Received: 15 May 2023

- -
Revised: 7 December 2023

https://doi.org/10.1016/j.jtha.2024.01.010

ORIGINAL ARTICLE
Accepted: 2 January 2024

Density-based lipoprotein depletion improves extracellular

vesicle isolation and functional analysis

Laura Botelho Merij1,2 | Luana Rocha da Silva3 | Lohanna Palhinha4 |

Milena Tavares Gomes1,2 | Paula Ribeiro Braga Dib1,2 | Remy Martins-Gonçalves4 |
Kemily Toledo-Quiroga3 | Marcus Antônio Raposo-Nunes4 |
Fernanda Brandi Andrade1,2 | Sharon de Toledo Martins5 |
Ana Lúcia Rosa Nascimento6 | Vinicius Novaes Rocha7 | Lysangela Ronalte Alves5 |
Patrícia T. Bozza5 | Monique Ramos de Oliveira Trugilho4,8 | Eugenio D. Hottz1,2

1
Laboratory of Immunothrombosis,
Department of Biochemistry, Institute of Abstract
Biological Sciences, Federal University of
Background: Blood plasma is the main source of extracellular vesicles (EVs) in clinical
Juiz de Fora, Juiz de Fora, Minas Gerais,
Brazil studies aiming to identify biomarkers and to investigate pathophysiological processes,
2
Programa de Pós-Graduação em Ciências especially regarding EV roles in inflammation and thrombosis. However, EV isolation
Biológicas, Institute of Biological Sciences,
from plasma has faced the fundamental issue of lipoprotein contamination, repre-
Federal University of Juiz de Fora, Juiz de
Fora, Minas Gerais, Brazil senting an important bias since lipoproteins are highly abundant and modulate cell
3
Laboratory of Toxinology, Oswaldo Cruz signaling, metabolism, and thromboinflammation.
Institute, Fiocruz, Rio de Janeiro, Rio de
Janeiro, Brazil
Objectives: Here, we aimed to isolate plasma EVs after depleting lipoproteins, thereby
4
Laboratory of Immunopharmacology, improving sample purity and EV thromboinflammatory analysis.
Oswaldo Cruz Institute, Fiocruz, Rio de Methods: Density-based gradient ultracentrifugation (G-UC) was used for lipoprotein
Janeiro, Rio de Janeiro, Brazil
5
depletion before EV isolation from plasma through size-exclusion chromatography
Gene Expression Regulation Laboratory,
Carlos Chagas Institute, ICC-Fiocruz, (SEC) or serial centrifugation (SC). Recovered EVs were analyzed by size, concentration,
Curitiba, Paraná, Brazil cellular source, ultrastructure, and bottom-up proteomics.
6
Laboratory of Ultrastructure and Tissue,
Results: G-UC efficiently separated lipoproteins from the plasma, allowing subsequent
Department of Histology and Embryology,
State University of Rio de Janeiro, Rio de EV isolation through SEC or SC. Combined analysis from EV proteomics, cholesterol
Janeiro, Rio de Janeiro, Brazil
quantification, and apoB-100 detection confirmed the significant reduction in lipopro-
7
Laboratory of Veterinary Pathology and
teins from isolated EVs. Proteomic analysis identified similar gene ontology and cellular
Histology, Department of Veterinary
Medicine, Institute of Biological Sciences, components in EVs, regardless of lipoprotein depletion, which was consistent with
Federal University of Juiz de Fora, Juiz de
similar EV cellular sources, size, and ultrastructure by flow cytometry and transmission
Fora, Minas Gerais, Brazil
8
Center for Technological Development in
electron microscopy. Importantly, lipoprotein depletion increased the detection of less
Health, Fiocruz, Rio de Janeiro, Rio de abundant proteins in EV proteome and enhanced thromboinflammatory responses of
Janeiro, Brazil
platelets and monocytes stimulated in vitro with EV isolates.

Manuscript handled by: Dr Patricia Liaw

-
Final decision: Dr Patricia Liaw, 02 January 2024

Laura Botelho Merij and Luana Rocha da Silva contributed equally to this study.

Monique Ramos de Oliveira Trugilho and Eugenio D. Hottz contributed equally to this study.

© 2024 International Society on Thrombosis and Haemostasis. Published by Elsevier Inc. All rights reserved.

J Thromb Haemost. 2024;▪:1–17 jthjournal.org 1

2
- MERIJ ET AL.

Correspondence
Eugenio D. Hottz, Programa de Pós- Conclusion: Combination of G-UC+SEC significantly reduced EV lipoprotein contami-
Graduação em Ciências Biológicas, Institute
nation without interfering in EV cellular source, gene ontology, and ultrastructure,
of Biological Sciences, Federal University of
Juiz de Fora, Juiz de Fora, 36036-330, MG, allowing the recovery of highly pure EVs with potential implications for functional
Brazil.
assays and proteomic and lipidomic analyses.
Email: eugeniohottz@gmail.com and
eugenio.hottz@ufjf.br
KEYWORDS
Monique Ramos de Oliveira Trugilho,
blood plasma, extracellular vesicles, EV isolation, EV proteomics, lipoproteins
Center for Technological Development in
Health, Fiocruz, Rio de Janeiro, 21040-361
RJ, Brazil.
Email: mrotrugilho@hotmail.com and
monique.trugilho@fiocruz.com

Funding information
This work was supported by grants from the
Conselho Nacional de Desenvolvimento
Científico e Tecnológico (CNPq),
Coordenação de Aperfeiçoamento de
Pessoal de Nível Superior (CAPES),
Fundação de Amparo a Pesquisa do Estado
de Minas Gerais (FAPEMIG, to E.D.H.), and
Instituto Serrapilheira (to E.D.H.).

1 | INTRODUCTION Several methods have been described to isolate EVs from plasma,
including serial centrifugation (SC), density-based gradients, and size-
Extracellular vesicles (EVs) are lipid-bound vesicles able to transport a exclusion chromatography (SEC) [16,19]. Nevertheless, none of them
variety of cargos from their secreting cells, including RNAs, proteins, alone has been proven to be capable of completely eliminating lipo-
and bioactive lipids, to be delivered to target cells via biofluids, protein contamination [19–21]. We propose a protocol for lipoprotein
thereby regulating many biological responses [1,2]. Owing to studies depletion from plasma using density-based gradient ultracentrifuga-
that revealed their unique roles in cell-to-cell communication, EVs tion (G-UC) prior to EV isolation. This approach proved efficient in
have attracted strong interest over the past decades, fueled by their significantly reducing low-density lipoprotein (LDL), very low-density
potential clinical utility as prognostic markers [3] and their participa- lipoprotein (VLDL), and most of the high-density lipoprotein (HDL)
tion in physiological and pathologic contexts [4]. EVs participate in the from plasma samples without damaging the EVs. Subsequent EV
immune signaling network [5], inflammatory amplification [6,7], he- isolation using SEC resulted in highly pure EVs by eliminating the
mostasis, and pathologic thrombosis [8], among others. Analysis of EVs remnant HDL and part of the plasma proteins. Combined analysis
in biofluids may reveal pathologic processes and the metabolic state of from biochemical, immunological, and proteomic approaches
the body in several conditions [9], including cancer [10] and cardio- confirmed the elimination of contaminants. Downstream character-
vascular [11], neurodegenerative [12], and infectious diseases [13]. ization of EV isolates showed no bias in EV cell source, ultrastructure,
However, studies aiming to characterize and functionally assess EVs and cellular component terms in gene ontology, regardless of lipo-
from biofluids still face the problem of co-isolated contaminants [14]. protein depletion. Importantly, lipoprotein elimination enhanced the
Blood plasma is available with minimal invasive sampling and is abundance of other proteins in the EV proteome and increased
the main source of EVs in clinical research. Plasma lipoproteins are functional responses of platelets and monocytes to EVs in vitro,
considered the main contaminant in isolated EVs because they share revealing major impacts in investigating EV function, cargo, and
similar size and density with all EV subsets [15–17]. In addition, li- biomarker assessment.
poproteins are highly abundant and may mask the response to other
bioactive components in EV samples. Conversely, lipoproteins partic-
ipate in the immunopathogenesis of several cardiovascular and 2 | METHODS
inflammatory diseases, especially when oxidized, acting as damage-
associated molecular patterns to activate pattern recognition 2.1 | Blood collection and plasma preparation
receptors and induce thromboinflammatory responses [18]. Thus,
eliminating lipoprotein contamination in EV samples is essential to Peripheral blood was obtained from healthy volunteers, as approved
investigate the participation of EVs in physiological and pathological by the Institutional Review Board of Federal University of Juiz de Fora
processes, thus avoiding any bias from lipoprotein bioactivity. (HU-UFJF, 2.223.542). All participants volunteered by providing
MERIJ ET AL.
- 3

