates are formed in a step-wise process Monomers can get misfolded, leading to the formation of soluble toxic oligomers/protofibrils and eventually less harmful fibrils, which are deposited as insoluble aggregates (Lewy bodies)

Full size image

There is no clear correlation between the extent of Lewy body pathology and disease severity (Harding and Halliday 2001), indicating that the fibrils are not the most pathogenic α-synuclein species. It has even been suggested that Lewy bodies could be cytoprotective by sequestering toxic soluble forms of α-synuclein in aggregates (Tanaka et al. 2004). Numerous ex vivo and in vivo studies have demonstrated that the prefibrillar α-synuclein oligomers and protofibrils are particularly prone to confer toxicity. For example, such species have been shown to be cytotoxic (Outeiro et al. 2008; Karpinar et al. 2009), disrupt cellular membranes (Danzer et al. 2007; Winner et al. 2011), cause mitochondrial damage (Chinta et al. 2010; Luth et al. 2014) and synaptic dysfunction (Pacheco et al. 2015), as well as elicit neuroinflammatory reactions (Zhang et al. 2005; Su et al. 2008; Wilms et al. 2009). Moreover, certain PD-causing α-synuclein point mutations seem to be pathogenic by increasing the tendency for formation of large soluble oligomeric forms (Conway et al. 2000).

Propagation of α-Synuclein Pathology

An increasing body of evidence suggests that α-synuclein pathology may spread within the affected brain. When PD brains which had been subjected to intra-striatal transplantation of dopaminergic cells from normal fetuses more than a decade earlier were examined at autopsy, they displayed α-synuclein-positive Lewy body like inclusions also in the transplants (Kordower et al. 2008; Li et al. 2008). The grafts thus seemed to have been affected by the presence of pathological α-synuclein in the host tissue, which indicates a spread of pathological α-synuclein from affected host neurons to healthy adjacent grafted cells.

In accordance with these findings, it has been demonstrated both in animal models and on cultured cells that prefibrillar species of α-synuclein can transfer between cells (Desplats et al. 2009; Hansen et al. 2011; Luk et al. 2012a, b; Aulic et al. 2014; Ulusoy et al. 2015). For example, in a recent study, wild-type rats displayed α-synuclein inclusions in the dorsal motor nucleus of the vagus nerve 12 h following injection of PD brain lysate into the intestinal wall, a sign of rapid protein transfer to neurons (Holmqvist et al. 2014). This finding suggests that retrograde transport of pathologically altered α-synuclein between the mesenteric plexus and the CNS may be a pathophysiological mechanism also in the human situation, since PD patients often display α-synuclein deposits in gut biopsies (Braak et al. 2006; Sanchez-Ferro et al. 2015). Interestingly, these findings could provide an attractive pathophysiological explanation to the observation that obstipation and other gastrointestinal symptoms are common at early disease stages in this patient group. Thus, several lines of evidence suggest that intercellular spreading of dysfunctional α-synuclein could be relevant for PD pathology.

Extracellular Vesicles as Vehicles for Cell-to-Cell Transfer of α-Synuclein

A number of observations suggest that the cell-to-cell spreading of α-synuclein may occur via exosomes and other EVs. In 2010, it was proposed that neuron-like cells overexpressing α-synuclein can secrete it both within exosomes and directly into the cell media (Emmanouilidou et al. 2010). In a study by Danzer and colleagues (Danzer et al. 2012), bifluorescence complementation (BiFC) assays were employed using human neuroglioma cells and primary mouse neurons to demonstrate that α-synuclein oligomers appear in the extracellular space both as free proteins and in an exosomal fraction. Moreover, by subjecting the samples to detergents, the authors could demonstrate that approximately half of the α-synuclein associated with EVs could be found within these structures, with the other half decorating the outside of the vesicles. Additionally, it could be demonstrated that the EV-associated α-synuclein from cells transfected with an expression construct for α-synuclein was efficiently taken up by non-transfected recipient cells. Cells exposed to vesicular α-synuclein also displayed a greater degree of caspase 3 and 7 activation as compared to cells that had been incubated with the free protein fraction, indicating that vesicular α-synuclein also may be cytotoxic (Danzer et al. 2012). This study only investigated the fusion proteins of α-synuclein with either hemi-Gaussia luciferase or hemi-Venus (modified yellow fluorescent protein) and it is therefore not clear whether non-modified forms of α-synuclein can also be transferred via similar mechanisms and have the same effects on recipient cells. However, the results point toward the possibility that cells may transfer harmful, aggregation prone species of α-synuclein via release and uptake of EVs.

