Exploring the significance of lipids in Alzheimer's disease and the potential of extracellular vesicles

Lipids play a significant role in maintaining central nervous system (CNS) structure and function, and the dysregulation of lipid metabolism is known to occur in many neurological disorders, including Alzheimer's disease. Here we review what is currently known about lipid dyshomeostasis in Alzheimer's disease. We propose that small extracellular vesicle (sEV) lipids may provide insight into the pathophysiology and progression of Alzheimer's disease. This stems from the recognition that sEV likely contributes to disease pathogenesis, but also an understanding that sEV can serve as a source of potential biomarkers. While the protein and RNA content of sEV in the CNS diseases have been studied extensively, our understanding of the lipidome of sEV in the CNS is still in its infancy.


INTRODUCTION
Since their discovery more than 30 years ago, it has become clear that small extracellular vesicles (sEV) play a role in the pathogenesis of various neurological disorders [1][2][3][4][5].A subset of sEV are exosomes.
While the protein and RNA content and function of sEV have been subject to intense investigation, only a limited number of studies have been published on sEV lipids [26,31,[36][37][38][39][40][41][42] and even fewer on CNS sEV lipids [43,44].In this review, we describe what is currently known about the lipid composition of EVs, with a focus on CNS-derived sEV.
Changes in lipid metabolism and lipid-regulating enzyme activity exist in many neurological disorders, including Alzheimer's disease (AD) [45][46][47][48][49][50][51].We provide a comprehensive summary of the known lipid changes in the brain, CSF, and blood in AD and their association with disease pathogenesis.We pose the question, 'could the lipid content of sEV provide insight into biological pathways and aid in the diagnosis of AD or other neurological disorders?' and highlight the developments and challenges of sEV isolation for the purposes of lipid profiling.
It is lipids that are key to maintaining sEV morphology and enabling sEV (and their cargo) to travel in biofluids without degradation [60,66,69].
Only a handful of studies to date have examined the biological activity of sEV lipids [61,[71][72][73].Extracellular vesicle PE and PS are known to participate in membrane dynamic modulation and facilitate sEV-cell membrane fusion.PE lipids are present on both leaflets of the sEV membrane in an asymmetric manner while the localization and dynamics of PS lipid reorganization within the membrane bilayer are still unclear [10,57,58,[74][75][76].When PS lipids localize to the sEV outer membrane leaflet, they are recognized by PS receptors (TIM1/4, Annexin 5) on recipient cells, facilitating sEV uptake/fusion and molecular transfer [24,74,75,77].
The lipid content of vesicles in blood and bronchoalveolar lavage fluid (BAL) has been investigated, however, whether these vesicles are sEV is unclear.A study examining serum vesicle lipids in pancreatic cancer found LPC(22:0), PE(16:0/18:1), and alkenyl-ether (plasmalogen-) containing PC(P-14:0/22:2) associated with disease stage and tumour diameter, with PE(16:0/18:1) correlating with survival rate [81].Plasma vesicle eicosatrienoic acid (C20:3) has been proposed as a potential biomarker for severe acute pancreatitis [84] and plasma vesicle lipids are suggested to differentiate between early and late stage of non-small cell lung cancer [85].Significant changes in glycerophosphoglycerol (PG), ceramide-phosphate, and ceramide have been reported in vesicles isolated from the BAL of asthmatics patients A simplified illustration of the endocytic biogenesis pathway and lipid cargo of small extracellular vesicles (sEV).A subset of sEV are referred to as exosomes.They are formed from invagination of the limiting endosomal membrane of the late endosome /multivesicular body (MVB).The MVB fuse with plasma membrane and release the internalised vesicles as sEV into the extracellular environment.sEV lipid bilayer is key to maintaining vesicle morphology and enabling sEV (and their cargo) to travel in biofluids without degradation [60,66,69].sEV are enriched in cholesterol, sphingomyelin (SM), ceramide, glycerophosphoserine (PS), ether glycerophosphoethanolamine (PE), lysophosphatidylethanolamine (LPE) and lysophosphatidylcholine (LPC) relative to the parental cell.Created with BioRender.com.with modification.
and SM(34:1) is thought to be increased in asthmatic patients exposed to second-hand smoke [86].

