Secreted Extracellular Vesicle Molecular Cargo as a Novel Liquid Biopsy Diagnostics of Central Nervous System Diseases

Secreted extracellular vesicles (EVs) are heterogeneous cell-derived membranous granules which carry a large diversity of molecules and participate in intercellular communication by transferring these molecules to target cells by endocytosis. In the last decade, EVs’ role in several pathological conditions, from etiology to disease progression or therapy evasion, has been consolidated, including in central nervous system (CNS)-related disorders. For this review, we performed a systematic search of original works published, reporting the presence of molecular components expressed in the CNS via EVs, which have been purified from plasma, serum or cerebrospinal fluid. Our aim is to provide a list of molecular EV components that have been identified from both nonpathological conditions and the most common CNS-related disorders. We discuss the methods used to isolate and enrich EVs from specific CNS-cells and the relevance of its components in each disease context.


Introduction
Extracellular vesicles (EVs) are cell-derived, membrane-bound structures, secreted into the extracellular environment by most, if not all, cell types. There is mounting evidence that EVs take part in control of normal physiological processes, in homeostasis control and in cell-to-cell communication within tissues and organs. Based on development of the EV field, different types of vesicles have been described, and include apoptotic bodies which range in size from 50 to 5000 nm, microvesicles (50-1000 nm), exosomes (40-200 nm) and the more recently recognized exomeres [1][2][3][4].
The EVs have started to attract attention also as potential disease biomarkers and as new ways to study the control of cell biology. They seem to offer diagnostic capacity of organ dysfunctions via the body fluids as the components of the "liquid biopsy" type of sample. For example, solid tumors exhibit the elusive nature of transformed cells. They Int. J. Mol. Sci. 2021, 22, x FOR PEER REVIEW 3 of 33

Methods and Factors for Isolation of Central Nervous System Extracellular Vesicles in Liquid Biopsies
Currently, as EVs are known to be able to cross the BBB, they are considered as potential biomarkers for monitoring CNS disorders as well as specific treatment responses. Of particular interest is the ability to characterize EVs based on their source, which may provide insight into the target disease. Although the majority of EVs research is performed through cell culture, the number of reports aiming at identifying biological fluids' EVs originated in different cell types is rising. Multiple methods (including ultracentrifugation, immunomagnetic beads, and size exclusion chromatography) can be used to separate mixed EVs (from multiple cell types) from different biological fluids [8]. This step may be taken further by using cell-specific protein markers to enrich for EVs of specific cell types from a mixed population of vesicles.
Specifically, EVs from the CNS are secreted from almost all cell types, including various types of neurons, astrocytes, oligodendrocytes, microglia and Schwann cells, and endothelial cells. EVs carry a variety of molecules related to neuronal function and neurotransmission in the brain. This contributes to the communication between nerve cells (for example, neuron-glia interaction), synaptic plasticity, and neuron development. Interestingly, Fiandaca et al. developed an immunoprecipitation-based method to isolate EVs rich in neuron sources from blood. This technology has been also used by others and consists of two steps: a preliminary separation of total EVs from plasma or serum samples using commercially available polymers for high-throughput particle precipitation, and im-

Methods and Factors for Isolation of Central Nervous System Extracellular Vesicles in Liquid Biopsies
Currently, as EVs are known to be able to cross the BBB, they are considered as potential biomarkers for monitoring CNS disorders as well as specific treatment responses. Of particular interest is the ability to characterize EVs based on their source, which may provide insight into the target disease. Although the majority of EVs research is performed through cell culture, the number of reports aiming at identifying biological fluids' EVs originated in different cell types is rising. Multiple methods (including ultracentrifugation, immunomagnetic beads, and size exclusion chromatography) can be used to separate mixed EVs (from multiple cell types) from different biological fluids [8]. This step may be taken further by using cell-specific protein markers to enrich for EVs of specific cell types from a mixed population of vesicles.
Specifically, EVs from the CNS are secreted from almost all cell types, including various types of neurons, astrocytes, oligodendrocytes, microglia and Schwann cells, and endothelial cells. EVs carry a variety of molecules related to neuronal function and neurotransmission in the brain. This contributes to the communication between nerve cells (for example, neuron-glia interaction), synaptic plasticity, and neuron development. Interestingly, Fiandaca et al. developed an immunoprecipitation-based method to isolate EVs rich in neuron sources from blood. This technology has been also used by others and consists of two steps: a preliminary separation of total EVs from plasma or serum samples using commercially available polymers for high-throughput particle precipitation, and immunoprecipitation with biotinylated antibodies against the neuron surface marker neuronal markers L1 cell adhesion molecule (L1CAM, CD171) [9]. Furthermore, L1CAM and the GluR2/3 subunits of glutamate receptors have been used for the identification of EVs that may be released from developing and mature hippocampal neurons [10,11]. Hence, using this method, EVs of neuronal origin may be enriched and evaluated based on proteins, lipids and nucleic acids.