Mechanisms by Which Extracellular Vesicles may Contribute to α-Synuclein Pathology

It is likely that aggravation of α-synuclein pathology could result either from an increased release of α-synuclein and promotion of cell-to-cell transfer, or from decreased release/secretion resulting in cellular accumulation of the protein. An appropriate balance between retention and release of different α-synuclein species, as well as the mode of release, may thus be important factors in maintaining synaptic integrity and cellular homeostasis. The balance between the cell’s internal protein degradative systems and the secreted or EV-mediated release of α-synuclein has been explored. Alvarez-Erviti and colleagues demonstrated that inhibition of lysosomal protein degradation by either bafilomycin A1 or ammonium chloride resulted in an increase in the levels of exosomal α-synuclein released from cultured neuroblastoma cells (Alvarez-Erviti et al. 2011).

Additional PD-related proteins may contribute to pathology by affecting the secretion/release and uptake of α-synuclein. Patients with mutations in LRRK2 (Leucine-Rich Repeat Kinase 2) manifest dominantly inherited forms of PD, with a distinct clinical profile (Paisan-Ruiz et al. 2004; Zimprich et al. 2004). To date, at least nine different pathogenic LRRK2 mutations have been described (reviewed in (Li et al. 2014)). Importantly, most familial LRRK2 cases display Lewy body pathology in the brain and a majority of experimental studies have indicated a pathogenic interaction between LRRK2 and α-synuclein (Qing et al. 2009; Guerreiro et al. 2013), although the apparent molecular link between LRRK2 and α-synuclein has not yet been elucidated.

Several lines of evidence suggest that LRRK2 may be involved both in the release and uptake of intracellular proteins via exosomes or other vesicle-related mechanisms. For example, LRRK2 has been demonstrated to interact with Rab5b, which regulates endocytic vesicle trafficking within cells. Primary neurons that either overexpressed LRRK2 or had that gene knocked out showed reduced reuptake of synaptic proteins, a defect that could be rescued by the simultaneous overexpression of Rab5b (Shin et al. 2008). Moreover, when Xiong and colleagues overexpressed wild-type LRRK2 in primary hippocampal neurons, it resulted in a reduction of both endocytosis and exocytosis of synaptic proteins in hippocampal neurons (Xiong et al. 2010). Although not proven, it could be speculated that such mechanisms may be vesicle-mediated and explain the increased cellular retention and accumulation of α-synuclein caused by dysfunctional LRRK2. Along those lines, another study suggested that the R1441C LRRK2 mutant causes an increased number and size of MVB and that dysfunctional LRRK2 can alter the dynamics of vesicular release (Alegre-Abarrategui et al. 2009).

Another PD-related protein, ATP13A2, is expressed in dopaminergic neurons of the substantia nigra and patients with ATP13A2 mutations at the PARK9 locus develop a rare autosomal recessive parkinsonian disorder known as Kufor-Rakeb syndrome (Ramirez et al. 2006). Overexpression of ATP13A2 in cultured neuroglioma cells promotes release of α-synuclein in EVs, whereas loss of function mutations in ATP13A2 leads to decreased vesicular release and thereby increasing cellular α-synuclein retention and accumulation (Tsunemi et al. 2014).

In addition to neurons, glial cells may also release and take up vesicles containing α-synuclein. Lee and colleagues demonstrated that astrocytes in culture were able to internalize α-synuclein present in the conditioned media from SH-SY5Y neuroblastoma cells overexpressing human α-synuclein (Lee et al. 2008, 2010). Although these studies did not specifically investigate the vesicle-containing fractions in the media, it seems likely that at least a portion of the neuroblastoma cell-produced α-synuclein may have been secreted via EVs.

Taken together, these different observations suggest that both increased and decreased EV secretion of α-synuclein can contribute to pathology, by either promoting spread or retention of dysfunctional protein species.