WHAT IS KNOWN ABOUT CNS DERIVED sEV LIPIDS?
Extensive evidence suggests altered lipid metabolism and abnormal activity of lipid-regulating enzymes in the context of neurological disorders, including AD [45][46][47][48][49][50][51], Parkinson's disease [87][88][89][90][91][92]; frontotemporal dementia [93]; multiple sclerosis (MuS) [43,94], and Lewy body disorders (LBD) [44].At the time of writing, only a handful of studies have reported the lipid profile of CNS-derived sEV [43,44,95] and the potential function of sEV lipids or lipid-regulating proteins [26,43,44,[95][96][97].Our group has shown that lipid dyshomeostasis in AD is also evident in sEV isolated from frontal cortex tissue and that brain-derived extracellular vesicles (BDEV) are enriched in PS and ether-PS lipids [95].Pieragostino et al. investigated CSF sEV lipids of MuS patients with a particular focus on SP lipids, namely SM [43].Acid sphingomyelinase, ASMase, a key enzyme in sphingolipid metabolism hypothesized to be involved in MuS, is also found enriched and active in MuS patient CSF sEV [43].Another study carried out by Kurzawa-Akanbi et al. reported LBD CSF sEV were heavily loaded with ceramides, a characteristic of LBD [44].From the few studies thus far, it is becoming apparent that the lipids and lipid-regulating proteins in sEV can report on the biological changes that occur as a consequence of the cellular impairments that characterize some neurological conditions [26,44,[95][96][97].
Together these studies showed that sEV in the CNS have a similar lipid content to sEV from other tissues, but they are enriched in lipids pertinent to the physiological or pathological state of the CNS.Of note, these studies demonstrate the benefit of analysing sEV over gross tissue/CSF for enhancing lipid signals [43,44,95].Improved detection of lipids will lead to greater insight into the biological/biochemical changes that arise either as causes or consequences of disease mechanisms.These include the role of sEV in disease progression and further understanding into whether sEV lipids drive pathology and or report on preclinical disease.CNS disorders hallmarked by lipid dysregulation [97,98] including AD [45][46][47][48][49][50][51], are likely to benefit from the insight to be gained from profiling sEV lipids.Below, we provide an overview of what is currently known about lipid dysregulation in AD and suggest that sEV could serve as indicators of AD-associated lipid pathobiology and candidate biomarkers to aid in disease diagnosis.

