Abstract
Endothelial cells (ECs) experience shear stress associated with blood
flow. Such shear stress regulates endothelial function by altering cell
physiology. Since most cell culture protocols and media compositions
are designed for static cultures and experiments with ECs are
predominantly conducted under these non‐physiological conditions, a
model for culturing ECs under flow conditions is developed, which more
closely mimics their physiological environment. This approach also
enables the isolation of EVs while minimizing FCS‐derived contaminants.
In this study, a comprehensive assessment of how physiologically
relevant cultivation conditions influence the vesicle composition and
function of ECs is provided. A detailed investigation is conducted for
the effect of different cell culture media on morphology and marker
expression of human umbilical cord endothelial cells (HUVECs) and EVs,
and optimize the conditions to culture ECs under flow, tailoring them
specifically to facilitate the efficient isolation of EVs using a
hollow‐fiber system model. These EVs are then characterized and
compared to those isolated from traditional static culture conditions.
Overall, this study presents a model on isolating EC‐derived EVs under
conditions that closely mimic physiological environments, and
characterization at their proteome, gene expression, and microRNA
profile.
Keywords: endothelial cells, extracellular vesicles, microRNAs, network
analysis, proteomics, shear stress
__________________________________________________________________
Schematic illustration of the current study. A) Multiple cell culture
media are used for HUVEC cultures under both static and flow
conditions. Cells are analyzed for morphology, endothelial markers, and
gene expression. B) EVs are isolated from static and flow HUVEC
cultures, and investigated for particle characterization, transcriptome
and proteome profiles. Created using BioRender.com.
graphic file with name SMTD-9-2401841-g006.jpg
1. Introduction
Extacellular vesicles (EVs) are nanosized membrane‐bound structures
released by almost all types of cells into their external environment.
Eukaryotic EVs are usually classified into three main categories, based
on their size and mode of production.^[ [50]^1 ^] Microvesicles are
formed by the outward budding of membrane vesicles from the cell
surface.^[ [51]^2 ^] Exosomes originate from the endocytic pathway
through the ‘outward’ budding of the late endosomal membrane.
Initially, they accumulate in structures known as multivesicular bodies
(MVBs), which later fuse with the plasma membrane, releasing their
contents as exosomes into the extracellular space.^[ [52]^3 ^] The
third major type of eukaryotic EVs called apoptotic bodies are produced
from cells undergoing programmed cell death by outward budding from the
surface of apoptotic cell.^[ [53]^4 ^]
The significance of EVs was for a long time underestimated, with EVs
being initially referred to as cellular ‘dust’.^[ [54]^5 , [55]^6 ^] It
is now well recognized that EVs carry a range of bioactive molecules
which play a crucial role in intercellular communication, influencing
various physiological and pathological processes on their recipient
cells.^[ [56]^7 , [57]^8 ^] While originating from the packaging of
cytoplasmic contents, EVs are known to harbor numerous proteins, and
the presence of these proteins can provide valuable insights into the
biogenesis and physiological functions of EVs.^[ [58]^9 ^] Moreover,
the encapsulated RNAs within vesicles can significantly influence
recipient cells by transferring between different cell types. This
transformation may manifest as the production of novel proteins in the
case of mRNA transfer or the regulation of gene expression with
miRNAs.^[ [59]^10 ^]
EVs produced by human cells are present in various biological fluids,
facilitating the delivery of their cargoes not only to neighboring
cells within the tissue microenvironment but also over long distances
throughout the bodies of multicellular organisms.^[ [60]^11 ^] In this
context, EVs derived from endothelial cells (ECs), are of particular
interest due to their role in vascular homeostasis and their potential
as biomarkers for vascular diseases.^[ [61]^12 ^] The interaction of
endothelial EVs with target cells has various effects on cardiovascular
diseases and is dependent on the condition of the donor cells and the
molecular cargo within the EVs.^[ [62]^13 ^] However, the
characteristics and composition of EC‐derived EVs under different
physiological conditions remain less studied. Shear stress, a
mechanical force exerted by blood flow, plays a crucial role in
maintaining endothelial cell function and vascular homeostasis,^[
[63]^14 ^] and significantly influences EC behavior, including
signaling,^[ [64]^15 ^] gene expression,^[ [65]^16 ^] and cell
morphology.^[ [66]^17 ^] This force is critical for maintaining
vascular health and can vary across the vasculature in both magnitude
and pattern. It is widely recognized that atheroprotective wall shear
stress in arteries generally ranges from 10 to 40 dyn cm^− ^2.^[
[67]^18 , [68]^19 , [69]^20 , [70]^21 ^] While previous studies have
investigated how shear stress affects EC function,^[ [71]^22 ^] its
impact on EV characteristics is less explored. Therefore, understanding
the effects of shear stress on EC‐derived EVs using experimental models
is essential for elucidating the mechanisms underlying vascular health
and disease.
