Isolation of mitochondria-derived mitovesicles and subpopulations of microvesicles and exosomes from brain tissues

Abstract

Extracellular vesicles (EVs) are nanoscale vesicles secreted into the extracellular space by all cell types, including neurons and astrocytes in the brain. EVs play pivotal roles in physiological and pathophysiological processes such as waste removal, cell-to-cell communication and transport of either protective or pathogenic material into the extracellular space. Here, we describe a detailed protocol for the reliable and consistent isolation of EVs from both murine and human brains, intended for anyone with basic laboratory experience and performed in a total time of 27 h. The method includes a mild extracellular matrix digestion of the brain tissue, a series of filtration and centrifugation steps to purify EVs and an iodixanol-based high-resolution density step-gradient that fractionates different EV populations, including mitovesicles, a newly identified type of EV of mitochondrial origin. We also report detailed downstream protocols for the characterization and analysis of brain EV preparations using nanotrack analysis, electron microscopy and western blotting, as well as for measuring mitovesicular ATP kinetics. Furthermore, we compare this novel iodixanol-based high-resolution density step-gradient to the previously described sucrose-based gradient. Although the yield of total EVs recovered is similar, the iodixanol-based gradient better separates distinct EV species compared to the sucrose-based gradient, including subpopulations of microvesicles, exosomes and mitovesicles. This technique allows for quantitative, highly reproducible analyses of brain EV subtypes under normal physiological processes and pathological brain conditions, including neurodegenerative diseases such as Alzheimer’s disease and Parkinson’s disease.

EDITORIAL SUMMARY

This protocol describes the isolation from brain tissue of extracellular vesicle (EV) subpopulations, including microvesicles, exosomes and mitochondria-derived mitovesicles, using a high-resolution (iodixanol) density step-gradient. EV characterization and analysis are also presented.

INTRODUCTION

Extracellular vesicles (EVs) are constitutively generated and released into the extracellular environment by all cell types, from prokaryotes to plant and human cells¹. Classically, EVs were divided into two main subgroups: microvesicles — also known as ectosomes or oncosomes — and exosomes. Microvesicles are larger (100–1000 nm) EVs that are generated by the outward budding of plasma membrane microdomains enriched in cholesterol². Exosomes are smaller (50–150 nm) EVs that originate from the endosomal compartment and that are released after fusion of fully matured late endosomes — also called multivesicular bodies — with the plasma membrane¹. We recently demonstrated the existence of a third type of EV that derive intracellularly from mitochondria we named mitovesicles³. The development of the method described here permitted the separation of mitovesicles (50–300 nm) from other small EVs in a specific iodixanol step-gradient fraction, enabling their identification and subsequent study³.

During the last two decades new data has demonstrated that the biological activity of EVs is complex, as they can modulate several pathways once taken up by recipient cells, including inflammation, cell survival and cell migration⁴. Consequently, EVs are now thought to be important players in cell-to-cell communication¹. In the context of brain physiology, EVs have neuroprotective functions by delivering neurotrophic factors between cells^5,6. In addition, numerous studies have demonstrated that EVs can have a critical role in the development of neurodegenerative diseases. For instance, EVs carry key pathogenic proteins, such as the prion protein^7–9, huntingtin and its toxic mRNA¹⁰, tau¹¹, and the amyloid β precursor protein (APP) and its metabolites^12–14, propagating their spread throughout the brain. Thus, EVs have pleiotropic roles in neurodegenerative disorders¹⁵: they attenuate neuronal stress through the release of intracellular noxious materials^16–18, but can contribute to the propagation of diseases by transporting neurotoxic molecules throughout the brain¹³.

EV subpopulations derived from different intracellular and plasma membrane compartments have different roles in the homeostasis of brain cells, and consequently they can have differential contribution to the development of neurodegenerative diseases³. Accordingly, investigating EV heterogeneity and complexity would shed light on both physiological and pathological processes in the brain, and give new insight into the progression of complex brain diseases such as prion diseases, Alzheimer’s disease, and Parkinson disease. Studies have already shown specific reciprocal relationships between endosomal and exosomal pathways in this context¹⁵. For instance, it was revealed that endosomal dysfunction that occurs at early stages of Down syndrome, Alzheimer’s disease^6,17,19 and spinocerebellar ataxia²⁰ can affect secretion of exosomes into the brain and cerebellum extracellular space^16–18,20. The converse was shown in the brain of human and murine carriers of the apolipoprotein allele ε4 — the major genetic risk factor for sporadic Alzheimer’s disease — where dysregulated exosome secretion precedes endosomal pathology^18,21. The secretion of microvesicles is unaffected in Down syndrome, whereas mitovesicle biology is altered, likely linked to mitochondrial dysfunction³.

Potential applications of EVs isolated from the extracellular space of the brain of individuals with neurodegenerative disease include the development of new diagnostic and prognostic platforms for blood circulating EVs. The existence of a marker that can be used to immunocapture brain-derived exosomes or microvesicles from blood is still controversial and the ability to isolate brain-derived EVs from blood or cerebrospinal fluid (CSF) is under debate²². However, there is no doubt that peripheral circulating EVs may be a game changer in the early diagnosis of neurological diseases. In this context, the study of mitovesicles opens a novel research field. It was shown that platelets release mitochondrial components, interpreted as ‘extracellular free mitochondria’, without a surrounding plasma membrane²³. However, public repositories of human plasma proteomes such as the Human Protein Atlas^24–26 https://www.proteinatlas.org/ show that extracellular mitochondrial datasets do not contain the full list of proteins that are found in mitochondria. The list of mitochondria proteins that are detectable in blood overlaps with the brain mitovesicle proteome (lacking all mitochondrial proteins involved in anabolic processes)³. Moreover, mitochondrial bodies released into the blood have a smaller size when compared to entire mitochondria (approximatively 100–200 nm)²³. These data suggest that these mitochondrial bodies are mitovesicles and provide experimental evidence for the existence of mitovesicles in the blood. In the context of neurodegeneration, the study of mitovesicle biology in the brain parenchyma can provide novel directions to identify biomarkers in blood. For instance, it was shown that circulating EVs isolated from plasma of Alzheimer’s disease patients have higher levels of mitochondrial mRNAs (mtRNAs) when compared to age-matched controls devoid of pathology²⁷. Given that mitochondrial damage is a typical phenotype found in Alzheimer’s disease, and given that mitovesicles contain mtRNAs and mitochondrial DNA (mtDNA), this evidence is consistent with our data in the brain³, as we demonstrated higher secretion of mitovesicles under conditions of mitochondrial dysfunction and higher levels of mitovesicles in the brain parenchyma of a mouse model of Down syndrome³.

Development of the protocol and comparison with other methods

Methods for the isolation of EVs from extracellular fluids, such as cerebrospinal fluid, blood plasma, and cell culture media are well-established and routinely and successfully performed worldwide¹. One of the most widely used method is size-exclusion chromatography (SEC), a technique that enriches particles and molecules in different fractions according to their size. Given that the diameter of EVs and soluble proteins differs by several orders of magnitude, SEC is appropriate to separate a total pool of EVs from extracellular proteins²⁸. However, EV subpopulations differ at most by only one order of magnitude and, although small microvesicles tend to be bigger than other EVs³, the diameters of different brain EV subpopulations, including small microvesicles, exosomes, and mitovesicles, are largely overlapping³. Therefore, due to its limited resolution, separation of small EV subpopulations remains a challenge using SEC²⁸.

In order to investigate the physiological and pathological roles of EVs in the brain in vivo, a method that reliably and reproducibly isolates small EVs from brain tissue and separates them into subgroups is imperative. Indeed, the mechanisms underlying biogenesis and secretion of EVs in the brain as well as changes that occur in EV species in neurodegenerative and neurodevelopmental disorders can only be identified in brain tissue. Better characterization of brain EVs would contribute to determining appropriate molecular targets to manipulate EV levels and content in order to minimize their impact on the risk of developing pathology. Furthermore, detection of abnormalities identified in brain serum EVs or CSF EVs early in the disease process may serve as biomarkers for at-risk individuals.

We were the first group to develop a method to isolate EVs from murine and human post-mortem brains using a sucrose-based step-gradient^12,16,29. Several research groups proposed modifications to our original protocol but have largely used our same overall strategy involving a short enzymatic digestion of the brain tissue to loosen the extracellular matrix (ECM), followed by differential centrifugation and a sucrose gradient^30–32. However, sucrose has several intrinsic chemical disadvantages when used as a density medium. Body fluids have an osmolarity between 250–300 mOsm/L³³ and sucrose is isosmotic and isotonic with body fluids only in the concentration range of 0.25–0.3 M, which corresponds to the first layer of the gradient (fraction a and part of fraction b). All the other layers of the gradient are hyperosmotic, including the sucrose solutions forming the interlays between fractions b, c, and d, where most of the EVs float (Table 1). The hyperosmotic nature of the sucrose gradient can cause vesicle shrinkage and can be of concern if biologically active EVs are required for later analyses. Moreover, sucrose-based solutions with similar molarities cannot be layered without mixing and this property hinders the generation of high-resolution step-gradients with sucrose as a density medium. Accordingly, sucrose is only partially effective in separating EV subpopulations.

Table 1 |.

Density range of fractions separated by sucrose (the fractions containing EVs are shown in bold)

Fraction	Molarity (M)	Density range (g/mL at 20 °C)
a	0.25	Lower than 1.033
b	0.25–0.60	1.033–1.078
c	0.60–0.95	1.078–1.114
d	0.95–1.30	1.114–1.169
e	1.30–1.65	1.169–1.213
f	1.65–2.00	1.213–1.267
g	Higher than 2.00	Higher than 1.267

We have recently modified our method of fractionation by using iodixanol to create the density column, demonstrating a successful separation between different subtypes of brain EVs using this alternative medium^3,20. Iodixanol is isosmolar with body fluids in a wide range of dilutions, is inert, has a relatively low viscosity and permits the formation of fractions with a closer density range than sucrose (Table 2), thus separating EVs with a higher resolution power. We describe this novel protocol in detail below and compare it with the sucrose-based density gradient. We also outline four different approaches to characterize brain EVs including nanotrack analysis (NTA), electron microscopy, western blot analysis, and measuring mitovesicular kinetics of ATP and show representative data for both iodixanol- and sucrose-based EV fractions. The described protocol uses freshly removed or previously frozen murine hemi-prosencephalon (hereafter hemibrain) but it is appropriate for murine cerebella and human brain tissues.

Table 2 |.

Density range of fractions separated by iodixanol (the fractions containing EVs are shown in bold)

Fraction	Percentage	Density range (g/mL at 20 °C)
1	Lower than 5	Lower than 1.054
2	From 5 to 7	1.054–1.064
3	From 7 to 9	1.064–1.074
4	From 9 to 11	1.074–1.083
5	From 11 to 13	1.083–1.093
6	From 13 to 15	1.093–1.103
7	From 15 to 20	1.103–1.127
8	From 20 to 40	1.127–1.223
9	Higher than 40	Higher than 1.223

Overview of the procedure, limitations, and possible solutions

The first stage of the protocol is to either dissect hemibrains or cerebella from mice (Steps 1–14) or to obtain postmortem human brain tissue. Subsequently, the brain EV isolation procedure is comprised of four main stages: mild enzymatic digestion of the tissue to release interstitial fluid (Steps 15–23), filtration and low-speed centrifugation (Steps 24–35), ultracentrifugation (Steps 36–40) and discontinuous step-gradients (Step 41, Fig. 1). Lastly, brain EV preparations are characterized using several techniques (Steps 42A-E). Steps 15 to 23 of this protocol are similar to techniques used to isolate intact neurons and glia from adult rodent brains^34,35. We adapted these procedures for EV isolation by discarding the cell-rich pellet and saving the EV-rich supernatant after a 300g centrifugation of the digested tissue, which is further cleared by a 40 μm mesh filtration.

Fig. 1 |. Overview of the procedure used to isolate brain EV subpopulations.

(a) Schematic representation of the procedure used to isolate EVs from the right murine hemibrain. (b) Representative photographs of the most important steps during brain EV isolation (from steps 15 to 40). (c) Representative photographs of the most important steps during the separation of brain EV subpopulations (from step 41A and step 41B). A sucrose low-resolution gradient (left) and an iodixanol high-resolution gradient (center) loaded with brain EVs are shown before and after the overnight ultracentrifugation at 200,000g. The right panel shows the collection of the iodixanol EV fractions and the appearance of the EV pellets before resuspension. All animal procedures were performed following the National Institutes of Health guidelines with approval from the Institutional Animal Care and Use Committee at the Nathan S. Kline Institute for Psychiatric Research.

The enzymatic digestion of tissue with papain is routinely used to loosen the ECM and release cells^35,36 and minimizes the rupture of cellular plasma membranes. This is particularly important in an EVs isolation strategy to prevent contamination with intracellular vesicles. Enzymatic digestion of brain tissue with papain is essential, but may intrinsically carry a disadvantage, given that papain lacks substrate specificity³⁷ and therefore may cause digestion of EV surface proteins as well, affecting downstream analysis of the EV proteome.

Nevertheless, our protocol guarantees high EV yield that avoids major degradation products¹² due to several adjustments we have made compared to standard brain cell isolation procedures. First, our protocol includes a shorter 15 min papain treatment compared to the 30 min treatment used for neuron isolation³⁵. Second, we add protease inhibitors immediately at the end of the digestion that block papain hydrolase activity with high potency, including E64 which is an irreversible specific papain inhibitor³⁸. The rigorous control of papain digestion is even more important in the context of neurodegenerative studies. Specific protein cleavage products such as APP metabolites are crucial in the pathology of Alzheimer’s disease and their presence in brain EVs may indicate a mechanism for the disease spreading^12,14. Accordingly, confounding elements including the non-specific cleavage by papain must be avoided to validate the presence of these fragments and study them. Western blot and total protein stain analyses of brain EVs isolated using our protocol show the presence of full-length proteins and the absence of low-molecular weight smears (Figs. 2 and 3), indicating a lack of major degradation products being generated by papain. However, the possibility that papain cleaves proteins on the surface of EVs during EV extraction should be ruled out by treatment of the brain tissue with alternative proteases that have been used for brain cell isolation, including accutase³⁹, dispase, and the neutral protease from Clostridium histolyticum ⁴⁰. The presence of the same bands with different preparing enzymes (that have different substrate specificity) may confirm the non-artifactual nature of the result. Of note, brain ECM has a unique composition as compared to the ECM of systemic organs and lacks common ECM components such as collagen and fibronectin⁴¹. Therefore, while successfully applied for the isolation of EVs from tumoral masses^32,42, the use of collagenases is not recommended for isolation of EVs from brain tissue.

