Extracellular Vesicles for Research on Psychiatric Disorders

Abstract

Extracellular vesicles (EVs) have gained increasing attention as underexplored intercellular communication mechanisms in basic science and as potential diagnostic tools in translational studies, particularly those related to cancers and neurological disorders. This article summarizes accumulated findings in the basic biology of EVs, EV research methodology, and the roles of EVs in brain cell function and dysfunction, as well as emerging EV studies in human brain disorders. Further research on EVs in neurobiology and psychiatry may open the door to a better understanding of intercellular communications in healthy and diseased brains, and the discovery of novel biomarkers and new therapeutic strategies in psychiatric disorders.

Introduction

Extracellular vesicles (EVs) are small lipid bilayer membrane vesicles secreted by various cells in the body under both physiological and pathological states.^1–5 EVs contain various molecules from their cell of origin, such as nucleic acids (RNAs and DNAs), proteins, and other cellular components (eg, lipids, carbohydrates, and metabolites).^1,2 By transferring these components, EVs can modify the function of recipient cells.^1,2 Recent extensive studies are now revealing the molecular mechanisms underlying EV biogenesis, secretion, and uptake.^3,6 EVs are also being vigorously studied in neurobiology, and accumulating evidence supports EV-mediated intercellular communication locally in the brain and between the brain and periphery.^2,7 In parallel, emerging evidence also indicates the potential utility of EVs as biomarkers for several brain disorders, such as Alzheimer’s disease (AD).^8–10 Thus, EV research in psychiatry may lead to the discovery of novel biomarkers. This review article aims to concisely deliver the current status of EV research and its translational applications.

EV Biology

EV Subtypes and Biogenesis/Release

EVs are a heterogeneous group of vesicles, such as exosomes and microvesicles, which are classified by their biogenesis mechanism and/or by size, density, or cell-type origin^3,5,11 (figure 1A). Exosomes are membrane vesicles of 30–150 nm in diameter that are generated inside multivesicular bodies (MVBs)/multivesicular endosomes (MVEs) in the endosomal system¹² and secreted on the fusion of MVBs with the plasma membrane.^3,6,13 Microvesicles bud directly from the plasma membrane and vary widely in size (50–1000 nm in diameter).³

Fig. 1.

Extracellular vesicles (EVs) in intercellular communication in the brain. (A) Representative mechanisms of EV biogenesis and uptake. Exosomes are formed as intraluminal vesicles (ILVs) within multivesicular bodies (MVBs) in the endosomal system. As MVBs are trafficked to and fuse with the plasma membrane, releasing ILVs into the extracellular space as exosomes. Microvesicles bud outward from plasma membranes via multiple mechanisms, such as those depending on the tumor susceptibility gene 101 protein (TSG101)-arrestin 1 domain-containing protein 1-dependent manner. EVs are taken up by recipient cells via multiple mechanisms, including endocytosis and phagocytosis. (B, C) EV-mediated intercellular communications in healthy (B) and diseased (C) brains. EVs are released by virtually any type of brain cells and modify the functions of recipient cells by transferring their cargo (eg, proteins, messenger RNAs, and microRNAs). Shown are summary of several previous findings described in the text. Images for these figures were adapted from Servier Medical Art (by Servier) (https://smart.servier.com/), licensed under CC BY 3.0 (https://creativecommons.org/licenses/by/3.0/).

Exosomes are formed as intraluminal vesicles (ILVs) within MVBs in the endosomal system by the endosomal-sorting complexes required for transport (ESCRT) machinery.^3,6,13 The ESCRT machinery regulates cytoplasmic cargo sorting, membrane deformation/invagination, cargo insertion into ILVs, and membrane fission. More recently, the syndecan–syntenin–ALIX pathway, which is ESCRT independent, has been described.^14,15 MVBs are trafficked to and fuse with the plasma membrane, releasing ILVs into the extracellular space as exosomes. A variety of mechanisms are involved in exosome release, such as those mediated by a Rab family of small GTPases, Rab27a, and Rab27b.¹⁶ Microvesicles, on the other hand, bud outward from plasma membranes via multiple mechanisms. One such mechanism generates microvesicles in a tumor susceptibility gene 101 protein (TSG101)-arrestin 1 domain-containing protein 1 (ARRDC1)-dependent manner.^17,18