Extracellular Vesicles may Induce and Release Toxic Oligomeric Forms of α-Synuclein

Although it remains an open question as to which cellular compartments are responsible for the oligomerization of α-synuclein, one study found that EV biology may have a central role. When recombinant α-synuclein monomers were incubated together with EV fractions from cultured neuroblastoma cells, the formation of oligomers was accelerated as compared to when monomeric α-synuclein was incubated alone (Grey et al. 2015). Moreover, the authors hypothesized that this reaction was mediated by different classes of phospholipids that are integral and enriched in the vesicular membranes and therefore investigated their direct involvement in the formation of oligomers. The ganglioside lipids, GM1 and GM3, were found to promote α-synuclein oligomerization in a manner similar to that observed for the vesicular preparations (Grey et al. 2015).

The formation of α-synuclein oligomers can be initiated ex vivo by a number of substances (Hong et al. 2008; Nasstrom et al. 2011), some of which may also have a physiological relevance in vivo. For instance, one study indicated that γ-synuclein, another member of the synuclein family, can form heterodimers with α-synuclein and act as a seed for the formation of α-synuclein aggregates (Surgucheva et al. 2012). Moreover, the same study found that γ-synuclein can be secreted via EVs, pointing toward the possibility that this compartment may be a crucial site for such cross-seeding reactions (Surgucheva et al. 2012).

If exosomes and other EVs are mediating transfer of pathogenic oligomeric α-synuclein between cells, they would most likely need to be internalized to cause cellular dysfunction and cell death. Although there is, as of yet, no in vivo proof of such mechanisms, one of the studies mentioned above provides ex vivo evidence that α-synuclein oligomers can be transferred between human CNS-derived cells via EVs (Danzer et al. 2012).

Previous research has also suggested that disruption of different intracellular processes may affect the formation of oligomers. One study found that inhibition of the autophagy-lysosomal pathway by bafilomycin A led to a higher toxicity in α-synuclein overexpressing cells with a concomitant decrease in α-synuclein fibril formation (Klucken et al. 2012). In addition to suggesting that impairment of intracellular degradation can increase the formation of toxic α-synuclein oligomers, this study provides additional evidence to the notion that such prefibrillar species are responsible for cellular damage.

Extracellular Vesicle-Associated α-Synuclein as a Diagnostic Biomarker

Several studies have explored the possibility of utilizing the presence of α-synuclein in biofluids to develop a biomarker for status and progression of PD and other α-synucleinopathies. Overall, most studies have found a decrease in total α-synuclein levels in cerebrospinal fluid (CSF) in patients, whereas a majority of the plasma-based studies have indicated α-synuclein levels to be slightly increased in patients vs healthy controls (reviewed in Kasuga et al. 2012). Moreover, the CSF and plasma studies are not entirely consistent and have also demonstrated a substantial overlap in levels between the groups, which seems to make soluble total α-synuclein not very useful as a biomarker for diagnostic purposes. A somewhat more promising study was published by Tokuda and colleagues, suggesting that PD patients have elevated CSF levels of α-synuclein oligomers (Tokuda et al. 2010) with one study to date replicating these results (Park et al. 2011).

The discouraging outcome of most CSF and plasma-based studies has prompted researchers to develop novel methods for detecting different forms of α-synuclein in biological fluids. In 2014, Shi and colleagues sought to isolate CNS-derived exosomes from human plasma and compared the levels of total α-synuclein in such preparations in hundreds of PD patients and healthy controls (Shi et al. 2014). By adopting a magnetic bead-based capture assay, in which the beads were conjugated with a monoclonal antibody against the neural cell adhesion molecule L1CAM, they aimed to specifically enrich for vesicles that displayed this purportedly CNS-selective marker.