Fig. 2 |. A 0.22 μm filtration step is crucial for EV quality and yield.

(a) Representative cryo-EM photomicrograph of EVs obtained prior to loading on a column (precolumn EVs) when the 0.22 μm filtration step is omitted. Note the presence of cell debris and electron-dense membranous aggregates. Scale bar: 200 nm. (b) Representative cryo-EM photomicrograph of precolumn EVs after the sample is passed through a 0.22 μm filter. No debris or aggregates are found. Scale bar: 200 nm. (c-e) Representative western blot analyses of protein markers for small microvesicles (Annexin A2), large microvesicles (β-actin), exosomes (Cd63) (c), and of mitovesicles (Hsp60) (d) when precolumn EVs are passed through 0.22 μm or 0.45 μm filters from different manufacturers (see the ‘Experimental Design’ section of the Introduction for further details). Lamin A/C, Mfn2, and Polg are contaminant cellular proteins that are not present in EVs. (e) Total protein staining with SyproRuby of precolumn brain EVs and brain homogenate (BH) are also shown. Precolumn EV lysates passed through 0.22 μm filters are indicated with numbers (1 to 5) while precolumn EV lysates passed through 0.45 μm filters are indicated with capital letters (A to C). Each number or letter corresponds to an isolation performed with uniquely manufactured filters. Although we tested in total seven types of 0.22 μm filters and three types of 0.45 μm filters, we were able to complete the isolation for only five types of 0.22 μm filters as the other two clogged instantaneously during the process (see the ‘Experimental Design’ section of the Introduction for further details). 1: CELLTREAT’s standard filter; 2: Corning’s pre-filtered filter; 3: Corning’s standard filter; 4: Cytiva’s Whatman Puradisc filter; 5: Enertech Solutions’ standard filter; A: 0.45 μm Enertech Solutions’ pre-filtered filter; B: 0.45 μm Enertech Solutions’ standard filter; C: 0.45 μm Cytiva’s Whatman GD/X (with a patented double pre-filter designed for hard-to-filter samples). For specifications, more information and catalogue numbers of each filter please refer to the ‘Experimental Design’ section of the Introduction. Equal volumes of precolumn brain EV lysates (3 μL) were loaded in each lane. The molecular weight markers shown in (c-e) are expressed in kilodaltons (KDa). All animal procedures were performed following the National Institutes of Health guidelines with approval from the Institutional Animal Care and Use Committee at the Nathan S. Kline Institute for Psychiatric Research.

Fig. 3 |. High-resolution iodixanol-based column fractionates subtypes of EVs that are not separated by sucrose-based column.

(a) SyproRuby total protein staining of a PVDF membrane after overnight (~16 h) transfer of iodixanol brain EV fractions 1 to 8. 20 μL (left) or 10 μL (right) EVs fractions were loaded after lysis in RIPA and prepared as indicated in the text. The arrowheads point to an area in the bottom part of the membrane enriched in lipids (and, thus, not stained). (b-c) Representative western blot analyses of EVs isolated from adult murine right hemibrains and separated through either a sucrose step gradient (b) or an iodixanol step gradient (c). EV fractions are blotted for the microvesicular Annexin A2, exosomal Cd63 and Alix as well as mitovesicular PDHE1-α. The molecular weight markers shown in (b) and (c) are expressed in kilodaltons (KDa). All animal procedures were performed following the National Institutes of Health guidelines with approval from the Institutional Animal Care and Use Committee at the Nathan S. Kline Institute for Psychiatric Research.

Experimental design

Starting material

The procedure described in this protocol is designed for the isolation of EVs from a whole murine hemibrain weighing 150–170 mg. Accordingly, all data presented here originate from this tissue. We have successfully used this protocol for the isolation of EVs from both male and female murine brains between 3 and 24 months of age of various murine backgrounds, including C57BL/6J (refs. ^3,18,20), B6EiC3SnF1/J (abbreviated 2N; refs. ^3,16), B6EiC3Sn-Rb(12.Ts17¹⁶65Dn)2Cje/CjeDnJ (abbreviated Ts2; refs. ^3,16) and B6;SJL Mixed Background (refs. ^12,14,43). We found age-specific and sex-specific differences in the number and composition of brain EVs (unpublished observations [AU: For the unpublished observations citation, please add authors’ initials or, if not authors of this paper, full names] and therefore recommend that all studies include brains from both females and males at a specific age. We typically isolate EVs from right hemibrains and homogenize the left hemibrains of the same mice for the determination of the intracellular expression levels of proteins found in EVs and/or other proteins of interest. Left/right hemibrain asymmetry is not exclusive to humans but also present in rodents, including, for instance, a left-hemisphere dominance for vocalization in mice⁴⁴. Thus, using always the same hemibrain for EV isolation minimizes intrinsic variability between experiments. Taking these precautions, we typically include from three to eight mice per group for statistically meaningful results.

Although the protocol was originally designed for murine hemibrains, it can be successfully applied for the isolation of EVs from postmortem human frontal cortex tissues³ and murine cerebella²⁰ with minimal adjustments. We found that the yield of EVs recovered per mg of tissue — quantified both as the amount of EV protein in each fraction and as the number of particles quantified by NTA — is significantly lower when EVs are isolated from cerebella²⁰ when compared to whole murine hemibrains³. Therefore, it is recommended to isolate EVs from at least two (ideally three) murine cerebella combined. The number of EVs recovered from the human frontal cortex Brodmann area 9 (BA9) is also lower than the number of EVs recovered from an equivalent amount of murine hemibrain and therefore EVs should be isolated from at least 300 mg of human tissue. Similar to murine studies, isolation of EVs from human postmortem brain tissues must take into consideration several variables for statistically meaningful results, including, but not limited to, brain region, sex, genetic background, and age. When comparing each of these variables, isolation of EVs should be conducted at the same time in at least one sample per group. For instance, if the effect of sex is studied, brain EVs from males and females should be isolated together. Brain tissues from brain banks provided in dry ice should be immediately transferred to a −80 °C freezer where they are kept frozen until processing. Immediately prior to EV isolation, pieces of tissue equivalent to 300 mg should be scraped off from the frozen sample in order to avoid defrosting and re-freezing of the tissue.

We do not show here data using either murine cerebella or human postmortem brain tissues as a source of EVs, but the procedure yields to similar results in these tissues as described here for murine hemibrains.

Filtration

In order to minimize contamination from intracellular components and to enrich for small EVs, our protocol includes a filtration step with a 0.2 μm surfactant-free, cellulose acetate (SFCA) filter. When this step is omitted, a large amount of tissue debris and aggregates co-precipitate with the EVs, as revealed by cryogenic electron microscopy (cryo-EM) (Fig. 2a–b). A disadvantage of this approach is that microvesicles larger than 200 nm may remain trapped in the 0.2 μm SFCA filter. This causes the enrichment of small microvesicles in the crude EV pellet, which may not be representative of the total EV population. This aspect can be partially addressed by the use of a more permissive 0.45 μm SFCA filter. For instance, when two microvesicle markers, Annexin A2⁴⁵ and β-actin⁴⁶ are analyzed in a crude EV pellet obtained after filtration with either 0.2 μm or 0.45 μm SFCA filters, Annexin A2 is preferentially found in small microvesicles while β-actin is found only in larger microvesicles (Fig. 2c–e), consistent with what was previously reported⁴⁶. However, when a 0.2 μm is not used after the 0.45 μm filtration step, an increased amount of contamination is observed in the EVs, as revealed by the finding of the N-terminally truncated form of Lamin A/C which is an apoptotic body marker⁴⁷, along with β-actin. Accordingly, we do not recommend the use of only 0.45 μm filters for routine use unless larger microvesicles need to be specifically studied, and particular caution needs to be exercised in the interpretation of the data because of a possible contamination of unrelated vesicles.

Moreover, we found that filters that share the same specifications but are produced by different manufacturers have a differential impact on the yield and purity of the EVs. 0.2 μm Nalgene 25 mm diameter standard filters (cat. no. 723–2520, Thermo Fisher Scientific) and 0.22 μm EZFlow 25 mm diameter standard filters (cat. no. 379–2215-OEM, Foxx Life Sciences) clog almost instantaneously when the preparation is pushed through them and cannot be used for EV isolation. Corning’s 28 mm diameter standard filters (cat. no. 431219, Corning), Cytiva’s Whatman Puradisc 30 mm diameter filters (cat. no. 10462200, Cytiva), CELLTREAT’s 30 mm diameter standard filters (cat. no. 229765, CELLTREAT Scientific Products) and 0.2 μm Enertech Solutions’ 30 mm diameter standard filters (Enertech Solutions) were experimentally found to give good, comparable results in terms of purity and yield. Pre-filtered 0.2 μm filters, including the Corning’s 0.2 μm pre-filtered, 28 mm diameter SFCA-PF filters (cat. no. 431218, Corning), can also be used but with a lower yield of the EVs recovered. Do not use in the same study different types of filters and filters with a different size pore (for instance 0.45 μm, see Fig. 2c–e).

Downstream characterization

Isolated EVs must be thoroughly characterized to ensure lack of intracellular contaminants and enrichment of EV subtypes, as suggested by the MISEV (minimal information for studies of extracellular vesicles) 2018 guidelines⁴⁸. The characterization of EVs includes NTA, electron microscopy (TEM or cryo-EM), and western blotting. Moreover, EV membrane potential, oxidative status, and bioenergetics of metabolically active EVs, including brain mitovesicles, can be studied using commercial kits but with several adjustments³ and should include the most appropriate negative controls, for instance, the presence of a selective chemical inhibitor of the electron transport chain, including FCCP or OA.

MATERIALS

Biological materials

• One frozen murine hemibrain or two frozen murine cerebella combined ! CAUTION All procedures should be performed according to national and local regulations. All relevant guidelines must be followed. All animal procedures reported in this paper were performed following the National Institutes of Health guidelines with approval from the Institutional Animal Care and Use Committee at the Nathan S. Kline Institute for Psychiatric Research. ▲ CRITICAL The data presented here were obtained from the right hemibrains of 12-month-old C57BL/6J male mice, but this protocol can be applied to both male and female murine hemibrains between 3 and 24 months of age of all murine backgrounds. We strongly suggest to always isolate EVs from the same hemibrain (see the ‘Experimental Design’ section of the Introduction for more details).

• 200–300 mg frozen human brain tissue ! CAUTION All procedures should be performed according to national and local regulations. All relevant guidelines must be followed. Informed consent for obtaining postmortem brain tissue must be provided from all subjects.

Reagents

▲ CRITICAL The specific company from which reagents were purchased is not strictly important. The list provided below is intended to show only examples of reagents that were successfully and routinely used for brain EV isolation in our laboratory. Products that share the same characteristics but are manufactured by a different company can also be used. If, to the best of our knowledge and expertise, it is imperative to use a product of a specific company, for instance, the 0.2 μm SFCA filters, then it will be explicitly stated.

General laboratory reagents

• Phosphate-buffered saline without calcium and magnesium, pH 7.4 (PBS, Corning, cat. no. 21–040-CMX12)

• Sodium chloride (NaCl, Fisher Chemical, cat. no. S271–10)

• Tris base (Fisher Bioreagents, cat. no. BP152)

• Hydrochloric acid (HCl, Fisher Chemical, cat. no. A144–212) ! CAUTION Corrosive, toxic if inhaled, irritant. Highly volatile, use only under a fume hood.

Murine brain dissection

• Anesthetic, veterinary grade (Isoflurane, Henry Schein Animal Health, cat. no. 11695–6776-2) ! CAUTION Irritant. Highly volatile, use only under a fume hood.

Crude EVs purification

• Lyophilized papain, vials (Worthington, cat. no. LK003178) ! CAUTION Irritant. Avoid breathing dust. Wear protective gloves/protective clothing.

• Hibernate A (Brainbits, cat. no. HA) ! CAUTION Open only under a laminar flow hood to avoid potentially harmful bacterial contamination.

• Antipain dihydrochloride (Sigma-Aldrich, cat. no. A6191)

• Leupeptin (Sigma-Aldrich, cat. no. L2884)

• Pepstatin A (Sigma-Aldrich, cat. no. P4265)

• E-64 (Sigma-Aldrich, cat. no. E3132)

• Phenylmethanesulfonyl fluoride (PMSF, Sigma-Aldrich, cat. no. P7626) ! CAUTION Corrosive, toxic if swallowed, irritant.

• Dimethylformamide (DMF, Sigma-Aldrich, cat. no. D4551) ! CAUTION Flammable, toxic if inhaled or in contact with skin, irritant.

• Ethanol (EtOH, Fisher Bioreagents, cat. no. BP2818) ! CAUTION Flammable, irritant.

Separation of brain EVs subpopulations

• OptiPrep density gradient medium (Sigma-Aldrich, cat. no. D1556)

• Sucrose (Sigma-Aldrich, cat. no. S5016)

• HEPES 1M (Gibco, cat. no. 15630080)

Nanotrack Analysis (NTA)

• 100 nm calibration beads (Particle Metrix, cat. no. 110–0020)

Electron microscopy

• 20% (wt/vol) Paraformaldehyde (PFA) aqueous solution (Electron Microscopy Sciences, cat. no. 15713) ! CAUTION Flammable, toxic if inhaled, swallowed, or in contact with skin, irritant. May cause genetic defects and cancer. Highly volatile, use only under a fume hood.

• Sodium cacodylate buffer, 0.2 M (Electron Microscopy Sciences, cat. no. 11652) ! CAUTION Toxic if swallowed.

• Uranyl acetate (Electron Microscopy Sciences, cat. no. 22400) ! CAUTION Fatal if swallowed, irritant. May cause genetic defects and cancer.

Analysis of EVs by western blot (WB)

• Sodium deoxycholate (Sigma-Aldrich, cat. no. D6750) ! CAUTION Toxic if swallowed.