ALZHEIMER'S DISEASE (AD)
AD is a neurodegenerative condition responsible for 60%-80% of dementia cases worldwide [99].Patients experience memory loss and changes in personality and behaviour.Unfortunately, patients are generally diagnosed after the onset of clinical symptoms [100][101][102] and limited treatment options exist [103].
The cause of AD is multifactorial and although a variety of genetic, lifestyle, and environmental factors have been implicated, age is the number one risk factor [104][105][106][107]. Mutations in the amyloid precursor protein (APP), presenilin-1 and presenilin-2 are associated with early onset familial AD [106].The ApoE-ε4 allele is regarded as the major genetic risk factor for late-onset AD, with carriers of ApoE-ε4 having a higher risk of developing dementia than ε3 allele carriers and carriers with the protective ε2 allele [104][105][106][107]. Lifestyle and environmental risk factors that contribute to the likelihood of developing AD include diet, educational attainment, physical exercise, and brain injury, amongst others [102].The importance of any one of these environmental factors in increasing or decreasing the risk of AD will differ from person to person.
Although an extensive array of factors in varied combinations may result in AD, two pathological hallmarks in the brain define the disease: amyloid-β (Aβ) plaques and neurofibrillary tangles (NFT).Aβ plaques accumulate outside neurons and are primarily composed of aggregated Aβ40/42 peptides generated from the cleavage of the APP [108,109].NFT, on the other hand, are intra-neuronal and primarily composed of hyper-phosphorylated tau protein [110].Prior studies have revealed total-tau and phosphorylated tau are associated with cognitive decline in mild cognitive impairment (MCI) and AD [111][112][113][114][115].
No single biochemical test can diagnose AD.The National Institute on Aging and Alzheimer's Association (NIA-AA) have emphasized diagnostic guidelines focusing on differential diagnosis of three stages of AD; preclinical [130], MCI due to AD [131], and dementia due to AD [132].Current AD diagnosis consists of neuropsychological and pathophysiological assessments.Neuropsychological assessments, including the broadly accepted clinical dementia rating (CDR) [133] and the minimental state examination (MMSE) [134,135], are employed to evaluate an individual's cognitive performance.These tests are also utilized to stage disease progression.Pathophysiological assessments include the detection of biomarkers, mainly Aβ40, Aβ42, total tau and phosphorylated tau (p-tau) species in CSF and blood, and imaging (PET and MRI) [136][137][138][139][140][141][142].While biochemical measurements and imaging can be used to accurately diagnose dementia due to AD, they are not routinely performed due to factors such as resource accessibility and cost [100][101][102].
The pathophysiological process of AD occurs decades before the appearance of symptoms and clinical diagnosis [143][144][145][146].This long 'preclinical' phase is an opportunity for therapeutic intervention; however, early diagnosis (and available treatments) is required for this to occur.Recent years have seen considerable breakthroughs in detecting, identifying and quantitating Aβ species, total tau and p-tau species [115,[147][148][149][150][151][152][153][154][155][156] as well as protein markers, that is, glial fibrillary acidic protein and neurofilament light protein [157,158], in CSF and blood.However, several challenges remain, such as the variability in acceleration/deceleration rate of changes in molecules of interest, the complexity and the variable biomarker baselines among individuals, and the specificity of biomarkers.There remains an urgency to develop efficient and accurate blood-based biomarker strategies for clinical and pre-clinical AD diagnosis and to identify new therapeutic targets.

DISRUPTION OF LIPID HOMEOSTASIS IN THE BRAIN IN AD AND ASSOCIATION WITH DISEASE PATHOGENESIS
Lipidomic studies, primarily on post-mortem tissues, suggest that dysregulation of lipid metabolism is a hallmark of AD [46-51, 129, 159-165] (see Table 1 for a summary of published studies).
An overall decrease in AD in the GP lipid category has been reported in the temporal and frontal lobes [46,[166][167][168].The majority of studies report decreased plasmalogen-PE and -PC levels in multiple cortical regions and cerebellum in MCI and end-stage AD [164,166,[169][170][171][172][173][174][175], with one study reporting increased plasmalogen-PE in the superiormiddle frontal gyrus and the superior temporal gyrus via nuclear magnetic resonance (NMR) [168].For the GL lipid category, an overall increase in monoglyceride (MG) and DG is observed in MCI and AD post-mortem frontal cortex [164,165], an increase of DG lipids is further evident in the recent study in the neocortex brains, accompanied with an increase of triglycerides (TG) lipids [175].
In the CNS, SP lipids are involved in signalling cascades, synaptic function, cholinergic function, signal transmission, and neuronal growth (axonal growth).The SM/ceramide cascade is impaired in AD [176] but there is little agreement between studies on the relative expression of SM and ceramide lipid species [47,177].The level of SM has depended on the brain region examined [46,168,178] and could be attributed to the density of myelinated axons in white and grey matter [47,179,180].An increase in ceremide is a consistent finding in the frontal cortex [177,181], the grey matter of the frontotemporal cortex [47], and the middle frontal gyrus [178], with specificity in terms of fatty acyl chain composition [180].There is evidence suggesting saturated ceramides are present in Aβ plaques in the superior temporal gyrus (Braak stage VI) [182].Increased ceramide is linked to mitochondrial dysfunction, oxidative stress, neuronal apoptosis and Aß generation [163,178,[183][184][185], which implicates a role for ceramides in disease pathogenesis.The enzymes involved in SM/ceramide pathways are dysregulated in AD, which is another possible explanation for the enhanced biosynthesis of ceramide [47,177,186].Sphingosine-1phosphate (S1P), a neuroprotectant against Aβ-induced apoptosis, is downregulated in AD and suggested to enhance apoptosis [47,184,187].Sulfatide depletion has been reported in MCI (CDR 0.5) [179] and (Braak stage ≥ II) [166].Degradation of sulfatides is suggested to cause hypo-myelination, resulting in neuronal dysfunction, shrinkage, and cholinergic dysfunction [47,179,187].
The complex gangliosides, GT1b, GD1b, GD1a, and especially GM1, which tightly bind Aβ42 [192], are generally down-regulated in AD [189,193].Cholesterol, a major component of myelin sheaths and lipid rafts, is altered in AD [46,178,180,194] with an increase in cholesterol proposed to enhance Aβ production and secretion [129,195,196] and contribute to memory impairment [196].It is reported that cholesterol accelerates the binding of Aβ to GM1 [197], forming an Aβ-GM1 cluster that not only causes membrane damage but also seeds Aβ accumula-TA B L E 1 Lipid dysregulation in AD brain.