In their study, the PD samples displayed a mean of about 25 pg α-synuclein/ml plasma in the L1CAM-positive subpopulation of EVs, whereas age-matched healthy controls displayed only 12 pg α-synuclein/ml in similarly prepared plasma samples. This difference between patients and controls was statistically significant (p < 0.001) with only a small overlap between the two groups. When the free fraction of α-synuclein was measured directly in plasma of the same patients and controls, there were no differences between groups. Finally, CSF was available from a subset (PD = 100, C = 100) of the subjects included for the plasma analyses. Measurements of total α-synuclein levels in these CSF samples demonstrated a decrease (as noted in a majority of studies) in α-synuclein concentration among PD patients, but with no correlation to the increase of α-synuclein in L1CAM-positive plasma EVs (Shi et al. 2014). Thus, these data seem to suggest that α-synuclein detected in CSF and in CNS-derived plasma EVs may reflect different pools of the protein. For example, sequestration and enrichment of α-synuclein in EVs may be a means for neurons to eliminate an excess of protein via the circulation, whereas its decreased presence in CSF could be an indirect effect of the increased aggregation (and retention) of α-synuclein inside the brain.

Enrichment of CNS-derived subpopulations of EVs from patient plasma has also been performed to evaluate novel biomarkers for AD (Fiandaca et al. 2014). In this study, AD patients (n = 57) were found to have significantly higher levels of both tau, hyperphosphorylated tau (p-tau) and amyloid β 42 (Aβ42) in CNS-derived EVs as compared to healthy controls (n = 57) (p ≤ 0005, p < 0.0001, and p < 0.0001, respectively) (Fiandaca et al. 2014).

Also CSF contains EVs derived from the brain, which can be investigated for disease biomarkers. No estimates of exosomal α-synuclein levels in CSF have been published to date, but in a recent report Gui and colleagues investigated CSF-derived EV samples from a cohort of PD and AD patients and found that several mRNA species, including those for α-synuclein and APP, were downregulated in both AD and PD patients compared to controls (Gui et al. 2015). Moreover, the authors explored the miRNA profile in these EVs and were able to validate that two miRNAs (miR-1 and miR-19b-3p) were reduced, whereas four (miR-153, miR-409-3p, miR-10a-5p, and let-7g-3p) were increased in PD vs control CSF samples (Gui et al. 2015).

Researchers have also started to explore saliva as a source of biomarkers. Free α-synuclein can be detected in this biofluid (Devic et al. 2011), but no studies to date have explored the possibility of targeting saliva EVs for assessment of α-synuclein in PD patients. However, a recent study investigated the miRNA profile in EV fractions of 15 young (median age 21 years) versus 13 aged (median age 66 years) individuals and found that one miR-24-3p was significantly upregulated in saliva from the older subjects (Machida et al. 2015). This study needs further confirmation before it can be concluded that this or other miRNAs can be correlated to the process of aging.

Urine is another example of an easily accessible biofluid that harbor EVs, but this type of sample has so far mainly been explored in the context of urogenital tract and prostate disorders (reviewed in Gamez-Valero et al. 2015 and Dijkstra et al. 2014). However, one published study has explored urine EVs in PD (Ho et al. 2014). While α-synuclein could not be found, the authors successfully detected the other PD-related proteins LRRK2 and DJ-1. When comparing the levels of these two proteins between patients and controls, no overall differences could be seen although both markers displayed a gender-dependent profile and DJ-1 levels showed an age-dependent increase in male patients (Ho et al. 2014).

Methodological Considerations Regarding α-Synuclein Detection in Extracellular Vesicles

A general method to isolate EV fractions for analyses of α-synuclein and other cargos is based on either a low speed centrifugation or a filtration step (to remove cell debris and other larger particles), or a combination thereof, before ultracentrifugation (see Witwer et al. 2013 for protocols and considerations). The ensuing pellet can then be subjected to a buffer either with or without detergents (to allow for analysis of cargo on the outside vs a combination of the content inside and outside of the vesicles). For example, one of the studies referred to above (Emmanouilidou et al. 2010) adopted a protocol in which the cell culture-derived samples were first centrifuged at 4000×g for 10 min at 4 °C, after which the supernatant was further centrifuged at 100,000×g for 2 h at 4 °C. The second supernatant was then collected for analysis of free α-synuclein, whereas the second pellet was reconstituted in radioimmunoprecipitation assay (RIPA) buffer (50 mM Tris–HCl, pH 7.6, 150 mM NaCl, 1 % NP-40, 0.5 % Nadeoxycholate and 0.1 % SDS) to generate samples in which the content of α-synuclein in EVs was analyzed. To verify the presence of vesicles in this pellet fraction, electron microscopy and Western blotting against vesicular proteins (Alix-1 and flotillin) was applied (Emmanouilidou et al. 2010).