• Triton X-100 (Sigma-Aldrich, cat. no. T8787) ! CAUTION Toxic if swallowed, irritant.

• Ethylenediaminetetraacetic acid (EDTA) disodium salt, dihydrate (Fisher Chemical, cat. no. S312–500) ! CAUTION Irritant.

• 2-Mercaptoethanol (β-Me, Sigma-Aldrich, cat. no. M6250) ! CAUTION Flammable, toxic if inhaled, swallowed, or in contact with skin, irritant. Highly volatile, use only under a fume hood.

• Bromophenol blue (Sigma-Aldrich, cat. no. B0126)

• Glycerol (Sigma-Aldrich, cat. no. G5516)

• Sodium dodecyl sulfate, 10%, Ultrapure (SDS, Invitrogen, cat. no. 24730020) ! CAUTION Flammable, toxic if inhaled or swallowed, irritant.

• Glycine (Fisher Bioreagents, cat. no. BP381)

• Methanol (MetOH, Fisher Chemical, cat. no. A412) ! CAUTION Flammable, toxic if inhaled, in contact with skin, or swallowed.

• Acetic Acid (Fisher Chemical, cat. no. A38–212) ! CAUTION Flammable, corrosive. Highly volatile, use only under a fume hood.

• Protein marker for gel electrophoresis, dual color (Bio-Rad, cat. no. 1610374).

• 4–20% Tris-HCl precast protein gels, 26 wells, 15 μL capacity (Criterion, Bio-Rad, cat. no. 3450034)

• 0.45 μm PVDF membranes (Immobilon-P, Millipore, cat. no. IPVH00010)

• Cellulose paper sheets, electrophoresis-grade (Fisherbrand, cat. no. 05–714-1)

• Tween 20 (Sigma-Aldrich, cat. no. P1379)

• Sodium azide (NaN₃, Fisher Bioreagents, cat. no. BP922) ! CAUTION Fatal if inhaled, in contact with skin, or swallowed. Toxic.

• Bovine serum albumin, ≥98% (BSA, Sigma-Aldrich, cat. no. A7906)

• Non-fat dry milk, blotting grade (Bio-Rad, cat. no. 1706404)

• β-actin (gene name Actb) antibody (clone 8H10D10; Cell Signaling Technology, cat. no. 3700; RRID: AB_2242334)

• Alix (gene name Pdcd6ip) antibody (clone E6P9B; Cell Signaling Technology, cat. no. 92880; RRID: AB_2800192)

• Annexin A2 (gene name Anxa2) antibody (clone EPR13052B; Abcam, cat. no. ab178677)

• Cd63 (gene name Cd63) antibody (clone EPR21151; Abcam, cat. no. ab217345; RRID: AB_2754982)

• Lamin A/C (gene name Lmna) antibody (clone E-1; Santa Cruz Biotechnology, cat. no. sc-376248; RRID: AB_10991536)

• Mitochondrial heat shock protein Hsp60 (gene name Hspd1) antibody (clone EPR18245; Abcam, cat. no. ab190828; RRID: AB_2814692)

• Mitochondrial DNA polymerase γ, catalytic subunit (gene name Polg) antibody (clone EPR7296; Abcam, cat. no. ab128899; RRID: AB_11145308)

• Mitofusin-2 (gene name Mfn2) antibody (clone 6A8; Abcam, cat. no. ab56889; RRID: AB_2142629)

• Pyruvate dehydrogenase E1, subunit α (PDHE1-α, gene name Pdha1) antibody (clone D-6; Santa Cruz Biotechnology, cat. no. sc-377092; RRID: AB_2716767)

• Tumor susceptibility gene 101 (Tsg101, gene name Tsg101) antibody (polyclonal; Proteintech, cat. no. 14497–1-AP; RRID: AB_2208090)

• Donkey anti-rabbit IgG antibody (horseradish peroxidase-conjugated; Jackson ImmunoResearch, cat. no. 711-035-152; RRID: AB_10015282)

• Goat anti-mouse IgG antibody (horseradish peroxidase-conjugated; Jackson ImmunoResearch, cat. no. 115-035-062; RRID: AB_2338504)

• SyproRuby total protein stain (Lonza, cat. no. 50565)

• ECL substrate (Pierce, cat. no. 32106)

• Femto ECL maximum sensitivity substrate (Pierce, cat. no. 34096)

ATP assay and mitovesicle kinetics

• Potassium phosphate monobasic (KH₂PO₄, Fisher Chemical, cat. no. P285–500)

• Magnesium acetate tetrahydrate (MgAc₂·4H₂O, Fisher BioReagents, cat. no. BP215–500) ! CAUTION Flammable (dust).

• Adenosine 5′-diphosphate, sodium salt (ADP, Sigma-Aldrich, cat. no. A2754)

• Sodium pyruvate (Pyr, Sigma-Aldrich, cat. no. P5280) ! CAUTION Irritant.

• l-(−)Malic acid (Mal, Sigma-Aldrich, cat. no. M7397) ! CAUTION Irritant.

• Carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP, Sigma-Aldrich, cat. no. C2920) ! CAUTION Irritant, corrosive, toxic if in contact with skin or swallowed.

• Antimycin A from Streptomyces sp. (Anti-A, Sigma-Aldrich, cat. no. A8674) ! CAUTION Fatal if swallowed.

• Oligomycin from Streptomyces diastatochromogenes (Oligo, Sigma-Aldrich, cat. no. O4876) ! CAUTION Flammable, toxic if inhaled or swallowed.

• Dimethyl sulfoxide (DMSO, Sigma-Aldrich, cat. no. D8418) ! CAUTION Flammable.

• ATP determination kit (Thermo Scientific, cat. no. A22066) ! CAUTION Firefly luciferase (Component B) may be toxic if inhaled ! CAUTION Dithiothreitol (DTT, Component C) may be toxic if in contact with skin or swallowed.

Equipment

General laboratory equipment and plasticware

• 0.1–10 μL low-retention pipette tips (Fisherbrand, cat. no. 02–717-134)

• 2–200 μL low-retention pipette tips (Woodpecker, Thomas Scientific, cat. no. P2101-N)

• 100–1000 μL low-retention pipette tips (VWR, cat. no. 76323–456) ▲ CRITICAL although tips purchased from other manufacturers can be used, we strongly suggest the use of low-retention tips instead of regular tips in all steps to minimize sample loss and variability. Resuspended EVs and density fractions are sticky and tend to remain attached to the side of regular tips.

• 1 mL graduated transfer pipets, sterile (Gosselin, Corning, cat. no. PAS1A-01)

• 1 mL slip tip syringes (BD, cat. no. 309659)

• 10 mL eccentric tip syringes (BD, cat. no. 305462)

• 10 mL Luer-Lok syringes (BD, cat. no. 309604)

• 10 cm Petri dishes (Falcon, Corning, cat. no. 351029)

• 26 G needles (BD, cat. no. 305111)

• Kimwipes Delicate Task Wipers (Kimberly-Clark Professional, cat. no. 34155)

• High precision balance (Mettler Toledo, cat. no. XPR303SN)

• Benchtop pH meter (Thermo Scientific, cat. no. STARA2116)

• Dry bath incubator (Isotemp, Fisherbrand, cat. no. 88–860-022)

• Water bath incubator (Isotemp, Fisherbrand, cat. no. FSGPD05)

Murine brain dissection

• Thick wax sheets (Electron Microscopy Sciences, cat. no. 50-948-964)

• Single edge razor blades (Stanley, cat. no. 11–515)

• Spoonulet spatula (Fisherbrand, cat. no. 14-375-20)

• Dissecting fine-pointed forceps (Fisherbrand, cat. no. 08–875)

• Standard dissecting scissors (Fisherbrand, cat. no. 08-951-20)

• Metzenbaum dissecting scissors (Surgical Design, cat. no. GM128S)

Crude EVs purification

• Single edge razor blades (Stanley, cat. no. 11–515)

• 40 μm cell strainers (Fisherbrand, cat. no. 22–363-547)

• 0.2 μm surfactant-free cellulose acetate (SFCA) filters for syringes (Corning, cat. no. 431219) ▲ CRITICAL Even if sharing the same specifications, we found that filters manufactured by different companies may have a differential impact on the yield and purity of the EVs recovered (Fig. 2)

• 70 mL ultracentrifugation polycarbonate bottles (Beckman Coulter, cat. no. 355622)

• Type 45Ti, titanium fixed-angle rotor (Beckman Coulter, cat. no. 339160)

• Optima XE-90 floor-type ultracentrifuge (Beckman Coulter, cat. no. A94471)

• Allegra X-30R tabletop-type refrigerated centrifuge (Beckman Coulter, cat. no. B06320)

Separation of brain EVs subpopulations

• 14 mL open-top, thin wall, ultra-clear tubes (Beckman Coulter, cat. no. 344060)

• 6.5 mL open-top, thick wall, polycarbonate tubes (Beckman Coulter, cat. no. 355647)

• MLA-80 fixed angle rotor (Beckman Coulter, cat. no. 367096) or Type 70.1Ti, titanium fixed-angle rotor (Beckman Coulter, cat. no. 342184)

• SW 40Ti, titanium swinging-bucket rotor (Beckman Coulter, cat. no. 331301)

• Optima XE-90 floor-type ultracentrifuge (Beckman Coulter, cat. no. A94471)

• Optima MAX-XP tabletop-type ultracentrifuge (Beckman Coulter, cat. no. 393315)

NTA

• ZetaView (Particle Metrix, cat. no. PMX-220)

• 500 mL-capacity 0.22 μm vacuum filter/storage bottle system (Corning, cat. no. 430758)

Electron microscopy

• R 2/2 copper 200 mesh Quantifoil grids (Ted Pella, cat. no. 656–200-CU)

• Formvar-covered carbon-coated copper, 200 mesh grids (Ted Pella, cat. no. 01810)

• TEM Grid holder block (Ted Pella, cat. no. 16820–25)

• Cryo-transfer holders (Gatan, cat. no. 915)

• TEM single tilt holder (ThermoFisher Scientific, cat. no. MOTILTROTSAMPLHOLD)

• PELCO easiGlow glow discharge cleaning system (Ted Pella, cat. no. 91000)

• Vitrobot Mark IV (ThermoFisher Scientific, cat. no. VITROBOT)

• Cryo-EM tweezers (Mitegen Nanosoft, cat. no. MCEMNSCT01)

• Talos L120C electron microscope with Ceta Camera (ThermoFisher Scientific, cat. no. N/A)

Analysis of EVs by western blot (WB)

• Standard power supply (PowerPac Basic, Bio-Rad, cat. no. 1645050)

• Vertical electrophoresis cell (Criterion, Bio-Rad, cat. no. 1656001)

• Blotter with plate electrodes (Criterion, Bio-Rad, cat. no. 1704070)

• Laboratory bench rockers (Labnet, cat. no. S2025-D-B)

• Imaging system for fluorescence and chemiluminescence (iBright CL1500, ThermoFisher Scientific, cat. no. A44114)

ATP assay and mitovesicle kinetics

• SpectraMax M multi-mode microplate reader (Molecular Devices, cat. no. N/A)

• Flat, untreated, white, polystyrene, 96-well microplates (Corning, cat. no. 3912) ▲ CRITICAL The ATP assay proposed in this protocol is a luminescent assay. Transparent plates jeopardize the results as the light generated in one well can pass through and affect the signal of the wells in close proximity. Black plates reflect light, and this causes signal dispersion. The wells in white ELISA plates are engineered for protein binding and absorption and this may alter the results.

Software

• NTA acquisition: ZetaView (version 8.05.12 SP1, Particle Metrix)

• Signal acquisition for the ATP assay: SoftMax Pro (version 6.5.1, Molecular Devices)

• Densitometry: ImageJ⁴⁹ (version 1.53c, National Institute of Health; https://imagej.nih.gov/ij/download.html)

• Statistics and graphing: Prism (version 9.2.0., GraphPad; https://www.graphpad.com/)

Reagent setup

Papain solution

Under a laminar flow hood, resuspend the lyophilized papain contained in a single vial (~140 U/vial) with 7.5 mL Hibernate A, for a final papain concentration of ~20 U/mL. Each vial is enough for either two murine hemibrains, six murine cerebella or two postmortem human brain tissues, as it can successfully digest up to 500–600 mg brain tissue. Move 3.5 mL of this solution into a 15 mL conical tube. Prepare one conical tube per hemibrain. Preheat in a water bath at 37 °C for 15 min prior to the start of the procedure ▲ CRITICAL Papain may lose enzymatic activity over time, therefore, always prepare fresh papain resuspensions just before use. If more than two hemibrains are analyzed, mix together the papain solutions from different vials before aliquoting the 3.5 mL per hemibrain. This will avoid batch-to-batch variability in enzymatic activity.

Protease and papain inhibitor stock solutions

To prepare a 1 mM E64 1000X stock solution, resuspend 1 mg of E64 in 2.8 mL ddH₂0, aliquot, and freeze at −20 °C. For 1000X LAP (leupeptin, antipain, and pepstatin A), resuspend 5 mg leupeptin in 1 mL DMF by pipetting up and down several times, transfer the solution to another vial containing 5 mg antipain dihydrochloride, resuspend again by pipetting up and down, and finally transfer the mixture to a third vial containing 5 mg pepstatin A. Resuspend, aliquot, and freeze at −20 °C. To prepare 100 mM PMSF 100X stock solution, resuspend 174 mg of PMSF in 10 mL pure EtOH. Aliquot, and freeze at −20 °C. Protease and papain inhibitor stock solutions are stable for several months at −20 °C ! CAUTION PMSF is toxic ▲ CRITICAL PMSF is very unstable in aqueous solutions and degrades rapidly, therefore, do not resuspend it in water. PMSF can precipitate over time in EtOH: always double check the presence of crystals in stock solutions and vortex them thoroughly before adding to your samples.

Papain stop solution

Prepare a papain stop solution containing 5 μg/mL LAP, 1 mM PMSF, and 1 μM E64 in Hibernate A. Prepare 6.5 mL of this solution per hemibrain ! CAUTION PMSF is toxic ▲ CRITICAL PMSF is very unstable in Hibernate A and degrades rapidly. Therefore, prepare a fresh papain stop solution just prior to usage and quickly move it into ice.