Although neuron-derived EVs are a common target in several studies, researchers are also interested in isolating EVs originating from other CNS cells, aiming also to translate via the EVs the origin of CNS-related diseases. The above-mentioned methodology can also be used to enrich for EVs from other cellular origins with antibodies against specific cell surface components. For example, for astrocytic-derived EVs enrichment, glial fibrillary acidic protein (GFAP), glutamine aspartate transporter (GLAST), and glutamine synthetase (GLUL) have been suggested to be specific enough to obtain CNS related EVs [12]. Additionally, for EVs originated in oligodendrocytes, myelin proteolipid protein (PLP) and 2 ,3 -cyclic nucleotide 3 -phosphodiesterase (CNP) have been proposed as specific enough to identify the cell from which the EVs derived [13] (Figure 2). munoprecipitation with biotinylated antibodies against the neuron surface marker neuronal markers L1 cell adhesion molecule (L1CAM, CD171) [9]. Furthermore, L1CAM and the GluR2/3 subunits of glutamate receptors have been used for the identification of EVs that may be released from developing and mature hippocampal neurons [10,11]. Hence, using this method, EVs of neuronal origin may be enriched and evaluated based on proteins, lipids and nucleic acids. Although neuron-derived EVs are a common target in several studies, researchers are also interested in isolating EVs originating from other CNS cells, aiming also to translate via the EVs the origin of CNS-related diseases. The above-mentioned methodology can also be used to enrich for EVs from other cellular origins with antibodies against specific cell surface components. For example, for astrocytic-derived EVs enrichment, glial fibrillary acidic protein (GFAP), glutamine aspartate transporter (GLAST), and glutamine synthetase (GLUL) have been suggested to be specific enough to obtain CNS related EVs [12]. Additionally, for EVs originated in oligodendrocytes, myelin proteolipid protein (PLP) and 2′,3′-cyclic nucleotide 3′-phosphodiesterase (CNP) have been proposed as specific enough to identify the cell from which the EVs derived [13] (Figure 2).

Figure 2.
Schematic representation of the brain microenvironment, its main cellular components and how the brain-bloodbarrier permeability allows for extracellular vesicles to reach circulation. Created with BioRender.com.
The application of specific immunoaffinity-based methodology for the isolation of CNS-related EVs in biological fluids, applied to different CNS diseases research, has led to promising results [9,14]. Once the enriched fraction of EVs is obtained, either surface or cargo content can be analyzed using different approaches according to the molecular target of interest, such as proteomics (e.g., mass spectrometry, Western blot or enzymelinked immunosorbent assays) and/or transcriptomics methodologies (e.g., high-throughput RNA sequencing or RT-qPCR). This opens a new avenue for research, using CNSderived EVs as vehicles for disease monitoring. Schematic representation of the brain microenvironment, its main cellular components and how the brain-bloodbarrier permeability allows for extracellular vesicles to reach circulation. Created with BioRender.com. are rather imperfect, because their dynamic range does not cover the entire course of the disease, and their classification accuracy is lower than the level accepted for clinical practice, thus, precluding their routine implementation. Thus, the use of CNS-enriched EVs and their molecular content as biomarkers for these specific pathological conditions may revolutionize the clinical management of these patients. Indeed, a growing number of studies within this field have been published in the last few years. To simplify, a summary of the main publications retrieved from the database query can be found in Tables 1 and 2, referring to either cargo or surface molecules, respectively [9,12, (Figure 3).