Cell culture supernatants are the most used source of EV isolation.^[
[72]^23 ^] Fetal calf serum (FCS) is a commonly used supplement in cell
culture media as it provides a rich source of nutrients, growth
factors, and hormones necessary for cell growth and proliferation.
However, its use in EV isolation has been a subject of controversy due
to the potential for FCS‐derived components to co‐isolate with EVs and
interfere with downstream applications.^[ [73]^24 , [74]^25 ^] To avoid
these concerns, several alternatives to FCS‐containing medium have been
proposed for EV isolation purposes, including serum‐free and
EV‐depleted FCS medium,^[ [75]^26 ^] or using supplements like
insulin‐transferrin‐selenium (ITS) solution.^[ [76]^27 , [77]^28 ^]
However, it is recommended to monitor the changes in cell behaviour and
evaluate the background of the analytes of interest to ensure that the
chosen method does not affect EV characteristics.^[ [78]^29 ^]
Here, we identified the optimal media for isolating EVs from primary
human umbilical vein endothelial cells (HUVECs), suitable for both
static and laminar flow culture conditions. To mimic physiological
conditions, we utilized a hollow fiber cartridge to apply laminar shear
stress to HUVECs. Additionally, we characterized EVs from static and
flow cultures based on morphology, particle size, and content using
miRNA sequencing and proteomics approaches. Our results suggest that
endothelial EV content differs under the regulation of laminar flow;
thus, affecting EV‐mediated mechanisms.
2. Experimental Section
2.1. Cell Culture
HUVECs were isolated from fresh umbilical cords from female individuals
(Klinikum Saarbrücken, Germany, consent of the Local Ethics Committee,
permission no. 131/08) under sterile condition using 0.1 g L^−1
collagenase for digestion (Roche) at 37 °C. To stop the digestion,
veins were rinsed with Earle`s medium 199 (PAA, # P04‐07500) containing
10% FCS (#F7524, PAA), 100 U mL^−1 penicillin G, and 100 µg mL^−1
streptomycin (#P4333). After centrifugation (10 min, 200 g) cells were
resuspended in 5 mL endothelial cell growth medium with supplement mix
(# C‐22010, Promocell) containing 10% FCS, 100 U m^−1 penicillin G, 100
µg m^−1 streptomycin, and 0.1% kanamycin (#K0254, Sigma), and
cultivated at 37 °C and 5% CO2 in a 25 cm^2 cell culture flask. After
one day, cells were washed three times with PBS (phosphate buffered
saline, 7.20 g L^−1 NaCl, 0.43 g L^−1 KH2PO4, 1.48 g L^−1 Na2HPO4) and
cultivated until confluence. Cells were cryopreserved in passage #1 and
used for further experiments.
2.2. Laminar Flow
In this work, two different systems were used to generate laminar flow.
A parallel plate flow chamber, which not only provided morphological
monitoring of the cells, also was suitable for preliminary experiments
in a small scale; and a hollow fiber cartridge for cell culture in a
larger scale for EV isolation. Details are mentioned bellow:
To assess morphology, viability, immunofluorescence, and gene
expression analysis of flow cultures, the following system was utilized
as described previously^[ [79]^30 ^] with minor modifications:
Sterilized glass slides (76 × 26 × 1 mm, Roth) were incubated for 30
min in 3 mL collagen (#11179179001, Roche) (50 µg mL^−1 in 0.2% acetic
acid) in 4‐well plates. Then, slides were washed with PBS and after
drying for 30 min, cells were seeded onto the glass slides.