Sucrose density gradient solutions

Prepare a sucrose working solution (SWS) by dissolving 6.85 g of sucrose in 10 mL of 20 mM HEPES pH 7.4 (final sucrose concentration = 2 M). Subsequently, prepare the step gradient solutions as indicated in the table below ▲ CRITICAL The volumes shown in the table are intended for one column. Multiply the volumes proposed by the number of columns if more columns are used. Always prepare fresh sucrose solutions as they can become contaminated if stored for a long time.

Final sucrose concentration	Volume of SWS	Volume of 20 mM HEPES pH 7.4
0.25 M	0.31 mL	2.19 mL
0.60 M	0.75 mL	1.75 mL
0.95 M	1.19 mL	1.31 mL
1.30 M	1.63 mL	0.88 mL
1.65 M	2.06 mL	0.44 mL
2.00 M	2.50 mL	-

Iodixanol density gradient solutions

Commercial OptiPrep solutions usually contain 60% (wt/vol) of iodixanol in water and need to be pre-equilibrated in the right buffer system before usage. Accordingly, prepare solution A by dissolving 0.17 g of sucrose in 2 mL of 60 mM Tris-HCl pH 7.4 (final sucrose concentration = 0.25 M). Add 1 mL of solution A to 5 mL of OptiPrep in order to obtain 6 mL of OptiPrep working solution (OWS, corresponding to 50% iodixanol in 10 mM Tris-HCl pH 7.4). Prepare solution B by dissolving 1.28 g of sucrose in 15 mL Tris-HCl 10 mM pH 7.4 (final sucrose concentration = 0.25 M). Prepare the step gradient solutions as indicated in the table below. Vortex all solutions thoroughly ▲ CRITICAL The volumes shown in the table are intended for one column. Multiply the volumes proposed by the number of columns used ▲ CRITICAL solution B may layer on top of OWS. Mix thoroughly and vortex to ensure that the density gradient solutions are homogeneous. Always prepare solutions A, B and OWS fresh for best results.

Final iodixanol concentration	Volume of OWS	Volume of solution B
40% (wt/vol)	1.60 mL	0.40 mL
20% (wt/vol)	0.80 mL	1.20 mL
15% (wt/vol)	0.60 mL	1.40 mL
13% (wt/vol)	0.52 mL	1.48 mL
11% (wt/vol)	0.44 mL	1.56 mL
9% (wt/vol)	0.36 mL	1.64 mL
7% (wt/vol)	0.28 mL	1.72 mL
5% (wt/vol)	0.25 mL	2.25 mL

2X RIPA buffer

Prepare a 2X RIPA buffer stock solution as indicated below. Sodium deoxycholate requires a long stirring period to fully dissolve. RIPA 2X can be stored at 4 °C for several months ! CAUTION Sodium deoxycholate, SDS and EDTA are toxic.

Reagent	Volume/mass	Final concentration
Triton-X100	1 mL	2.00% (vol/vol)
Sodium deoxycholate	1 g	2.00% (wt/vol)
10% (wt/vol) SDS	1 mL	0.20% (wt/vol)
NaCl	0.877 g	300 mM
200 mM Tris-HCl pH 7.4	25 mL	100 mM
500 mM EDTA	0.2 mL	2 mM
ddH2O	Up to 50 mL	-

6X sample buffer

Prepare a 6X sample buffer (SB) stock solution with the components indicated below. Do not add additional ddH₂O. Mix the first four components in a 15 mL conical tube and place it in a water bath at 50 °C for 15 min. At this temperature, glycerol is less viscous and will permit to dissolve the SDS powder. Shake and vortex the tube until all the powder is completely dissolved. Finally, add bromophenol blue to the mixture. Aliquots kept at −20 °C are stable for several years ! CAUTION β-Me and SDS are toxic ▲ CRITICAL Do not add bromophenol blue together with the other reagents, only at the end. Adding bromophenol blue at the beginning of this process will turn the solution dark blue. This may hinder the ability to visualize SDS particles and to check if the powder is dissolved completely.

Reagent	Volume/mass	Final concentration
Glycerol	5 mL	50% (vol/vol)
SDS (powder)	1.2 g	9% (wt/vol)
β-Me	3 mL	30% (vol/vol)
2 M Tris-HCl pH 6.8	1.87 mL	375 mM
Bromophenol blue	10 mg	0.03% (wt/vol)

5x/6x running/transfer buffer

Prepare the 5x/6x running/transfer buffer as follows. Adjusting the pH of this solution is not needed (it should be ~8.3). This solution is stable for several months at 4 °C.

Reagent	Volume/mass	Final concentration
Tris base	30.3 g	125 mM
Glycine	187.7 g	1.25 M
ddH20	Up to 2 L	-

Running buffer

Prepare the running buffer working solution as follows ! CAUTION SDS is toxic ▲ CRITICAL The volumes shown in the table are intended for one gel. Multiply the volumes proposed by the number of gels loaded. Prepare fresh.

Reagent	Volume	Final concentration
5x/6x running/transfer buffer	100 mL	1X
10% (wt/vol) SDS	5 mL	0.1% (wt/vol) SDS
ddH20	395 mL	-

Transfer buffer

Prepare the transfer buffer as indicated in the table. Refrigerate at 4 °C for at least 1 h before use ! CAUTION MetOH is toxic ▲ CRITICAL The volumes shown in the table are intended for one transferring apparatus, regardless of whether it contains one or two gels. If more tanks are loaded, multiply accordingly. Prepare fresh.

Reagent	Volume	Final concentration
5x/6x running/transfer buffer	250 mL	1X
MetOH	300 mL	20% (vol/vol)
ddH20	950 mL	-

20X Tris-buffered saline (TBS)

Dissolve the components listed below in 800 mL ddH₂0, adjust pH to 7.5 with HCl and bring the volume to 1 L. Stable at 4 °C for at least several months ! CAUTION HCl is toxic and highly volatile, adjust pH under a fume hood ▲ CRITICAL The solution requires several mL of HCl to be correctly adjusted, therefore dissolving directly with 1 L of ddH₂0 is not recommended. The solution is close to the saturation point, and therefore, an initial volume of ddH₂0 lower than 800–900 mL may result in an incomplete dissolution of the components.

Reagent	Volume/mass	Final concentration
Tris base	121 g	1 M
NaCl	175.2 g	3 M
HCl	(Variable, up to pH 7.5)	-
ddH20	Up to 1 L	-

TBS-Tween 20 (TBST)

Add 1 mL of Tween 20 to 1 L of TBS 1X (final Tween concentration = 0.1% (vol/vol)). Stir and save at 4 °C, stable for at least several months ▲ CRITICAL Tween 20 is highly viscous, using 2 mL serological pipets instead of a P1000 may help to pipette it more accurately.

Blocking buffer

Either dissolve non-fat dry milk in TBST at a final concentration of 5% (wt/vol) or dissolve BSA in TBST at a final concentration of 3% (wt/vol), depending on the primary antibody used. Prepare fresh, as no preservative is included.

Primary antibody solutions

Dissolve NaN₃ in 3% BSA blocking buffer to a final concentration of 0.1% (wt/vol) NaN₃. This solution can be used to dilute the primary antibody to the desired concentration and is stable for several months at 4 °C (NaN₃ is a preservative) ! CAUTION NaN₃ is highly toxic ▲ CRITICAL The concentration of the primary antibody varies from one antibody to the other and depends on the signal-to-noise ratio and blotting performances (high background or saturation of the bands with higher concentrations). Routinely, when working with untested antibodies, we start from the dilution of 1:1,000 and then we further dilute or concentrate the primary antibody according to the results obtained by the immunoblotting of a brain EV lysate or a brain homogenate. The dilutions and blotting conditions for the antibodies used in this protocol are indicated below.

Antigen	Dilution	Incubation time and temperature
β-actin	1:10,000	1 h at room temperature
Alix	1:1,000	Overnight (~16 h) at 4 °C
Annexin A2	1:10,000	1 h at room temperature
Cd63	1:3,000	Overnight (~16 h) at 4 °C
Lamin A/C	1:1,000	Overnight (~16 h) at 4 °C
Hsp60	1:1,000	Overnight (~16 h) at 4 °C
Polg	1:500	Overnight (~16 h) at 4 °C
Mfn2	1:1,000	Overnight (~16 h) at 4 °C
PDHE1-α	1:3,000	Overnight (~16 h) at 4 °C
Tsg101	1:500	Overnight (~16 h) at 4 °C

ATP assay buffer (AAB)

Prepare AAB as indicated in the table below. ▲ CRITICAL Always prepare fresh just prior to usage ! CAUTION MgAc₂ is flammable.

Reagent	Volume/mass	Final concentration
Sucrose	0.77 g	225 mM
KH2PO4	60 mg	44 mM
MgAc2·4H2O	26.8 mg	12.5 mM
500 mM EDTA	0.12 mL	6 mM
ddH20	Up to 10 mL	-

Mitochondrial toxin stock solutions

Dissolve FCCP in DMSO to a final concentration of 20 mM (1000X stock solution). In a separate tube, prepare the OA solution by dissolving both Anti-A and Oligo in DMSO to a final concentration of 4 mM for Anti-A and 2 mM for Oligo (1000X stock solution). Aliquot and freeze at −20 °C, stable for at least several months ! CAUTION FCCP, Anti-A and Oligo are toxic (see Reagents) ▲ CRITICAL FCCP, Anti-A and Oligo are light sensitive, protect from light at all times ▲ CRITICAL The natural color of the 20 mM FCCP stock solution is pale yellow. When oxidized, it becomes bright yellow. Old FCCP aliquots may naturally get oxidized, do not use aliquots that change color.

ATP Master Mix

Prepare the ATP Master Mix following the table below. Always prepare fresh. ! CAUTION Firefly luciferase (Component B of the ATP determination kit) and DTT (Component C) are toxic ▲ CRITICAL 10 mL of solution are enough for ~50 wells, therefore, adjust volumes according to the number of wells used ▲ CRITICAL The ATP determination kit provides a reaction buffer to detect ATP; however, it does not stimulate oxidative phosphorylation and is not appropriate for kinetic studies as a standalone reagent. Therefore, we modified the composition of the final solution by adding precursors that are crucial to boost oxidative phosphorylation and the Krebs cycle, including ADP, malate, and pyruvate (see the table below) ▲ CRITICAL DTT is unstable and freezing and thawing cycles can deteriorate its reducing power. Therefore, dissolve the 25 mg DTT powder provided in the kit by adding 1.62 mL of ddH₂0 and prepare aliquots of 120 μL. Thaw each vial only once and use it for one experiment only, then discard the remnants (if any) ▲ CRITICAL D-luciferin (Component A) and firefly luciferase (Component B) are light-sensitive. Keep away from light as much as possible. ▲ CRITICAL Always add the firefly luciferase (Component B) as the last component and mix it very gently, e.g., by inversion. Do not vortex as this can denature the enzyme. ▲ CRITICAL This method to measure ATP is highly sensitive, and any contamination of exogenous ATP can be detected. Avoid direct contact between skin and any component of the mix, plasticware or plate used during the procedure. Wear gloves to avoid ATP contamination.

Reagent	Volume/mass	Final concentration
10 mM D-Luciferin (Component A of the kit)	0.5 mL	0.5 mM
0.1 M DTT (Component C of the kit)	0.1 mL	1 mM
20X Reaction Buffer (Component E of the kit)	0.5 mL	1X
0.1 M ADP	0.1 mL	1 mM
0.1 M Pyr	0.1 mL	1 mM
0.1 M Mal	0.1 mL	1 mM
ddH20	Up to 10 mL	-
Firefly luciferasea (Component B of the kit)	2.5 μL	1.25 μg/mL

^aAdd at the end.

Equipment setup

Murine brain dissection surgical table

Fill a Petri dish with dry ice. Lay a thick wax sheet on top of it. Let the wax cool for at least 5 min before sacrificing the mouse. The surface of the wax sheet will be the refrigerated surgical table where murine brains will be dissected prior to snap freezing.

NTA

Before starting the acquisition, always calibrate the machine using the 100 nm reference beads provided by the manufacturer. The ZetaView software permits the customization of acquisition parameters. For EV samples, set the parameters as follows: sensitivity 85; min size 10; max size 1000; shutter 100; frame rate 30; trace length 15; bin size 5 nm; positions per single reading 11. For the beads, set a sensitivity of 65 ▲ CRITICAL The sensitivity must be set at a minimum of 85 to identify smaller EVs that are dimmer as they scatter light less efficiently. To measure the diameter accurately, do not set the shutter at less than 100.

PROCEDURE

Murine brain dissection ●TIMING 4 min per brain

▲ CRITICAL Although human brain tissues can be used instead of murine hemibrains and cerebella as a source of EVs, this protocol is primarily designed for the isolation of EVs from murine hemibrains. Accordingly, the procedure of how human postmortem brain samples are prepared is beyond the scope of this protocol and will not be detailed. See the ‘Experimental Design’ section of the Introduction for more details.

! CAUTION Experiments using animal and human tissues must follow national and local regulations. The murine anesthesia method used may vary but must be approved by the local animal care committee. Informed consent must be provided when obtaining human postmortem tissues.

• 1Prepare the surgical table before starting the procedure, as indicated in the Equipment setup section, and place 1.5 ml microcentrifuge tubes, properly labeled, in abundant dry ice at least 5 min before starting the dissection.
• 2Anesthetize the mouse following the method approved by your local animal care committee.

  ! CAUTION Anesthetic agents used for veterinary anesthesia, including isoflurane (see Reagents), are toxic for humans and should be handled under a fume hood.

• 3Spray the fur of the mouse around the neck and head region with 70% ethanol and decapitate the mouse. Quickly move the decapitated head onto the wax sheet of the surgical table.

  ! CAUTION Step 3 to Step 11 should be performed as fast as possible to preserve the integrity of the samples, typically in no more than 4 min. ▲ CRITICAL STEP To preserve the integrity of the tissue, brains must be kept as cold as possible during the whole procedure. Keep the brains on the surgical table at all times, which is refrigerated by the presence of dry ice beneath it.

• 4Roll up a Kimwipe and place it at the base of the head, between the skull and the skin, close to the foramen magnum. Delicately move it upwards with both thumbs, so that the wipe always touches the cranium. This procedure will remove the skin covering the skull. quickly (few seconds) without using scissors.
• 5Enter with appropriate scissors into the foramen magnum and make a horizontal cut at each side of the skull, from the medulla oblongata to the eye.

