Extracellular vesicles and Alzheimer’s disease in the novel era of Precision Medicine: implications for disease progression, diagnosis and treatment

Abstract

Extracellular vesicles (EVs), secreted membranous nano-sized particles, are critical intercellular messengers participating in nervous system homeostasis, while recent evidence implicates EVs in Alzheimer’s disease (AD) pathogenesis. Specifically, small EVs have been shown to spread toxic proteins, induce neuronal loss, and contribute to neuroinflammation and AD progression. On the other hand, EVs can reduce amyloid-beta deposition and transfer neuroprotective substances between cells, mitigating disease mechanisms. In addition to their roles in AD pathogenesis, EVs also exhibit great potential for the diagnosis and treatment of other brain disorders, representing an advantageous tool for Precision Medicine. Herein, we summarize the contribution of small EVs to AD-related mechanisms and disease progression, as well as their potential as diagnostic and therapeutic agents for AD.

1. Introduction

After two decades of failed therapeutic trials and one of the lowest success rates for drug development in the entire field of medicine, the Alzheimer’s disease (AD) scientific community has been increasingly focusing on the early stages of AD and the long pre-symptomatic phase of the disorder to prevent or attenuate brain and synaptic damage before it becomes irreversible. During the long pre-symptomatic phase of AD, which may last up to 20 years (Dubois et al., 2016), different pathogenic processes and factors such as genetics, inflammation, lifetime stress, insulin resistance, and co-morbidities may influence disease progression and related brain damage (Golde, 2022). The emerging view of Precision Medicine suggests that AD should not be approached as a unitary biologic entity; instead, efforts should focus on a better understanding of variables that affect AD initiation and progression, which will improve our comprehension of the disease etiopathogenesis and pathophysiology (Berkowitz et al., 2018). This may lead to the identification of AD sub-types and the development of tailored novel treatments, as well as preventive interventions. In this context, extracellular vesicles (EVs), a class of nanosized secreted membranous particles that includes exosomes and microvesicles, are increasingly shown to play important roles in AD pathomechanisms and contribute to the propagation of AD pathology across the brain. Moreover, their ability to cross the blood-brain barrier (Chen et al., 2016; Saint-Pol et al., 2020), coupled with their cargo-protecting capacity and molecular signature similar to that of their originating cell, make exosomes and other small EVs unique biomarker tools in the new era of Precision Medicine that is considered to stimulate the development of 1) prognostic tests to implement therapies at the pre-symptomatic phase; 2) diagnostic tools with increased discrimination between neurological disorders; 3) cell-specific drug delivery systems for a personalized approach to complex disorders such as AD.

2. The biogenesis and secretion of EVs

Under healthy conditions, the nervous system and, more specifically, the brain is characterized by coordinated responses and homeostatic balance generated by proper communication between various cell types. Cells coordinate their actions by communicating with one another to maintain brain homeostasis. Conversely, if any of the processes underlying brain homeostasis is dysregulated, pathology can ensue.

In the past, intercellular communication in the nervous system was thought to be mediated mainly through synapses, specialized structures where cellular appendices of one cell lie in close proximity to their functional counterparts on another cell. Now, cell-to-cell communication is known to also occur through a variety of additional mechanisms: paracrine transmission of hormones and neurotransmitters (Barker and Smith, 1980); gap junctions that connect the cytoplasm of neighboring cells; tunneling nanotubes that join distanced cells; and most recently, extracellular vesicles (EVs), which are nanosized cargo-loaded vesicles released from cells. The first description of such vesicles occurred in the 1960s when Bonucci observed that chondrocytes secret small vesicles of 100 nm in size (Bonucci, 1967), and Wolf showed that platelets release EVs with clotting activity (Wolf, 1967), which confirmed previous suggestions of their existence (Chargaff and West, 1946; Hougie, 1955; O’Brien, 1955). Later, in the 1980s, studies on reticulocyte transferrin receptor turnover established that exosome biogenesis is essential for membrane-quality control (Pan and Johnstone, 1983; Harding and Stahl, 1983). Finally, at the turning of the millennium, Amigorena’s group first proposed a role for exosomes in the communication between cells of the immune system, thus making their debut as mediators of cell-cell communication (Zitvogel et al., 1998).

According to their biogenesis, extracellular vesicles were originally classified into three categories: microvesicles, apoptotic bodies, and exosomes (see Fig. 1). Briefly, exosomes derive from the endolysosomal pathway and the fusion of multivesicular bodies (MVBs) with the plasma membrane, while microvesicles and apoptotic bodies arise from plasma membrane budding and blebbing, respectively. However, as current isolation and characterization methods are not able to separate and distinguish between EVs based on their biogenesis, the International Society of Extracellular Vesicles released the “Minimal information for studies of extracellular vesicles 2018” (MISEV2018) guidelines, suggesting a different categorization of EVs based on physical characteristics (size or density), biochemical composition or cell of origin (Théry et al., 2018). Recently, a novel non-membranous nanovesicle, termed exomere, was reported. The isolation of these smaller than 50 nm nanovesicles was possible by applying asymmetric-flow field-flow fractionation (AF4), and their proteomic characterization revealed proteins associated with hypoxia, coagulation and metabolism, microtubule-associated proteins, and proteins related to mTOR signaling and glycolysis (Zhang et al., 2018).

Fig. 1.

Extracellular vesicle life cycle. Extracellular vesicles (EVs) are membranous nano-sized particles released from the cell and are classically classified into exosomes, microvesicles, and apoptotic bodies. On the donor cell, exosomes derive from the endolysosomal pathway. Following endocytosis, early endosomes are generated and mature into late endosomes that will suffer further internal invagination to form intraluminal vesicles (ILVs). The MVBs fuse with the plasma membrane to release ILVs into the extracellular space as exosomes. Microvesicles are released through membrane budding, while apoptotic bodies are released by dying cells through the blebbing of the plasma membrane. Exomeres are a novel entity whose biogenesis is still being uncovered. Upon EVs release (the majority of studies have focused on small EVs – a mixture of exosomes and microvesicles), they are internalized by the recipient cell through several pathways such as receptor-mediated endocytosis (Dubois et al., 2016), macropinocytosis (Golde, 2022), phagocytosis (Berkowitz et al., 2018) and membrane fusion (Chen et al., 2016). Following their uptake by the recipient cell, EV cargo may undergo clearance or induce a cellular response.

EV biogenesis processes have often been categorized as ESCRT-dependent or ESCRT-independent, based on the contribution of the endosomal sorting complex required for transport (ESCRT) proteins. However, studies have shown that these systems may work synergistically to regulate exosome biogenesis (Babst, 2011; Gurunathan et al., 2021). For instance, different EV subpopulations might derive from different cellular pathways, while the cell type and its homeostatic state could also determine which cellular pathway is used for EV production. By far, the ESCRT-dependent mechanisms are the most extensively studied (Colombo et al., 2013; Katzmann et al., 2001), with ESCRT-independent mechanisms recently gaining attention after recent reports showed that lipids such as ceramide, as well as tetraspanins and heat shock proteins, can contribute to EV biogenesis (Stuffers et al., 2009; Matsuo, 2004). The fusion of MVBs with the plasma membrane to release exosomes is a complex process involving cytoskeletal proteins, molecular motors and switches, and fusion machinery (Hessvik and Llorente, 2018; Raposo and Stoorvogel, 2013).