Some microvesicles are similar in size and function to exosomes, and separating exosomes from microvesicles is technically challenging. Thus, there is a trend to describe EVs based on size, such as small EVs, rather than their origin, such as exosomes or microvesicles.^3,11 Extensive studies are still in progress regarding the biogenesis and release of different EV subtypes. Newer tools, such as CD63 tagged with pH-sensitive green fluorescent protein (GFP), may allow live imaging of EV release at single-cell level and thus offer new opportunities to dissect the mechanisms underlying vesicle release.¹⁹ Additional new mechanisms are very likely to be discovered in the near future and contribute to better understanding of differential origins of vesicle subtypes.

Loading of molecules into EVs seems to be regulated by various mechanisms.^20,21 For example, studies suggest that RNA-binding proteins, such as Ago2 and hnRNPA2B1, regulate the loading of microRNAs (miRNAs) into EVs.^22,23 Although some studies showed that the loading of miRNAs into EVs is selective,²⁴ other studies found no selective bias in EV miRNAs compared with miRNAs in their cells of origin.²⁵ Further studies are required to advance our understanding of the loading mechanisms of molecules during EV biogenesis.

Biological Function of EVs

EVs, in particular exosomes, were originally described as a mechanism for sheep and avian reticulocytes to remove unused membranes during their maturation.^26–28 Subsequently, it was shown that mouse and human B-lymphoblastoid cells secrete exosomes expressing major histocompatibility complex (MHC) class II, which stimulate antigen-dependent T-cell proliferation.²⁹ Dendritic cells also release EVs expressing peptide-MHC and activate antigen-specific effector T cells.³⁰

Secreted EVs also bind to recipient/target cells; however, the molecular and cellular basis of target cell specificity is unclear, except for a few emerging studies.^3,6,13,31 EVs are taken up by recipient cells via multiple endocytosis pathways and/or phagocytosis.^3,6,13 EV-mediated transfer of molecular cargo, such as RNAs and proteins, has been shown to modify gene expression, activate immune responses, and deliver pathological products to recipient cells. For example, miRNAs in EVs modify gene expression in recipient cells under various contexts, such as cancer progression, immune responses, and brain function.^{1,2,8,24,32–36} In addition, cancer-derived EVs have been shown to activate innate immune signaling via recognition of their cargo miRNAs by toll-like receptors in cocultured macrophages.³⁵ It should be noted, however, that the functional transfer of EV genetic material, such as miRNAs, is technically challenging to demonstrate, particularly in vivo.²⁰ In fact, most of the currently available evidence is based on the findings of the effects of isolated EVs on target cells, far from physiological contexts and subject to contaminants in isolated EV samples. A newer method such as the Cre-loxP-based tracking system will contribute to further understanding of EV-based transfer of molecular cargos.

Although EVs are heterogeneous, consisting of exosomes, microvesicles, and other types of larger vesicles,³⁷ their functional difference is not clearly understood. For details on EV basic biology and current challenges, readers are advised to refer to recent excellent reviews.^3,6,13

EV Isolation and Analysis

EV Isolation

Multiple methods are used to isolate EVs from body fluids and tissues.³⁸ Because the method of EV isolation impacts the population of vesicles tested and downstream experiments,^39,40 it is critical to choose an appropriate method and rigorously characterize the isolated EVs so that EV-related findings are reproducible. Here, we describe several typical EV isolation methods and their advantages/disadvantages (table 1).

Table 1.