Central Nervous System Disease-Related Biomarkers in Extracellular Vesicles
CNS-related disorders, including neurodegenerative diseases, such as Alzheimer's (AD) and Parkinson's (PD) diseases, malignant tumors, such as gliomas, and psychiatric disorders, are very different pathological conditions, but share common features: anatomical origin and difficult early diagnosis. The existing liquid biopsies and imaging biomarkers are rather imperfect, because their dynamic range does not cover the entire course of the disease, and their classification accuracy is lower than the level accepted for clinical practice, thus, precluding their routine implementation. Thus, the use of CNS-enriched EVs and their molecular content as biomarkers for these specific pathological conditions may revolutionize the clinical management of these patients. Indeed, a growing number of studies within this field have been published in the last few years. To simplify, a summary of the main publications retrieved from the database query can be found in Tables 1 and 2, referring to either cargo or surface molecules, respectively [9,12, (Figure 3). Alzheimer's disease (AD) is neurodegenerative disorder characterized by memory loss and behavioral changes. EVs have been studied as a potential diagnostic marker for AD. This disease has a complex progression including early development of neuronal dysplasia, angiogenic changes, release of inflammatory mediators by CNS glial and peripheral immune cells, and development of extracellular amyloid-β (Aβ) fibrils that deposit in the brain as amyloid plaques, and impair synaptic plasticity [98]. At the molecular level, the extracellular deposition of insoluble amyloid-β Aβ peptide plaques (39-43 amino acids produced by amyloid precursor protein, APP, peptides) occurs, which can interact with regulatory proteins to phosphorylate the microtubule-associated protein tau. The increase of phosphorylated tau protein (P-S396-tau, P-T181-tau) and Aβ42 detected in plasma EVs is associated with increased risk for AD [98]. Of note, both Aβ and tau are secreted by neurons through the release of EVs [12,26,36,56,90]. Moreover, some authors also studied molecules found in EVs from patients suffering from mild cognitive impairment. Their memory has deficits that do not significantly affect daily functioning, but are often regarded and treated as early phase AD [52,55,99]. Alzheimer's disease (AD) is neurodegenerative disorder characterized by memory loss and behavioral changes. EVs have been studied as a potential diagnostic marker for AD. This disease has a complex progression including early development of neuronal dysplasia, angiogenic changes, release of inflammatory mediators by CNS glial and peripheral immune cells, and development of extracellular amyloid-β (Aβ) fibrils that deposit in the brain as amyloid plaques, and impair synaptic plasticity [98]. At the molecular level, the extracellular deposition of insoluble amyloid-β Aβ peptide plaques (39-43 amino acids produced by amyloid precursor protein, APP, peptides) occurs, which can interact with regulatory proteins to phosphorylate the microtubule-associated protein tau. The increase of phosphorylated tau protein (P-S396-tau, P-T181-tau) and Aβ42 detected in plasma EVs is associated with increased risk for AD [98]. Of note, both Aβ and tau are secreted by neurons through the release of EVs [12,26,36,56,90]. Moreover, some authors also studied molecules found in EVs from patients suffering from mild cognitive impairment. Their memory has deficits that do not significantly affect daily functioning, but are often regarded and treated as early phase AD [52,55,99].   Levels of α-synuclein in EVs were substantially higher in PD patients than in HC. A significant correlation between α-synuclein found on EVs and disease severity was observed.
[18] α-synuclein levels were lower in the PD group than in the ET and HC. Levels were lower in the NTD group than in the TD group. α-synuclein was found to moderately aid in PD diagnosis (AUC = 0.675) and had a potential to diagnose NTD (AUC = 0.761). α-synuclein levels in NDEVs were significantly higher in patients with early-stage PD compared with HCs. Longitudinally increased α-synuclein were associated with higher risk for motor symptom progression in PD. [88] Aß1-42, p- Tau Flow cytometric assay ApoE and Aß-42-positive particle concentrations were reduced in middle and older age subjects, whereas apoAI increased with age. ApoAI and annexin V levels were reduced in MCI and/or AD patients vs. HCs.
[31] EV proteins cystatin C and CD14 were related to CWMLs and the progression of brain atrophy in patients with manifest vascular disease.
[16] T-tau and p-tau181 levels of CSF-derived EV were positively correlated with the t-tau and p-tau181 levels of total CSF in patients, respectively, but not in the HCs. A total of 291 dysregulated proteins were identified by comparing ABE patients with HCs, by mass spectrometry. S100a9, S100a7, lTF and DEFA1 were further validated. 2000× g (20 at RT) + 13,000× g (2 at RT) + flow cytometric assay SCZ patients displayed more phosphatidylserine+ EVs in CSF compared with HCs.
[47] Several combinations of two or three miRNAs were able to distinguish active from quiescent disease with greater than 90% accuracy. Additional miRNAs associated with stable remission, and a positive response to fingolimod in patients with active disease prior to treatment.