3. Sorting of cargo into EVs

EV cargo is composed of bioactive molecules of great biological significance (e.g., proteins, lipids, nucleic acids) that have the capacity to influence the recipient cell. In addition, their composition greatly depends on the type and conditions of the parental cell and thus, reflects the proteomic, genomic, and homeostatic identity of that cell. The processes that govern protein or miRNA recruitment into exosomes are still poorly understood. It has been suggested that their cargo is passively accrued; however, recent studies have demonstrated multiple sorting systems that actively load molecules into exosomes (Fu et al., 2020). Protein loading into EVs uses different cellular mechanisms, including ESCRT components, ceramide, tetraspanins, and heat-shock proteins. ESCRT machinery recognizes ubiquitinylated proteins and is considered a crucial transporter of proteins to exosomes (Fu et al., 2020). Interestingly, recent evidence suggests a more prominent role of ESCRT-independent mechanism for cargo sorting. For instance, studies have suggested a mechanism of lateral cargo segregation dependent on lipid raft microdomains and lipids such as ceramide (Zhang et al., 2019). Similarly, tetraspanin-enriched microdomains containing CD81 were also shown to play a role in receptor sorting into small EVs (Perez-Hernandez et al., 2013). Furthermore, miRNAs loading into EVs may follow multiple sorting mechanisms (Janas et al., 2015). Similar to protein loading, miRNA sorting may be dependent on ESCRT-machinery (Gibbings et al., 2009). For instance, it has been described that miRNAs sequence motifs target them to exosomes. Indeed, heterogeneous nuclear ribonucleoprotein A2B1 (hnRNPA2B1) recognizes these motifs and controls exosomal loading (Villarroya-Beltri et al., 2013), thus mutations on these motifs or changes in the sumoylation levels of hnRNPA2B1 protein affect miRNA sorting to exosomes. In addition, the 3′-end miRNA sequence was also suggested to contribute to this sorting (Koppers-Lalic et al., 2014). Another identified pathway is related to ceramide and neural sphingomyelinase 2, as expression of this enzyme relates to miRNA exosomal content (Kosaka et al., 2013). In addition, RNA-binding proteins, Ago2, miRNA induced silencing complex (miRISC) were shown to interact with miRNA and contribute to its differential sorting into exosomes (van Niel et al., 2018).

Similarly, the sorting of DNA and nuclear proteins into exosomes is a matter of intensive research and remains largely unclarified. It was recently described a role for micronuclei (markers of genomic instability) for nuclear content loading into EVs by interacting with MVBs precursors and tetraspanins (Yokoi et al., 2019). Thus, although the mechanisms involved in cargo sorting into EVs require further exploration, it is evident that EVs’ cargo packaging occurs both randomly and selectively. Our understanding of these mechanisms may enable the curated target of EV loading processe for future patient-tailored therapeutic approaches.

4. EV uptake

Following their secretion, EVs facilitate intercellular communication locally and systemically. This message-relaying system occurs through different mechanisms (see Fig. 1). First of all, there is tropism between donor and recipient cells, where a conserved cellular signature acts as a recognition motif for the uptake by the same cell type (Sancho-Albero et al., 2019). Moreover, EVs can carry membrane-specific motifs that enable their recognition in distant locations. Upon reaching the cell of interest, EVs may interact with cell surface receptors, activating intracellular signaling cascades. Alternatively, they might be uptaken by the recipient cell through macropinocytosis, endocytosis, membrane fusion, or receptor-ligand mediated interactions (McKelvey et al., 2015). As will be discussed below, EVs are suggested to be involved in the propagation of disease pathology between synaptically-connected cells and brain areas in different brain pathologies. Thus, understanding the mechanisms of secretion and uptake of EVs may contribute to the understanding of the initiation of neurodegenerative and neurological disorders towards the identification of novel therapeutic targets that may slow down or halt disease progression. On the other hand, a better understanding of exosome uptake mechanisms may aid in the development of more reliable and targeted exosome delivery for therapeutic purpose as their ability to cross the blood-brain barrier, together with their reduced immunogenicity and cargo-transport properties, make EVs promising “tools” for efficient and biocompatible drug delivery (Whitford and Guterstam, 2019; Schiffelers et al., 2012).

5. EVs in Alzheimer’s disease brain pathology

AD constitutes a worldwide public health crisis with a significant societal and financial burden. Indeed, it is estimated that approx. 50 million people worldwide suffer from AD and other types of dementia; this number will double by 2030 and surge to more than 115 million by 2050 (Prince et al., 2013). Moreover, it is estimated that the worldwide costs of dementia were US$818 billion in 2015, with 86% being spent in high-income countries (Wimo et al., 2017). Nowadays, AD is widely accepted as a multifactorial neurodegenerative disorder characterized by progressive cognitive impairment, including memory loss and behavioral changes. Among several non-modifiable risk factors for AD, age is the most significant, while certain genetic factors cause an autosomal-dominant form of early-onset AD (e.g., APP and presenilin mutations) and other non-deterministic genetic factors increase the risk for late-onset/sporadic disease (e.g., APOE) (Querfurth and LaFerla, 2010; Reiman et al., 2005). Education level is the most prominent among the modifiable risk factors, with years of education contributing to cognitive reserve and resilience to AD (Stern, 2012), probably by modulating synaptic density. Moreover, lifetime chronic stress and elevated glucocorticoids appear to increase the risk for AD (Rasmuson et al., 2001; O’Brien et al., 1996), with GC levels correlating with the rate of cognitive decline (Lucassen et al., 2014; Lupien et al., 1998) and influencing AD pathogenic mechanisms (Sotiropoulos et al., 2019; Justice, 2018; Mohammadi et al., 2021).

AD histopathological hallmarks include extracellular senile plaques of amyloid β (Aβ) and intracellular neurofibrillary tangles of insoluble hyperphosphorylated Tau aggregates. In addition, synaptic loss is a major feature of this disease and the most significant correlate of cognitive decline. Similarly, Tau pathology degree and spatial distribution also correlate with disease severity and AD symptoms (Bejanin et al., 2017; Ossenkoppele et al., 2016). A recent human study has provided insight into the association between human Tau pathology and synaptic loss and altered synaptic function (Coomans et al., 2021), as it had previously been found in murine AD models (Lasagna-Reeves et al., 2011; Zhou et al., 2017). AD pathology is suggested to initiate at the entorhinal cortex, propagating to the hippocampus and eventually to the rest of the neocortex in a prion-like spreading pattern through mechanisms not yet fully understood. However, several studies have demonstrated the seeding potential of both free and exosomal Aβ and Tau, which play a role in the propagation of AD brain pathology (McAllister et al., 2020; Thompson et al., 2016). Different mechanisms are suggested to be involved in AD brain pathology, namely the disruption of cellular proteostasis (Labbadia and Morimoto, 2015), including protein folding and clearance mechanisms (Iliff et al., 2014; Tarasoff-Conway et al., 2015), astrogliosis, and inflammation. Below, we will focus on the relationship of EVs with these disrupted cellular processes (see Fig. 3), as well as their involvement in the propagation of Tau and Aβ (see Fig. 2).

Fig. 3.

Role of extracellular vesicles in the spreading of Аβ and Tau pathology. Due to their endocytic nature, extracellular vesicles (EVs) reflect the genomic and proteomic status of the parental cell. For this reason, neuron-derived EVs (neuro-EVs) and glia-derived EVs (glia-EVs) may contain pathological forms of Tau and Aβ, which are capable of triggering AD pathology in the recipient cells. Besides proteins, EVs carry miRNAs and mRNAs that will also induce a cellular response into the recipient cell. The spreading of EVs can be released from the diseased neuron (donor cell) and be uptaken by the healthy neuron (recipient cell). Alternatively, glia cells can also uptake neuro-EVs and, in their turn, can secrete exosomes (glia-EVs) which will be uptaken by neurons. In addition, as EVs can cross the blood-brain barrier, both neuro- and glia-EVs are found in the peripheral blood, where they can be differentially isolated using immunoprecipitation for specific cell-specific transmembrane markers, e.g., L1CAM for neuro-EVs.

Fig. 2.