PA
• Reduction in AD superior temporal gyrus grey matter [168].

PS
• Reduction in AD inferior parietal lobule and in occipital cortex [168].

PG
• Reduction in AD superior temporal gyrus grey matter [168].

Plasmalogen-PE
• Plasmalogen-PE deficiency present in frontal, parietal, temporal and cerebellar white matter and grey matter in early stage of AD (CDR 0.5) with no further depletion in white matter in CDR 1, 2 and 3 samples while further depletion was observed with the progression of AD in grey matter in all examined brain regions except for cerebellar cortex [171].
• Deficiency of plasmalogen-PE to PE ratio in AD mid-temporal cortex and in cerebellar grey matter [169].
• Elevation of plasmalogen-PE in the AD superior-middle frontal gyrus and superior temporal gyrus [168].
• Increased SM in middle frontal gyrus (MFG) grey matter and no change in MFG white matter [180].
• Decreased SM in superior temporal cortex white matter in late stage, no change in early stage [179].
• Decreased soluble cytosolic SM and no change of membrane SM in AD frontotemporal grey matter [47].

Sphingosine and S1P
• Increased soluble cytosolic sphingosine and decreased soluble cytosolic S1P in AD frontotemporal grey matter while no change was observed in either sphingosine or S1P in membrane fraction [47].

Hexosyl-ceramide
• glucosylceramide and galactosylceramide were found to be increased in the prefrontal cortex [46].
• Decreased GM1 and GD1a in AD temporal cortex grey matter [189] Glycerolipids MG, DG and TG • Increased pool of DG lipids in AD prefrontal cortex and selected triglyceride (TAG) species in AD entorhinal cortex [46].

Cholesterol
• Increased cholesterol in AD cerebral cortex [194] and middle frontal gyrus grey matter [178,180], with a trend of increase as disease progresses in frontal cortex [178].
• No change observed in AD prefrontal cortex or entorhinal cortex [46].
TA B L E 2 Potential lipid biomarkers reported in CSF of AD patients.

PC metabolites
• Increased choline metabolites, phosphocholine, free choline and PC in AD CSF suggested PC breakdown in AD brain [256].