As for clinical biofluid-based samples, different protocols have been adopted to generate EVs. Shi and colleagues used a three-day protocol based on an overnight incubation of 300 µl plasma (previously centrifuged sequentially at 2000 and 15,000×g to remove cell debris) with magnetic beads conjugated to an antibody against the neural cell adhesion molecule L1CAM (10 µg, clone UJ127, Abcam, Cambridge, MA) (Shi et al. 2014). After the incubation, beads were washed four times in PBS and the remaining bead-antibody-vesicle complex was lysed in 1 % Triton-X. The resulting lysate was subjected to an ELISA measuring total levels of α-synuclein (Luminex, Austin, TX). It remains possible that this procedure, in addition to EVs, also could isolate protein aggregates containing both L1CAM and α-synuclein.

In the study by Fiandaca and colleagues, where amyloid β was measured on EV-preparations from AD patients and controls (Fiandaca et al. 2014), the adopted protocol differed somewhat from the one described above for α-synuclein (Shi et al. 2014). Instead of magnetic beads, EVs were collected from plasma using the ExoQuick method (System Biosciences, Mountain View, CA), after which they were incubated with a biotinylated version of an L1CAM antibody (mouse anti-human CD171, clone 5G3; eBioscience, San Diego, CA). The samples were collected by adding streptavidin-agarose resin, followed by centrifugation at 200×g to generate a pellet that was suspended in glycin–HCl with 0.1 % Tween 20 before the ELISA measurement. Also with this procedure, EVs as well as protein aggregates containing both L1CAM and α-synuclein would be measured.

The prospect of developing protocols to measure diagnostic biomarkers in CNS-derived EVs in patient plasma is a daunting task. Depending on the protein that is targeted, the concentrations will often be much lower in vesicular subpopulations as compared to their levels in the surrounding plasma and in vesicles released from blood-related cells. For the biomarker candidates in neurodegenerative disorders, such as PD and AD, the plasma concentrations of free protein are relatively high (mid pg/ml to low ng/ml for α-synuclein, Aβ, tau, and p-tau) as compared to their reported levels in EVs (low pg/ml for both α-synuclein and Aβ). Thus, the levels of free proteins in plasma are several orders of magnitude higher, making the EV fraction sensitive to contamination from the free proteins. The final preparation could thereby easily contain residual proteins from the plasma with free protein that can attach to either the vial walls or to the vesicles or form a conglomerate aggregate that can interfere with the measurements.

In order to better understand how important the mode of preparation of EVs is to the potential contamination of residual plasma, we decided to perform two proofs of principle experiments. For both of these, we adopted previously published protocols to extract EVs from plasma under various conditions. The samples were then subjected to an established ELISA for the measurement of total α-synuclein (Fagerqvist et al. 2013). Although we believe that oligomeric α-synuclein may be the more relevant PD biomarker, we first wanted to use this pan-α-synuclein ELISA to measure the total levels (which also include the oligomeric species).

For the first experiment, we adopted modified versions of the three-day protocol described above (Shi et al. 2014). Peripheral venous blood was collected into purple top EDTA-BD Vacutainer tubes (BD Franklin Lakes, NJ), using standard phlebotomy procedures. In order to avoid platelet activation through cooling, tubes were centrifuged at 2000 rcf for 5 min at room temperature (Hoffmeister et al. 2003). Immediately after centrifugation, the plasma was aliquoted into 500 μL aliquots in low-retention 1.5 mL tubes (Fisher Scientific, CA), using low-retention pipette tips. Aliquots were bar-coded, electronically tracked, and stored ready-to-use at −80° within 4 h of blood draw. Time of blood draw, time of last meal prior to phlebotomy, time to centrifugation, and time to freezing were monitored for quality-control purposes for all samples.