Comparison of Typical Extracellular Vesicle (EV) Isolation Methods

Methods	Major advantages	Major disadvantages
Differential ultracentrifugation	A well-established method	Time-consuming procedure
		Results in heterogeneous EV populations
		Contamination of non-EV materials
		Ultracentrifugation step might damage EVs and might lead to aggregation
Density-gradient ultracentrifugation	Isolate EV subpopulations	Time-consuming procedure
Precipitation	Easy to perform	Results in heterogeneous EV populations
	Easily applicable to high-throughput clinical studies	Contamination of non-EV materials
Size-exclusion chromatography (SEC)	Isolate EV subpopulations based on size	Requires optimization and familiarity with methods
	Less contamination of non-EV materials	Not high throughput
Antibody/column-based isolation	Isolate EV subpopulations based on their membrane/surface protein signatures	Results in only subpopulations of EVs
	Less contamination of non-EV materials
	Applicable to high-throughput clinical studies
Tangential flow filtration (TFF)	Isolate EV subpopulations based on size	Requires sequential TFF or in combination with other methods (eg, SEC)
	Less contamination of non-EV materials
	Applicable to large amounts of sample (eg, culture medium)
Asymmetric flow field-flow fraction	Isolate EV subpopulations based on size	Requires optimization and familiarity with methods
	Less contamination of non-EV materials

Ultracentrifugation.

Differential ultracentrifugation is the most widely used method to separate EVs from cell culture supernatant.⁴¹ However, isolated EV pellets are known to contain other substances, such as proteins and lipids, as contaminants. Isolated EVs are also heterogeneous, containing various different types of vesicles. In addition, the ultracentrifugation may damage vesicles and lead to aggregation.^42,43 Density-gradient ultracentrifugation allows EVs to be further fractionated, which can identify biologically active EV subpopulations.^44,45 Because of low yield, ultracentrifugation, specifically density-gradient ultracentrifugation, may not be suitable for clinical specimens with limited volumes.

Precipitation.

Most commercially available kits (eg, ExoQuick) rely on a precipitation method to isolate EVs.³⁹ Although this method can quickly enrich EVs even from clinical specimens with limited volumes,⁴⁶ it also coprecipitates many contaminants substantially. Thus, the findings based on precipitation-based EV collection need to be carefully interpreted.

Size-Exclusion Chromatography.

Size-exclusion chromatography (SEC) allows for the fractionation of EVs and other molecules based on physical size.^47,48 Although this enables purer EV isolation, other vesicles of similar size, such as cholesterol vesicles, may be co-isolated. In addition, this method is not high throughput, preventing its application for high-throughput clinical studies.

Antibody/Column-based Isolation.

Antibodies against EV surface protein markers such as tetraspanins (eg, CD63) are used to pull down EVs from biological materials.⁴⁴ Isolated EVs are more purified than ultracentrifugation or precipitation methods, but their yield is much lower than these methods. Similarly, several affinity column-based isolation methods have been used to capture EVs.^48,49 A major limitation in these antibody/column-based isolation approaches is that only subpopulations of EVs can be collected.

Other Emerging Methods.

Tangential flow filtration (TFF) is used to separate vesicles or remove protein and lipid contaminants by using a porous membrane. TFF combined with SEC can generate a high yield of EVs with high purity and can process large amount of sample.⁵⁰ TFF is thus suitable for EV isolation from large amount of culture media for clinical application; however, to isolate EVs of a specific size or population, sequential TFF or in combination with SEC is required.^51,52 Asymmetric flow field-flow fractionation is another method that has been shown to highly purify EV subpopulations by size. Resulting EV subpopulations were reported to exhibit unique protein, lipid, DNA, RNA, N-glycosylation, and biophysical properties.⁵³

EV Characterization and Quantification

There are multiple ways to characterize and quantify EVs. To promote rigor and reproducibility in EV research, the International Society for Extracellular Vesicles detailed the minimum requirements for reporting EV-related findings in 2014.^4,54 It recommends that EVs should be characterized as single vesicles, the presence of EV-enriched proteins and absence of intracellular vesicles markers should be assayed, and EV concentration for functional experiments should be reported. Typical characterization/quantification methods are described next.

Electron Microscopy.

Transmission electron microscopy (TEM) is widely used to visualize isolated EVs to verify their size and morphology and also to examine the presence of contaminants.⁴ Some studies use TEM data to determine the size distribution of EVs.⁵⁵

Single-Particle Analysis.

Single-particle analyses, such as nanoparticle tracking analysis (NanoSight, ZetaView) and resistive pulse sensing (qNano), allow for size distribution profiling of EV samples. Because the size of most EVs is below the detection threshold of conventional flow cytometry (FCM), FCM analysis of single EV vesicles is generally challenging.⁵⁶ Newer technologies such as high-resolution FCM are being tested for single-vesicle analysis.^57–59

Biochemical Assays.