SM
• Increased in pre-clinical patients compared to non-demented controls but no change in mild or moderate patients compared to controls [258,259].
• All examined SM species were positively correlated with all Aβ species and total-tau [162].
• Increased ceramide levels in AD CSF compared to age matched neurological controls [181].
The brain contains polyunsaturated fatty acids (PUFA), cholesterol, and has a high oxygen level for energy consumption, making it susceptible to oxidative stress, and subsequent oxidative modification.
Plasmalogens exhibit a protective feature by suppressing amyloidogenesis and neuroinflammation induced by lipopolysaccharide in a mouse model [249] and PUFA-containing plasmalogens are suggested to attenuate nitric oxide production in microglia cells [250].It has also been reported that PUFA-containing plasmalogens induce ferroptosis (a type of programmed cell death dependent on iron) that has been implicated in AD [215,[251][252][253][254].Although the molecular mechanism and the biological function of plasmalogens in the brain are not fully understood, changes in peroxisome function and plasmalogen levels could be both biomarkers and therapeutic targets for AD [255].

LIPID CHANGES IN AD CSF AND BLOOD
Changes in the lipid content of CSF (Table 2) and blood (Table 3) occur in preclinical and clinical AD.In CSF, PC lipids and the PC substrates, phosphocholine and choline, are increased [256] and the levels of ceramide and SM lipids, as well as specific PC lipids positively correlate with CSF Aβ42, tau, and p-tau181 [162,181,257].Kosicek et al.
reported increases in multiple CSF SM species in MCI but no change in mild or moderate AD compared to cognitively normal controls [258,259].A significant reduction in sulfatide has also been reported in AD CSF [260], consistent with that reported in the brain [179,187].
Serum biomarker discovery studies have identified specific lipids capable of distinguishing AD from healthy control individuals in discovery studies (Table 3).Saturated and short chain PC, LPC and a group of lipid peroxidation products are up-regulated, while PE, especially plasmalogen-PE, are decreased in AD serum [174,[261][262][263][264]. A longitudinal study spanning nine years showed that serum SM and TA B L E 3 Potential lipid biomarkers reported in serum and plasma from AD patients.

PC
• Increased saturated and short chain fatty acids containing PC lipids with decreased PUFA-PC [261].

SM and ceramides
• High level of SM and ceramide lipids is associated with memory impairment [160].

TG
• Negative correlation between PUFA-TG species with AD neuropathology and brain atrophy in MCI and AD patients compared to control in the ADNI study [265].
• No change between AD and control but decreased plasmalogen-PE level was observed a year later in the same AD cohort [271].
• No change in ceramide level between AD vs control while lower levels of very long chain ceramides, C22:0 and C24:0, were found in MCI patients; Among MCI patients, higher level of ceramides C22:0 and C24:0 predicted further cognitive decline [273].

SM and ceramides
• Increased ratio of ceramide to SM containing same fatty acyl chain in AD [159].

MG and DG
• • Increased MG and DG in MCI patients [165].

Free fatty acid
• • A general increase of free fatty acids was observed in AD plasma [271].

