Are exosomes the vehicle for protein aggregate propagation in neurodegenerative diseases?

Aggregation process is a nucleation-dependent process, which requires a nucleus formation during the lag phase [22]. When pre-formed aggregates are present, the aggregation process bypasses the lag phase and rapidly produces aggregates using the pre-formed aggregates as seeds (Fig. 1). This has been the molecular principle underlying the prion infectivity and transmissibility. The same principle can be applied to interneuronal transmission of the aggregated proteins other than prion. The seeding-dependent mechanism has not been completely proven; there are other possibilities by which the protein aggregates are propagated through the interneuronal aggregate transfer [27, 28]. Although the precise mechanism of the transmission remains ill-defined, there are ample experimental supports for interneuronal aggregate transmission in both cells and animals. However, in order for the cytoplasmic protein aggregates to propagate among the neuronal cells, they have a significant challenge: the aggregates have to be secreted from one neuron and transferred to another.

Such problem of this barrier has begun to be addressed as some experiments showed that tau, α-synuclein, and other aggregated proteins were secreted from neurons [7]. α-synuclein and tau have been shown to be secreted through an endoplasmic reticulum/Golgi-independent pathway, which are collectively referred to as unconventional secretory pathway [2, 14, 24‐26, 34, 43]. Physiological functions of the secreted protein aggregates are unknown. The extracellular aggregates may regulate synaptic activity [9] and inflammatory responses [23]. However, the dichotomy between physiological and pathological functions of these extracellular aggregates remains fuzzy. Among the unconventional secretory pathways, the exosome-associated exocytosis has attracted much attention because this particular mechanism has been shown to be involved in the secretion of many disease-linked proteins, including prion, Aβ, α-synuclein and tau (see below), as well as in both physiological and pathogenic processes in human health and disease.

Cells secrete a variety of membranous vesicles during their lives. There are many types of extracellular vesicles: apoptotic bodies, microvesicles/membrane particles, exosomes, etc. (Fig. 2). Each type of these vesicles has its own size, marker proteins, and different secretion pathway. Microvesicles are large (>100 nm diameter) membranous vesicles produced by shedding/budding/blebbing from the plasma membrane of various cell types [19, 31, 36, 38, 39]. These vesicles have irregular shapes and their biomarkers, such as integrins, selectins, and CD40 [31]. Membrane particles are also originated from the plasma membrane and are round-shaped vesicles in 50-80 nm diameter. CD133 (prominin-1) is their biomarker, not CD63 [30]. On the other hand, exosomes are intraluminal vesicles (ILVs) within the multivesicular bodies (MVBs) secreted when MVBs are fused with the plasma membrane. They are 50-100 nm in diameter identified by methods such as electron microscopy (EM) and nanoparticle tracing analysis (NTA). Although there are descriptions on exosomes and other extracellular vesicles, currently there is no consensus criteria to define different types of EVs [32, 47]. Extracellular vesicles (particularly microvesicles and exosomes) often overlap in size and share some surface markers (e.g., tetraspanins CD9, CD63, CD81, etc.) [19, 29, 31, 36, 38, 39]. The cup-shaped morphology of exosomes is known to be an artifact of TEM fixation [41]. More studies with improved procedure for exosome preparation would be necessary for identification of specific markers for exosomes.

Exosomes are purified by the specific procedure, which involve differential ultracentrifugation with cell culture supernatant followed by a rate-zonal centrifugation (Table 1) [41]. Exosomes are located at the 1.10–1.20 g/mL density fraction [19, 31, 38, 39]. Although this procedure has been widely used, some contamination of serum components could not be avoided [32]. There are alternative procedures available, including size exclusion, immunoaffinity isolation, and polymeric precipitation [45]. However, the benefits of these procedures over the conventional centrifugation procedure has not been established yet. Exosomes and other extracellular vesicles may function as vehicles for transportation of macromolecules from one cell to another. Exosomes contain not only proteins but also RNAs and DNAs. Transfer of these macromolecules may play roles in several biological processes, such as innate and acquired immunity and angiogenesis, as well as in human diseases, such as neurodegenerative diseases and carcinogenesis and metastasis [47].

Recent studies suggest that exosomes play important roles in interneuronal transmission of pathogenic proteins and neurodegeneration. The first example was provided in a prion model. Both cellular prion (PrPc) and scrapie form of prion (PrPsc) proteins were released into the extracellular space in association with exosomes [16]. The exosome-associated PrPsc was infectious [16]. Later, it was shown that some Aβ peptides generated from the β- and γ-cleavage of amyloid precursor protein (APP) in endosomes were routed to MVBs and secreted by the exosomal pathway [35]. Furthermore, exosomal proteins were found in the amyloid plaques of AD patient brains [35]. Injection of the exosome preparations into 5xFAD mouse brains promoted aggregation of Aβ1–42 [13]. Inhibition of exosome-associated exocytosis by intraperitoneal injections of GW4869, an inhibitor of neutral sphingomyelinase-2 (nSMase2; a ceramide synthesis regulator which is essential for subtype of exosome biogenesis in MVBs) reduced the Aβ1–42 plaque load in vivo [13]. These suggest that exosomes have their role in AD pathology, triggering aggregation and deposition of amyloid plaques.

Tau was also shown to be secreted via the exosomal pathway [37, 44]. Exosomal tau secretion was identified in primary neurons and tau overexpressing N2a cell culture media. Exosome-associated phospho-tau (AT270-positive) was present in human CSF [37]. The exosomes with associated tau proteins were taken up by neurons and microglia, and induced tau inclusions [44]. Exosomes from CSF samples were also able to promote tau aggregation in cultured N2a cells [44]. Microglia also secrete tau via exosomes and tau propagation is reduced significantly by inhibition of exosome synthesis [4]. When microglial nSMase2 was silenced by siRNA or pharmacologically inhibited with GW4869, exosome-associated secretion of tau and propagation of tau pathology were reduced [4].