Because there are no established pan-EV markers, a combination of several markers is necessary for verification of EVs. For example, the combination of exosome lumen-enriched proteins (such as TSG101 or ALIX) and exosome membrane-enriched proteins (such as CD9, CD63, or CD81), as well as the absence of intracellular vesicle markers, such as Calnexin, is suggested to verify the presence of exosomes.

Quantification.

EVs are quantified by counting the number of vesicles and/or measuring the amount of proteins co-isolated with EVs. For vesicle counting, the aforementioned single-particle analysis is used. For protein measurement, conventional Bradford and bicinchoninic acid (BCA) assays are widely used.

Analysis of EV Cargo

The effects of EVs on target cells depend on the contents of EVs. Accordingly, the analysis of EV contents, such as nucleic acids, proteins, lipids, and metabolites, is central to understanding the biological role of EVs. High-throughput technologies, such as RNA sequencing, proteomics, and metabolomics, are widely used to characterize key EV cargo in various basic and translational studies.^37,44,60 Notably, many studies have revealed that EVs are extremely heterogeneous and that not all EVs contain the same sets of cargo. For example, it was reported that the number of given messenger RNA (mRNA) and miRNA molecules per vesicle is only one or less.^37,61 Thus, only some EVs may contain RNA molecules whereas others may not.

EVs in Brain Health and Disease

EV Production and Intercellular Communication in the Brain

EVs can be produced by virtually any cell type in the brain and affect brain cell functions in both an autocrine and a paracrine manner. In addition, peripheral EVs can influence brain cells. Here, we summarize some of the previous findings relevant for brain function and dysfunction (figures 1B and 1C). It should be noted that most studies used in vitro cell culture system and relatively few studies examined the biological role of EVs in the brain.

Neuron-derived EVs.

In vitro rat primary cultured neurons release EVs containing L1 cell adhesion molecules, GPI-anchored prion proteins, and the GluR2/3 subunits of glutamate receptors.⁶² Neuronal EV release is enhanced by the presence of bicucullin, a gamma amino butyric acid type-A (GABA_A) receptor antagonist and attenuated by addition of cyanquixaline (CNQX), an α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor antagonist, and MK-801, an N-methyl-d-aspartic acid (NMDA) receptor antagonist to bicucullin.⁶³ EVs secreted on synaptic activation are preferentially taken up by neurons.⁶⁴ These findings suggest that neuronal EV release is regulated by synaptic glutamate activity and mediates interneuronal communication. Neurons also secrete EVs enriched in miRNAs and other small noncoding RNAs, particularly miR-124a, which increases GLT1 protein expression and glutamate uptake in targeted astrocytes in vitro and in vivo.⁶⁵ Very recently, it has been reported that Arc, a neuronal protein critical for synaptic function and learning and memory, self-assembles into virus-like capsids and is released from neurons as EVs in flies and rodents.^66,67 Arc EVs contain Arc mRNA and transfer it into target neurons, where Arc mRNA undergoes activity-dependent translation and modulates synaptic plasticity. Retinoic acid receptor β agonist-treated neurons secrete exosomes, which modulate astrocyte function around regenerating axons in a rat model of spinal cord injuries.⁶⁸ Neurons also secrete and propagate prion (PrP) proteins in exosomes.⁶⁹ Lysosomal dysfunction in neurons triggers the secretion of atypical EVs, enriched for undigested lysosomal substrates.⁷⁰

Astrocyte-derived EVs.