EVs interplay with different cellular mechanisms found to be disrupted in AD. 1) Protein degradation - The dysregulation of protein homeostasis due to impaired lysosome degradation may increase MVBs number in cells which may lead to an increase of EVs secretion, which in turn contributes to the enhanced Aβ deposition; 2) Neuroinflammation - Inflammatory activation associated with cytokines release leads to EVs secretion from astrocytes (astro-EVs) which, in this turn, will activate microglia contributing further to the pro-inflammatory state and ultimately to neurodegeneration. 3) Neurovascular unit homeostasis - blood-brain barrier dysfunction and increased permeability is found in AD, probably due to damage of the neurovascular unit. While the role of EVs in neurovascular dyshomeostasis in AD is under intense investigation, recent evidence suggests that neuron-derived EVs (neuro-EVs) carrying specific miRNAs (e.g., miR-132) are able to regulate endothelial integrity. 4) Synaptic plasticity and homeostasis – EVs are important cell-to-cell messengers, and emerging evidence suggests that EVs secreted by neurons, astrocytes, and microglia have been shown to regulate different levels of synaptic homeostasis, including synaptic pruning and maintenance, neurotransmitter release and reuptake, proteins of postsynaptic density; all important components of learning and memory that are damaged in AD. 5) Neuroprotection - Glial cells contribute to neuronal homeostasis and may tip the balance towards neuronal malfunction under pathological conditions. For instance, astrocyte-derived EVs (astro-EVs) transfer synapsin and neuroglobin to neurons, contributing to neurite growth. Moreover, microglia-derived EVs (micro-EVs) and astro-EVs are also shown to stimulate oligodendrocyte precursor cells, promoting (re)myelination. In their turn, oligodendrocytes-derived EVs (Oligo-EVs) and micro-EVs may support neuronal metabolic activity and axonal transport, providing different ways through which EVs are involved in neuroprotection and neurodegeneration.

5.1. Disruption of protein homeostasis and EVs

The main routes for cellular protein degradation are the ubiquitinproteasome system (UPS), the endolysosomal pathway, and autophagy (Cuervo et al., 2010). Proteasomes are protein complexes wherein misfolded proteins are degraded following their targeting by ubiquitination (Tanaka, 2009). The endolysosomal pathway engulfs cytosolic and membrane content targeting it to multivesicular bodies (MVBs), which may fuse with lysosomes for content degradation. Lysosomes, the terminal compartments of the endolysosomal and autophagy pathways, have an acidic lumen for the degradative activity of hydrolases. Finally, autophagy, activated by cellular stress (e.g., starvation, accumulation of misfolded proteins), is responsible for the degradation of macromolecules and organelles.

Neuronal endosomal traffic jams were first described in AD when abnormal early endosomal enlargement was detected at near-diagnostic precision (Cataldo et al., 2000). Since then, abnormalities of the endolysosomal pathway have been commonly reported, even before protein aggregation is observed, suggesting a key etiopathogenic role of degradative compartments in the downstream molecular alterations leading to neurodegeneration (Nixon et al., 2008). It’s interesting to note that chronic stress, a risk factor for AD, was shown to impair these degradative pathways contributing to the accumulation of hyperphosphorylated Tau and facilitating its aggregation (Vaz-Silva et al., 2018; Silva et al., 2019). Interestingly, exosomes derive from the endolysosomal pathway and represent an alternative pathway for the disposal of unwanted molecules besides lysosomal degradation. Indeed, EV secretion is enhanced when neuronal lysosomes are impaired (Guix et al., 2017; Miranda et al., 2018), which, in turn, may contribute to the release of toxic Aβ species and the formation of Aβ plaques (Eitan et al., 2016). Supporting this concept, inhibition of neutral sphingomyelinase 2, an enzyme required for exosome biogenesis, has been shown to decrease amyloid deposition (Dinkins et al., 2016). Recently, brains of patients with the ApoE ε4 allele were shown to have reduced levels of TSG101 and Rab35, molecules involved in the ESCRT pathway (part of the endolysosomal mechanism of degradation) and also implicated in exosome biogenesis and secretion. These findings suggest that the ApoE ε4 genotype may predispose to decreased exosome release, in turn inducing endolysosomal damage and leading to neuronal vulnerability and a higher risk of AD (Peng et al., 2019). On the other hand, the EV-delivery of toxic Tau species also contributes to the malfunction of degradative organelles, further contributing to Tau accumulation (Pedrioli et al., 2021).

5.2. EVs in Aβ and Tau propagation

The abovementioned studies show a clear role for the endolysosomal and other degradative pathways in AD pathology, supporting a potential role for exosomes and other EVs in the spreading of toxic forms of Aβ and Tau. Accumulating evidence suggests that Tau spreading in the brain occurs via prion-like mechanisms between synaptically-connected brain regions (Clavaguera et al., 2009; Clavaguera et al., 2020), e.g., from the entorhinal cortex to the hippocampus and then to the neocortex (Wang et al., 2020). The specific mechanisms underlying the propagation of pathological Tau remain unclear; however, Tau secretion from neurons has been described to occur through different routes, i.e., free secretion, or through nanotubes or inside EVs (Zhang et al., 2021). Indeed, there is ample evidence of the presence of Tau in EVs and its ability to induce Tau aggregation in recipient cells. For instance, plasma neuronal-derived EVs from AD patients have been shown to seed Tau aggregation and induce AD-like neuropathology in wild-type mouse brains (Winston et al., 2016). Recently, brain-derived EVs of AD patients were also shown to induce Tau lesions in vivo, supporting the hypothesis that brain derived-EVs participate in the prion-like propagation of Tau pathology (Leroux et al., 2022). Moreover, EVs isolated from the brains of Tau transgenic mice or cultured primary neurons expressing mutant Tau can stimulate the aggregation of Tau in recipient cells and increase Tau phosphorylation and soluble Tau oligomer formation (Baker et al., 2016; Polanco et al., 2016; Wang et al., 2017a). Recent proteomic evidence further consolidated our knowledge of the involvement of different cell types in AD pathology (Bai et al., 2021), which is mimicked by brain-derived EVs multi-omic analysis (Muraoka et al., 2021; You et al., 2022; Cohn et al., 2021). Indeed, not only neuronal EVs but also microglial or astrocytic EVs have been shown to drive Tau pathology propagation. For instance, overexpression of BIN1, the second most statistically-significant locus associated with late-onset AD, promoted Tau release via microglial EVs and exacerbation of Tau pathology in P301S-Tau transgenic (Tg) mice (Crotti et al., 2019). Recently, it was also shown that plaque-associated microglia hyper-secrete EVs, contributing to Tau propagation in a humanized AD mouse model (Clayton et al., 2021). Moreover, inhibition of P2X purinoceptor 7 (P2RX7), an ATP-evoked Na⁺/Ca²⁺channel predominantly expressed in microglia, suppresses disease phenotypes in the P301S-Tau Tg mice, probably by decreasing the release of microglial EVs (Ruan et al., 2020). Moreover, Asai and colleagues showed that microglia phagocyte and release Tau via EVs, while inhibition of either microglia function or exosome biogenesis was shown to halt Tau pathology propagation in P301S-Tau Tg mice (Asai et al., 2015).

Similarly, EVs derived from AD patients’ brain that contained Aβ oligomers were found to transfer such toxic Aβ species to recipient neurons in culture-inducing toxicity (Sardar Sinha et al., 2018). Concerning the role of astrocyte-derived EVs, recent evidence showed that, after engulfing Aβ fibrils resistant to lysosomal degradation (Söllvander et al., 2016), astrocytes caused neurotoxicity by releasing EVs that contained an N-terminus-truncated form of Aβ42, which is even more resistant to degradation, more prone to aggregation and more toxic (De Kimpe et al., 2013). Moreover, microglial EVs can also contain toxic forms of Aβ1–42 and Aβ1–40, supporting a detrimental role for microglia in AD, which can mediate the spreading of neurotoxic Aβ species throughout the brain (Joshi et al., 2014). Moreover, it is important to note that Aβ spreading is also dependent on neuronal EVs that appear to accelerate Aβ amyloidogenesis (Yuyama et al., 2012). On the other side, neuronal EVs were also shown to exhibit a protective role by inhibiting Aβ oligomerization and enhancing microglia-mediated Aβ clearance in vitro (Yuyama et al., 2012). By trapping Aβ through membrane glycosphingolipids and then transporting it into microglia, neuronal EVs were also shown to decrease Aβ pathology (Yuyama et al., 2014).