written informed consent. Peripheral blood samples were drawn into 000 × g depending on the length of the gradient and rotor angle
anticoagulant acid-citrate-dextrose (ACD) and centrifuged at 700 × g (Beckman Coulter), as informed in Supplementary Table S1.
for 20 minutes to obtain the platelet-poor plasma (PPP), which was Lipoprotein-depleted plasma after G-UC was submitted to dialysis
subsequently centrifuged at 2500 × g for 15 minutes to deplete against phosphate-buffered saline (PBS) overnight under slow homog-
remaining platelets, apoptotic bodies, membrane fragments, and cell enization (1-kDa cutoff dialysis system, GE Healthcare #80648394). A
debris to obtain the plasma (Figure 1A). Quantification of the triphasic density gradient was also used for lipoprotein separation and is
remaining platelets in PPP (86 ± 26/μL) and plasma (3 ± 1.5/μL) is described in Supplementary Methods and Supplementary Table S2.
shown in Supplementary Figure S1. The protease inhibitor cOmplete
Mini EDTA-free (Roche) was applied in samples that would later be
analyzed by proteomics. All plasma samples were immediately stored
2.3 | SEC
at −20 ◦ C.

SEC was prepared by filling a PD10 column with 10 mL of resin

2.2 | G-UC Sepharose CL-2B (Sigma Aldrich) according to the study by Böing et al.
[20]. Each column was activated with 50 mL of 0.22-μm filtered sterile
Biphasic density gradient was assembled, as reported previously PBS, and 1 mL of plasma was individually loaded into the chroma-
[22,23], with plasma mixed with 0.5-g/mL potassium bromide for den- tography column. Eighteen milliliters of PBS was applied to maintain
sity adjustment topped with saline solution (0.9% w/v NaCl) until filling column flow (Figure 1C). A total of 35 fractions containing 0.5 mL of
the tube (Figure 1B). Gradients were ultracentrifuged at 150 000 to 250 the eluate were collected and stored at −20 ◦ C.

F I G U R E 1 Workflows for EV isolation

through the different approaches. (A)
Plasma was obtained from healthy
volunteers after centrifugation and
processed for EV isolation with or without
(B) preprocessing with a density-based G-
UC for LP depletion. EVs were isolated
through (C) SEC, (D) SC, (E) G-UC+SEC, or
(F) G-UC+SC. EV, extracellular vesicle; G-
UC, gradient ultracentrifugation; LP,
lipoprotein; SC, serial centrifugation; SEC,
size-exclusion chromatography.
4
- MERIJ ET AL.

2.4 | SC FACSCanto II in low flow rate through FACSDiva software. Minimal

information [27,29] to report EV flow cytometry data (MIFlowCyt-EV)
Large EVs were obtained through differential centrifugation according is available in Supplementary Table S3.
to the International Guidelines for EVs [16,19]. Briefly, plasma sam-
ples were centrifuged at 19 000 × g for 2 hours at 4 ◦ C to sediment
large EVs. The supernatant was removed to a final volume of 100 μL
2.7 | Sample preparation for mass spectrometry
and washed with 1400 μL of filtered PBS. This procedure was
analysis
repeated 2 more times (Figure 1D). EVs suspended in PBS were stored
at −20 ◦ C.
Fractions from SEC were pooled or analyzed individually as shown in
Supplementary Figure S3. Samples were first dried using vacuum
concentrators (SpeedVac Plus SC210A, Savant) and suspended in
2.5 | Nanoparticle tracking analysis
RapiGest SF (Waters) at 0.1% (w/v) in 50 mM ammonium bicarbonate.
The protein concentration of each sample was determined by absor-
EV quantification along the 35 fractions collected by SEC was inferred
bance at 280 nm with a Nanodrop 2000 (Thermo Fisher Scientific).
by nanoparticle tracking analysis (NTA) using a Nano-Sight LM10
Fifty micrograms of total protein was prepared, as described previ-
(Malvern Panalytical) instrument with an O-ring top plate with a limit
ously [30]. Proteomic experiments were performed with EVs isolated
of detection from 10 to 1000 nm. Analysis was performed in the NTA
from 4 healthy volunteers and analyzed through LC-MS/MS in tech-
V.3.4.3 program following the manufacturer’s guidelines. All fractions
nical triplicates.
were diluted 1:10 in 0.22-μm–filtered PBS to a final volume of 1 mL.
After reduction in dithiothreitol (3 hours at 37 ◦ C, 10 mM final
Videos were obtained with a green laser (wavelength: 532 nm) in an
concentration) and alkylation in iodoacetamide (30 minutes in the
sCMOS camera operating at 30 frames per second, and the camera
dark at room temperature, 25 mM final concentration), trypsinization
level parameter was adjusted to 12. Each sample was analyzed by 60-
(overnight at 37 ◦ C, 1:50 [m/m]) was stopped by adding trifluoroacetic
second videos (technical replicates), with particle count and size range
acid to a final concentration of 1%. The digested peptide mixture was
determinations made at a detection limit of 5.
desalted with Poros R2 resin (POROS R2, Applied Biosystems) [30].
Samples were resuspended in 20 μL of 1% (v/v) formic acid (FA),
normalized to an approximate concentration of 0.250 μg/μL, and
2.6 | EV flow cytometry
stored at −20 ◦ C until mass spectrometric analysis.

Phosphatidylserine-positive large EVs were analyzed through flow

cytometry regarding Annexin V binding and a size of <1 μm. Forty
microliters of each EV sample was labeled with fluorescein isothio- 2.8 | Mass spectrometry analysis
cyanate (FITC)- or phycoerythrin (PE)-conjugated Annexin V (cat #
51-65874X, BD Pharmingen, 1:10 dilution). Polystyrene spherical The nLC-nESI MS/MS analysis was performed on an Ultimate 3000
particles of known light refractive index (RI) and size (Rosetta; Exo- (Dionex) coupled to a Q-Exative Plus (Thermo Scientific). Desalted
metry) were used to calibrate and gate the events for <1 μm tryptic peptides (0.5 μg/injection) were loaded into a guard column (2
considering the known RI of latex beads and the previously described cm × 100-μm internal diameter × 3-μm particle size Magic C18 AQ,
RI of EVs. We used the Rosetta Calibration v2.05 software to convert Michrom Bioresources), followed by an analytical column (38.5 cm
light scattering into nanometers based on Mie theory calculation, as PicoFrit Self-Pack, New Objective × 75-μm internal diameter × 1.9-μm
shown in Supplementary Figure S2 [24–27]. EV concentration was particle size; ReproSil-Pur 120 C18-AQ, Dr Maisch). Mobile phase A
determined using polystyrene beads of know concentration (#431, (0.1% v/v FA in water) and mobile phase B (0.1% v/v FA in acetonitrile)
Polysciences Inc) by calculating the EVs-to-beads ratio in each ac- were used in a separation gradient from 2% to 40% B for 152 minutes;
quired sample, as reported previously [28], and represented as EV/μL the concentration was increased to 80% B in 2 minutes and main-
in the isolated EV sample. In selected experiments, EV main cellular tained isocratically for 2 additional minutes. The lens voltage was set
sources were investigated by labeling 30 μL of each EV sample with to 60 V. Full scan MS mode was acquired with a resolution of 70 000
PE-conjugated Annexin V, FITC-conjugated anti-CD41 (cat # 555466, (FWHM for m/z: 200; automatic gain control set to 1 × 106). Up to 12
clone HIP8/mouse IgG1k, BD Pharmingen, 1:10 dilution), and allo- of the most abundant precursor ions from each MS scan (m/z: 300-
phycocyanin (APC)-conjugated anti-CD235 (cat # 551336, clone GA- 1500) were sequentially subjected to fragmentation by higher energy
R2 (HIR2)/mouse IgG2bk, BD Pharmingen, 1:10 dilution). EVs collisional dissociation. Fragment ions were analyzed (MS2 scan) at a
labeled with each antibody separately were used for appropriate color resolution of 15 000 and automatic gain control set to 5 × 104.
compensation, and isotype-matched immunoglobulin G (IgG) conju- Samples were analyzed in technical triplicate, and data were acquired
gated with the same fluorochrome (BD Pharmingen) was used as a using Xcalibur software (version 3.0.63). The mass spectrometry data
negative control. After incubation, labeled samples were diluted 10 have been deposited in the ProteomeXchange consortium via the
times in Annexin V binding buffer and acquired in flow cytometer BD PRIDE partner repository under the identifier PXD043684.
MERIJ ET AL.
- 5