EVs from astrocyte-enriched cell cultures contain synapsin I and promote neurite outgrowth and neuronal survival on high neuronal activity and/or oxidative stress.⁷¹ Astrocytes also protect cerebellar neurons against oxidative stress-induced damage by releasing EVs containing PrP.⁷² Astrocytes also produce toxic EVs. Cultured rodent astrocytes produce apoptosis-inducing EVs on incubation with amyloid-β peptides and neurotoxic EVs when they overexpress mutant copper-zinc superoxide dismutase 1 (SOD1).^73,74 Human astrocytes infected with human immunodeficiency virus (HIV) also produce EVs containing the viral protein Nef.⁷⁵ In tumor biology, astrocyte-derived EVs reduce the expression of PTEN in tumor cells, promoting brain metastasis.⁷⁶ In addition, astrocyte-derived EVs released in response to intracerebral inflammation enter peripheral circulation to reach peripheral tissues, such as the liver, lung and spleen, and trigger peripheral inflammation.⁷⁷ Certain disease-associated proteins such as amyloid-β and mutant huntingtin were also shown to inhibit astrocyte EV production.^78,79

Oligodendrocyte-derived EVs.

Oligodendrocyte-derived EVs contain myelin-related proteins such as proteolipid protein (PLP), 2′3′-cyclic-nucleotide-phosphodiesterase (CNP), myelin basic protein (MBP), and myelin oligodendrocyte glycoprotein (MOG).^80,81 Oligodendrocyte-derived EVs were shown to have an autoinhibitory property on their growth, differentiation and myelin formation, and are taken up by microglial macropinocytosis.⁸² Disruption of NPC intracellular cholesterol transporter 1p, the gene disrupted in Niemann-Pick disease type C1, increases cholesterol content in oligodendrocyte-derived EVs.⁸³

Microglia-derived EVs.

Activated microglia produce miRNA-enriched EVs that regulate the expression of key synaptic proteins and dendritic spine loss, accompanied by reduced miniature excitatory postsynaptic current (mEPSC) frequency and amplitude.⁸⁴ In addition, microglia-derived EVs are shown to carry endocannabinoids, N-arachidonoyl-ethanolamine (AEA), and inhibit presynaptic transmission in GABAergic neurons.⁸⁵ In another study, it was found that adenosine triphosphate (ATP)-stimulated microglia release microvesicles containing interleukin-1β (IL-1β) in an acid sphingomyelinase-dependent manner⁸⁶ and enhance excitatory synaptic activities in vitro and field potentials evoked by visual stimuli in vivo.⁸⁷ Microglia-derived EVs also contribute to the propagation of tau proteins in the rodent brain, which is relevant for AD.⁵⁵

EVs from Other Cell Types.

In the neonatal brain, subventricular zone (SVZ) neural stem cells (NSCs) generate and release EVs, which preferentially target microglia and modify their cytokine release, leading to negative feedback control of NSC proliferation.⁸⁸ In the adult brain, hypothalamic stem/progenitor cells contribute greatly to EV miRNAs in the cerebrospinal fluid. A decline in these EV miRNAs is reportedly associated with aging.⁸⁹ The choroid plexus produces EVs on systemic inflammatory responses induced by peripheral immune challenge.⁹⁰ Choroid plexus-derived EVs reach the brain parenchyma and induce inflammatory gene expression in astrocytes and microglia.

EV Transfer From Periphery into Brain.

EVs derived from a variety of cells, such as cancer cells, mesenchymal stem cells, and hematopoietic cells, were shown to traverse the blood–brain barrier,^91–95 suggesting that they may mediate signaling from the periphery directly to the brain. Systemic injection of an exosome release inhibitor, GW4869, in mice is also shown to cause deficits in memory formation.⁹⁶ However, GW4869, which inhibits neutral sphingomyelinase (nSMase), is suggested to disrupt neuronal function by altering ceramide contents and expression of AMPA and NMDA receptors. Further studies are required to investigate the impact of decreased EV release on neuronal function and behavior. More recently, systemic injection of mesenchymal stem cell-derived EVs has been reported to rescue cognitive impairments after traumatic brain injury in mice.⁹⁵ It is not clear, however, whether and how EV-mediated signals from peripheral tissues/cells influence neuronal function. Studies using genetically engineered cells and EVs that express specific fluorescent markers¹¹ will be useful to determine the functional changes of EV target cells in the brain.