Although the vast majority of evidence monitoring the transmission of protein aggregates in neurodegenerative disorders focuses on the brain, studies have also described this transmission from the periphery to the central nervous system. For example, intraperitoneal administration of brain extracts containing Aβ or Tau triggers their accumulation in the brain, while parabiosis between Aβ transgenic and wild-type mice results in Aβ accumulation in the brain of wild-type animals suggesting that Аβ or Tau circulating in the blood can enter the central nervous system (Peng et al., 2020). However, the role of EVs in this periphery-to-central nervous system propagation remains unclear.

5.3. Exosomes and neuronal stem cells in brain homeostasis

Emerging evidence suggests a role for EVs and exosomes originated from neuronal stem cells (NSCs) as possible players of brain homeostasis. NSCs are multipotent cells characterized by self-renewal, paracrine transmission, and inducible differentiation into three cell lineages of the central nervous system: neurons, astrocytes, and oligodendrocytes (Ling et al., 2020).

NSCs-derived EVs appear as key “players” in the EVs-based cross-talk among the different adult neurogenic niches such as the hypothalamus, hippocampal subgranular zone, and sub-ventricular zone (SVZ) of the lateral ventricles (Sardar Sinha et al., 2018). A recent study supports the essential role of NSC-derived EVs in regulating the transition of NSCs from the quiescence to proliferation by regulating protein translation related to the cell cycle. In addition, proteomic analysis of the cargo of NSCs-derived EVs revealed that they are enriched in growth-factor proteins, while treatment of neuronal progenitor cells (NPCs) with NSC-derived EVs promotes NPC proliferation (Ma et al., 2019).

Despite their participation in paracrine communication, EVs from NSCs are also proposed as promising “actors” under pathological conditions with therapeutic potential (Lee et al., 2018a). An increasing number of studies show that NSC-derived EVs carrying a “pallet” of proteins and miRNAs are involved in biological processes necessary for maintaining brain homeostases, such as inflammation, oxidation, and apoptosis (Upadhya et al., 2020a). For instance, SVZ NSC-derived EVs have immunomodulatory and inflammatory properties acting as microglia morphogens, e.g., activating the STAT1 pathway in target cells (Ruan et al., 2021). Moreover, NSC-derived EVs were also shown to have a neuroprotective effect in vitro against Parkinson’s disease (Lee et al., 2022).

5.4. EVs role in synaptic plasticity and neuronal function

Complex behavioral responses, such as learning and memory, require the dynamic refinement of synaptic connections by eliminating specific synapses in a process known as synaptic pruning (Goda and Davis, 2003). Glial cells have been shown to play an important role in this process (Eroglu and Barres, 2010) in a complement-dependent manner and are triggered by neuronal activity (Hong et al., 2016; Schafer et al., 2012). Indeed, neuronal EVs secreted by depolarized PC12 cells increased the expression of C3 (complement system protein of immune response) in microglial cells, resulting in enhanced clearance of neurites (Bahrini et al., 2015). Moreover, neuronal exosomes carrying PRR7 (a transmembrane protein) were shown to contribute to the elimination of excitatory synapses in neighboring neurons by i) blocking the secretion of Wnt via exosomes (secreted Wnt plays a critical role in synapse formation and maintenance) and ii) promoting proteasomal degradation of post-synaptic density (PSD) proteins (Lee et al., 2018b). Also, astrocytic EVs regulate neuronal protein expression and dendritic complexity by delivering several miRNAs (miR-26a-5p, miR-29c, miR-130a) to neurons (Lafourcade et al., 2016; Zou et al., 2015; Zhang et al., 2016). In addition, in response to ATP and IL-10 (anti-inflammatory cytokine), astrocytic EVs are thought to exhibit a role in neuronal plasticity as they are enriched in proteins involved in neurite outgrowth, axonal guidance, synaptogenesis, and long-term synaptic potentiation regulating gap junction, and CREB signaling (Datta Chaudhuri et al., 2020; Patel and Weaver, 2021; Peng et al., 2022). Besides being involved in the regulation of synapse pruning and maintenance, EVs also influence synaptic transmission. Different studies have shown a role for EVs in regulating postsynaptic retrograde signaling (Korkut et al., 2013) and presynaptic modulation. Furthermore, microglial EVs have been shown to increase neurotransmitter release, enhance excitatory neuro-transmission by transferring endocannabinoids to neurons, and promote sphingolipid metabolism (Antonucci et al., 2012; Gabrielli et al., 2015). EVs also play a role in the “tripartite synapse”, in which neuronal EVs transfer miR124-3p to astrocytes, thereby inactivating other miRNAs and increasing the expression of astrocytic GLT1 transporter (Morel et al., 2013; Men et al., 2019), which, in turn, favorites for glutamate uptake from the synaptic cleft. Noteworthy, EV release itself is stimulated by synaptic activity and glutamate in neurons and oligodendrocytes (Lachenal et al., 2011) and by ATP or inflammatory stimulus in other glial cells.

In addition to regulating synaptic transmission, EVs also regulate cellular mechanisms involved in neuroprotection and myelination. For example, it is shown that astrocytic EVs facilitate neuroprotection (Upadhya et al., 2020b; Guitart et al., 2016; Pascua-Maestro et al., 2019) by transferring neuroglobin (Venturini et al., 2019) and synapsin-1 (Wang et al., 2011) to neurons that contribute to neurite outgrowth and neuronal survival. Microglial and astrocytic EVs were also found to promote (re)myelination by stimulating oligodendrocyte precursor cell migration and differentiation (Lombardi et al., 2019; Willis et al., 2020). In addition, oligodendroglial EVs protect neurons from stress conditions, supporting higher metabolic activity and axonal transport (Frühbeis et al., 2013; Frühbeis et al., 2020; Mukherjee et al., 2020; Krämer-Albers et al., 2007; Fröhlich et al., 2014). Moreover, oligodendroglial EVs communicate with other glial cells for clearance and immune surveillance of the microenvironment. Indeed, microglia cells are the scavengers of the nervous system, phagocyting toxic substances, and EVs. While the uptake of neuronal EVs loaded with misfolded proteins may activate microglial cells, glioma EVs and oligodendroglial EVs are shown to maintain microglia in an immunologically silent state (Paolicelli et al., 2019; Fitzner et al., 2011; Abels et al., 2019; Maas et al., 2020). Microglial EVs have also been shown to contribute to the metabolic support of neurons, providing supplementary energy substrates during synaptic activity (Potolicchio et al., 2005). In addition, under several conditions, EVs from microglia were shown to contribute to neuronal survival by transferring miR-124, which induces neurite outgrowth (Song et al., 2019; Huang et al., 2018; Li et al., 2019).

As mentioned previously, one of the hallmarks of AD is synaptic loss and impaired synaptic homeostasis. The mechanisms that lead to this dysfunction have yet to be understood. Aβ oligomers and plaques are known to disrupt synaptic function (Laurén et al., 2009; Walsh et al., 2002; Shankar et al., 2008), and some studies indicate that EVs neutralize these effects (Yuyama et al., 2012; Bulloj et al., 2010; An et al., 2013), while others suggest that they have a role in mediating Aβ neurotoxicity (Elsherbini et al., 2020a; Elsherbini et al., 2020b). Indeed, Aβ in astrocytic EVs is internalized by neurons and was shown to induce mitochondrial clustering and dysfunction, increasing DRP1 levels and activating caspases, ultimately leading to neuronal death (Elsherbini et al., 2020a; Elsherbini et al., 2020b).