2.9 | Platelet and monocyte stimulation in vitro 2.10 | Data analysis

Platelets and monocytes were isolated as described in the Supple- GraphPad Prism program, version 8 (GraphPad software Inc), was
mentary Methods. Isolated platelets were stimulated with EVs ob- used to perform the statistical analysis. Samples were analyzed for
tained through SEC or G-UC+SEC at 5 or 25 μg of protein/mL for 2 normal distribution using the Shapiro–Wilk and Kolmogorov‒Smirnov
hours at 37 ◦ C in the presence of 10-μg/mL polymyxin B (Sigma tests. Comparisons between 2 groups were performed using the
Aldrich). Platelets were centrifuged (900 × g, 10 minutes), the Mann–Whitney U-test for nonparametric distributions or Student’s t-
supernatants were recovered and stored at −80 ◦ C until the moment test for parametric distributions.
of use, and the platelets were labeled for flow cytometry as Raw mass spectrometry files were analyzed using PatternLab for
described in the Supplementary Methods. Monocytes were Proteomics V4 [31]. For peptide identification, the peptide spectrum
stimulated with EVs isolated through SEC or G-UC+SEC at 5 or 25 match method was employed using the Comet algorithm and data-
μg of protein/mL for 18 hours at 37 ◦ C in 5% CO2 atmosphere in the base from Homo sapiens obtained from the neXtProt consortium [32]
presence of 10-μg/mL polymyxin B (Sigma Aldrich). The monocytes (containing 42.135 protein sequences), and a target-decoy strategy
were centrifuged (300 × g, 10 minutes), the supernatants were was chosen to infer the false discovery rate (FDR). The identifications
◦
recovered and stored at −80 C until the moment of use, and the were filtered by the Search Engine Processor (SEPro) module into
monocytes were labeled for flow cytometry as described in the PatternLab V4. XCorr identification metric values, DeltaCN, Delta-
Supplementary Methods. Mass, z-score, number of corresponding peaks, and secondary

F I G U R E 2 EV elution profile after SEC. Plasma from healthy volunteers was applied in SEC columns, and the EVs in each fraction were
analyzed through (A) NTA or (B) flow cytometry (≤1 μm/Annexin V+). (C) SEC EV-enriched fractions (7-11) were pooled and analyzed through
flow cytometry alongside EVs from the same healthy volunteers isolated through SC (19 000 × g, 120 minutes). The concentration of large EVs
(≤1 μm/Annexin V+) is shown for each condition. Bars and dots represent the mean ± SEM of 3 to 7 independent experiments with plasma
from different donors. *P < .05 compared with SC. EV, extracellular vesicle; NTA, nanoparticle tracking analysis; SC, serial centrifugation; SEC,
size-exclusion chromatography.
6
- MERIJ ET AL.

F I G U R E 3 EV contaminants after
isolation through SEC. Fractions resulting
from SEC were evaluated for the presence
of contaminants (lipoproteins and plasma
proteins). The concentrations of (A) HDL
and (B) total cholesterol and total proteins
within SEC fractions were determined
through specific colorimetric assays. (C)
MSpec related to albumin, ApoB-100, and
ApoA-I and A-II within SEC fractions. Dots
represent the mean ± SEM of 3 to 7
independent experiments with plasma from
different donors. HDL, high-density
lipoprotein; Mspec, mass spectra; SEC, size-
exclusion chromatography.

ranking values were used to generate a Bayesian discriminator, and a org/). From this total database, only proteins from EVs extracted
cutoff score was established to accept a FDR of <1% based on the from human plasma studies were considered. Gene ontology
number of decoy tags. Trypsin specificity was set with a maximum of 2 analyses for cellular components and molecular function of EV pro-
missed cleavages, cysteine carbamidomethylating was set as a fixed teome identified by each method were performed in FunRich analysis
modification, and methionine oxidation was set as a variable modi- tools.
fication. The principle of maximum parsimony was applied to
generate the final list of identified proteins.
Spectral count quantification was generated in the Project Or- 3 | RESULTS
ganization module in PatternLab V4 to evaluate the differential
abundance of proteins identified in EVs isolated by SEC compared 3.1 | EVs isolated through SEC have persistent
with G-UC+SEC. The conditions for the association of mass spectra to lipoprotein contamination
proteins were only unique peptide spectral count.
Total proteins identified were compared with the Vesiclepedia SEC is a method commonly used to isolate EVs from plasma [15,19].
EV database [33] using the FunRich program tools (http://funrich. Consistent with previous reports [20,29], NTA analysis showed a
MERIJ ET AL.
- 7

F I G U R E 4 Characterization of lipoproteins and EVs in fractions from density-based G-UC. Plasma was applied in a biphasic density gradient
and processed through ultracentrifugation. (A) The concentrations of total cholesterol, triacylglycerol, and albumin in each fraction. (B) The
concentration of albumin and large EVs (≤1 μm/Annexin V+) in each 2 fractions pooled together. (C) EV-enriched fractions (21-26) were
pooled and precipitated through SC (19 000 × g, 120 minutes) alongside native plasma from the same healthy volunteers. The concentration of
large EVs (≤1 μm/Annexin V+) and the main cellular source (CD235: erythrocytes; CD41: platelets) are shown for each condition. Bars and
dots represent the mean ± SEM of 3 to 4 independent experiments with plasma from different healthy volunteers. EV, extracellular vesicle;
G-UC, gradient ultracentrifugation; SC, serial centrifugation.

SEC elution profile with large particles (200-1000 nm) and small phosphatidylserine-positive large EVs by calibrating the flow
particles (<200 nm) enriched between fractions 7 and 12 cytometry reads as size [24–26] and using latex beads of known
(Figure 2A). To confirm whether larger particles detected in the concentration as reference to quantify the vesicles [23,24].
SEC eluate were large EVs, we employed flow cytometry to analyze We observed an enrichment of phosphatidylserine-positive large
8
- MERIJ ET AL.

F I G U R E 5 Combination of density-based G-UC plus SEC significantly reduces lipoprotein contamination in isolated EVs. (A, B)
Concentration of total cholesterol and albumin in SEC fractions with or without preprocessing with G-UC to lipoprotein elimination. (C)
Volcano plots of all shared protein entries and their abundance in EVs isolated through SEC or G-UC+SEC. Each dot represents a protein
mapped according to its log2-fold change on the ordinate axis and its −log10 P value on the abscissa axis. Red dots indicate proteins that satisfy
MERIJ ET AL.
- 9