[49] MiR-301a levels on EVs were upregulated in glioma patients compared to HC, and correlated with ascending pathological grades. MiR-301a levels were significantly reduced after surgical resection of primary tumors and increased again during GM recurrence, and were independently associated with overall survival.
[40] The expression levels of miR-21, miR-222 and miR-124-3p in EVs of patients with high grade gliomas were significantly higher than those of low grade gliomas and HC, and were decreased in samples obtained after surgery. Parkinson's disease (PD) is other common neurodegenerative disorder. Although PD can be described more correctly as a syndrome caused by different genetic and epigenetic alterations, a shared downstream result is the degeneration of dopamine-releasing axon terminals in the striatum and of corresponding neurons in the substantia nigra, which leads to impairment of motor and speech skills [100]. EVs have also been used as a diagnostic platform for PD. Indeed, increased α-synuclein levels in plasma EVs were associated with PD, and disease clinical severity, although results are inconsistent [18,72,88]. Moreover, reduced levels of CSF's EVs containing apolipoprotein A1 were associated with a higher risk of PD [21].
Multiple sclerosis (MS) and amyotrophic lateral sclerosis (ALS) are very different demyelinating diseases. MS is an autoimmune disease that affects the myelin sheath, which insulates nerve cell fibers in the brain and spinal cord [101], whereas ALS is a motor neuron disease that mainly affects the brain and spinal cord motor neuron cells [102]. For both diseases, several efforts have been made to find disease-related biomarkers in EVs isolated from patients' biological fluids, with a special emphasis in non-coding RNAs.
Non-coding RNAs (ncRNAs) are also generally found in circulating EVs. There are several ncRNA categories, commonly classified according to their size: the long ncRNAs (lncRNAs) with more than 200 nt, and the small ncRNAs (sncRNAs), including microRNAs (miRs), which present less than 200 nt [103,104]. These ncRNAs have been studied as specific disease "signatures" found in EVs in various disorders including MS/ALS. In fact, for all major group of CNS-related diseases mentioned in this review, published data on EVs ncRNAs have been assessed. For instance, in a recent study by Ebrahimkhani S. et al., the utility of serum exosome miRs as disease biomarkers of MS patients under treatment with fingolimod was assessed, and they found that several combinations of two or three miRNAs could discriminate active from quiescent disease with more than 90% accuracy [79]. Likewise, for ALS, Banack et al. found that eight miRNAs, isolated from neural-enriched EVs, significantly distinguished ALS patients from controls, and thus might assist in early diagnosis of this disease [75].
Gliomas are glial cell-derived brain tumors, classified according to cell type, including ependymomas (ependymal cells), oligodendrogliomas (oligodendrocytes) and astrocytomas (astrocytes). The most common and malignant primary brain tumor, the grade IV astrocytoma/glioblastoma multiform belongs to this group of tumors [105]. Glioblastomaspecific or glioblastoma-rich protein and genetic material can be detected in EVs isolated from the biological fluids of glioblastoma's patients. The usefulness of using EVs as platforms for the analysis of specific biomarkers, which can help predict the disease outcome and prognosis of GM patients, have been recently established. For example, EGFRvIII is the oncogenic form of epidermal growth factor receptor (EGFR), never found in normal tissues. According to various reports, EGFRvIII protein, mRNA and DNA have been detected in plasma EVs isolated from patients with glioblastoma carrying EGFRvIII, but not in EVs isolated from noncancer patients [28,41,44]. Similarly, multiple studies have also shown that miR-21 levels are higher in EVs isolated from the serum and CSF of patients with glioblastoma. This finding is supported by the higher levels of miR-21 also described in clinical glioblastoma specimens [20,27,46]. Traumatic brain injury (TBI), also known as intracranial injury, is a brain damage caused by external forces [106]. TBI classification can be based on severity (from mild to severe TBI), mechanism (closed or penetrating head injury), or other characteristics (for example, occurring in a specific location or widespread area). TBI usually causes neurological sequels, which can be seen even in mild TBI. Mild TBI may lead to acute symptoms, including chronic traumatic encephalopathy, cognitive impairment, dementia, movement disorders, and motor neuron dysfunction [107]. One of the main clinical challenges in TBI is to accurately identify its occurrence, and to determine the extent of CNS damage. Peltz et al. showed that in war veterans with a history of TBI, CNSenriched exosome concentration of pTAU, NfL, IL-6, TNFa are associated with cognitive impairment, which may assist clinicians in choosing a more appropriate treatment schedule and follow-up [89].