In Parkinson’s model, α-synuclein was also secreted via the exosomal pathway in a calcium-dependent manner [15]. These exosomes can be transferred to recipient cells via endocytosis [12], and the transfer was increased when lysosomes were impaired [3]. Additionally, exosome-associated α-synuclein oligomers were taken up by recipient cells more efficiently and had higher toxicity than free α-synuclein oligomers [10]. Exosomes isolated from the CSF of PD and DLB patients were shown to contain pathogenic species of α-synuclein and were able to induce oligomerization of soluble α-synuclein [40]. Injection of exosomes isolated from DLB patients into the brain led to spreading of a-synuclein aggregates [33].

Recently, another way that exosomes contribute to the disease propagation was suggested. Exosomes can accelerate in vitro oligomerization of recombinant α-synuclein monomer and increase toxicity of these proteins [18]. This study also showed that the accelerated oligomerization was partly due to the lipid contents; gangliosides of the exosomal membranes. Similarly, Aβ assembly is markedly accelerated by incubation with the exosome fraction from the PC12 cell culture media [48].

Although there is a body of literature supporting the role of exosomes in the interneuronal aggregate propagation, there are some issues that need clarification. First, not all studies prepared exosomes in a proper way. As mentioned above, a pure exosome preparation requires both differential ultracentrifugation and rate-zonal centrifugation. Differential ultracentrifugation alone results in mixture of many different types of extracellular vesicles. Even after rate-zonal centrifugation, the exosome preparation still may contain contaminated non-exosomal vesicles, which necessitates electron microscopy analysis with immune-gold labeling [41]. As summarized in Table 2, in some studies, exosome preparations were obtained from the differential ultracentrifugation without the rate-zonal centrifugation with sucrose gradient being performed. These preparations would contain not only exosomes, but also other extracellular vesicles such as microvesicles and membrane particles.

Another issue is that pathogenic proteins in association with exosomes represent only a minor part of the total secreted pathogenic proteins [14, 35]. Only a little fraction (<1%) of total secreted Aβ into the extracellular space is shown to be exosome-associated [35]. In case of α-synuclein, less than 3% of the total secreted protein seems to be associated with exosome [14]. However, low representation in quantity does not necessarily indicate insignificant functional roles. Danzer et al. [10] showed that exosome-associated α-synuclein was more toxic to cells than the exosome-free protein. However, since the study used a larger quantity of exosome-associated α-synuclein proteins than actual, the effects of actual concentration of exosome-associated α-synuclein remains unknown.

Perhaps, the most critical critique of the exosome theory of aggregate propagation may be that this theory relies mostly on phenomenology in that at least some secreted pathogenic proteins are present in the exosomes. There is little functional evidence which demonstrates the involvement of exosome-associated proteins in aggregate propagation. Genetic modifications of exosome formation and trafficking in cells and animals would be one way to resolve this issue. In light of this, the study by Hasegawa et al. [20] was alerting. In this study, when exosome biogenesis was inhibited by silencing the expression of VPS4, a protein necessary for exosome formation, secretion of α-synuclein was increased, rather than decreased. In the same study, the authors failed to detect α-synuclein proteins in exosome preparations from human CSF. However, this study did not show the effect of VPS4 silencing on intercellular a-synuclein transmission, leaving the possibility of exosome still being important for aggregate transmission.

Several mechanisms other than the exosome pathway have been proposed to explain cell-to-cell aggregate transmission (Fig. 3). The endosome recycling pathway is one of those mechanisms. This pathway verified its functional importance in the cellular amyloid precursor protein trafficking and Aβ generation [46]. Another unconventional secretion that might be involved in the interneuronal aggregate transmission is exophagy, which refers to the exocytosis mediated by fusion of autophagosome/amphisome with the plasma membrane. The release of α-synuclein monomer and aggregates was mediated by exophagy, when the autophagosome-lysosome fusion was impaired [2, 14]. In PC12 cell, tubulin polymerization-promoting protein (TPPP/p25α) co-localized with α-synuclein in autophagosomes and inhibited the fusion of autophagosomes with lysosomes [14]. This led to α-synuclein secretion into the media through exophagy [14]. This study also showed that the secretion was modulated by Rab27a, a regulator of the late endosomal and amphisomal exocytosis [14].

Recently, another unconventional exocytosis mechanism has been discovered and was named as misfolding-associated protein secretion or MAPS. This mechanism was activated when the proteasomal degradation was impaired [43]. Polyubiquitylated proteins were recruited to the cytosolic surface of ER through the interaction with USP19, an ER membrane-associated deubiquitylase [26, 43]. The misfolded proteins were then deubiquitylated and packaged in the late endosome-like compartments before secreted [26, 43]. The MAPS pathway is probably a part of cellular protein quality control system, removing misfolded proteins from cells; however, this pathway might also play an important role in aggregate propagation.

Alternatively, direct intercellular transmission of pathogenic proteins through the tunneling nanotubes was recently proposed [1, 42]. Tunneling nanotubes are protrusions of actin enriched structures extended from the plasma membrane. They can be as thick as 50 to 200 nm in diameter and be as long as up to several cell diameters [11]. In this case, the cytoplasm of neurons are directly connected, which makes exocytosis of aggregates obsolete. Although the evidence for transport of proteins and organelles through tunneling nanotubes has been accumulating in culture, the presence of tunneling nanotubes in vivo, especially in the brain, has not been proven yet.