EVs in fractions 7 to 11 (Figure 2B), which is consistent with the Next, we investigated the position of phosphatidylserine-positive
NTA analysis. We then pooled EV-enriched fractions (7-11) to large EVs after G-UC using flow cytometry. Large EVs were detected
evaluate EV concentrations compared with large EVs isolated in the plasma fractions (21-26) alongside albumin (Figure 4B). We then
through SC (19 000 × g, 120 minutes; Figure 1C, D). As expected, pooled the plasma fractions after G-UC and isolated large EVs through
we observed a significant dilution of the EVs isolated through SEC SC (19 000 × g, 120 minutes; Figure 1F) to analyze EV concentrations
(Figure 2C). and main cell sources (platelet and erythrocyte) compared with EVs
To assess EV contamination, we investigated the position of isolated through SC only. As shown in Figure 4C, there were no sig-
plasma proteins and lipoproteins in the SEC elution fractions. Similar nificant changes in the concentration and cellular source of
to previous reports [20,29–31], there was no overlap of HDL and phosphatidylserine-positive large EVs after G-UC. These data show
plasma proteins with the fractions containing large EVs (Figure 3A, efficient depletion of LDL and VLDL and an important reduction in
B). However, we still identified higher levels of cholesterol in the HDL from plasma, allowing subsequent EV isolation.
EV-rich fractions, indicating persistent contamination with LDL and/
or VLDL in EVs isolated through SEC (Figure 3B). In a comple-
mentary way, we carried out proteomic analysis of SEC fractions 1
3.3 | Gradient ultracentrifugation plus SEC
to 14. Fractions 1 to 5 were analyzed together because of the low
significantly reduces EV lipoprotein contamination
individual protein amounts, and fractions 6 to 14 were analyzed
individually (Supplementary Figure S3A). Total protein identified
As G-UC did not entirely eliminate plasma HDL (Figure 4) and HDL
based on the maximum parsimony concept resulted in the identifi-
was effectively separated from EVs through SEC (Figure 2), we pro-
cation of an average of 290 proteins (Supplementary Table S4). Our
posed the use of the G-UC+SEC combination (Figure 1E) to eliminate
results showed a significant spectral count of apoB-100 in EV-
contamination with all lipoproteins. Specific cholesterol quantification
enriched fractions (7-11; Figure 3C and Supplementary Table S4),
showed that the combination of G-UC+SEC significantly reduced the
confirming persistent contamination by VLDLs and/or LDLs. Albu-
cholesterol contamination from EV-enriched SEC fractions (7-11;
min spectra were detected in all fractions, although at low fre-
Figure 5A, B). Lipoprotein elimination was also evidenced by an
quency in the EV-enriched ones, with a significant increase in
important reduction of apoB100 in the proteome analysis of EVs
fractions 12 to 14. Spectral counts referring to HDL (ApoA-I and
isolated through G-UC+SEC (pooled fractions 7-11; Figure 5C, D and
ApoA-II) were also observed in fractions 7 to 14, although less
Table). Other apolipoproteins, such as ApoC-I, ApoC-II, ApoC-III,
expressively.
ApoD, and ApoM, confirmed a higher abundance of lipoproteins in
EVs isolated by SEC only (Table and Supplementary Table S4B, C, and
F). Similarly, ApoB100 elimination was also achieved in EVs isolated
3.2 | Gradient ultracentrifugation separates plasma through G-UC+SC compared with SC alone (Figure 5E). These data
lipoprotein from EVs show that plasma processing through G-UC results in an efficient
reduction in the apoB-100 spectral count, approaching zero, regard-
G-UC is widely used for the isolation of lipoproteins [34,35]. We less of the sequential isolation method (Figure 5C–E and
employed a biphasic density–based gradient to deplete lipoproteins Supplementary Table S5).
from plasma samples before isolating EVs (Figure 1B). We subjected Our results also demonstrate very low contamination by albumin
plasma to G-UC and collected 26 fractions, as reported previously in the EV-enriched fractions, which was not affected by G-UC
[23]. Our data confirm the efficiency of G-UC in separating VLDL, LDL, (Figure 5A, E). In addition, the proportion of total spectral counting
and most of the HDL (triglyceride-rich and/or cholesterol-rich frac- referring to albumin was lower in G-UC+SEC (32.2%) than in samples
tions) from plasma (albumin-rich) fractions (Figure 4A and processed through G-UC+SC (54.9%), while SEC and SC alone
Supplementary Figure S5A). However, a small contamination of reached 19.9% and 27%, respectively. Therefore, G-UC+SEC also
cholesterol HDL remained within the plasma (Figure 4A and reduced the albumin spectral count compared with G-UC+SC.
Supplementary Figure S5A). Similar results were observed with a To confirm the proteomic data of apoB-100 elimination, we
triphasic density–based gradient (Supplementary Figures S4 and S5). investigated the presence of apoB-100 in the EV-enriched fractions

neither the fold-change cutoff nor the FDR cutoff α (0.05), green dots depict protein entries that satisfy the fold-change cutoff but not FDR α,
and orange dots indicate proteins that satisfy both fold-change and FDR α, but present low fold changes. Blue and yellow dots represent
protein entries that satisfy all statistical filters. (D) Frequency of proteins related to apolipoproteins, coagulation, and complement pathways in
EV samples isolated through SEC or G-UC+SEC. (E) MSpec related to albumin, ApoB-100, and ApoA-I and II among EV proteomes isolated
through SC, SEC, and G-UC followed by SC (G-UC+SC) or SEC (G-UC+SEC). Dots represent the mean ± SEM of 4 independent experiments
with plasma from different donors. (F) Western blot analysis of apoB-100 in SEC fractions combined or not with G-UC. (G) Dot blot analysis of
apoB-100, CD63, and CD81 in pooled EV-enriched SEC fractions (7-11) combined or not with G-UC. Fraction 35 was used as negative control
for the EV markers. Blots are representative of 3 to 4 independent experiments with EVs from different healthy donors. EV, extracellular
vesicle; G-UC, gradient ultracentrifugation; Mspec, mass spectra; SC, serial centrifugation; SEC, size-exclusion chromatography.
10
- MERIJ ET AL.

TABLE Differentially expressed proteins identified in the proteome of EVs isolated through SEC vs G-UC+SEC.

Fold change P value Description Gene name Volcano plot

2.41 .0182 Complement component C8 beta chain C8B Enriched in G-UC+SEC

2.25 .0129 Phosphatidylinositol-glycan-specific phospholipase D GPLD1

2.03 .0266 Complement component C9 C9

1.95 .0026 Complement component C7 C7

1.84 .0017 Vitronectin VTN

1.60 .0296 Alpha-1-antitrypsin SERPINA1

1.50 .0004 Immunoglobulin kappa constant IGKC

1.47 .0149 Serotransferrin TF

1.41 .0275 Carboxypeptidase N catalytic chain CPN1

1.36 .0141 Prothrombin F2

−1.36 .0136 Apolipoprotein D APOD Enriched in SEC

−1.66 .0185 Coagulation factor XIII B chain F13B

−1.77 .0173 Apolipoprotein C-I APOC1

−2.04 .0027 Apolipoprotein M ApoM

−3.18 .0029 Apolipoprotein C-II APOC2

−3.78 .0085 Apolipoprotein C-III APOC3

−7.97 .0167 Apolipoprotein(a) LPA

−18.19 .0003 Apolipoprotein B-100 ApoB

Fold change P value Description Gene name Volcano plot

3.06 .0040 Alpha-1-antitrypsin SERPINA1 Enriched in G-UC+SEC

2.82 .0112 Soluble scavenger receptor cysteine-rich domain SSC5D

2.31 .0012 Complement factor H-related protein 1 CFHR1

2.21 .0330 Coagulation factor X F10

1.65 .0248 Immunoglobulin kappa variable 2-40 IGKV2-40

−1.54 .0100 Fibrillin-1 FBN1 Enriched in SEC

−2.47 .0026 Serum amyloid A-4 protein SAA4

The top protein list shows proteins that reach statistical significance in both fold change and P value; the bottom protein list shows protein entries that
reach both statistical filters but with lower abundance, as determined by an additional stringency filter.
EV, extracellular vesicle; G-UC, gradient ultracentrifugation; SEC, size-exclusion chromatography.

from SEC or G-UC+SEC through Western blot and dot blot. Our data apolipoproteins were largely enriched in SEC, with high statistically
confirmed that apoB-100 was significantly reduced in the EV-enriched significant fold change, especially ApoB-100 with an 18-fold change
fractions when plasma was subjected to G-UC+SEC (Figure 5F–G and (Figure 5C, D, Table, and Supplementary Table S5F). The reduction of
Supplementary Figure S6). Altogether, we confirmed that the combi- highly abundant apolipoproteins through G-UC+SEC significantly
nation of G-UC, followed by SEC, importantly reduced lipoprotein enhanced the detection of other proteins in EV samples (Figure 5C, D,
contamination from isolated EVs. Table, and Supplementary Table S5F). Proteins that were enhanced in
samples isolated through G-UC+SEC meeting both fold change and P
value statistical filters included proteins related to coagulation and
3.4 | Lipoprotein depletion enhances the detection complement pathways, with 1.5- to 3-fold increase, while proteins
of less abundant proteins by proteome meeting both statistical filters in SEC-isolated EVs were majorly
apolipoproteins (Figure 5C, D, Table, and Supplementary Table S5F).
The majority of the proteins detected in the proteome were As shown in Figure 5D, a relative enhancement of other proteins is
commonly expressed in EVs isolated through SEC and G-UC+SEC observed after the important reduction in highly abundant apolipo-
(330 common proteins). When analyzing these common proteins, proteins. Similar results were observed for proteins reaching
MERIJ ET AL.
- 11

increased fold change but not the P value cutoff (green protein list in diameter, consistent with small EVs. Notably, EVs isolated through SC
Figure 5C and Supplementary Table S5F). presented a more crumpled shape, while SEC-isolated EVs showed
EVs isolated by SEC or G-UC+SEC had 30 exclusive proteins greater preservation of their membranes and ultrastructure
identified in each proteome (Supplementary Table S5F). In the pro- (Figure 6E). Importantly, preprocessing through G-UC did not alter EV
teome of EVs isolated through SEC, specific proteins identified in all 4 size (Figure 6E and Supplementary Figure S2D), membrane structure,
samples were apolipoprotein C-IV (APOC4), putative apolipoprotein- and ultrastructure following isolation through SC or SEC (Figure 6E).
like 2 (LPAL2), cathelicidin antimicrobial peptide, and coagulation Consistent with preserved membrane structures in EVs isolated by
factor VIII. Conversely, L-selectin (SELL), transferrin receptor protein SEC, gene ontology analysis showed greater conservation of identified
1 (TFRC), proteins from extracellular matrix, and coagulation factors proteins and enriched pathways related to membrane function (CD44,
IX and XI were exclusively identified in EVs isolated through CD99, CHL1, COL18A1, COL6A3, COMP, JUP, LAMA2, LAMC1,
G-UC+SEC (Supplementary Table S5F). These are in agreement with PKP1, RELN, SVEP1, TNXB, VCAN) in EVs isolated through SEC and
the enrichment of coagulation-related proteins in EV proteome anal- G-UC+SEC (Supplementary Figure S7A), while plasma proteins and
ysis alongside elimination of apolipoproteins. platelet structural proteins were predominantly identified in the
proteome of SC and G-UC+SC (Supplementary Figure S7B). Alto-
gether, our data show that G-UC before EV isolation provides elimi-
3.5 | Gradient ultracentrifugation and SEC isolate nation of lipoprotein contamination without significantly biasing EV
EVs with no bias in gene ontology, size, and cellular source, ultrastructure, cellular components, and gene
ultrastructure ontology.