5.5. EVs and brain inflammation

Genome-wide association studies of AD patients have reported that more than 2/3 of single nucleotide polymorphisms occur in genes that encode proteins expressed in microglia and are related to inflammation, endosomal pathways, and cholesterol metabolism (ElAli and Rivest, 2016). Indeed, AD has been associated with chronic innate inflammation in the CNS, where activated microglia and reactive astrocytes increase the production of cytokines and chemokines and lead to the infiltration of immune cells, culminating in secondary neurodegeneration (Calsolaro and Edison, 2016; Czirr and Wyss-Coray, 2012). Although the initial activation of microglia is beneficial for clearing toxic Aβ from the AD brain, over time, the chronic stimulation of microglia by Aβ may also be deleterious, leading to prolonged inflammation, excessive Aβ deposition, and acceleration of the neurodegenerative process (Sala Frigerio et al., 2019; Hickman et al., 2008). Astrocytic EV release is promoted by pro-inflammatory cytokines (Wang et al., 2017b), which recruits peripheral leukocytes (Dickens et al., 2017), indicating a role for astrocytic EVs in the modulation of neuroinflammation. Astrocytic EVs containing miR-138 can activate microglia via TLR7 signaling (Liao et al., 2020), while those enriched in miR-873a-5p or miR-9 have been shown to attenuate microglia-mediated neuroinflammation (Long et al., 2020; Yang et al., 2018a). In addition, exosomes from reactive astrocytes are enriched in miRNAs that target neurogenesis and synaptogenesis (Datta Chaudhuri et al., 2020; Chaudhuri et al., 2018), increasing neuronal uptake and reducing neurite outgrowth, branching, and neuronal firing (You et al., 2020), as well as having axonotrophic effects following spinal cord injury (Adolf et al., 2019). These findings suggest a beneficial role for astrocytic EVs in neuronal function under inflammatory conditions. Microglial EVs and their cargo reflect the phenotype of the parent cells and may promote inflammation under some circumstances (depending on cytokine cargo, e.g., IL1-β, TNF-α) (Yang et al., 2018b). Previous studies have demonstrated that abnormal elevation of glutaminase 1 in microglia promotes their switch to a pro-inflammatory phenotype and exosome release in an AD mouse model (Gao et al., 2019). In addition, active microglia deliver EV-associated microRNA 146-a-5p to neurons promoting synaptic weakening (Prada et al., 2018) and neuronal death (Beneventano et al., 2017; Tang et al., 2016). On the other hand, neuron-derived EVs with high levels of miR-21-5p trigger neuroinflammation by activating microglia (Yin et al., 2020). Recently, EVs from mesechrymal stem cells (MSC-EVs) were also shown to inhibit glial activation through miRNAs, reversing brain inflammation (Markoutsa et al., 2022; Losurdo et al., 2020). In summary, neuroinflammation is an important aspect of AD pathogenesis, and whether exosomes promote or inhibit inflammatory responses depends on their cargoes and the cellular conditions under which they are released. Clearly, the mechanisms underlying beneficial and detrimental effects of EVs on inflammation in the context of Alzheimer’s disease need further investigation.

5.6. Neurovascular homeostasis and EVs

The neurovascular unit (NVU) is an anatomical structure composed of neurons, glial cells, brain endothelial cells, pericytes, and smooth muscle cells. Its function is to promote an effective cerebral blood flow, maintaining neuronal metabolic activity and a functional blood-brain barrier (BBB) (Soto-Rojas et al., 2021). Dysfunction of the NVU is an early event in AD and a reliable predictor of cognitive deficits (Thal et al., 2003; Nation et al., 2019; van de Haar et al., 2016). Aβ deposits have been associated with deterioration of the NVU, decreased cerebral lymphatic and blood flow, leading to reduced Aβ clearance (Bakker et al., 2016; Kimbrough et al., 2015). Similarly, Tau perivascular deposits have also been reported with Tau-mediating vascular defects preceding neuronal loss (Canepa and Fossati, 2021). Little is known about the role of EVs in NVU homeostasis, although there are reports that EVs contribute to neurovascular repair processes (Zhang et al., 2015; Dong et al., 2022). In addition, EV-associated miR-132 from neurons has been shown to regulate endothelial cadherin and maintain vascular integrity (Xu et al., 2017). Recently, astrocytic-EVs derived from AD brains were shown to impair neuroglial and vascular components (González-Molina et al., 2021). A deeper understanding of the crosstalk between NVU elements and the role of EVs in NVU homeostasis could promote novel interventions to halt or slow disease progression.

6. Technical difficulties related to the isolation and study of brain EVs

The literature shows that neuronal and glial EVs impact brain homeostasis and function. The seemingly contradictory roles of EVs in these processes derive from the fact that it is challenging to study exosomes secreted by neurons or glial cells in vivo. Therefore, most of the studies to date have taken in vitro approaches to explore the role of EVs in AD pathomechanisms, which may not fully recapitulate the endogenous mechanisms in the brain. In addition, the conditions in which EVs are collected and isolated greatly influence their content and biological function (Moloudizargari et al., 2021). Therefore, the collection and characterization of physiologically relevant exosomes are of the utmost importance. However, it is still challenging to identify, isolate and quantify exosomes efficiently, accurately, and selectively. For instance, the first protocol for brain exosome isolation proposed in 2012 (Perez-Gonzalez et al., 2012), and its following adaptations (Vella et al., 2017), rely on tissue digestion. Although it has provided invaluable insight into the role of EVs and, particularly, exosomes in brain pathology, this method of exosome isolation is based on brain tissue digestion which may contaminate the exosome fraction by cellular disruption/damage; here, intracellular components (e.g., immature vesicles intraluminal vesicles (ILVs) that could otherwise be targeted for degradation instead of exocytosis – see Fig. 1) and artificially produced vesicles may be isolated together with EVs/exosome fraction (Vella et al., 2017). As it is not possible to discriminate between ILVs that would be secreted (exosomes) from those that would be degraded, tissue dissociation approaches may lead to less pure or contaminated EVs yield, leading to the conflicting, incongruent, or inaccurate characterization of exosomes profile. Our team has recently developed a novel protocol for the purification of small EVs that are spontaneously released by brain tissue. This novel protocol allows for the isolation of small EVs from different brain areas in both mouse and human, avoiding the mechanical disruption and subsequent enzymatic digestion required for the current method of EV isolation from brain tissue, which may contaminate the EV fraction with immature vesicles and influence EV cargo.

Moreover, the seemingly opposing roles for EVs described above may be due to the fact that these effects are mediated by a different sub-population of EVs. Thus, great efforts are being undertaken to provide novel tools for the study of such subpopulations of vesicles. Novel techniques such as asymmetric-flow field-flow fractionation allow for differential isolation of subpopulations of small vesicles (Zhang et al., 2018), and a new commercial device, ExoView, allows for the study of specific subpopulations based on their membrane proteins – e.g., tetraspanin-specific, neuronal or microglial specific (Silva et al., 2021). In addition, novel “exosome reporter” mice have enabled the isolation and characterization of cell type-specific EVs in vivo (McCann et al., 2020), the temporal and spatial labeling of exosomes (Luo et al., 2020), and the study of neuron-glia communication (Men et al., 2019). Moreover, in vivo tracking of EVs has enabled the dissection of their trafficking and communication routes – indeed, several labeling methods have been developed for this purpose (Gangadaran et al., 2018; Carpintero-Fernández et al., 2017). Another particularly interesting technique is immunoprecipitation, which takes advantage of the membranous content of EVs to capture specific populations; for instance, it is common to use L1CAM or NCAM to immunoprecipitate neuron-derived EVs, while astrocytic EVs precipitate with GLAST. This is essential to study the role of specific EV populations in a given biological process, including brain pathology, and their particular contribution to AD biomarker discovery (Delgado-Peraza et al., 2021).

7. EVs as a tool for clinical studies and medical practice

Due to their small size, EVs are able to cross the blood-brain barrier and the CSF-brain barrier, although unequivocal demonstration of this crossing under healthy conditions is largely lacking (Chen et al., 2016; Haqqani et al., 2013; Matsumoto et al., 2017). Nevertheless, EVs with brain cell cargos have been found in several biofluids (e.g., CSF, plasma, urine, saliva). This feature provides a unique opportunity for biomarker research, as we can easily access brain-derived EVs, which may mirror brain pathology in the periphery, e.g., peripheral blood (Delgado-Peraza et al., 2021). In addition, their low immunogenicity also makes them perfect candidates for customized drug delivery systems.