Sterol lipids
Cholesterol esters • The level of long chain cholesteryl esters followed the trend of decrease from CTL to MCI and AD [311].
ceramide levels had potential acting as predictive biomarkers for memory impairment [160].In the Alzheimer's Disease Neuroimaging Initiative (ADNI) serum study, PUFA-TG negatively correlated with AD neuropathology and brain atrophy in MCI and AD patients compared to controls [265].
The majority of lipidomic studies have been performed on patient plasma (relative to CSF or serum) (Table 3).In plasma, a few PC lipids, mainly the PUFA containing species, are decreased in AD [266][267][268] with an increase in PC(40:4) reported by Proitsi et al. [269].Ether lipids, mainly alkyl-ether PC/PE (PC-O and PE-O) and alkenyl-ether PC/PE (PC-P and PE-P), were down-regulated in The Australian Imaging, Biomarker & Lifestyle Flagship Study of Ageing (AIBL) and ADNI AD cohorts [270].Dysregulation in PE and plasmalogen lipid metabolism are worth examining further to pinpoint if alterations in these pathways could serve as targets for therapeutic intervention, or if changes in these lipids are simply a consequence of the disease [271].Alterations in the levels of plasma SM and ceramides were also reported [159,270,272].Importantly, they are altered in MCI and associated with cognitive decline and hippocampal volume loss [161,273].In the AIBL and ADNI cohorts, ceramides containing different acyl chains correlated with AD, irrespective of their sphingoid base, with negative correlation observed in C22:0 and C24:0 species and positive correlation observed in C18:0, C20:0 and C24:1 species [270].The ratio of very long chain to long-chain ceramides, for example, C24:0/C16:0 and C20:0/C16:0, were inversely associated with the risk of developing incident dementia and AD [272].However, the ratio of ceramides C24:0/C16:0 was also found to be negatively correlated with coronary artery disease and acute coronary syndromes in three different patient cohorts [274], indicating that lipid changes in blood can be attributed to multiple factors.The PUFA-TG species, especially, C22:6 containing TG(58:8), together with alkyl-triglyceride (TG-O) lipids, were negatively associated with AD in the AIBL and ADNI cohorts [270], consistent with findings in serum (ADNI) [265].

COULD THE LIPID CONTENT OF sEV PROVIDE INSIGHT INTO BIOLOGICAL PATHWAYS AND AID IN THE DIAGNOSIS OF ALZHEIMER'S DISEASE OR OTHER NEUROLOGICAL DISORDERS?
The first report suggesting sEV may contribute to AD was published by Rajendran et al. in 2006 [275].Since this time, the field has expanded with numerous discovery studies on the function of sEV in AD and their potential as a source of protein and RNA biomarkers.The protein and RNA content and function of sEV in AD will not be reviewed here as it has been covered by others in detail [31,276,277].
While there are numerous studies on the protein and RNA content of sEV, there are few studies on sEV lipids in AD (Table 4).Recently, our group undertook a comprehensive and semi-quantitative lipid profiling of sEV isolated from human post-mortem frontal cortex [95] (Table 4) using an established protocol to isolate BDEV [278].In our study, we showed that BDEV were enriched in PS lipids, specifically ether-PS lipids compared to brain tissue, regardless of disease condition and we identified lipids in BDEV that distinguished AD from neurological control tissue.AD BDEV contained decreased PUFA-containing lipids, including PS(40:6), PE (40:6) and LPE (22:6) containing DHA, LPE (22:4) containing docosatetraenoic acid, and PC (38:4) and PE(38:4) lipids containing arachidonic acid, consistent with that observed in AD tissue [95].Plasmalogen-PE lipids, including PE(P-36:2) and PE(P-38:4), were significantly upregulated in AD BDEV compared to controls [95].This lipidomic data also suggested remodelling of the sphingolipid metabolism pathway in a N-acyl chain dependent manner [95].The predominant sulfatide (d42:2), mainly (d18:1/24:1) was found decreased in AD BDEV [95], consistent with general sulfatide depletion reported in brain tissue [179] and CSF [260] in early stage disease.Lipid dysregulated in the frontal cortex was recapitulated in BDEV, however since BDEV are less complex than brain tissue, the differences, particularly in PUFA-lipids and plasmalogen-PE, between AD and neurological control were more pronounced and statistically significant.
Cohn et al. used a similar isolation approach [278] to examine sEV in the parietal cortex in AD, specifically examining microglia derived sEV (CD11b enriched BDEVs) [96].In agreement with our study, Cohn et al. also reported a decrease in phospholipids harbouring DHA in AD microglial BDEV.They additionally reported upregulation of the most abundant lipid species of LBPA and monohexosylceramide (Table 4).
LBPA is an endo-lysosomal specific lipid, its presence in sEV likely reflects impairments in the endo-lysosomal pathway [4,5].
It is well known that sEV, specifically exosomes, are formed in the endocytic pathway and are packaged with proteins and lipids that almost exclusively come from the endosomal, autophagy, and lysosomal (EAL) pathways [5,11].Crosstalk between the exosome biogenesis and EAL pathways contributes to cellular homeostasis in the form of coordinated release of exosomes and modulation of their cargo depending on the needs of the cell.Alterations in the EAL pathways are well-recognized early neuropathological features of AD, marked by prominent enlargement of endosomal compartments, progressive accumulation of autophagic vacuoles and lysosomal deficits [279][280][281].
Therefore, the composition of the released exosomes might provide insight into the interactions between EAL compartments and enables detection, outside the cell, of pathway specific changes in AD.
Studies by Cohn et al. [96] and our team [95] suggest that sEV could be used as a tool for integrating the EAL pathways and identifying molecular species in the blood that originate from these intracellular pathways.Huynh et al. identified a number of plasmalogen-PE species in AD plasma, some of which were also found in AD BDEV, and reported that they were negatively associated with AD (AIBL and ADNI cohorts) [95,270].They also reported N-acyl chain dependent changes in sphingolipids in AD plasma [270].Considering the complexity of blood composition, studying peripheral sEV lipids might significantly reduce unwanted variability (ʻnoiseʼ) and enhance the sensitivity for detecting the specific lipids of interest.