Flow cytometric analyses of isolated EVs indicated a small but

nonsignificant decrease in the yield of Annexin V-positive large EVs 3.6 | Lipoprotein depletion increases
isolated through G-UC+SEC compared with SEC alone (Figure 6A). thromboinflammatory response to EV isolates
These data were confirmed through flow cytometry analysis of pooled
EV-enriched fractions 7 to 11 (Figure 6B) using gold standard analysis We then inquired whether lipoprotein contamination affects platelet
that fulfill the MIFlowCyt-EV requirements (Supplementary Figure S2 and monocyte responses to EV isolates. First, we stimulated platelets
and Supplementary Table S3). Similar results were observed regarding with EVs for 2 hours and investigated platelet activation and secretion
the small EVs markers CD63 and CD81, which were slightly reduced (Figure 7A–E). Interestingly, EVs isolated through G-UC+SEC stimu-
in EVs isolated through G-UC+SEC (Figure 5F and Supplementary lated higher integrin αIIb/β3 activation (PAC-1) and P-selectin
Figure S6A). These data suggest a small EV loss in samples isolated (CD62P) surface translocation compared with EVs isolated through
through G-UC+SEC, which is acceptable considering the important SEC only (Figure 7A, B). Similarly, lipoprotein depletion enhanced
reduction in apolipoprotein contaminants (Figure 5). platelet secretion of the granule-stored factors sCD40L, RANTES/
To ensure that lipoprotein depletion through G-UC did not bias CCL5, and PF4/CXCL4 in response to EV isolates (Figure 7C–E).
EV populations after isolation, we analyzed EV gene ontology through Platelet CD63 expression and the secretion of PDGF and sCD62 were
proteomics and ultrastructure through transmission electron micro- not observed in response to EVs (Supplementary Figure S8A–C).
scope. Proteomic data analysis generated a final list of identified Regarding isolated monocytes, lipoprotein-depleted EV preparations
proteins based on the concept of maximum parsimony with similar induced higher tissue factor (TF) expression and the secretion of
protein numbers identified in EVs isolated through SC or SEC, proinflammatory cytokines, such as TNF-α, IL-1β, and IL-8/CXCL8
regardless of previous lipoprotein elimination through G-UC (Figure 7F–I). Stimulation with EVs did not regulate monocyte CD16
(Supplementary Table S6). Importantly, follow-up analysis confirmed expression and had significantly reduced IL-6 and IL-10 secretion at
no significant changes in the classification of EV proteins by cellular higher concentration (25 μg/mL; Supplementary Figure S8D–F).
component terms in gene ontology (Figure 6C). In the analysis per- However, the secretion of both IL-6 and IL-10 were slightly increased
formed by FunRich software, the proteome from EVs isolated by SEC by lipoprotein-depleted EV preparations at lower concentration (5 μg/
or G-UC+SEC were derived very similarly from “extracellular” (77.2% mL; Supplementary Figure S8E, F). Collectively, these data indicated
vs 77.91%), “exosomes” (56.4% vs 60.08%), “extracellular region” that lipoprotein elimination improved platelet and monocyte response
(44.8% vs 44.57%), “cytoplasm” (39% vs 37.60%), “lysosome” (30.1% to EV isolates in vitro.
vs 33.72%), and “extracellular space” (28.19% vs 27.52%), all terms
with a P value of <.001. Additionally, we used the Vesiclepedia
database filtered for plasma EVs to observe the intersection with the 4 | DISCUSSION
proteome of EVs from the different methods (Figure 6D), and no
significant difference was observed between EVs isolated after lipo- Previous attempts to isolate EVs from plasma have shown expressive
protein depletion through G-UC. contamination with lipoproteins [19], which is a setback in EV studies
Transmission electron microscope images showed samples because lipoproteins are highly abundant in EV preparations and may
comprising majorly round-shaped vesicles smaller than 200 nm in also modulate metabolic and thromboinflammatory responses
12
- MERIJ ET AL.

F I G U R E 6 EV gene ontology, size, and ultrastructure are not changed by lipoprotein depletion. (A) Concentration of large EVs in each
fraction of SEC combined or not with density-based G-UC for lipoprotein elimination. (B) EV-enriched fractions (7-11) were pooled and
analyzed through flow cytometry regarding large EV concentration. (C) GO analysis for the most significant terms mapped on the database of
EVs isolated by SC, SEC, and G-UC plus SC (G-UC+SC) or SEC (G-UC+SEC). Bars represent the mean ± SEM of 4 to 8 independent experiments
with plasma from different healthy volunteers. (D) Intercession between the EV proteome from different methods and the Vesiclepedia
database. (E) Representative transmission electron microscope images of EVs isolated through SC, SEC, G-UC+SC, or G-UC+SEC. Scale bars
represent a total length of 200 nm. Images are representative of EVs isolated from 4 independent experiments with plasma from different
healthy volunteers. EV, extracellular vesicle; GO, gene ontology; G-UC, gradient ultracentrifugation; SC, serial centrifugation; SEC, size-
exclusion chromatography.
MERIJ ET AL.
- 13

F I G U R E 7 Lipoprotein elimination enhances platelet and monocyte responses to EV isolates. (A–E) Platelets were isolated from a healthy
volunteer and stimulated with EVs from 4 independent healthy donors for 2 hours. (A, B) The percent of platelets expressing (A) activated
integrin αIIb/β3 (PAC-1) or (B) P-selectin (CD62P) after stimulation with EVs isolated though SEC combined or not with density-based G-UC
for lipoprotein elimination (G-UC+SEC). (C–E) The concentration of RANTES/CCL5 (C), soluble CD40L (D), and PF4/CXCL4 (E) secreted by
platelets stimulated with EVs isolated through SEC or G-UC+SEC. (F–I) Monocytes were isolated from a healthy volunteer and stimulated with
EVs from 4 independent healthy donors overnight. (F) The percent of monocytes expressing TF after stimulation with EVs isolated through SEC
or G-UC+SEC. (G–I) The concentration of TNF-α (G), IL-1β (H), and IL-8/CXCL8 (I) secreted by monocytes stimulated with EVs isolated through
SEC or G-UC+SEC. Bars represent mean and SEM of platelets or monocytes stimulated with EVs isolated from 4 independent plasma donors.
EV, extracellular vesicle; G-UC, gradient ultracentrifugation; SEC, size-exclusion chromatography; TF, tissue factor.

[18,36–39]. Here, we demonstrate a protocol to obtain EVs with Onódi et al. [42] using iodixanol density gradient to isolate EVs. In that
minimal lipoprotein contamination through the processing of plasma work, EVs were separated from the majority of lipoproteins, but HDL
by G-UC before EV isolation by SEC or SC (Figure 1). G-UC is a well- and albumin contaminations were still abundant [42]. Similarly, EVs
known method for isolating lipoproteins from plasma [34,35]. isolated from plasma after KBr- or OptPrep-based density ultracen-
Different density-based gradients have also been used to isolate EVs trifugation remained contaminated by HDL in other studies [21,41].
themselves [19,35,40,41]. In this work, we tested biphasic or triphasic Similar results were observed with different isolation methods based
density–based gradients as a step to deplete lipoproteins from plasma on density, precipitation, size exclusion, and membrane and protein
before EV isolation. Our results confirmed that G-UC was efficient in affinity, which consistently presented a non-EV protein cluster,
separating lipoproteins from plasma without jeopardizing EVs. How- including serum proteins and HDL-related apolipoproteins [41]. This
ever, albumin and small amounts of HDL remained contaminants in suggests that a combination of approaches may be necessary to
the EV-enriched plasma. These results are similar to those reported by eliminate lipoprotein contamination from EV isolates.
14
- MERIJ ET AL.