Psychiatric disorders, like schizophrenia (SCZ) or chronic depression, are estimated to affect a major percentage of the world population and are very difficult to diagnose and manage. Schizophrenia is a neurological disorder characterized by behavioral deficits, associated with impaired locomotor activity and cognitive defects. The diagnosis of schizophrenia is associated with demonstrable alterations in brain structure and changes in dopamine and glutamate neurotransmission in the cortex [108]. In this field, there is a demand for the discovery of biomarkers, which, if successful, can support clinicians in personalized treatment strategies. In this context, Du Y. et al. showed that the expression levels of specific miRs in EVs isolated from the blood of SCZ patients were sensitive to long-term medication [58].
Stroke, which can be categorized into ischemic or hemorrhagic stroke, affects 13.7 million people worldwide each year, and is the second leading cause of death, with 5.5 million deaths each year [109]. A stroke occurs when the blood supply to part of the brain is interrupted or reduced, limiting brain tissue from obtaining oxygen and nutrients, and generally causing brain damage [110]. Several authors explored the importance of the innate immune response as a contributor to the inflammatory response after stroke [111][112][113]. In this regard, Kerr et al. used a technique designated Simple Plex assay (an immunoassay in a microfluidic cartridge) to determine the presence and concentration of some inflammasome proteins in serum EVs of stroke patients [38]. The authors found that protein levels of ASC (caspase-recruitment domain) remained higher in serum-derived EVs from stroke samples, when compared to controls, and performed well as a potential biomarker for this pathological condition [38].
Major depressive disorder (MDD) is a debilitating mental disease characterized by persistent low mood. It affects the behavior as well as various physical functions, such as appetite and sleep [114]. Although the understanding of the neurobiology of MDD has improved in recent years, the knowledge of the mechanisms that may explain the most relevant aspects of the disease is limited. Nonetheless, MDD was associated with smaller hippocampal volumes, changes in activation or connectivity of neural networks, and also changes in the main neurobiological systems that mediate the stress response, including the hypothalamicpituitary-adrenal (HPA) axis, autonomic nervous system, and immune system [115,116]. The identification of susceptibility biomarkers that might identify individuals prone to depression and disease recurrence would improve treatment and allow for recurrence prevention. A series of studies by different research teams have implicated a systemic metabolic dysfunction known as insulin resistance in the pathophysiology and treatment of disorders of mood and cognition, including MDD [117,118]. Therefore, the insulin-receptor (IR) investigated in these studies was identified in brain areas associated with mood and cognition [119]. Indeed, Nasca et al. recently found that insulin-receptor substrate-1 (IRS-1) was highly abundant in neuronal-derived EVs isolated from plasma samples of MDD patients, compared to nonpathological controls, and that higher IRS-1 levels were associated with suicidality and anhedonia in those individuals. This data could lead to an improved strategy in the treatment of MDD patients [86].

Future Perspectives and Conclusions
Research based on EVs, specifically neuronal-derived EVs, is rapidly moving forward. We have now recognized a wide range of biological processes mediated by these molecules, and its importance into cell-cell communication, as well as disease spread within the CNS. With increasing knowledge about EVs from specific neurological diseases and the advances in technologies used to analyze these nanostructures, research is now moving towards clinical translation for biomarker platforms. For this, established and validated biomarkers are needed. Indeed, various single reports based on small cohorts have been published, but very few were validated by larger, cross-sectional investigations. The promising results so far point towards a future where brain-derived EVs could be used for not only for diagnosis of CNS-related disorders, but also for patient monitoring, and for studying the influence of the release of this specific EVs in behavioral and mental health related conditions.
Overall, EVs produced in the brain and circulating in peripheral fluids have become an incredible window to the brain, which is expected to assess the disease status of a given patient with great accuracy and simplicity. In the near future, one can anticipate that a relevant biomarker platform may move quickly into clinical context, to provide an enormous amount of information that may assist in clinical patients' follow-up. This may occur at the physician's office, as many technological solutions exist today to monitor such biomarkers using point-of-care devices. In a long-term perspective, one may anticipate that the information deriving from EVs may also act as an early disease warning system. It is only natural that the biological/biochemical information of such circulating EVs shall be better and more significant in time, considering the huge developments expected in the fields of nanotechnology, bioengineering and artificial intelligence. Specifically, when these are combined for the same purpose, even lower concentrations of a given pool of biomarkers in very complex samples might be detected. Finally, it will be indeed a breakthrough when such advanced biomarker panel from circulating EVs (that may well derive from CNS) becomes a tool to prevent disease progression.