  ▲ CRITICAL STEP From Step 5 to Step 8, make sure not to damage the brain with the scissors, tweezers and/or spatula. Avoid touching the brain parenchyma with your gloved hands. The extraction of intact brains requires some level of expertise that can be obtained with practice.

• 6Cut the skull on the midline in a sagittal orientation, starting from the medulla oblongata to the nose.
• 7Using forceps, grab the left part of the skull from the posterior side and flip it. This will expose the brain underneath. Repeat the same step with the right part of the skull.
• 8Remove the brain out of the skull with a small spatula and delicately place the brain on the wax sheet of the surgical table. Discard the head and make sure the brain is clean, with no fur or skull bits on it.

  ? TROUBLESHOOTING

• 9Make a coronal cut with a razor blade between the cerebellum and the forebrain. Repeat the same operation in the front of the brain to detach the olfactory bulbs.
• 10Make a sagittal cut over the midline to separate the two hemibrains.
• 11With a spatula, delicately move the olfactory bulbs, the cerebellum and each hemibrain into separate 1.5 ml refrigerated tubes prechilled in dry ice (see Step 1). Leave the tubes in dry ice for at least 2–3 min.
• 12Place an empty 1.5 ml microcentrifuge tube on a high precision balance and set it to zero.
• 13Remove the empty 1.5 ml microcentrifuge tube and weigh each tube containing the olfactory bulbs, the cerebellum and each hemibrain.

  ▲ CRITICAL STEP The instructions below include normalization of EV levels to brain weight. Therefore, accurate measurements here will guarantee more reliable data. We recommend a precision of at least ±10 mg but preferably ±1 mg. Each hemibrain should weigh in the range of 150–170 mg and each cerebellum 50–70 mg.

• 14Place the tubes at −80 °C.

  ■ PAUSE POINT Hemibrains, cerebella and olfactory bulbs are stable at −80 °C for at least several months. To ensure reproducibility, keep all brains frozen for at least an overnight period (~16 h) at −80 °C.

Crude EV purification ●TIMING 5–6 h (2 to 4 hemibrains; if more tissues are processed, time can be longer)

▲ CRITICAL All the following points are intended as a protocol to isolate EVs from right murine hemibrains. Nonetheless, 300 mg human brain tissue or 2–3 murine cerebella can be used instead as a source of brain EVs, as indicated in the ‘Experimental Design’ section of the Introduction).

• 15Turn on a benchtop refrigerated centrifuge for conical tubes and a floor-type ultracentrifuge to give it time to reach and stabilize at 4 °C. The Type 45Ti and the SW40Ti rotors should be kept in a cold room or in a 4 °C refrigerator at least an overnight period (~16 h) before starting this procedure.
• 16Prepare a fresh papain solution, as indicated in the Reagent setup.
• 17Place the tubes containing 3.5 mL of papain solution (one tube for each hemibrain being processed) in a water bath at 37 °C for 15 min prior to the start of the isolation procedure.

  ▲ CRITICAL STEP The 15 minutes waiting time is critical to reach the right temperature but do not leave the enzyme for too long at 37 °C as it may lose activity over time.

• 18During the 15 min incubation period of the papain solution, prepare the papain stop solution as specified in Reagent setup. Quickly place the tube on ice. Prepare 6.5 mL solution per hemibrain.

  ! CAUTION PMSF is toxic.

• 19Delicately place either a frozen murine hemibrain, two cerebella, or human brain tissue sample in a Petri dish.

  ! CAUTION when human tissue samples are manipulated, always perform the procedures indicated following the local and national guidelines with approval from your institution.

  ▲ CRITICAL STEP Move the brain sample from the −80 °C freezer to dry ice, and from dry ice directly to the Petri dish to start the isolation. To minimize the natural degradation of the tissue, do not defrost the tissue on ice before the isolation. Moreover, the brain is sticky, and a thawed brain will attach to the side of the microcentrifuge tube.

  ▲ CRITICAL STEP To minimize variability between experiments, always isolate EVs from the same side of the murine brain (we usually isolate from the right hemibrain) to avoid left or right brain asymmetry and discrepancies⁴⁴. See the ‘Experimental Design’ section of the Introduction.

• 20With a transfer pipette, add 3 drops of the papain solution kept at 37 °C (see Step 17) on top of the tissue and coarsely chop it with a single edge razor blade. This procedure should result in brain pieces of approximatively 1–5 mm³ in order to increase the surface contact area of the tissue with the papain.

  ! CAUTION Handle blades with extra care to avoid cuts and scratches.

  ▲ CRITICAL STEP Do not mince or over-chop the tissue and do not use Dounce homogenizers to avoid damage to cells and contamination of the extracellular fluid with intracellular material.

• 21Using the transfer pipette, transfer the coarsely chopped tissue into the 15 mL conical tube containing the rest of the papain solution.
• 22Move the 15 mL conical tubes back into the 37 °C water bath for 15 min to complete the digestion. Gently turn upside down briefly the tubes every 5 min.

  ▲ CRITICAL STEP To preserve the sample do not vortex and do not shake or mix the tubes vehemently during papain digestion. Just flip the tube once every 5 min to guarantee homogeneity of the mixture.

• 23Stop the enzymatic activity by adding 6.5 mL of ice-cold papain stop solution (see Step 18). Gently pipette the solution up and down 20 times using a 10 mL disposable pipet to further loosen the tissue.

  ▲ CRITICAL STEP From this point forward, all the remaining steps must be performed at 4 °C on ice to thermically block papain activity for the rest of the procedure. There is no need to move the samples into a cold room. Samples can be kept in an ice bucket on the laboratory bench.

• 24Centrifuge at 300g for 10 min at 4 °C using a refrigerated benchtop centrifuge. In the meantime, load a 40 μm cell strainer onto the neck of a 50 mL conical tube.
• 25Pass the supernatant through the 40 μm mesh to eliminate cells, debris, and undigested material. The resulting filtrate should be homogenously turbid. Discard the pellet.

  ? TROUBLESHOOTING

• 26Plug a 26G needle to a 10 mL syringe and collect all the liquid in the 50 mL conical tube using this syringe. Flip the syringe, discard the needle, and replace it with a 0.2 μm SFCA filter.

  ! CAUTION Handle needles with care to avoid cuts and scratches.

  ▲ CRITICAL STEP This step is critical for the quality of the preparation, and the omission of this step causes contamination of the extracellular fluid with intracellular material (Fig. 2). Excessive force during the process would result in rupture of the EVs and/or contamination by intracellular material. Therefore, we recommend collecting the preparation from the tube with slow upward pressure of the plunger and the use of three standard SFCA filters per mouse hemibrain instead of one to avoid too much force during the filtration step. Due to differences in yield and content, use the same type of filter for all samples isolated for each project.

• 27Push about 3 mL solution through the plunger, recovering the filtered solution in a 50 mL conical tube. Unplug the filter and save it.
• 28Change the filter and repeat the operation twice, until all the solution (~9.5 mL) is filtered.

  ▲ CRITICAL STEP After the filtering process, the solution should be clear.

  ▲ CRITICAL STEP The three used filters will all retain liquid. To ensure reproducibility and the highest yield possible, do not discard the used filters at this step (the remaining liquid will be recovered through the ‘air washing steps’ in Steps 29–30).

• 29Press a full syringe of air into one of the three used filters and recover the liquid that comes out (approximatively 0.5 mL) into the 50 mL conical tube with the rest of the solution.
• 30Unplug the filter and repeat Step 29 twice with the other two filters. You can now discard the used filters.
• 31Centrifuge the combined resulting solution (Steps 26–29 + 29–30) at 2,000g for 10 min at 4 °C using a refrigerated benchtop centrifuge.
• 32Transfer the supernatant into a 70 mL ultracentrifugation polycarbonate bottle. Add ice-cold PBS to bring the total volume up to 50 mL.

  ▲ CRITICAL STEP Impurities in PBS and the presence of extra ions and metals may result in improper pelleting and buffering of the solution. We highly recommend the use of a commercially available PBS suitable for cell culture (without calcium and magnesium) for this step and all the following steps requiring PBS.

• 33Weigh the polycarbonate bottles with a precision scale and adjust the volume inside each tube by adding a few drops of PBS with a transfer pipette. At the end of this step, all bottles should have the same final weight (with a ±20 mg tolerance).

  ! CAUTION The improper use of an ultracentrifuge may be dangerous. All 70 mL ultracentrifugation polycarbonate bottles should be filled around ¾ of their volume. Minimal imbalances in the weight of the tubes/bottles may result in major failures; accordingly, the use of precision scales to get the most accurate possible estimation of the weights is highly recommended. The equalization of the bottle weights should include the O-rings, the black plastic caps, and the red aluminum caps.

• 34Centrifuge at 10,000g for 30 min at 4 °C. If using a Type 45Ti rotor, this corresponds to 11,000 rpm (k-factor: 2,218).

  ▲ CRITICAL STEP It is highly important to include bottle O-rings and not just rotor O-rings to preserve the samples and the centrifuge. The absence of bottle sealing O-rings may result in loss of the sample and contamination of the centrifuge vacuum pump.

• 35Carefully pipette the supernatant into another 70 mL ultracentrifugation polycarbonate bottle and discard the pellet.

  ▲ CRITICAL STEP Ensure not to disturb the pellet or accidentally transfer any part of the pellet with the supernatant.

• 36Repeat steps 32 to 33 and then centrifuge at 100,000g for 70 min at 4 °C. If using a Type 45Ti rotor, this corresponds to 36,000 rpm (k-factor: 207). The final pellet will contain the crude EVs preparation.

  ! CAUTION The improper use of an ultracentrifuge may be dangerous. See step 33 for details.

• 37Pour off the supernatant without disturbing the pellet.
• 38Wash the pellet by resuspending it in 50 mL of PBS kept at 4 °C. Repeat steps 32 to 33.
• 39Centrifuge again at 100,000g for 70 min at 4 °C and pour off the supernatant without disturbing the pellet.

  ! CAUTION The improper use of an ultracentrifuge may be dangerous. See step 33 for details.

• 40Invert the tube and collect the leftover supernatant with a micropipette. Delicately wipe the internal part of the tube with an appropriate tissue to dry all the remaining PBS without touching the pellet. Keep the crude EV pellet at 4 °C.

  ▲ CRITICAL STEP The EV pellet is firmly attached to the side of the tube and there is no risk of losing it when inverting the tube. Be sure to remove all the PBS to minimize volume variability and to avoid changing the concentration of the first layer of the following density gradient column.

  ■ PAUSE POINT Some researchers may not be interested in separating brain EV subpopulations or may not have the expertise to properly layer density gradient columns. Therefore, the crude unpurified EV pellet (hereafter ‘precolumn EVs’) of Step 40 can be resuspended in 500 μL of PBS and be frozen at −80 °C (stable for at least several months). For the separation of brain EV subpopulations, do not freeze the samples and proceed to Step 41. The freezing step may alter EV properties and density: do not use previously frozen precolumn EVs on a density gradient to separate EV subpopulations.

Separation of brain EV subpopulations through a low-resolution (sucrose) or a high-resolution (iodixanol) density step-gradient ●TIMING 6 h (but may vary depending on the number of samples) plus 16 h overnight centrifugation

▲ CRITICAL Preparation of good-quality density gradients requires high levels of expertise and training. Using a Pipet-Aid controller plugged with 2 mL serological pipets instead of a P1000 is preferred for the potential smoother movements of the liquids upon ejection.

• 41
  During the centrifugation step to obtain the crude EV pellet (Step 35), prepare either sucrose density gradient solutions or iodixanol density gradient solutions as indicated in the Reagent setup. Below, follow either option A for a sucrose gradient or option B for an iodixanol gradient. Option B is preferable under most circumstances to separate different subtypes of EVs – and it is the only option available to isolate mitovesicles. Nonetheless, option A is relatively less difficult to perform as sucrose solutions are easier to layer and is less time-consuming given that half of each column is prepared before the crude EV pellet (Step 35) is obtained. Additionally, while option A generates three EV fractions per brain sample, eight EV fractions are generated by option B. Therefore, option A may be preferred for specific applications, such as when more than four brain samples are isolated at the same time.