Different approaches have been employed to tackle the issue of collected plasma. A limitation of this study was that platelets were not
lipoprotein contamination. Recently, Mladenovic et al. [17] demon- completely eliminated from plasma, with remaining platelets of 1 to 4
strated that moderate plasma acidification to precipitate lipoproteins per μL. A plasma filtration step before EV isolation has been shown to
achieved a 60% reduction in contamination. But, despite enriching successfully remove reminiscent platelets [55,56] and must be added
the yield of EVs, lipoproteins were not completely eliminated [17]. to this protocol in the future. New studies will be necessary to
A dual-mode chromatography combining SEC and cation exchange to investigate how less ideal preanalytical conditions such as other an-
separate particles through size and charge has achieved an important ticoagulants, blood-processing time, plasma storage time, and freeze-
reduction in ApoB100 (VLDL and LDL), but not ApoA1 (HDL), in iso- thaw cycles [52–54] may affect the efficiency of EV isolation through
lated EVs [43]. Mørk et al. [44] used anti-ApoB antibodies conjugated G-UC+SEC.
to magnetic beads to remove VLDL and LDL from EV samples. An important advantage of reducing the contamination with
Nonetheless, magnetic immunodepletion resulted in EV loss, and highly abundant apolipoproteins (1.4- to 18-fold change) was the
ApoB was not completely eliminated. This may have occurred due to increased detection of other proteins in EV proteome (1.5- to 3-fold
insufficient anti-ApoB antibodies, which must be standardized for change). A similar phenomenon has been observed by Karimi et al.
future use [44]. In addition, contamination with HDL persisted after [57] and Van Deun et al. [43] when separating EVs from lipoproteins.
immunodepletion since HDL does not express ApoB [44]. Using KBr- This enrichment in protein abundance may directly contribute to a
based density ultracentrifugation, we achieved near-complete elimi- better understanding of EV molecular processes and functional re-
nation of VLDL and LDL, and reduced HDL from plasma prior to EV sponses. Despite a small loss in EV yield, our data show that EVs
isolation. New studies are still needed to locate the different lipo- recovered after G-UC+SEC are still similar to EVs isolated through
proteins after ultracentrifugation with more osmotic friendly density SEC alone in terms of cell source, gene ontology, and ultrastructure as
gradients, such as OptPrep, since the hyperosmotic environment of well as the obvious advantage of contaminant elimination. Botha et al.
KBr gradient may affect EV isolation and structure. Despite that, no [58] have recently shown that 20% to 40% of phosphatidylserine-
important alteration was observed after G-UC in our transmission positive events in EV flow cytometry are ApoB-positive, therefore,
electron microscope experiments. To tackle the remnant HDL lipoprotein contaminants. This is in agreement with lipoprotein par-
contamination after G-UC, we proposed the combination with SEC, ticles probably standing for a lot of events in NTA [59]. This also
which is known to separate EVs from HDL and albumin [20,45–48]. suggests that the small loss in large EVs after G-UC may stand not
SEC is a widely used method for isolating EVs [20]. Although only for EV loss but also the removal of phosphatidylserine-positive
SEC is able to provide EVs with minimal HDL contamination, VLDL lipoproteins. Importantly, lipoprotein depletion increased EV-induced
and LDL are consistently observed, as demonstrated in this and platelet and monocyte activation in vitro. This may have involved
previous studies [20,45,46,48]. In addition, SEC-isolated EVs still the enrichment of complement and coagulation proteins observed in
present a small contamination with plasma proteins [46–48]. Our EV samples isolated though G-UC+SEC. Because apolipoproteins are
results confirm these findings through a combination of immunologic highly abundant in SEC-isolated EVs, eliminating the contaminants
and biochemical approaches to detect cholesterol, ApoB-100, and may have increased the concentration of other bioactive components
albumin among EV-enriched SEC fractions. It is worth mentioning in G-UC+SEC–obtained EVs as we have normalized the EV stimuli by
that despite diluting the EV sample, SEC better preserved the EV their protein concentration. However, these experiments were per-
plasma membrane ultrastructure. These findings are in agreement formed with samples from healthy volunteers, and different clinical
with previous reports [20,40]. Although Askeland et al. [45] have conditions may affect platelet and monocyte responses to both EVs
reported a more complex proteome in vesicles isolated by high- and lipoproteins [18,36–39,60]. Future research evaluating the impact
speed SC than in those isolated by SEC, one cannot exclude higher of lipoprotein elimination in EVs from infectious, inflammatory, and
contamination with plasma proteins in EVs isolated through centri- metabolic diseases will be of paramount importance to assess EV roles
fugation. In addition, a limitation of EV isolation through SC is the in disease pathogenesis, especially in conditions with increased lipo-
need of multiple washing steps, which may result in EV loss during protein levels and oxidation, such as atherosclerosis and obesity.
the isolation [49]. In conclusion, we propose a protocol that eliminates lipoprotein
The combination of G-UC+SEC was highly efficient in reducing contamination with high efficiency, ensuring highly pure EVs without
the contamination with all lipoprotein classes, known as the main EV compromising their cellular source, gene ontology, size, and ultra-
contaminants [15,50,51]. By performing immunologic and biochemical structure. Although it may seem laborious, contamination by lipo-
approaches to detect cholesterol and apoB100, we achieved a sur- proteins is an important bias for many approaches, such as functional
prisingly efficient reduction in contamination, approaching 0 in EV-rich assays and proteomic and lipidomic analyses, and it is much more
fractions. Furthermore, it must be considered that by using G-UC, it is reliable and reproducible to eliminate confounding factors such as li-
possible to isolate plasma lipoproteins in addition to the lipoprotein- poproteins from EV isolates.
depleted EVs in separated fractions, allowing better use of clinical
samples. Different preanalytical approaches have been used to ACKNOW LEDGMEN TS
generate plasma samples for EV clinical research [52–55]. In the The authors thank the Laboratório Integrado de Pesquisa (LIP) from
present study, we isolated EVs from ACD-anticoagulated freshly the Programa de Pós-graduação em Ciências Biológicas (PPGCBio/
MERIJ ET AL.
- 15