EV cargo with biomarker potential for Alzheimer’s disease diagnosis spans multiple biomolecules, i.e., proteins, miRNA, mRNA, and crDNA, reflecting the biological processes associated with AD brain pathology. Due to their cargo-protecting abilities, EV-related proteins have been studied for their diagnostic potential (Arioz et al., 2021) as well as their capacity to distinguish pathologies or offer a prognosis for dementia (Jia et al., 2021). Therefore, small EVs, such as exosomes, represent a form of liquid biopsy that seems to be a powerful tool in the new era of Precision Medicine (Hampel et al., 2019) and of particular importance for the early detection of AD. Moreover, as described above, EVs seems to participate in disease progression and may also be promising candidates for novel therapies as will be discussed below.

8. The biomarker potential of EV miRNAs

miRNAs are small non-coding RNAs that modulate RNA expression. Recently, the dysregulation of miRNA has been associated with AD pathological status (Olsson et al., 2016), and while they may be found freely in body fluids, they are significantly more stable in small EVs, such as exosomes (Cheng et al., 2014; Kim et al., 2017). Thus, a growing number of studies have looked at the exosomal miRNA potential for diagnosis (Xing et al., 2021; He et al., 2021; Dong et al., 2021; Dong et al., 2020). Interestingly, the delivery of some miRNA via exosomes has also been studied with promising therapeutic results and without short-term side effects (He et al., 2021).

So far, exosomal miRNA studies in AD have been cross-sectional, focusing on plasma, serum, and CSF exosomes allowing the distinction of AD from other neurodegenerative disorders (e.g., Parkinson’s disease [PD], Dementia with Lewis Bodies [DLB]), and healthy conditions. Indeed, Nie et al. (Nie et al., 2020) identified eight miRNA that were differently expressed in AD patients, with let-7e-5p and miR-548 being differently expressed in comparison with PD patients and thus, being potential differencing markers. Similarly, Gámez-Valero et al. (Gámez-Valero et al., 2019) have identified two miRNAs that differentiate AD from DLB patients (miR-451a and miR-21-5p), while four miRNAs are downregulated in AD patients compared to healthy controls. Interestingly, miR-23a-3p is downregulated in this study, while others (Nie et al., 2020; Serpente et al., 2020) have shown its upregulation in AD patients. These incongruencies may be due to the different isolation methods used and miRNA sequencing techniques (SEC and NGS (Gámez-Valero et al., 2019) vs. Exosome isolation Kit and small RNA seq (Nie et al., 2020)). Furthermore, a study by Lugli et al. (Lugli et al., 2015) proposes a panel of 7 miRNAs for the diagnosis of AD with an accuracy near 90%. Serpente et al. and Cha et al. (Serpente et al., 2020; Cha et al., 2019) have isolated neuronal-derived EVs from the plasma, identifying that miR-23a-3p, miR-223-3p, miR-190a-5p were upregulated in AD subjects, while miR-132 was able to distinguish AD patients from controls but not MCI-AD from healthy subjects. This may suggest that different miRNAs would be altered in different stages of dementia. Briefly, serum-derived exosomal miRNA have also demonstrated diagnostic potential with Cheng et al., proposing a model with miRNAs and other serum factors with 87% sensitivity and 77% specificity (Cheng et al., 2015). Similarly, Barbagallo et al. created a model to discriminate among different neurodegenerative processes using a panel of exosomal miRNAs (Barbagallo et al., 2020). Furthermore, Li et al. generated a model, including miRNAs and pSer396-Tau, that was able to distinguish AD from vascular dementia and MCI (Li et al., 2020). Another study by Yang et al. demonstrated the ability of miR-200b to distinguish between MCI and AD, while it provided novel evidence of the importance of miR-135a, miR-200a, and miR-42a in AD progression. Moreover, plasma EVs miR-233 in AD patients are shown to be correlated with cognitive scores of the patients (Mini-Mental State Examination and Clinical Dementia Rate scores) (Wei et al., 2018). Lastly, the most common biofluid used in AD diagnosis is CSF, from which exosomes have also been isolated and studied. Based on CSF exosome analysis, Riancho et al. identified exosomal miR-9-5p and miR-598 as potential AD diagnostic markers (Riancho et al., 2017), while another study by McKeever et al. provided evidence of miRNAs biomarkers in early- and late-onset AD patients (McKeever et al., 2018). A summary of the studies that identified exosomal miRNAs in AD patients is provided in Table 1.

Table 1.

Exosomal miRNAs as biomarkers for AD.

Study	Source	Down-regulated miRNAs	Up-regulated miRNAs	Known target genes (relevant to AD pathogenesis)
Riancho et al., 2017	CSF	miR-9-5p
	CSF	miR-16-2
McKeever et al., 2018	CSF	miR-16-5p		ADAM10; APH1A; APP; MME; CDK5R1
Gui et al., 2015	CSF	miR-29c		BACE1; GAPDH; LPL; CASP7; GSK3β
McKeever et al., 2018	CSF		miR-125-5p	APP
Gui et al., 2015	CSF		miR-132-5p
McKeever et al., 2018	CSF		miR-125b-5p
Gui et al., 2015	CSF	miR-136-3p
Liu et al., 2014	CSF	miR-193b
Gui et al., 2015	CSF	miR-331-5p
McKeever et al., 2018	CSF	miR-451a
Gui et al., 2015	CSF		miR-485-5p
Riancho et al., 2017	CSF	miR-598
McKeever et al., 2018	CSF	miR-605-5p
Nie et al., 2020	Plasma	let-7e-5p
Gámez-Valero et al., 2019	Plasma	let-7i-5p		APH1A; IDE; LRP1; CASP3
Nie et al., 2020 Serpente et al., 2020 Gámez-Valero et al., 2019	Plasma	miR-23a-3p	miR-23a-3p	ADAM10; PSEN1; NCSTN; APH1A; CASP3; CASP7
Lugli et al., 2015	Plasma	miR-23b-3p		ADAM10; APH1A; GAPDH; NCSTN; PSEN1; ASP3; CASP7
Lugli et al., 2015	Plasma	miR-24-3p		APH1A; NCSTN; CAPN1; GSK3β
Lugli et al., 2015	Plasma	miR-29b-3p		BACE1; GAPDH; LPL; CASP7; GSK3β
Serpente et al., 2020	Plasma	miR-100-3p
Nie et al., 2020	Plasma	miR-125a-5p
Lugli et al., 2015	Plasma	miR-125b-5p		APP
Gámez-Valero et al., 2019	Plasma	miR-126-3p		CAPN1
Cha et al., 2019	Plasma	miR-132		APH1A; CAPN2; CASP7; GSK3β
Lugli et al., 2015	Plasma		miR-138-5p
Lugli et al., 2015	Plasma	miR-139-5p
Lugli et al., 2015	Plasma	miR-141-3p		APH1B; GAPDH; PSEN1
Lugli et al., 2015	Plasma	miR-150-5p
Gámez-Valero et al., 2019	Plasma	miR-151a-3p		GAPDH; APH1A
Lugli et al., 2015	Plasma	miR-152-3p
Lugli et al., 2015	Plasma	miR-185-5p		BACE2; GAPDH; LPL; MME; GSK3β
Serpente et al., 2020	Plasma		miR-190a-5p
Nie et al., 2020	Plasma	miR-204-5p
Cha et al., 2019	Plasma	miR-212		CAPN2; GSK3β
Serpente et al., 2020	Plasma		miR-223-3p
Lugli et al., 2015	Plasma	miR-338-3p
Lugli et al., 2015	Plasma	miR-342-3p		APH1B; MME; LPL; GSK3β
Lugli et al., 2015	Plasma	miR-342-5p		GAPDH; LRP1
Nie et al., 2020	Plasma		miR-369-5p
Nie et al., 2020	Plasma	miR-375
Nie et al., 2020	Plasma		miR-423-5p
Lugli et al., 2015	Plasma		miR-659-5p
Nie et al., 2020	Plasma	miR-1468-5p
Lugli et al., 2015	Plasma	miR-3065-5p
Lugli et al., 2015	Plasma	miR-3613-3p
Lugli et al., 2015	Plasma	miR-3916
Lugli et al., 2015	Plasma	miR-4772-3p
Lugli et al., 2015	Plasma		miR-5001-3p
Li et al., 2020 Cheng et al., 2015	Serum		miR-15a-5p	ADAM10; APH1A; APP; BACE1; MME; CDK5R1
Li et al., 2020 Cheng et al., 2015	Serum	miR-15b-3p
Li et al., 2020 Cheng et al., 2015	Serum		miR-18b-5p	APH1B; GAPDH
Cheng et al., 2020 Li et al., 2020 Cheng et al., 2015	Serum		miR-20a-5p	APP
Barbagallo et al., 2020	Serum		miR-29a	APH1A; GAPDH; BACE1; BACE2; LPL; GSK3β; CASP7
Li et al., 2020 Cheng et al., 2015	Serum		miR-30e-5p	APP; NAE1; CASP3
Cheng et al., 2020	Serum		miR-32-5p
Barbagallo et al., 2020	Serum		miR-34b
Li et al., 2020 Cheng et al., 2015	Serum		miR-93-5p	APP; GAPDH; LRP1; GSK3β
Li et al., 2020 Cheng et al., 2015	Serum		miR-101-3p	ADAM10; APP; PSEN1; CAPN2; CASP3; GSK3β
Li et al., 2020 Cheng et al., 2015	Serum		miR-106a-5p	APP; ADAM17; CAPN2
Li et al., 2020 Cheng et al., 2015	Serum		miR-106b-5p	APP; CASP7
Yang et al., 2018c	Serum		miR-135a
Li et al., 2020 Cheng et al., 2015	Serum		miR-143-3p	ADAM10; LRP1
Yang et al., 2018c	Serum	miR-193b
Cheng et al., 2020	Serum		miR-219a
Wei et al., 2018	Serum	miR-223
Li et al., 2020 Cheng et al., 2015	Serum		miR-335-5p	APBB1; LRP1; CASP7
Li et al., 2020 Cheng et al., 2015	Serum	miR-342-3p		APH1B; MME; LPL; GSK3β
Li et al., 2020 (Cheng et al., 2015)	Serum		miR-361-5p	ADAM10; APH1B; LRP1
Cheng et al., 2020	Serum		miR-374a-5p
Yang et al., 2018c	Serum		miR-384
Li et al., 2020 Cheng et al., 2015	Serum		miR-424-5p	APBB1; APH1A; APP; LRP1; CDK5R1; GSK3β
Li et al., 2020 Cheng et al., 2015	Serum		miR-582-5p	IDE
Cheng et al., 2020	Serum		miR-585-5p
Cheng et al., 2020	Serum		miR-941
Li et al., 2020 Cheng et al., 2015	Serum	miR-1306-5p
Cheng et al., 2020 Li et al., 2020 Cheng et al., 2015	Serum		miR-3065-5p
Cheng et al., 2020	Serum		miR-3157-5p