FUTURE DEVELOPMENTS AND CHALLENGES
To harness the potential of peripheral sEV lipid profiling for clinical practice, certain developments and challenges need to be addressed.
One of the main challenges associated with sEV isolation from plasma or serum is the removal of co-isolated lipoproteins.Due to their common physical features, namely density and particle size, lipoproteins are often co-purified with sEV when using currently available 'EV' isolation techniques or kits [282,283].Lipoproteins are rich in lipids, which, when co-isolated with sEV, confound the identification of sEV specific lipid profiles [284,285].This is one of the main reasons that the precise lipid profile of EVs in blood is still unresolved.Several groups intending to profile the lipid content of sEV in serum or plasma, have likely TA B L E 4 Summary of findings in AD BDEV lipids.
• Increase in total plasmalogen-PE in AD BDEV.
• Increase in free cholesterol.
analyzed sEV in the presence of lipoproteins [286][287][288][289].For example, Peterka et al. isolated plasma 'sEV' via polymer precipitation, a method known to co-isolate lipoproteins [287], and not surprisingly reported an approximately 55%-82% increase in TG (mol% lipid abundance) in 'sEV' relative to plasma via different mass spectrometry platforms [290].Cholesteryl ester and TG lipids are predominant in lipoproteins [285,291].Chen et al. took a more stringent isolation approach, using serial ultracentrifugation and density gradient separation, however their isolation method most likely would have still co-isolated high-density lipoproteins (HDL), which have a similar density to sEV [288].In another study that used a commercial precipitation kit, negligible ApoA1 and ApoB proteins were detected in plasmaand serum-derived sEV compared to HDL and low-density lipoprotein (LDL) enriched particles.However, only GP and SP lipid categories were reported and the differential cholesteryl esters and GL lipid data were not reported [289].Size exclusion chromatography (SEC) [292,293] and serial ultracentrifugation [294] have also been used to isolate vesicles in blood for the purposes of sEV lipid profiling, however these techniques, are unable to separate sEV from HDL (using SEC alone) and other lipoproteins (using ultracentrifugation).
Of the studies published thus far on blood EVs, PS lipids have either not been detected, or are only present as a small percentage of the total lipid concentration [288-290, 292, 294].However, PS lipids are known to be highly enriched in sEV isolated from other sources [56,60,66,69,95].This discrepancy may relate to the source of sEV or the EV corona in plasma and serum [284,295,296].The minimal information for studies of extracellular vesicles (MISEV) 2018 guidelines, suggest using apolipoprotein A1/A2 (major components in HDL), apolipoprotein B (major components LDL/VLDL), and albumin levels to demonstrate the efficiency of contaminant removal (lipoprotein and plasma proteins) from sEV preparations [297].Removal of contaminates can be achieved when density gradient and size exclusion are used in tandem and while these methodologies together may reveal the true lipid content of sEV in blood, they are low throughput, so unsuitable for use in large scale discovery studies and clinical applications [287,298].Thus, new gen-eration, high throughput products capable of enriching sEV from blood plasma and serum without co-isolation of contaminants are needed.
It has been suggested that sEV can cross the blood-brain barrier, possibly via transcytosis [292,299,300].This provides the opportunity to profile the lipid content of BDEV in patient blood.This is of particular interest in CNS disorders characterized by impairments in lipid metabolism.Capturing legitimate BDEV from blood, however, has proven difficult.Several groups have isolated and characterized neuronal, astrocytic, or microglial exosomes (NDE, ADE, MDE, respectively) [301][302][303][304][305][306][307].The isolation of these populations has been via the use of a commercial polymer precipitation kit followed by immunocapture with cell type specific antibodies.Questions have arisen as to whether this technique isolates sEV of specific origin firstly because of lack of antibody specificity [308] and secondly the use of polymer precipitation which is widely known to isolate sEV of low purity.The field is currently reassessing targets for EV immuno-capture and exploring new methods to capture CNS cell type specific sEV from blood.The question will then be, 'are there sufficient numbers of the EV population of interest for downstream lipid analysis and detection of changes associated with disease or treatment?' Advances in mass spectrometry are beginning to enable highthroughput, sensitive, comprehensive, and quantitative detection of lipid species from clinically relevant biological samples [63,270].With further technological advances, we envisage that detection of oxidized lipids will also become easier.As oxidative stress is a hallmark of several neurological diseases [121,122,125,178,214], we predict that comprehensive profiling of oxidized lipids will advance our understanding of disease mechanisms.In the future, mass spectrometry-based lipidomics will become a powerful tool to facilitate comprehensive clinical profiling for disease diagnosis however its clinical application is currently limited for a number of reasons (see Meikle et al. for a comprehensive review on the subject [309]).One reason is the complex nature and number of the lipids in biological fluids, particularly blood plasma and serum.Complexity reduction of clinical samples, such as blood, could be achieved by enriching for sEV to remove non-EV associated lipids.We have shown that sEV not only have a unique lipid signature, but they also provide improved detection of lipids of interest, relative to gross or more complex tissues [95].