UFJF) multiuser platform; the Fiocruz Network of Technology Plat- [13] Giannessi F, Aiello A, Franchi F, Percario ZA, Affabris E. The role of
forms, for the use of core facilities (Proteomics/RJ RPT02A—mass extracellular vesicles as allies of HIV, HCV and SARS viruses. Viruses.
2020;12:571.
spectrometry); and the Centro Nacional de Biologia Estrutural e
[14] Webber J, Clayton A. How pure are your vesicles? J Extracell Vesicles.
Bioimagem (CENABIO) and Adélia Mara Belém Lima for technical 2013;2:19861.
assistance. This work was supported by grants from the Conselho [15] Liangsupree T, Multia E, Riekkola ML. Modern isolation and sepa-
Nacional de Desenvolvimento Científico e Tecnológico (CNPq), ration techniques for extracellular vesicles. J Chromatogr A.
2021;1636:461773.
Coordenação de Aperfeiçoamento de Pessoal de Nível Superior
[16] Coumans FAW, Brisson AR, Buzas EI, Dignat-George F, Drees EEE,
(CAPES), Fundação de Amparo a Pesquisa do Estado de Minas Gerais El-Andaloussi S, Emanueli C, Gasecka A, Hendrix A, Hill AF,
(FAPEMIG, to E.D.H.), and Instituto Serrapilheira (to E.D.H.). Lacroix R, Lee Y, van Leeuwen TG, Mackman N, Mäger I, Nolan JP,
van der Pol E, Pegtel DM, Sahoo S, Siljander PRM, et al. Methodo-
logical guidelines to study extracellular vesicles. Circ Res. 2017;120:
AUTH OR CONT RIBUTIONS
1632–48.
L.B.M. and L.R.S performed most of the experiments, data analyses, [17] Mladenović D, Khamari D, Kittel Á, Koort K, Buzás EI, Zarovni N.
and manuscript writing. L.P., M.T.G., P.R.B.D., R.M.G., K.T.Q., M.A.R.N., Acidification of blood plasma facilitates the separation and analysis
F.B.A., S.T.M., and A.L.R.N. performed part of the experiments and of extracellular vesicles. J Thromb Haemost. 2023;21:1032–42.
[18] Miller YI, Shyy JYJ. Context-dependent role of oxidized lipids and
data analysis. L.R.A. and V.N.R. performed data analysis. P.T.B.,
lipoproteins in inflammation. Trends Endocrinol Metab. 2017;28:
M.R.O.T., and E.D.H. performed experimental design and reviewed the
143–52.
manuscript. E.D.H. conceptualized the study. M.R.O.T. conceptualized [19] Théry C, Witwer KW, Aikawa E, Alcaraz MJ, Anderson JD,
and designed the proteomic experiments. M.R.O.T. and E.D.H. directed Andriantsitohaina R, Antoniou A, Arab T, Archer F, Atkin-Smith GK,
the study. All authors reviewed and critically edited the manuscript. Ayre DC, Bach JM, Bachurski D, Baharvand H, Balaj L, Baldacchino S,
Bauer NN, Baxter AA, Bebawy M, Beckham C, et al. Minimal infor-
mation for studies of extracellular vesicles 2018 (MISEV2018): a
DECL AR ATION OF COMPETING INT ERES TS position statement of the International Society for Extracellular
There are no conflicting interests to disclose. Vesicles and update of the MISEV2014 guidelines. J Extracell Vesicles.
2018;7:1535750.
[20] Böing AN, van der Pol E, Grootemaat AE, Coumans FAW, Sturk A,
R EF E REN CE S Nieuwland R. Single-step isolation of extracellular vesicles by size-
[1] O’Brien K, Breyne K, Ughetto S, Laurent LC, Breakefield XO. RNA exclusion chromatography. J Extracell Vesicles. 2014;3:23430.
delivery by extracellular vesicles in mammalian cells and its appli- [21] Yuana Y, Levels J, Grootemaat A, Sturk A, Nieuwland R. Co-isolation
cations. Nat Rev Mol Cell Biol. 2020;21:585–606. of extracellular vesicles and high-density lipoproteins using density
[2] Ratajczak J, Miekus K, Kucia M, Zhang J, Reca R, Dvorak P. Em- gradient ultracentrifugation. J Extracell Vesicles. 2014;3:23262.
bryonic stem cell-derived microvesicles reprogram hematopoietic [22] Silva AR, Pacheco P, Vieira-de-Abreu A, Maya-Monteiro CM,
progenitors: evidence for horizontal transfer of mRNA and protein D’Alegria B, Magalhães KG, de Assis EF, Bandeira-Melo C, Castro-
delivery. Leukemia. 2006;20:847–56. Faria-Neto HC, Bozza PT. Lipid bodies in oxidized LDL-induced
[3] Xu R, Greening DW, Zhu HJ, Takahashi N, Simpson RJ. Extracellular foam cells are leukotriene-synthesizing organelles: a MCP-1/
vesicle isolation and characterization: toward clinical application. CCL2 regulated phenomenon. Biochim Biophys Acta. 2009;1791:
J Clin Invest. 2016;126:1152–62. 1066–75.
[4] De Toro J, Herschlik L, Waldner C, Mongini C. Emerging roles of [23] Merij LB, Andrade FB, Silva AR, Hottz ED. Isolation of microvesicles
exosomes in normal and pathological conditions: new insights for from plasma samples avoiding lipoprotein contamination. Methods
diagnosis and therapeutic applications. Front Immunol. 2015;6:203. Mol Biol. 2022;2409:245–55.
[5] Robbins PD, Morelli AE. Regulation of immune responses by extra- [24] Welsh JA, Horak P, Wilkinson JS, Ford VJ, Jones JC, Smith D,
cellular vesicles. Nat Rev Immunol. 2014;14:195–208. Holloway JA, Englyst NA. FCMPASS software aids extracellular
[6] Ratajczak J, Wysoczynski M, Hayek F, Janowska-Wieczorek A, vesicle light scatter standardization. Cytometry A. 2020;97:569–81.
Ratajczak MZ. Membrane-derived microvesicles: important and un- [25] van der Pol E, Sturk A, van Leeuwen T, Nieuwland R, Coumans F,
derappreciated mediators of cell-to-cell communication. Leukemia. ISTH-SSC-VB Working group. Standardization of extracellular
2006;20:1487–95. vesicle measurements by flow cytometry through vesicle diameter
[7] Hottz ED, Lopes JF, Freitas C, Valls-de-Souza R, Oliveira MF, Bozza MT, approximation. J Thromb Haemost. 2018;16:1236–45.
Da Poian AT, Weyrich AS, Zimmerman GA, Bozza FA, Bozza PT. [26] van der Pol E, Welsh JA, Nieuwland R. Minimum information to
Platelets mediate increased endothelium permeability in dengue report about a flow cytometry experiment on extracellular vesicles:
through NLRP3-inflammasome activation. Blood. 2013;122:3405–14. communication from the ISTH SSC subcommittee on vascular
[8] Zifkos K, Dubois C, Schäfer K. Extracellular vesicles and thrombosis: biology. J Thromb Haemost. 2022;20:245–51.
update on the clinical and experimental evidence. Int J Mol Sci. [27] Welsh JA, Arkesteijn GJA, Bremer M, Cimorelli M, Dignat-George F,
2021;22:9317. Giebel B, Görgens A, Hendrix A, Kuiper M, Lacroix R, Lannigan J, van
[9] Shah R, Patel T, Freedman JE. Circulating extracellular vesicles in Leeuwen TG, Lozano-Andrés E, Rao S, Robert S, de Rond L, Tang VA,
human disease. N Engl J Med. 2018;379:958–66. Tertel T, Yan X, Wauben MHM, et al. A compendium of single
[10] Lazar S, Goldfinger LE. Platelets and extracellular vesicles and their extracellular vesicle flow cytometry. J Extracell Vesicles. 2023;12:
cross talk with cancer. Blood. 2021;137:3192–200. e12299.
[11] Yang J, Zou X, Jose PA, Zeng C. Extracellular vesicles: potential [28] Cointe S, Judicone C, Robert S, Mooberry MJ, Poncelet P,
impact on cardiovascular diseases. Adv Clin Chem. 2021;105:49–100. Wauben M, Nieuwland R, Key NS, Dignat-George F, Lacroix R.
[12] Xiao Y, Wang SK, Zhang Y, Rostami A, Kenkare A, Casella G, Standardization of microparticle enumeration across different flow
Yuan ZQ, Li X. Role of extracellular vesicles in neurodegenerative cytometry platforms: results of a multicenter collaborative work-
diseases. Prog Neurobiol. 2021;201:102022. shop. J Thromb Haemost. 2017;15:187–93.
16
- MERIJ ET AL.