  1. Low-resolution (sucrose) density gradient

     1. During the centrifugation step to wash EVs (Step 39), carefully layer one on top of the other the following sucrose solutions in a 14 mL thin-wall Ultra-Clear tube: 2 mL of 2 M, 2 mL of 1.65 M, 2 mL of 1.3 M, 1 mL of 0.95 M sucrose in HEPES 20 mM. Leave the column on ice until the centrifugation is completed. Always keep the column at 4 °C.
        ? TROUBLESHOOTING
     2. Resuspend the EV pellet obtained in Step 40 by pipetting up and down 1 mL of 0.95 M sucrose solution.
        ▲ CRITICAL STEP The EV pellet is firmly attached to the side of the tube. Therefore, always ascertain that the EV pellet is fully detached and dissolved uniformly in the density medium. Do not leave undissolved EV clumps.
     3. Add the 0.95 M sucrose-equilibrated EVs on top of the column.
     4. Layer on top of the 0.95 M sucrose-equilibrated EVs 2 mL of 0.6 M and finally 2 mL of 0.25 M sucrose solutions.
        ? TROUBLESHOOTING
     5. Centrifuge at 200,000g overnight for 16 h at 4 °C in a swinging-bucket rotor. If using a SW40Ti rotor, this corresponds to 40,000 rpm (max speed, k-factor: 137).
        ! CAUTION The improper use of an ultracentrifuge may be dangerous. See step 33 for details.
        ▲ CRITICAL STEP In order to balance the columns before loading the rotor, weigh the tubes, set the heaviest to zero and adjust the other one (if needed) with one or two drops of the 0.25 M sucrose solution used for the last layer. Do not use a transfer pipette for this procedure as it will create a turbulent flux of the liquid and disturb your gradient: use 2 mL serological pipets plugged to a Pipet-Aid controller instead, releasing the liquid on the side of the column as slowly as possible. If an odd number of brains is analyzed and an extra balance tube needs to be prepared, use the same solutions used for the columns in the same proportions. Do not use PBS, water, or other kinds of solvents/solutions to fill the tube as they are lighter than sucrose solutions and will not reach the same weight.
        ▲ CRITICAL STEP When loading the centrifuge, move the tubes the least possible amount to avoid sharp motions. If compatible with your ultra-centrifuge machine, a low-braking setting is preferable at the end of the overnight (~16 h) spin.
     6. Using a P1000 pipette, carefully collect 1 mL of the gradient from the top of the column (fraction a).
        ▲ CRITICAL STEP All the fractions are collected from the surface of the column with a tip tilted, touching the side of the tube. Do not insert the tip deep into the column and do not inject air while removing each fraction.
     7. Transfer three fractions (fractions b, c and d) of 2 mL each into different 6.5 mL open-top, thick-wall polycarbonate tubes pre-chilled at 4 °C.
        ▲ CRITICAL STEP The only fractions that contain EVs are b, c, and d (corresponding to the interphases 0.25 M-0.6 M, 0.6 M-0.95 M, and 0.95 M-1.3 M, respectively, Table 1). The rest of the gradient (fractions a, e, f, and g) is largely EV-free (Fig. 3b) and can be discarded^12,29.
     8. Add 3.5 mL ice-cold PBS to each 6.5 mL open-top, thick-wall polycarbonate tube containing a column fraction and mix by pipetting up and down several times. Do not vortex.
        ▲ CRITICAL STEP Due to the density and viscosity of the medium, EVs within the high-density fractions, e.g., fraction d, are more difficult to pellet and need to be mixed thoroughly. If mixing is insufficient, no pellet will be detected after centrifugation. If this is the case, split fraction d in two tubes, add PBS to reach 11.00±0.02 g of weight, mix again and repeat the centrifugation. The addition of extra PBS will lower the density and viscosity of the solution and facilitate pelleting.
     9. Weigh the tubes and fill them with a few drops of PBS until they reach 11.00±0.02 g.
        ▲ CRITICAL STEP The use of precision scales to get the most accurate possible estimation of the weights is highly recommended (±20 mg tolerance). 11 g is the weight of the tube when it is filled to about ¾ of its capacity, which is the ideal volume to avoid spills and cracks.
     10. Centrifuge at 100,000g for 70 min at 4 °C. If using an MLA80 rotor, this corresponds to 46,000 rpm (k-factor: 57), if using a Type 70.1Ti rotor, the speed is 40,000 rpm (k-factor: 94).
         ! CAUTION The improper use of an ultracentrifuge may be dangerous. See Step 33 for details.
     11. Pour off the supernatant and eliminate the residual liquid from the tube as described in Step 40.
     12. Resuspend the EV fractions in 30 μL ice-cold PBS. Prepare 5 μL aliquots and save at −80 °C for future applications.
         ▲ CRITICAL STEP For best results, always use fresh, previously unfrozen aliquots of EVs in PBS for preparing EV lysates and for either transmission electron microscopy (TEM)¹² or cryo-EM. Although fresh samples are preferable, NTA can also be performed on frozen aliquots of EVs in PBS ⁵, following provider’s instructions. Either frozen or fresh aliquots of EVs in PBS can be used for estimating EV protein amount using BCA¹⁶, proteomics¹⁶, transcriptomics³, lipidomics¹⁸, and RNA analyses¹⁸.
         ■ PAUSE POINT EV aliquots in PBS at −80 °C are stable for at least several months.
  2. High-resolution (iodixanol) density gradient

     1. Resuspend the EV pellet obtained in Step 40 in 1.5 mL ice-cold 40% (vol/vol) iodixanol solution. Delicately move this solution onto the bottom of a 14 mL thin-wall Ultra-Clear tube.
        ▲ CRITICAL STEP The EV pellet is firmly attached to the side of the tube, so the resuspension may be difficult, especially in 40% iodixanol, a highly viscous solution. Always ascertain that the EV pellet is fully detached and dissolved uniformly in the dense medium. Do not leave undissolved EV clumps.
     2. Carefully layer 1.5 mL of decreasing concentrations of iodixanol on top of it (20% first, then 15%, 13%, 11%, 9%, 7%). Always keep the column at 4 °C.
        ? TROUBLESHOOTING
     3. Layer 2 mL of a 5% iodixanol solution on top.
     4. Centrifuge at 200,000g overnight for 16 h at 4 °C in a swinging-bucket rotor. If using an SW40Ti rotor, this corresponds to 40,000 rpm (max speed, k-factor: 137).
        ! CAUTION The improper use of an ultracentrifuge may be dangerous. See Step 33 for details.
        ▲ CRITICAL STEP In order to balance the columns before loading the rotor, weigh the tubes, set zero to the heaviest and adjust the other one (if needed) with one or two drops of the 5% iodixanol solution used for the last layer. Do not use a transfer pipette for this procedure as it will create a turbulent flux of the liquid and disturb your gradient: use 2 mL serological pipets plugged to a Pipet-Aid controller instead, releasing the liquid on the side of the column as slowly as possible. If an odd number of brains is analyzed and an extra balance tube needs to be prepared, use the same solutions used for the columns in the same proportions. Do not use PBS, water, or other kinds of solutions to fill it as they are lighter than iodixanol solutions and will not reach the same weight.
        ▲ CRITICAL STEP When loading the centrifuge, move the tubes as little as possible to avoid sharp motions. If compatible with your ultra-centrifuge machine, a low-braking setting is preferred at the end of the overnight (~16 h) spin.
     5. Collect 1.25 mL of the gradient from the top (fraction 1) into a 6.5 mL thick-wall polycarbonate tube pre-chilled at 4 °C containing 4 mL ice-cold PBS. Mix by pipetting up and down.
        ▲ CRITICAL STEP All the fractions are collected from the surface of the column with a tip tilted, touching the side of the tube. Do not insert the tip deep into the column and do not inject air while removing each fraction.
     6. Continue collecting 1.5 mL fractions corresponding to the different iodixanol interphases (fractions 2 to 8) into 6.5 mL thick-wall polycarbonate tubes containing 4 mL ice-cold PBS. Mix by pipetting up and down.
     7. Weigh the tubes and fill them with a few drops of PBS until they reach 11.00±0.02 g.
        ▲ CRITICAL STEP The use of precision scales to get the most accurate possible estimation of the weights is highly recommended (±20 mg tolerance). 11 g is the weight of the tube when it is filled to about ¾ of its capacity, which is the ideal volume to avoid spills and cracks.
     8. Split fraction 8 into two tubes and fill them with PBS until they reach 11.00±0.02 g.
        ▲ CRITICAL STEP Due to the density and viscosity of the medium, EVs in fraction 8 are more difficult to pellet and need to be split into two tubes. The addition of extra PBS lowers the density and viscosity of the solution and facilitates pelleting.
     9. Centrifuge at 100,000g for 70 min at 4 °C. If using a MLA80 rotor, this corresponds to 46,000 rpm (k-factor: 57), if using a Type 70.1Ti rotor, the speed is 40,000 rpm (k-factor: 94).
        ! CAUTION The improper use of an ultracentrifuge may be dangerous. See Step 33 for details.
     10. For fractions 1 to 7, pour off the supernatant and eliminate the residual liquid from the tube as described in Step 40. For the two tubes containing fraction 8, carefully eliminate the supernatants using a P1000 pipette (but leaving ~500 μL of supernatant in each tube), resuspend the two pellets and combine them in one single tube. Adjust the weight to 11 g with PBS and centrifuge again as indicated in Step 41B(ix). Pour off the Fraction 8 supernatant and eliminate the residual liquid from the tube as described in Step 40.
         ▲ CRITICAL STEP While the pellets of iodixanol fractions 1 to 7 are firmly attached to the side of the polycarbonate tube, the fraction 8 pellet (which shows a distinctive pale pink/yellow color) is only loosely attached and can be lost when the supernatant is poured off.
     11. Resuspend each EV fraction in 30 μL ice-cold PBS or 30 μL AAB, depending on downstream applications. Prepare 5 μL aliquots and save at −80 °C for future applications.
         ▲ CRITICAL STEP For best results, always use fresh, previously unfrozen aliquots of EVs in PBS for preparing EV lysates and for TEM¹² or cryo-EM. Although fresh samples are preferable, NTA can also be performed on frozen EVs⁵, following provider’s instructions. Either frozen or fresh aliquots of EVs in PBS can be used for estimating EV protein amount using BCA¹⁶, proteomics¹⁶, transcriptomics³. lipidomics¹⁸, and RNA analyses¹⁸. Resuspend EVs in AAB and use fresh samples for metabolic assays, including ATP assays.
         ■ PAUSE POINT EV aliquots in PBS at −80 °C are stable for at least several months.

EV characterization and analysis

• 42
  Here we report detailed procedures adapted for brain EV preparations for NTA, electron microscopy characterization, western blot analysis, and for measuring the mitovesicular kinetics of ATP. These assays have been used to successfully characterize brain derived EVs^{3,5,12–14,16,18,20,29}.

  1. NTA ●Timing 1 h for 1 hemibrain

     1. Filter 500 mL PBS with a 0.2 μm Vacuum Filter and Storage Bottle System to degas it.
        ▲ CRITICAL STEP This step increases the accuracy of the reading, as air bubbles naturally present in PBS scatter light and are read as particles by the NTA machine.
     2. Dilute fresh sucrose or iodixanol EV fractions 1 to 1,000 in degassed PBS.
        ■ PAUSE POINT These stocks can be used fresh or can be frozen at −80 °C for future analyses. The use of fresh samples is preferred to avoid possible aggregation of EVs.
     3. Further dilute the fractions in degassed ddH₂0 to a final volume of 1 mL just prior to reading.
        ▲ CRITICAL STEP Dilution of EV stocks in ddH₂0 should be in the range of 1 to 10–100, in order to achieve a final dilution of the original stock of 1 to 10,000–100,000. The specific dilution for each fraction depends on the yield of the EVs recovered and needs to be empirically determined. The appropriate concentration of particles is the one that falls into the ideal reading range of the machine (100–400 particles per field of view when using a ZetaView machine).▲ CRITICAL STEP The use of ddH₂0 instead of PBS at this step is preferred as PBS may contain phosphate crystals that scatter light and that can be read as particles. Moreover, PBS cannot be used if the ζ-potential is estimated.
     4. Dilute the calibration beads 1 to 250,000 in degassed ddH₂0 and perform the auto-alignment and the auto-focus of the machine according to the manufacturer’s instructions.
     5. Wash the beads out of the machine using degassed ddH₂0 and set the appropriate parameters for reading EVs.
        ▲ CRITICAL STEP Take particular care to avoid injecting air bubbles while flushing the machine.
     6. Using a 1mL tuberculin syringe without the needle, inject the diluted fraction into the analyzer.
     7. Acquire the video and perform morphometric measurements.
        ▲ CRITICAL STEP Before the acquisition, always check that the particle drift is not higher than 30 nm/s. If it is too high, wait until the EVs slow down and the particle drift drops to normal values. If the drift is >30 nm/s in a specific channel, a ‘Max drift’ error is shown for that channel at the end of the acquisition.
        ▲ CRITICAL STEP Before the reading, visually inspect all channels for the absence of air bubbles and impurities to increase accuracy. Some impurities are firmly attached to the side of the chamber and will not detach even after repeated washings with ddH₂0. In this case, unplug the reading chamber and perform a deep cleaning with a detergent and EtOH.
        ? TROUBLESHOOTING
     8. Wash the chamber with degassed ddH₂0 and load the next fraction, repeating the same procedure as in Steps 41A(vi – vii).
     9. After the last fraction is analyzed, perform a wet or a dry cleaning of the machine.
        ▲ CRITICAL STEP A wet cleaning is preferred if the machine will be reused in 1–2 days, a dry cleaning should be performed for longer periods between use.
  2. TEM ●TIMING 1 h per sample including preparation and acquisition

     1. Add an appropriate amount of EVs (1–5 μL) to 10 μL of 0.2 M sodium cacodylate buffer and 2 μL of a 20% (wt/vol) PFA aqueous solution. Bring to a final volume of 20 μL with ddH₂0 (final concentration: 100 mM sodium cacodylate, 2% (wt/vol) PFA).
        ! CAUTION sodium cacodylate and PFA are toxic.
        ▲ CRITICAL STEP The amount of EVs to be fixed and stained depends on the number of EVs in each fraction and the yield of the EVs recovered. Generally, a range of 1 μL (sucrose fractions and iodixanol fractions 3 and 4) to 5 μL (iodixanol fraction 8) of EVs is enough for an optimal density of EVs per field of view (Extended Data Fig. 1).
        ■ PAUSE POINT Fixed EVs can be stored at 4 °C for several days
     2. Place the Formvar-covered mesh grids in the TEM grid holder.
     3. Glow discharge the grids at 20 mA for 30 sec (pressure of 0.39 mbar).
        ▲ CRITICAL STEP Standalone automated systems, for instance, the PELCO easiGlow glow discharge cleaning system, are preferred for reproducibility and minimization of variability between the samples.
     4. Use self-clamping tweezers to hold the grid so the Formvar-coated side is pointing up. Gently apply 3 μL of fixed EVs to the discharged grids and leave for 20 sec so that they can be adsorbed to the grid.
     5. Wick off the solution with a Whatman paper and add 1% (wt/vol) uranyl acetate to negatively stain EVs. Immediately remove the staining solution with Whatman paper.
        ! CAUTION Uranyl acetate is toxic.
        ▲ CRITICAL STEP Do not overstain EVs to avoid black clumps during acquisition.
     6. Repeat Step 41B(v) three times, then incubate the grids in 1% (wt/vol) uranyl acetate for 5 min.
     7. Wick off the solution with Whatman paper and allow the grid to air dry.
     8. Load the sample into a single tilt microscope holder.
     9. Image the grids using the TEM. The Talos L120C TEM uses a LaB6 filament and operates at 120 kV.
  3. cryo-EM ●TIMING Variable (depending on the number of samples, roughly 1 h per sample including preparation and acquisition)