Potential Utility of EVs as Biomarkers for Psychiatric Disorders

Recent studies in both rodents and humans suggest that brain-derived EVs are detected in the peripheral blood.^77,97 In human studies, EVs expressing neuronal cell adhesion molecules, NCAM or L1CAM, were detected in plasma.⁹⁷ L1CAM⁺ EVs were further shown to be enriched in proteins related to neurons. Despite the limitation that L1CAM and other neuronal proteins are not necessarily specific for neurons in the brain,⁹⁷ these neuronal marker-enriched EVs have successfully demonstrated the potential of EVs as tools for biomarker discovery. Future studies will need to determine whether these neuronal EVs can be used to assess the functional status of neurons in the brain.

The potential utility of miRNAs and proteins in EVs as biomarkers has been shown for several brain disorders, such as glioblastoma, AD, and multiple sclerosis.^8–10 In AD, 16 specific EV-miRNA signatures, together with age, sex, and apolipoprotein ε4 (Apoε4) allele status, were shown to predict AD with a high sensitivity and specificity.⁹ In addition, higher pSer312-IRS-1 and lower p-panTyr-IRS-1 in neuronal EVs in the peripheral blood were shown to be associated with brain voxel-based morphometry in patients with AD.⁹⁷ Moreover, decreased synaptic proteins, such as synaptophysin and synaptotagmin-2, were reported in neuronal EVs of patients with frontotemporal dementia and AD.⁹⁸

An earlier study on schizophrenia also suggested that circulating miRNAs in the blood may be differentially expressed in patients with psychotic disorders.⁹⁹ Because EVs are enriched in miRNAs,¹⁰⁰ this observation raises an interesting possibility that circulating EVs can be used for clinical evaluation of schizophrenia and related disorders. Future studies are required to determine whether EV contents can serve as biomarkers for psychosis and other psychiatric disorders.

Isolation of EVs from freshly collected body fluids or cell culture supernatants is ideal. Because cells/platelets contain many EV-sized particles and are disrupted by freeze/thaw cycles, it is recommended that they be removed from biological fluids before freezing to prevent contamination.¹⁰¹ However, this is challenging for human sample studies, as the samples are already frozen in most cases. Although some studies reported that isolated EVs are resistant to freeze/thaw cycles,¹⁰² it is important to minimize freeze/thaw cycles and compare samples that have undergone this process. Multiple studies have examined the stability of EVs and their contents in clinical specimens after short- and long-term storage at various temperatures.¹⁰³ Some studies reported that EV-associated miRNAs were highly stable for up to 5 years in frozen plasma whereas other studies found that longer storage at −80°C more than 8 years dramatically decreased miRNA recovery from EVs in frozen sera.^104,105 The stability of proteins and other molecular components in EVs needs additional study before a conclusion can be made. At this moment, it is not clear whether freeze/thaw cycles and/or storage conditions alter EV biological function. In fact, considerable inconsistency in EV characterization and analysis across these studies makes it difficult to compare the findings. The utilization of standardized EV characterization and analysis methods would contribute to further understanding of the impact of freeze/thaw cycles and storage conditions on EVs and their cargos.

Conclusions and Future Perspectives

Despite growing interest in the biology and utility of EVs in various biomedical science fields, research on the basic biology of EVs is still in its infancy. Currently, EV research has limitations both in its technical and biological aspects. Technically, no standard methods and/or protocols have been established to isolate particular EV subtypes from various biological samples.^39,101 Analysis of single EVs is also limited because separating and analyzing single vesicles is technically challenging. Biologically, further studies are needed to explain how, or whether, EVs specifically target recipient cells. It also remains unclear how EVs mediate interorgan communications, in particular those between the brain and the peripheral tissues. Much remains to be learned in this exciting field over the coming years. As basic EV research advances, the utility of EVs in translational research on schizophrenia and other psychiatric disorders will be further extended. Because EVs can potentially be engineered for the loading of therapeutic molecules,¹⁰⁶ EVs may also be harnessed as tools to deliver therapeutic agents into the brain for the treatment of psychiatric disorders in the future.

Funding

This work has been supported by the funds from National Institute of Mental Health (R00MH093458, R01MH113645) and Johns Hopkins University (Discovery Innovation Award from School of Medicine, Venture Discovery Award from Department of Psychiatry and Behavioral Sciences).

Conflict of Interest

The authors have declared that there are no conflicts of interest in relation to the subject of this study.