[29] Welsh JA, Van Der Pol E, Arkesteijn GJA, Bremer M, Brisson A, nanoparticle tracking analysis of extracellular vesicles. J Extracell
Coumans F, Dignat-George F, Duggan E, Ghiran I, Giebel B, Görgens A, Vesicles. 2017;6:1308779.
Hendrix A, Lacroix R, Lannigan J, Libregts SFWM, Lozano-Andrés E, [45] Askeland A, Borup A, Østergaard O, Olsen JV, Lund SM,
Morales-Kastresana A, Robert S, De Rond L, Tertel T, et al. MIFlowCyt- Christiansen G, Kristensen SR, Heegaard NHH, Pedersen S. Mass-
EV: a framework for standardized reporting of extracellular vesicle flow spectrometry based proteome comparison of extracellular vesicle
cytometry experiments. J Extracell Vesicles. 2020;9:1713526. isolation methods: comparison of ME-kit, size-exclusion chroma-
[30] Moraes ECDS, Martins-Gonçalves R, da Silva LR, Mandacaru SC, tography, and high-speed centrifugation. Biomedicines. 2020;8:246.
Melo RM, Azevedo-Quintanilha I, Perales J, Bozza FA, Souza TML, [46] de Menezes-Neto A, Sáez MJ, Lozano-Ramos I, Segui-Barber J,
Castro-Faria-Neto HC, Hottz ED, Bozza PT, Trugilho MRO. Prote- Martin-Jaular L, Ullate JM, Fernandez-Becerra C, Borrás FE, Del
omic profile of procoagulant extracellular vesicles reflects comple- Portillo HA. Size-exclusion chromatography as a stand-alone meth-
ment system activation and platelet hyperreactivity of patients with odology identifies novel markers in mass spectrometry analyses of
severe COVID-19. Front Cell Infect Microbiol. 2022;12:981. plasma-derived vesicles from healthy individuals. J Extracell Vesicles.
[31] Carvalho PC, Lima DB, Leprevost FV, Santos MD, Fischer JS, 2015;4:27378.
Aquino PF, Moresco JJ, Yates JR, Barbosa VC. Integrated analysis of [47] Baranyai T, Herczeg K, Onódi Z, Voszka I, Módos K, Marton N,
shotgun proteomic data with PatternLab for proteomics 4.0. Nat Nagy G, Mäger I, Wood MJ, El Andaloussi S, Pálinkás Z, Kumar V,
Protoc. 2016;11:102–17. Nagy P, Kittel Á, Buzás EI, Ferdinandy P, Giricz Z. Isolation of exo-
[32] Zahn-Zabal M, Michel PA, Gateau A, Nikitin F, Schaeffer M, Audot E, somes from blood plasma: qualitative and quantitative comparison
Gaudet P, Duek PD, Teixeira D, Rech de Laval V, Samarasinghe K, of ultracentrifugation and size exclusion chromatography methods.
Bairoch A, Lane L. The neXtProt knowledgeBase in 2020: data, tools PLoS One. 2015;10:e0145686.
and usability improvements. Nucleic Acids Res. 2020;48:D328–34. [48] Wang T, Turko IV. Proteomic toolbox to standardize the separation
[33] Pathan M, Fonseka P, Chitti SV, Kang T, Sanwlani R, Van Deun J, of extracellular vesicles and lipoprotein particles. J Proteome Res.
Hendrix A, Mathivanan S. Vesiclepedia 2019: a compendium of RNA, 2018;17:3104–13.
proteins, lipids and metabolites in extracellular vesicles. Nucleic Acids [49] Geeurickx E, Tulkens J, Dhondt B, Van Deun J, Lippens L,
Res. 2019;47:D516–9. Vergauwen G, Heyrman E, De Sutter D, Gevaert K, Impens F,
[34] Potts JL, Fisher RM, Humphreys SM, Gibbons GF, Frayn KN. Sepa- Miinalainen I, Van Bockstal PJ, De Beer T, Wauben MHM, Nolte-’t-
ration of lipoprotein fractions by ultracentrifugation: investigation of Hoen ENM, Bloch K, Swinnen JV, van der Pol E, Nieuwland R,
analytical recovery with sequential flotation and density gradient Braems G, et al. The generation and use of recombinant extracellular
procedures. Clin Chim Acta. 1994;230:215–20. vesicles as biological reference material. Nat Commun. 2019;10:3288.
[35] Redgrave TG, Roberts DCK, West CE. Separation of plasma lipo- [50] Sódar BW, Kittel Á, Pálóczi K, Vukman KV, Osteikoetxea X, Szabó-
proteins by density-gradient ultracentrifugation. Anal Biochem. Taylor K, Németh A, Sperlágh B, Baranyai T, Giricz Z, Wiener Z, Turiák L,
1975;65:42–9. Drahos L, Pállinger É, Vékey K, Ferdinandy P, Falus A, Buzás EI. Low-
[36] Rhoads JP, Major AS. How oxidized low-density lipoprotein activates density lipoprotein mimics blood plasma-derived exosomes and micro-
inflammatory responses. Crit Rev Immunol. 2018;38:333–42. vesicles during isolation and detection. Sci Rep. 2016;6:24316.
[37] Kugiyama K, Kerns SA, Morrisett JD, Roberts R, Henry PD. Impair- [51] Smalley DM, Root KE, Cho H, Ross MM, Ley K. Proteomic discovery
ment of endothelium-dependent arterial relaxation by lysolecithin in of 21 proteins expressed in human plasma-derived but not platelet-
modified low-density lipoproteins. Nature. 1990;344:160–2. derived microparticles. Thromb Haemost. 2007;97:67–80.
[38] Tjaden K, Adam C, Godfrey R, Hanley PJ, Pardali E, Waltenberger J. [52] Dhondt B, Pinheiro C, Geeurickx E, Tulkens J, Vergauwen G, Van
Low density lipoprotein interferes with intracellular signaling of Der Pol E, Nieuwland R, Decock A, Miinalainen I, Rappu P,
monocytes resulting in impaired chemotaxis and enhanced chemo- Schroth G, Kuersten S, Vandesompele J, Mestdagh P, Lumen N, De
kinesis. Int J Cardiol. 2018;255:160–5. Wever O, Hendrix A. Benchmarking blood collection tubes and
[39] Ahsan A, Han G, Pan J, Liu S, Padhiar AA, Chu P, Sun Z, Zhang Z, processing intervals for extracellular vesicle performance metrics.
Sun B, Wu J, Irshad A, Lin Y, Peng J, Tang Z. Phosphocreatine pro- J Extracell Vesicles. 2023;12:e12315.
tects endothelial cells from oxidized low-density lipoprotein-induced [53] Görgens A, Corso G, Hagey DW, Jawad Wiklander R,
apoptosis by modulating the PI3K/Akt/eNOS pathway. Apoptosis. Gustafsson MO, Felldin U, Lee Y, Bostancioglu RB, Sork H, Liang X,
2015;20:1563–76. Zheng W, Mohammad DK, van de Wakker SI, Vader P, Zickler AM,
[40] Gardiner C, Di Vizio D, Sahoo S, Théry C, Witwer KW, Wauben M, Mamand DR, Ma L, Holme MN, Stevens MM, Wiklander OPB, et al.
Hill AF. Techniques used for the isolation and characterization of Identification of storage conditions stabilizing extracellular vesicles
extracellular vesicles: results of a worldwide survey. J Extracell Ves- preparations. J Extracell Vesicles. 2022;11:e12238.
icles. 2016;3078:32945. [54] Gelibter S, Marostica G, Mandelli A, Siciliani S, Podini P, Finardi A,
[41] Veerman RE, Teeuwen L, Czarnewski P, Güclüler Akpinar G, Furlan R. The impact of storage on extracellular vesicles: a system-
Sandberg A, Cao X, Pernemalm M, Orre LM, Gabrielsson S, Eldh M. atic study. J Extracell Vesicles. 2022;11:e12162.
Molecular evaluation of five different isolation methods for extra- [55] Bettin B, Gasecka A, Li B, Dhondt B, Hendrix A, Nieuwland R, van
cellular vesicles reveals different clinical applicability and subcellular der Pol E. Removal of platelets from blood plasma to improve the
origin. J Extracell Vesicles. 2021;10:e12128. quality of extracellular vesicle research. J Thromb Haemost. 2022;20:
[42] Onódi Z, Pelyhe C, Terézia Nagy C, Brenner GB, Almási L, Kittel Á, 2679–85.
Manček-Keber M, Ferdinandy P, Buzás EI, Giricz Z. Isolation of high- [56] Bracht JWP, Los M, van Eijndhoven MAJ, Bettin B, van der Pol E,
purity extracellular vesicles by the combination of iodixanol density Pegtel DM, Nieuwland R. Platelet removal from human blood plasma
gradient ultracentrifugation and bind-elute chromatography from improves detection of extracellular vesicle-associated miRNA.
blood plasma. Front Physiol. 2018;9:1479. J Extracell Vesicles. 2023;12:e12302.
[43] Van Deun J, Jo A, Li H, Lin HY, Weissleder R, Im H, Lee H. Integrated [57] Karimi N, Cvjetkovic A, Jang SC, Crescitelli R, Hosseinpour Feizi MA,
dual-mode chromatography to enrich extracellular vesicles from Nieuwland R, Lötvall J, Lässer C. Detailed analysis of the plasma
plasma. Adv Biosyst. 2020;4:e1900310. extracellular vesicle proteome after separation from lipoproteins.
[44] Mørk M, Handberg A, Pedersen S, Jørgensen MM, Bæk R, Cell Mol Life Sci. 2018;75:2873–86.
Nielsen MK, Kristensen SR. Prospects and limitations of antibody- [58] Botha J, Handberg A, Simonsen JB. Lipid-based strategies used to
mediated clearing of lipoproteins from blood plasma prior to identify extracellular vesicles in flow cytometry can be confounded
MERIJ ET AL.
- 17

by lipoproteins: evaluations of annexin V, lactadherin, and detergent elevation myocardial infarction triggers platelet activation and ag-
lysis. J Extracell Vesicles. 2022;11:e12200. gregation. Blood. 2013;122:3632–41.
[59] Gardiner C, Ferreira YJ, Dragovic RA, Redman CWG, Sargent IL.
Extracellular vesicle sizing and enumeration by nanoparticle tracking
analysis. J Extracell Vesicles. 2013;2:19671.
SUPPL EMENTAR Y MAT ERIAL
[60] Chan HC, Ke LY, Chu CS, Lee AS, Shen MY, Cruz MA, Hsu JF,
Cheng KH, Chan HC, Lu J, Lai WT, Sawamura T, Sheu SH, Yen JH, The online version contains supplementary material available at
Chen CH. Highly electronegative LDL from patients with ST- https://doi.org/10.1016/j.jtha.2024.01.010