     1. Dilute EVs 1 to 10 in PBS, except for mitovesicles in iodixanol fraction 8, which should be diluted 1 to 2.
        ▲ CRITICAL STEP In wild-type mice and individuals devoid of pathology, the number of mitovesicles in the brain is very low when compared to other fractions, roughly 1% of the total EV population (Fig. 4d)³. A 1 to 10 dilution for iodixanol fraction 8 is not recommended as EVs may be too sparse and hard to find using the microscope.
     2. Glow discharge a R 2/2 Copper 200 mesh Quantifoil grid.
        ▲ CRITICAL STEP Standalone automated systems, for instance, the PELCO easiGlow glow discharge cleaning system, are preferred for reproducibility and minimization of variability between the samples.
     3. Use Vitrobot tweezers to pick up the grid and place in Vitrobot.
     4. Set the Vitrobot Mark IV parameters as follows: temperature 4 °C, humidity 100%, blot time 3 sec, blot force 0, wait time 10 sec and blot total 1.
     5. Cool Vitrobot cup with liquid nitrogen and add liquefying ethane to the central cup until it is full. Add ethane only when nitrogen bubbling has stopped.
        ▲ CRITICAL STEP Freezing can begin only after a thin layer of solid ethane is observed on the inner walls of the cup.
     6. Gently apply 3 μL of diluted EVs to the discharged grid.
     7. Initiate the Vitrobot so that blotting the sample with filter paper happens automatically and is subsequently plunged into the liquid ethane and frozen.
        ▲ CRITICAL STEP Removal of the Vitrobot tweezers from the dovetail on the plunger can be cumbersome, so take care when handling the sample. For ease of use, we recommend the alternative Mitegen Nanosoft cryo-EM tweezers that may help with sample transfer.
     8. Carefully transfer sample to storage box in liquid nitrogen.
     9. Transfer grid to the Gatan cryo-holder and insert into TEM.
     10. After sample insertion wait for 30 mins for sample to stabilize and avoid drift. Then in ‘low dose’ mode at low magnification find region of sample with good ice vitrification.
     11. Acquire the image at desired magnification (we usually set a magnification of 45000x).
         ▲ CRITICAL STEP Always use the ‘low dose’ mode configuration to minimize sample damage in the region of interest.
         ? TROUBLESHOOTING
  4. SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and western blotting ●TIMING Variable (24–40 h depending on the primary antibody incubation time)

     1. Prepare all the solutions needed for the SDS-PAGE, including the lysis buffer (RIPA) 2X, the 6X SB, and the running buffer.
     2. Add protease inhibitors to a proper amount of RIPA 2X for a final concentration of 2X inhibitors (concentration of the inhibitors in the RIPA 2X: 10 μg/mL leupeptin, 10 μg/mL antipain dihydrochloride, 10 μg/mL pepstatin A, 2 mM PMSF, 2 μM E64).
        ▲ CRITICAL STEP The amount of RIPA 2X to be mixed with protease inhibitors is adjusted according to the number of samples to be lysed. Consider approximatively 40 μL of RIPA 2X for each EV fraction and 550 μL of RIPA 2X for each precolumn EV sample used.
        ▲ CRITICAL STEP PMSF 100X should be diluted 1 to 50, LAP 1000X and E64 1000X should be diluted 1 to 500.
        ▲ CRITICAL STEP PMSF is very unstable in aqueous solutions, including in RIPA 2X. Add protease inhibitors just prior to use and keep the solution at 4 °C at all times.
     3. Add one volume of RIPA 2X to one volume of EVs in PBS.
        ▲ CRITICAL STEP Lysed EVs cannot be analyzed by other techniques, such as NTA, TEM or cryo-EM, and cannot be used for functional studies.
        ▲ CRITICAL STEP Although frozen aliquots of EVs in PBS are also applicable for western blotting, freshly isolated EVs guarantee the best results and the highest signal-to-noise ratio.
     4. Add 6X SB to the lysed EV sample in a 1 to 5 ratio.
        ▲ CRITICAL STEP 6X SB is extremely viscous at room temperature. Placing the 6X SB in a dry bath incubator at 95 °C for ~30 sec will make the solution more fluid and easier to pipette.
     5. Incubate samples at 95 °C for 5 min in a dry bath incubator and spin down for a few seconds using a benchtop centrifuge to collect all the liquid.
        ■ PAUSE POINT samples can now be directly loaded on a polyacrylamide gel or frozen at −20 °C for future use. Lysates are stable at −20 °C for several months. To ensure proper unfolding and running of the proteins during SDS-PAGE, re-incubate at 95 °C for 5 min before loading every time samples are thawed.
     6. Insert a precast polyacrylamide gel into a running cell for vertical electrophoresis and fill the chamber with running buffer, according to manufacturer’s instructions.
        ▲ CRITICAL STEP Make sure that the gel is not expired, that the SDS within the buffer is not precipitated and that the wells are fully covered with running buffer.
     7. Load the protein ladder(s) and no more than 10 μL of EV lysate per sample.
        ▲ CRITICAL STEP Given their small dimension, EVs, and particularly sucrose fraction b and iodixanol fractions 1, 2, 3, have a high lipid-to-protein ratio and are highly enriched in cholesterol³. Consequently, when an EV lysate is subjected to SDS-PAGE, lipids of small molecular mass (cholesterol is lighter than 0.4 KDa) run packed all together at the bromophenol blue frontline. This causes a ‘wavy’ run of the bottom part of the gel and a perturbation of the electrophoretic run of the small molecular weight polypeptides, visible with a total protein staining (such as SyproRuby or Ponceau S) of the blotted membrane (Fig. 3a). Accordingly, loading more than 10 μL is not recommended, unless a previous delipidation step is performed²⁹.
     8. Run for 2 h at 120 V.
        ! CAUTION The voltage used can be harmful under certain circumstances and the improper use of the power supply can be dangerous for human health.
     9. During the 2 h run, prepare transfer buffer and keep it at 4 °C.
        ▲ CRITICAL STEP The transfer buffer solution is exothermic upon addition of MetOH and will heat up. Refrigerate at 4 °C for at least 1 h before use.
     10. At the end of the run, incubate the PVDF membrane in MetOH for 45 s, wash it in ddH₂0 for 2 min and keep it in transfer buffer until further use.
         ▲ CRITICAL STEP After the incubation in MetOH, the membrane will float over water; rock the box containing the membrane to sink it. After the 2 min washing step, do not move the membrane to the transfer buffer if it is still floating. Change the water and wash the membrane until it sinks to the bottom of the box.
     11. Remove the gel from the cassette and equilibrate it in transfer buffer.
     12. Soak the Whatman papers and the sponges in transfer buffer and assemble the transfer sandwich.
         ▲ CRITICAL STEP Proteins bound to SDS are negatively charged and will move to the positive electrode. Ascertain that the gel faces the negative electrode (usually colored in black) and the membrane the anode (usually colored in red) before starting the transfer.
     13. Transfer overnight (~16 h) at 4 °C at 100 mA.
         ! CAUTION The current used can be harmful under certain circumstances and the improper use of the power supply can be dangerous for human health.
         ■ PAUSE POINT After transfer, membranes can be dried and blotted upon reactivation in MetOH - repeating step 42D(ix).
     14. Prepare all the solutions needed for western blotting (TBST, blocking buffer, primary antibody buffer) and an AM solution (7% (vol/vol) acetic acid, 10% (vol/vol) MetOH in ddH₂0) for SyproRuby staining.
     15. Remove the membrane from the transfer sandwich and wash away the residual transfer buffer with ddH₂0.
         ▲ CRITICAL STEP From this moment on, all steps are performed at room temperature with gentle agitation of the membrane on a bench rocker or on an orbital shaker, unless otherwise specified. Use a mild rocking setting, such as 50 rpm for the orbital shaker.
     16. Incubate the membrane for 15 min in AM solution.
     17. Wash the membrane with 4 changes of ddH₂0, 5 min each washing.
     18. Stain the membrane with the SyproRuby blot stain reagent for 15 min and wash away the excess dye with 3 washes of ddH₂0, 1 min each washing.
     19. Visualize and acquire an image.
         ▲ CRITICAL STEP The SyproRuby dye has two excitation maxima at ~280 nm and ~450 nm, and an emission maximum at ~620 nm. Accordingly, it can be visualized using either an UV transilluminator, a blue-light transilluminator or a laser scanner (typically used with lasers that emit from 450 to 540 nm). It can also be visualized with an integrated fluorescence and chemiluminescence acquisition system, such as the iBright 1500 machine. The acquisition with a CCD camera (and the appropriate filters) enhances the sensitivity when compared to direct observation by eye. The stain is highly photostable and does not photobleach during longer exposure times. Clean the transilluminator after every use to avoid accumulation of residual dye on the glass that can increase the background fluorescence.
         ▲ CRITICAL STEP It is crucial to stain proteins with SyproRuby and acquire the signal before the immunoblotting, as more than 90% of the stain is washed off the blot during the blocking step (when a PVDF membrane is used). No specific destaining steps are required.
     20. Block the membrane in the appropriate blocking buffer for 1 h at room temperature.
         ▲ CRITICAL STEP The blocking agents, including skim milk, BSA, and gelatin vary depending on the primary antibody used and should be chosen according to the signal-to-noise ratio and blotting performance.
         ■ PAUSE POINT If necessary, blots can stay in blocking buffer for several days at 4 °C before starting the primary antibody incubation.
     21. Wash the membrane in TBST for 10 min.
     22. Dilute the primary antibody and incubate either for 1 h at room temperature or overnight (~16 h) at 4 °C.
         ! CAUTION NaN₃ is toxic.
         ■ PAUSE POINT If necessary, blots can rock in primary antibody for several days at 4 °C. This step will slightly increase the signal when compared to an overnight incubation.
     23. Wash the membrane in TBST 3 times for 10 min.
     24. Incubate for 1 h at room temperature with the correct HRP-conjugated secondary antibody, previously diluted in blocking buffer.
         ▲ CRITICAL STEP The dilution of the secondary antibody varies and is usually indicated in the datasheet.
     25. Wash the membrane in TBST 3 times for 10 min.
     26. Mix 1:1 the ECL luminol solution with the peroxide buffer and incubate the membrane with this solution for 5 min at room temperature.
         ? TROUBLESHOOTING
     27. Expose and acquire bands using a chemiluminescence imaging system.
         ▲ CRITICAL STEP Both digital systems, including CCD camera-based imaging systems (such as the iBright 1500), and analog systems, including X-ray autoradiography films, are suitable for the acquisition step. Nevertheless, they have different advantages and disadvantages. X-ray films do not occupy any equipment during exposure, are in direct contact with the blots and can be left in the cassette for several minutes or even hours, if necessary. As such, analog systems are more sensitive than digital acquisitions and are usually preferred to detect low-abundance proteins and weak protein bands. On the other hand, the higher sensitivity of X-ray films implies that they are much more easily saturated when compared to CCD camera-based acquisitions. Accordingly, films have a more limited dynamic range and linearity when compared to digital acquisitions.
         ■ PAUSE POINT If necessary, after the acquisition, blots can be either incubated in a new primary antibody (repeating the same process starting from Step 42D(xxii)), or dried and reutilized in the future. Dried membranes are stable at room temperature for months.
  5. ATP assay and mitovesicle kinetics ●TIMING 2 h (for ~20 wells)
     ▲ CRITICAL Consider using 6 wells per sample, 2 wells for the blanks and 10 wells for the ATP standard curve.

     1. Prepare the AAB, the 1000X mitochondrial toxin stock solutions, and the ATP Master Mix.
        ! CAUTION DTT and the firefly luciferase can be toxic. MgAc₂ is flammable.
        ▲ CRITICAL STEP ADP is unstable at room temperature. Always prepare the ATP Master Mix fresh just prior to usage and keep it at 4 °C.
     2. Dilute the 5 mM ATP stock solution (provided as Component D with the ATP determination kit) in AAB to get five ATP standards: 10 nM, 50 nM, 100 nM, 500 nM, and 1 μM. The total volume of standards to be prepared depends on the frequency this assay is performed - one assay requires 20 μL of each standard.
        ▲ CRITICAL STEP ATP standards should not be left at room temperature. Prepare 25 μL aliquots and freeze at −20 °C for long term storage.
     3. Split the ATP Master Mix into two conical tubes. Add either FCCP or OA (a mixture of the two electron transport chain inhibitors Oligo and Anti-A, see Reagent setup) to a final 1X concentration to one of the tubes. Add an equivalent amount of DMSO to the other conical tube.
     4. Resuspend the mitovesicles (iodixanol fraction 8 EVs) in 30 μL of AAB.
        ▲ CRITICAL STEP AAB is crucial for the correct reading of mitovesicle kinetics. It contains a source of PO₄³⁻ and Mg²⁺ which are essential cofactors of the ATP synthase. Do not use EVs resuspended in PBS for this assay.
        ▲ CRITICAL STEP Always perform this assay on fresh EV preparations, do not use frozen aliquots.
     5. Measure the protein concentration of the samples to be tested using a high-sensitive protocol, such as through high-sensitive BCA.
     6. Adjust the volume of mitovesicles to a final concentration of 0.2 μg/μL in AAB. 10 μL of mitovesicles will be used per well during the assay.
        ▲ CRITICAL STEP The assay has an optimal reading range when ~2 μg of mitovesicles are used in each well. Especially in samples from young mice, it is possible that the starting concentration is lower than 0.2 μg/μL. If this is the case, pellet the mitovesicles again, resuspend in a volume of AAB lower than 30 μL and repeat the procedure.
     7. Load the 96-well white microplate either with 200 μL of the ATP Master Mix-DMSO per well or 200 μL of the ATP Master Mix-OA or 200 μL of the ATP Master Mix-FCCP for the negative controls.
        ▲ CRITICAL STEP To minimize the amount of sample used, this protocol is designed to have experimental duplicates per sample and a single reading for the negative controls (+FCCP or +OA). More wells can be loaded for experimental triplicates per sample. However, this will typically require that the whole amount of mitovesicles recovered from a murine hemibrain is used for this assay.
     8. Add 10 μL of AAB to the wells designed for the blanks, 10 μL of each ATP standard in duplicate to their assigned wells and 10 μL of mitovesicles in each dedicated well, both the ATP Master Mix-DMSO wells and the ATP Master Mix-OA/ATP Master Mix-FCCP wells.
     9. Insert into a plate reader and read every minute for 10 min at 28 °C.
        ▲ CRITICAL STEP At this temperature, oxidative phosphorylation is inhibited but the firefly luciferase activity is in its ideal range. Accordingly, the reading should be stable over the 10 min at this step and will represent the basal ATP levels found in mitovesicles.
     10. Increase the temperature to 37 °C and read every minute for 20 min.
         ▲ CRITICAL STEP The reading should increase in a sigmoid fashion by at least a 2–3-fold for mitovesicles incubated with the ATP Master Mix-DMSO and remain stable for all the other wells (Fig. 5).
     11. Export the data in Excel format. For the sample analysis, subtract the background luminescence found in the blanks, generate a standard curve with the ATP standards and calculate the concentration of ATP over time for each sample from the standard curve.