First, one plasma sample from a PD case (male, age 59) was extracted for L1CAM-positive vesicles (Fig. 3). We then compared three different strategies in parallel. As a first protocol (Protocol 1), corresponding to the original protocol, the bead-antibody-vesicle complex was washed four times in PBS with a change to a new Eppendorf tube before the lysis. As a second alternative (Protocol 2), we performed all four washes but without changing to a new tube before lysing the vesicles. For the third and final protocol (Protocol 3), we changed tubes between each washing step. All samples were then measured for total α-synuclein by an ELISA, using the α-synuclein-specific monoclonal antibody syn-1 (BD Biosciences, Franklin lake, NJ, USA) as the capturing antibody and the pan-α-synuclein polyclonal antibody FL140 (Santa Cruz Biotechnology, Santa Cruz, CA, USA) as the reporter (Fagerqvist et al. 2013).

Fig. 3
figure 3

Evaluation of three protocols for isolation of L1CAM-positive EVs from human plasma. a Three different protocols, with various degrees of stringency, were evaluated for the final preparation step of the enrichment of L1CAM-positive EVs from human plasma. In Protocol 1, the magnetic bead-L1CAM antibody-vesicle complex was processed according to the published protocol (Shi et al. 2014), in which the complex was washed four times in PBS and transferred to a new tube before lysis in 1 % Triton-X100. According to Protocol 2, the complex was washed four times in PBS, without changing to a new tube before lysis. Finally, in Protocol 3, the complex was washed four times with change of tubes after each washing step. b The lysates generated from the three protocols were assessed for total α-synuclein concentrations by ELISA

Full size image

When adopting the same protocol as in Shi et al., the average level of the analyzed duplicate was 6.57 pM (Fig. 3, Protocol 1). When performing the same number of washing steps, but omitting the change of tubes, the α-synuclein levels were considerably higher, 247.79 pM (Fig. 3, Protocol 2). Finally, when adopting the most stringent protocol variant, where each wash was followed by transfer of the sample to a new Eppendorf tube, there was no measurable α-synuclein (Fig. 3, Protocol 3). This experiment either suggests that a small remaining portion of free plasma α-synuclein can influence the read-out or supports the possibility that free α-synuclein can associate with the extracted vesicles.

As a second experiment, we wanted to analyze EV-associated α-synuclein from human plasma by adopting a protocol based on ultracentrifugation followed by fractionation of the ensuing pellet in a sucrose gradient separation (Lai et al. 2014) which is an established method of vesicle verification (Witwer et al. 2013). A volume of 2.0 ml of a pooled sample from four healthy controls (0.5 ml from each subject) + 0.3 ml of sterile PBS (which had been filtered twice through a 0.22 µM filter to remove particles) was subjected to ultracentrifugation at 100,000×g at +4 °C for  2 h to collect vesicles and aggregated proteins. Next, the pellet was resuspended in 500 µl of the double-filtered PBS. Here, we either carefully removed all the supernatant from the pellet (Fig. 4a, “Stringent”) or left approximately 5 µl of the supernatant on purpose (Fig. 4a, “Dirty”). Next, the sample was added to a sucrose gradient tube, containing four layers of sucrose (with the following concentrations: 60, 45, 30, and 8 %) (Lai et al. 2014). Following centrifugation of the sample at 232,500×g at +4 °C for 38 min, 10 different fractions were sequentially removed and subjected to another round of ultracentrifugation at 100,000×g at +4 °C for 2 h. Finally, each pellet was resuspended in 220 µl of sample buffer with and without detergent (0.1 % BSA in PBS or 0.1 % BSA in PBS with 1 % Triton-X and protease inhibitors (pH 7.4)) before measuring α-synuclein using the same total α-synuclein ELISA as for the first experiment (Fagerqvist et al. 2013).