9. EVs proteins as potential biomarkers for AD

Similar to exosomal miRNA content, the protein load of brain-derived exosomes constitutes a promising pool of molecules with great biomarker potential for AD (see Table 2 and Fig. 4). These EVs proteins include Aβ and Tau, synaptic proteins, lysosomal function, inflammation, transcription, and other processes (Kim et al., 2021). Besides their diagnostic value in the clinical phase, it is of extreme importance to uncover biomarkers for the early detection of AD. Thus, several groups have undertaken longitudinal studies that unraveled some potential markers for early detection of AD, namely lysosomal proteins (Goetzl et al., 2015), synaptic proteins (Jia et al., 2021), and others, including Aβ42, phosphorylated Tau, total Tau, that have prognostic value with a lead time of 10 years (Fiandaca et al., 2015; Kapogiannis et al., 2015; Eren et al., 2020; Kapogiannis et al., 2019). In line with the idea for better AD biomarkers for disease progression and treatment-response, synaptic proteins in exosomes are shown to decline as the disease progresses (Goetzl et al., 2018a; Chanteloup et al., 2019). Moreover, complement protein levels in astrocyte-derived exosomes may predict conversion from MCI to AD (Winston et al., 2019). Interestingly, Tau protein in neuron-derived EVs accompanies the changes in the scores of cognitive function and depression in AD patients (Mini-Mental State Examination and Geriatric Depression Scale, respectively (Nam et al., 2020). Changes in the levels of the above exosomal proteins upon treatment are promising game-changers in clinical trials for new drugs for Alzheimer’s disease.

Table 2.

Exosomal protein as biomarkers in AD.

Study	Exosomal proteins	Change	Specimen	Patients
Gu et al., 2020) (Jia et al., 2019 Abner et al., 2016 Goetzl et al., 2016a Winston et al., 2016 Fiandaca et al., 2015	Aβ42	Increased	Plasma	AD
Gu et al., 2020) (Jia et al., 2019 Kapogiannis et al., 2019 Goetzl et al., 2016a Abner et al., 2016 Winston et al., 2016 Fiandaca et al., 2015	p-T181-Tau	Increased	Plasma	AD
Jia et al., 2019 Fiandaca et al., 2015	t-Tau	Increased	Plasma	AD
Goetzl et al., 2016a Winston et al., 2016 Fiandaca et al., 2015	p-S396-Tau	Increased	Plasma	AD
Abner et al., 2016 Goetzl et al., 2015	cathepsin D	Increased	Plasma	AD & MCI
Goetzl et al., 2016a	BACE-1	Increased	Plasma	AD & MCI
Goetzl et al., 2016a	γ-secretase	Increased	Plasma	AD, MCI & early dementia
Goetzl et al., 2016a	sAPPβ	Increased	Plasma	AD, MCI & early dementia
Goetzl et al., 2016a	sAPPα	Increased	Plasma	AD, MCI & early dementia
Goetzl et al., 2015	Heat-Shock Protein 70 (HSP70)	Decreased	Plasma	AD
Jia et al., 2021 Abner et al., 2016 Winston et al., 2016	Neurogranin (NRGN)	Decreased	Plasma	AD
Jia et al., 2021 Agliardi et al., 2019	Synaptosome Associated Protein 25 (SNAP25)	Decreased	Blood	AD, amnestic MCI
Jia et al., 2021 Goetzl et al., 2016b	Growth Associated Protein 43 (GAP43)	Decreased	Blood	AD, amnestic MCI
Jia et al., 2021 Goetzl et al., 2016b	Synaptotagmin-1	Decreased	Blood	AD, amnestic MCI
Goetzl et al., 2016b	Synaptophysin	Decresead	Plasma	AD
Goetzl et al., 2016b	Synaptopodin	Decreased	Plasma	AD
Goetzl et al., 2016b	pS9-synapsin	Decreased	Plasma	AD
Goetzl et al., 2016b	Synapsin	Decreased	Plasma	AD
Kapogiannis et al., 2015	Total insulin receptor substrate 1 (t-IRS-1)	Decreased	Plasma	Preclinical AD
Kapogiannis et al., 2019 Kapogiannis et al., 2015	Phosphorylated-serine 312-type 1 insulin receptor substrate (p-S312-IRS-1)	Increased	Plasma	Preclinincal AD
Kapogiannis et al., 2019 Kapogiannis et al., 2015	ratio of P-serine 312-IRS-1 to P-pan-tyrosine-IRS-1 (p-S312-IRS-1/p-Y-IRS-1)	Increased	Plasma	Preclinincal AD
Goetzl et al., 2015	LAMP-1	Increased	Plasma	AD
Goetzl et al., 2018b	Complement proteins	Increased	Plasma	AD
Abner et al., 2016 Winston et al., 2016	REST	Increased Decreased	Plasma	AD
Goetzl et al., 2016b	MOG	Decreased	Plasma	AD
Goetzl et al., 2016a	GDNF	Decreased	Plasma	AD, MCI & early dementia
Goetzl et al., 2016a	GFAP	Decreased	Plasma	AD, MCI & early dementia
Goetzl et al., 2016a	GluSyn	Decreased	Plasma	AD, MCI & early dementia
Goetzl et al., 2016a	Neuron-specific enolase	Decreased	Plasma	AD, MCI & early dementia
Goetzl et al., 2016a	Septin-8	Decreased	Plasma	AD, MCI & early dementia
Gu et al., 2020	Metalloproteinase 9 (MMP-9)	Increased	Plasma	AD

Fig. 4.