CONCLUSION
There is great potential in sEV lipids, particularly in the aspect of diagnosing neurological disorders associated with lipid dyshomeostasis.To take full advantage of this potential, current limitations must be resolved.To overcome these limitations, future research needs to focus on developing high throughput products capable of enriching sEV from blood without co-isolation of contaminants, novel isolation methods to capture CNS cell type specific EVs, and the development of clinically applicable lipidomic platforms.Additionally, research should focus on understanding the role of sEV lipids in health and disease, as well as developing strategies to manipulate sEV lipids for therapeutic purposes.With the right combination of technological advances and scientific understanding, the potential offered by sEV lipids could be fully realized.
phase, and none have yet progressed to validation or clinical use.Most studies on human sEV lipid composition have come from the cancer field, of note colorectal, prostate, renal, and pancreatic cancer [62, 63, 78-81].Some in vitro studies include Lydic et al. that characterized the [62]d composition of a colorectal cancer cell LIM1215, and their derived sEV[62, 63], and reported an enrichment in total lipid content, a distinct sphingolipid profile, and alterations in fatty acyl chain length and saturation degree in sEV compared to the parental cells[63].An indepth lipidomic characterization of the metastatic prostate cancer cell, PC3, and their derived sEV by Llorente et al., reported that sEV are 8.4 times more enriched in lipids per mg protein compared to cells, specifically, glycosphingolipids, SM, cholesterol, and PS[62].