Fig. 4
figure 4

Evaluation of α-synuclein in EVs isolated by sucrose density centrifugation. a Two different variants were compared in terms of ultracentrifugation-based EV isolation from human plasma. Two ml of plasma from four healthy donors was centrifuged at 100,000×g, followed by sucrose gradient fractionation of the pellet at 232,500×g. Finally, each fraction was centrifuged at 100,000×g. According to the “Stringent” variant, as much plasma as possible after the first centrifugation step was removed. According to the “Dirty” variant, 5 µl of plasma was left on purpose with the pellet from the first centrifugation step. b The resulting pellets were dissolved in either PBS + 0.1 % BSA (“without detergent” or in PBS + 0.1 % BSA + 1 % Triton-x (“with detergent”) and measured by ELISA for total α-synuclein

Full size image

Of the two variant samples generated, for the stringently handled sample, there was only a small amount of α-synuclein (approximately 10 pM) seen in fraction 10 (i.e., the fraction with the highest density, corresponding to protein aggregates, whereas vesicles typically are enriched in fractions 7–9) (Fig. 4b). For the sample, where 5 µl of plasma was left on purpose together with the original pellet, there were clear ELISA signals in fractions 7–10, with the highest signal at approximately 30 pM seen in fraction number 8, corresponding to one of the vesicle fractions (Fig. 4b). When analyzing fractions from samples where the pellet had been resuspended in the detergent-containing buffer, the pattern was highly similar to the detergent-free results for the more stringent variant 1 (Fig. 4b). For the less stringent variant 2 (where 5 µl of plasma had been left on purpose), the α-synuclein levels appeared only slightly higher in fractions 7–8 (Fig. 4b). The most likely interpretation of these results is that there is very little α-synuclein in total EVs from plasma, and that extravesicular α-synuclein in the plasma can be collected as various aggregated species or associate with the outside of the vesicles in various aggregated states.

Taken together, our experiments emphasize the importance of a stringent sample handling for analyses of α-synuclein in plasma samples. Minute residual volumes of plasma, rich in α-synuclein, could have an impact on the detection of EV content. On the other hand, extensive washing procedures could potentially result in a wash-out of α-synuclein bound to the surface of EVs and thereby lead to false-negative results.

Summary and Conclusions

Several studies have investigated the presence and function of α-synuclein, as free proteins as well as in exosomes and other EVs. Current results suggest that α-synuclein may be released both as free proteins/aggregates and within or associated with vesicles. Moreover, evidence suggests that the vesicular protein is particularly prone to be taken up by other cells (Fig. 5). Such protein transfer may have a pathogenic significance as it has become more and more evident that pathological α-synuclein species may spread from cell-to-cell and thereby also between interconnected brain areas. Of particular interest is the observation that presumably toxic α-synuclein oligomers can not only be secreted, but also carried as cargo within or coating the EVs. Thus, such vesicles may contribute to an environmental relay that is responsible for both the formation and transport of purportedly toxic α-synuclein species in the context of neurodegenerative disorders.

Fig. 5
figure 5

Uptake and release of α-synuclein from neurons. Accumulating evidence suggests that α-synuclein can be both taken up and secreted by neurons, either as free protein or together with exosomes or other EVs. α-synuclein associated with such vesicles seems to mainly consist of toxic oligomers that can be found both within and on the outside of the vesicles. EV-associated α-synuclein may be particularly prone to propagate pathology between neurons, although the free protein may also have such properties

Full size image

The idea of detecting subpopulations of CNS-derived EVs that could better represent the presence of dysfunctional proteins in the brain should be further investigated. The published findings, suggesting that analysis of α-synuclein from CNS-derived EVs in plasma could serve as a novel biomarker for PD (Shi et al. 2014), are intriguing but need to be replicated before we know whether this strategy can be developed into a new diagnostic method. Our experiments, as described above, illustrate that a plasma-based extraction protocol has to be very carefully designed to avoid the risk of carrying small amounts of the analyte from the original sample to the detection step, or of removing α-synuclein from the outside of vesicles. Extensive washing, including change of tubes, appears to either abolish the signal or change the vesicular content of α-synuclein, suggesting that the protein may be stuck either on the inside of the vial walls or on the outside of the vesicles. Most of the proteins of interest in the study of neurodegenerative disorders, including α-synuclein, are particularly prone to bind to the inside of the test tubes, explaining why it is crucial to change tubes between each washing step.

To conclude, exosomes and other EVs appear to contribute to the spreading α-synuclein pathogenesis within the brain and could in the future potentially also be utilized for diagnostic purposes through detection in biofluids. In this context, it would be particularly relevant to monitor the toxic oligomeric α-synuclein species that appear to be central in the disease pathogenesis.