Fig. 4 |. NTA analyses of sucrose and iodixanol-based EV fractions.

EVs isolated from adult murine right hemibrains were separated by either a sucrose-based step gradient (a) or by an iodixanol step gradient (c) and subjected to NTA. The frequency of the distribution is normalized to the mode of each group and the curves are obtained using a four-point moving average. Fractions 1 to 3 (LDFs) and fractions 4 to 6 (IDFs) of the iodixanol step gradient are combined. The number of EVs, as estimated by NTA, for the sucrose fractions (b) and for the iodixanol fractions (d) are normalized to hemibrain weight. Data are expressed as billions of particles per mg of brain tissue.

Statistical test in (b) and (d): one-way ANOVA with Tukey’s multiple comparisons test. Number of independent isolations: n = 6 for all the experiments. Data shown in (b) and (d): mean ± S.E.M. (standard error of the mean). ** P < 0.01, *** P < 10⁻³, **** P < 10⁻⁴.

All animal procedures were performed following the National Institutes of Health guidelines with approval from the Institutional Animal Care and Use Committee at the Nathan S. Kline Institute for Psychiatric Research.

Fig. 5 |. Mitovesicles kept at 37 °C produce ATP through the electron transport chain.

Representative results obtained when EVs are isolated from a 12-month-old male mouse brain in iodixanol fraction 8 and treated as indicated in Step 42E. After 10 min at 37 °C, the amount of ATP detected typically doubles, whereas it remains low if the same EVs are treated with the electron transport chain inhibitors Oligo and Anti-A (+OA). The line shown is a sigmoidal interpolation, while the dots correspond to the raw data.

TIMING

Steps 1–2, preparation of the operating table, tubes, etc., and anesthesia: 15 min

Steps 3–14, murine brain dissection and partition: 4 min per brain

Steps 15–23, enzymatic digestion of the tissue: 40 min (can be more if several hemibrains are isolated at the same time)

Steps 24–31, filtration and low-speed centrifugation steps: 35 min (can be more if several hemibrains are isolated at the same time)

Steps 32–40, ultracentrifugation steps and preparation of the crude EV pellet: 3 h 30 min (can be more if several hemibrains are isolated at the same time)

Step 41, separation of brain EV subpopulations:

• Step 41A(i-xii), sucrose gradient: 18–19 h (can be more if several hemibrains are isolated at the same time)

• Step 41B(i-xii), iodixanol gradient: 20–21 h (can be more if several hemibrains are isolated at the same time)

Step 42, characterization of brain EVs:

• Step 42A(i-ix), NTA: ~1 h per hemibrain (including preparation of the stocks; can be more if several hemibrains are isolated at the same time)

• Step 42B(i-ix), TEM: ~1 h per sample

• Step 42C(i-xi), cryo-EM: ~1 h (can be more if several hemibrains are isolated at the same time)

• Step 42D(i-xxvii), western blot analysis: variable (depending on the primary antibody incubation time; as a general rule, not less than 24 h)

• Step 42E(i-xi), mitovesicle kinetics: ~2 h (including 30 min for the BCA assay)

TROUBLESHOOTING

Troubleshooting advice can be found in Table 3.

Table 3 |.

Troubleshooting table

Step	Problem	Possible reason	Solution
8	The brain is torn when removed from skull	The olfactory bulbs have not been released properly from the olfactory cavity and they pull out the rest of the brain when the operator tries to separate it from the skull	Carefully make two extra cuts at both sides of the olfactory cavity and one extra cut in front of it, just above the nose. Eliminate the front part of the skull and expose the bulbs
25	The preparation does not flow through the 40 μm mesh	There are too many tissue pieces in the preparation possibly because part of the 300g pellet was mistakenly picked up	Tap the conical tube from the side without touching the solution to help the air flow. If this does not help, carefully pipet up and down on the top of the mesh to move the tissue pieces clogging it
41A(I, iv) and 41B(ii)	Either the sucrose or the iodixanol solutions mix while preparing the column, and the layers are not formed	The solutions were pipetted too strongly one on top of the other or an automated density gradient maker machine was used	Avoid using gradient machines. Iodixanol densities are too similar and an automated machine may not be sensitive enough. Very gently tilt the column at ~30° and release the liquid close to the surface of the previous layer while facing the side of the column, not the liquid, so that the release is more gentle. Pipette very slowly, the stream while releasing the liquid should be constant but extremely mild. Never stop while releasing the liquid as an interruption and a reprise of the release will cause a turbulent flow that mixes the layers
42A(vii)	Blurry/pixelated images are visualized in one of the channels or too many particles are detected	Air bubbles or impurities were introduced into the NTA machine	Flush with a full 10 mL syringe of water to wash away impurities. If this is not enough, take 5 mL of water with a 10 mL eccentric tip syringe and inject the whole volume into the machine. The syringe should inject air first, and water later on. This will cause air washing before the water flow restarts
42C(xi)	Spherical, amorphous, or polygonal (such as hexagonal, or cubic) black structures show up in cryo-EM photos	Crystals are formed during the freezing or the acquisition step (Extended Data Fig. 2)	Use only high purity reagents (PBS or liquid ethane) that are free of contaminants. The use of a degassed commercial PBS suitable for cell culture is strongly recommended. Moisture, gaseous contaminants, and air particulates can also be sources of crystal deposition. Transfer your frozen sample as quickly as possible and freeze your sample in a low humidity room. During acquisition, do not focus the beam for too long on the region of interest as beam damage can cause crystal formation before sample loss
42D(xxvi)	After exposure, bands are not even	ECL did not cover the membrane uniformly	Ensure that enough ECL is made to fully cover the membrane (1 mL of ECL is enough for an area of no more than 17 cm2). Try to keep the membrane as flat as possible during the development step. Place the membrane on a smooth surface. The use of a few drops of TBST beneath the membrane may help it to adhere to the surface without air bubbles.

ANTICIPATED RESULTS

NTA characterization of sucrose and iodixanol EV fractions

We have shown elsewhere that our novel iodixanol-step gradient is able to separate EVs with different hydrodynamic diameters, as estimated by NTA^3,20. To compare sucrose and iodixanol fractions, we isolated EVs from C57BL/6J mouse brains and separated them either with a sucrose gradient or with an iodixanol gradient. The sucrose factions b, c, and d that contain EVs are similar in size and their peaks are particularly round-shaped, suggesting a mixture of EVs of different sizes is present within each fraction (Fig. 4a). No clear difference can usually be detected in the frequency of small and large EVs between fractions b, c, and d, although NTA reveals a small trend for fraction c to show a higher frequency of small EVs, consistent with its enrichment in exosomes. The number of EVs is typically not different between fractions b and c but is significantly lower in fraction d (Fig. 4b).

Larger differences in EV sizes are observed in the fractions obtained by the iodixanol-step gradient. For the sake of simplicity, fractions 1 to 3 (low-density fractions, or LDFs) and fractions 4 to 6 (intermediate-density fractions, or IDFs) are shown combined in Fig. 4c to allow a direct comparison to the respective density ranges of the sucrose gradient (see Table 1 and 2). The frequency of smaller EVs (<100 nm) is the highest in IDFs when compared to other fractions and the peak is at ~110 nm, consistent with the size of exosomes. Small microvesicles, that are found in LDFs, are larger than the EVs found in IDFs, given that the frequency of larger EVs (>150 nm) is the highest in these fractions and the peak is at ~150 nm. Fraction 8 EVs contains mitovesicles and shows a peak at ~130 nm (Fig. 4c). The iodixanol fractions that contain the highest number of EVs are fractions 3 and 4, while the fraction with the lowest number of EVs is fraction 8, which accounts for nearly 1% of the total EV population (Fig. 4d). Although this number is probably underestimated, considering some mitovesicle proteins could be detected in fraction 7 (Fig. 4c), these data suggest that the number of mitovesicles in brains devoid of pathology is particularly low when compared to microvesicles and exosomes, and is consistent with a model in which mitovesicle secretion is enhanced during the course of a disease by mitochondrial stress and accumulation of mitochondrial reactive oxygen species³.

Cryo-EM overperforms negative stain TEM in EV imaging

EVs are generally visualized by negative staining using standard TEM imaging, a procedure that is well-established, quick, relatively cheap, and easy to perform⁵⁰. However, EVs prepared for TEM are paraformaldehyde-fixed, stained with high molecular weight dyes (mainly uranium-based, such as uranyl acetate) and dehydrated during sample preparation (see Step 42B). These procedures cause artifactual changes in the endogenous morphological properties of EVs as they lose their native electron-density and acquire a cup-shaped morphology (Extended Data Fig. 1). By contrast, when visualizing EVs under cryo-EM, samples are preserved in their hydrated state, are not fixed, and are not stained (see Step 42C), guaranteeing a spherical appearance and superior image quality⁵⁰ (Fig. 2b). Thus, cryo-EM allows the accurate measurement of EV diameter and the classification of EVs according to their morphological features and intrinsic electron-density³, a goal impossible to achieve if EVs are fixed and stained prior to acquisition of their image.

Optimization of western blot analysis of brain EVs

As suggested in Step 42D(vii), the number of purified EVs loaded for western blot analysis is crucial for good quality blots (Fig. 3a). Purified EVs are enriched in lipids that have low molecular weight when compared to proteins (usually less than 1 KDa), including cholesterol that constitutes approximatively 40% of all the lipids found in microvesicles and exosomes³. The low molecular weight lipids tend to run packed together at the frontline of the electrophoresis gel. Consequently, when the amount of EVs is too high, the normal electrophoretic run of low molecular weight proteins is disrupted (Fig. 3a, left panel). This causes the appearance of an amino acid-free, lipid-rich, black spot at the bottom of the membrane upon application of fluorescent total protein stain, such as SyproRuby (Fig. 3a, arrowheads). This effect is more prominent in low-density fractions, due to the presence of EVs with a higher lipid over protein ratio³. To avoid or to limit this outcome, a lower number of EVs should be loaded (Fig. 3a, right panel) or a delipidation step should be considered, as we have shown elsewhere²⁹.

Western blot characterization of sucrose and iodixanol EV fractions

We described above two discontinuous step gradients to fractionate brain EV subpopulations. To compare the resolution power of iodixanol and sucrose, we isolated murine brain EVs and separated them in parallel either with the procedure described in Step 41A or B. Iodixanol gradients generated 8 EV-containing fractions, while sucrose-based gradients generated 3 EV-containing fractions. After using a sucrose-step gradient, Annexin A2, a marker of small microvesicles, is mainly found in fraction b and c, while exosomes are found mainly in fractions c and d, as indicated by the exosome markers Cd63 and Alix. The fraction with the highest density, fraction d, is particularly promiscuous as it contains a residue of small microvesicles, exosomes, and mitovesicles, as revealed by the mitovesicle marker PDHE1-α (Fig. 3b). By contrast, in EV fractions separated by the iodixanol column, microvesicles are enriched in fractions 1 to 3, exosome markers are mainly found in fractions 4, 5, and 6 (with a typical peak in fraction 5) and mitovesicles are resolved in fractions 7 and 8. In particular, fraction 8 contains no residual exosomes or small microvesicles but contains pure mitovesicles (Fig. 3c), an unachievable feat with sucrose-based columns. In addition, we recently published a thorough characterization of small brain EVs separated by the iodixanol gradient described here, reporting inclusion and exclusion protein markers, cryo-EM morphometry, lipid and mtDNA/mtRNA compositions³.

Characterization of brain mitovesicle kinetics

Mitovesicles are metabolically competent EVs, producing ATP in vitro, as we have previously demonstrated³. This production of ATP is fast and temperature-sensitive. When incubated at 37 °C with the proper substrates (see Step 42E), brain mitovesicles increase the amount of ATP within few minutes in a sigmoid fashion (Fig. 5), allowing the study of their kinetics with a fast and a relatively simple assay.

Conclusions

A protocol for the purification of brain EVs is described in detail. This protocol enables the separation of small plasma membrane-derived microvesicles, endosome-derived exosomes, and mitochondria-derived mitovesicles as well as separating subtypes of microvesicles and exosomes. Therefore, this procedure provides a valuable resource for the scientific community to investigate the variable nature of these different classes of bioactive EVs in a physiological context as well as in neurodevelopmental and neurodegenerative diseases.

Extended Data

Extended Data Fig. 1. Purified brain EVs fixed with PFA and stained with uranyl acetate show a distinctive cup shape morphology.

Representative photomicrograph of sucrose fraction c brain EVs isolated from a 12-month-old female mouse and visualized after negative stain. Note the cup shape of fixed EVs and the absence of contaminating material. Scale bar: 200 nm. All animal procedures were performed following the National Institutes of Health guidelines with approval from the Institutional Animal Care and Use Committee at the Nathan S. Kline Institute for Psychiatric Research.

Extended Data Fig. 2. Representative cryo-EM photomicrograph of contaminating crystals during acquisition of mitovesicles (iodixanol fraction 8 EVs).

Crystals are visualized as electron-dense bodies that are either amorphous (orange arrowheads) or polygonal, for instance, hexagonal or cubic (red arrowheads). The white arrow indicates a typical mitovesicle, which is characterized by the presence of a double membrane. The amorphous and polygonal dark structures in this case are the same contaminant, caused by moisture from the air that has frozen during the freezing process. Scale bar: 200 nm. All animal procedures were performed following the National Institutes of Health guidelines with approval from the Institutional Animal Care and Use Committee at the Nathan S. Kline Institute for Psychiatric Research.

DATA AVAILABILITY

All data needed to evaluate the conclusions of the paper are present in the paper. No datasets or custom code were generated in this study. All single data points are reported in the respective graphs, when possible (Fig. 4 and 5) and as Excel Source Data files. Raw, uncropped blots for Fig. 2 and 3 are provided as pdf Source Data files. Additional data related to this paper may be requested from the authors.