EVs proteins in AD diagnostics. Retrospective and cross-sectional studies of human cohorts have revealed the potential of EV contents as diagnostic and prognostic tools. While the AD biomarker potential of miRNA in EVs has recently started to be investigated, EV protein content has been assessed under multiple clinical conditions and tasks, including their diagnostic and prognostic potential. Based on previous clinical studies (see also Table 2), one group of EVs proteins was detected to be increased as dementia progresses (e.g., total Tau, specific Tau phospho-epitopes, Aβ, ubiquitinylated proteins) while another group of EVs proteins, mostly synapse-related proteins, are found to be decreased in AD. Further longitudinal studies of large and polycentric cohorts of healthy and AD patients are necessary in order to clarify the protein and miRNA cargo of EVs along the progress of AD and identify specific target or groups of targets of protein and/or miRNA EVs cargo with high biomarker value for AD.

10. EVs as AD drug targets and therapy delivery systems

Since EVs seem to play a role in disease propagation and progression, they may also be good candidates for novel therapeutic interventions. Indeed, genetic ablation or chemical inhibition of neutral sphingomyelinase 2 (nSMase2), an enzyme involved in EVs biogenesis, was found to reduce Aβ and Tau deposits, slowing down disease progression (Dinkins et al., 2016; Asai et al., 2015). Moreover, repetitive treatment with nSMase2 inhibitor GW4869 provided protection against behavioral and neuropathological deficits following immune system activation in AD preclinical studies (Sobue et al., 2018). In addition, knocking down ESCRT proteins involved in MVB biogenesis, TSG101, and VPS4A were found to reduce both the spread of Aβ oligomers and the downstream cellular toxicity (Sardar Sinha et al., 2018). Another study showed that inhibition of P2RX7 in microglial cells hindered exosome release and offered protection against Tau pathology (Ruan et al., 2020). Recently, dual nSMase2/Acetylcholinesterase Inhibitors were developed with promising results for the treatment of AD (Bilousova et al., 2020). For targeting disease-promoting EVs, potential points for intervention could include: 1) exosome biogenesis, 2) their release from parent cells, and 3) exosome uptake by recipient cells. In the cancer field, which has generated critical discoveries in the EVs field, several attempts have already been made or are ongoing in order to target the above steps involved in exosome biogenesis and spreading (Moloudizargari et al., 2021).

Moreover, EVs have also garnered interest as drug delivery agents for personalized medicine since their discovery. For instance, intracerebrally administered EVs can act as potent Aβ scavengers by trapping Aβ and promoting its degradation by microglia, representing a novel therapeutic intervention for AD (Yuyama et al., 2014). Moreover, their ability to carry bioactive molecules, to circulate in body fluids, and cross physiological barriers (e.g., BBB), as well as their low immunogenicity, make them appropriate tools to deliver chemical or biological compounds. So far, EVs have been engineered to carry chemotherapeutic agents, siRNAs, miRNAs, Cas9 protein, and other compounds. These processes involve medium conditioning with drugs of interest or techniques like electroporation, saponin permeabilization, hypotonic dialysis, and passive incubation (Rufino-Ramos et al., 2017; Sutaria et al., 2017). For instance, delivery of EVs derived from curcumin-treated cells can ameliorate AD neuropathology and memory impairment in an AD mouse model by inhibiting Tau hyperphosphorylation and neuronal death (Wang et al., 2019). Given the critical role of neuroinflammation in AD, the EV-based delivery of compounds that modulate brain inflammation but could not necessarily cross the BBB (such as curcumin) was a major breakthrough in the research field of AD (Zhuang et al., 2011). In addition, current efforts focus on the delivery of RNA species inside small EVs (Iranifar et al., 2019). For instance, a recent study showed that microglia-derived exosomes that function as a vehicle to deliver Tetraspanin 2 siRNA across the BBB are able to modulate the neuroinflammatory response (Reynolds and Mahajan, 2020). Moreover, the use of EVs as drug delivery tool is also focusing on their delivery to specific cell types. For instance, the exosomal delivery of BACE-1 siRNA to neurons, microglia, and oligodendrocytes leads to a 55% reduction of Aβ levels in the mouse brain (Alvarez-Erviti et al., 2011).

Another aspect of the therapeutical potential of EVs relies on the use of EVs from mesenchymal and adipocyte stem cells which have also exhibited beneficial effects for the treatment of AD (Hao et al., 2014; Perets et al., 2019; Liew et al., 2017). Mesenchymal stem cells (MSC) possess immunomodulatory and tissue regenerative properties and, therefore, have been suggested as therapeutic tools (Pittenger et al., 2019). MSC-derived exosomes have been shown to decrease Aβ levels but also inhibit the Aβ-driven downregulation of synaptic plasticity-related genes (Chen et al., 2021) or microglial activation (Kaniowska et al., 2022). Other studies have also shown that MSC-derived EVs have beneficial effects in in vitro AD models by protecting neurons from oxidative stress and synaptic damage induced by Aβ oligomers (Bodart-Santos et al., 2019; de Godoy et al., 2018). EVs from adipose stem cells (ADSCs-derived EVs) have also been shown to alleviate neuronal damage and promote neurogenesis in an AD mouse model (Ma et al., 2020). Moreover, ADSC-derived EVs were shown to carry enzymatically active Neprilysin (NEP), an Aβ-degrading enzyme that was proven to decrease levels of Aβ in the brain of an AD model, reducing cell apoptosis (Lee et al., 2018a; Katsuda et al., 2013). Lastly, EVs derived from stem cells of other origins, e.g., human umbilical cord MSCs (hUMSCs) and bone marrow mesenchymal stem cells (BM-MSC), were recently shown to exhibit beneficial effects against Aβ (Yang et al., 2020; Jeong et al., 2021).

Despite the above important advances in the field, EV-based therapies still have a long way to go, as many therapeutic approaches fail in clinical trials. For instance, while autologous injection of EVs leads to very low immunogenicity, EVs produced by cell lines or genetically distinct individuals may induce immune responses. As mentioned in the section “Technical difficulties related to the isolation and study of brain EVs,” our characterization methods have profiled a heterogeneous population of small EVs of similar size, including microvesicles and exosomes, and thus it remains to be identified specific (surface or cargo) content that can help us discriminate these two EV subpopulations which could, in turn, improve our engineering strategies for the production and therapeutic delivery of EVs. Another significant hurdle is the scalability of EVs production for treatment in large cohorts of AD patients, as we lack an efficient isolation method to obtain large quantities of pure EVs (e.g., from cell cultures or biofluid free of molecular and cellular contaminants). Moreover, there is an ongoing debate, in the clinical context, related to the optimal dosage (whether the EV number or content in EVs) and the route of EVs administration, as well as the timeframe for EVs treatment which is critical to define the most efficient therapeutic scheme (Lv et al., 2017). In addition, the long-term biological safety of EVs still needs to be assessed in order to investigate their potential adverse effects and efficacy of the administration in AD patients.

11. Conclusion

EVs represent an important vehicle for intercellular messaging in the brain. while emerging evidence suggest that EVs regulate multiple physiological processes that are impaired in the AD brain. Cell-type-specific EVs have differential effects on the progression of the disease, with some ameliorating and others worsening it. the development of novel EV-related tools and animal models will advance our EV studies in vivo, improving our understanding of how EVs contribute to both physiological and pathological mechanisms of the brain. Thus, a better understanding of EVs biogenesis and spreading, as well as cargo sorting, will help our research efforts to regulate and engineer EVs against AD neurodegeneration, as well as facilitate the discovery of specific and accurate biomarkers for early detection of AD and dementia-precipitating processes, representing a valuable tool in the new era of Precision Medicine.