HUVEC‐seeded slides were incorporated into the parallel plate flow
chambers (Figure [80]1A). The chambers were then linked to a
peristaltic pump (403U/VM purple/white, Watson Marlow) and filled with
different media (Figure [81]1B). Laminar flow rates were regulated to
fit a shear stress of 20 dynes cm^−2 and the flow was unidirectional.
Figure 1.
Figure 1
[82]Open in a new tab
A) One parallel plate flow chamber with cell‐seeded glass slide. B)
Schematic illustration of seeded glass slide connected to the
peristaltic pump used for morphology, viability, immunofluorescence,
and gene expression analysis. Created using BioRender.com.
The medium flow rate determines the degree of laminar shear stress. To
calculate the flow rate (Q) for reaching the shear stress (τ) of 20
dynes cm^−2, the following formula was used:
[MATH: τ=6Qμbh
2 :MATH]
(1)
τ = shear stress (dynes cm^−2), Q = flow rate (cm^3 s^−1), μ =
viscosity (0.01 dynes s cm^−2),^[ [83]^31 ^] b = channel width (1.9
cm), h = channel height (= thickness of the middle part of the chamber
(1.15 mm) – thickness of the glass slide).
A hollow fiber cartridge (#C2025, FiberCell system) with the FiberCell
Systems Duet Pump^[ [84]^32 ^] was used to culture HUVECs for EV
isolation experiments (Figure [85]2 ).^[ [86]^33 ^] Prior to loading
the HUVECs into the cartridge, the following preparations were
performed according to the manufacturer's instructions.
Figure 2.
Figure 2
[87]Open in a new tab
A) One individual cartridge and tubing. B) Schematic illustration of
the cartridge and its cross section used to culture HUVECs for EV
isolation experiments. Created using BioRender.com.
1. Activation: fibers were activated by injecting 70% absolute ethanol
using a luer‐lock syringe (#EP97.1, B. Braun, Germany). After ethanol
being in contact with the fibers for at least 1 min, excess ethanol was
drained, and fibers were rinsed with sterile water.
2. Coating: 1 mg mL^−1 collagen was injected into the fibers (5–10 mL)
and incubated for 30 min. Then the fibers were washed by injecting PBS.
3. Calibration: complete medium was circulated through the fibers for 1
h at 37 °C with degree 10 on the pump, while the extra capillary space
was filled with complete medium as well.
4. Seeding was performed according to the manufacturer's instructions.
Laminar flow rates were set to achieve a shear stress of 20 dynes cm^−2
according to the following formula provided by the manufacturer:
[MATH: τ=4QηπR
3 :MATH]
(2)
τ = shear stress (dynes cm^−2)
Q = fluid flow rate (mL s^−1) (per fiber)
η = viscosity (dyne s cm^−2)
R = internal radius (0.07 cm)
2.3. Morphological Assessment
HUVECs were seeded at 200 000 cells per well in a 6 well plate (2 mL
medium per well), and 500 000 cells per sterilized glass slide (76 × 26
× 1 mm, Roth) in a 4 well plate (4 mL medium per well). Cells were
incubated overnight to attach. The next day, old medium was removed and
replaced with the test medium (Table [88] 1 ) after PBS wash. Cells
were incubated for 72 h under static conditions or under 20 dynes cm^−2
shear flow. Cells under static culture were imaged with an Incucyte^®
S3 system every 24 h to monitor morphological changes. Cells under flow
condition were imaged with a digital camera (Cannon EODS 400D) attached
to a Zeiss AXIOVERT 40 CFL inverted microscope before and after
starting the flow.
Table 1.
Media options.
10% FCS medium (Co) # C‐22010, Promocell containing 10% FCS (#F7524,
PAA)
Endopan medium # P04‐0065K, PAN‐Biotech
EGM™ BulletKit™ (Lonza) # CC‐3162, Lonza
ITS solution # 41400045, Gibco™
[89]Open in a new tab
2.4. Immunofluorescence Staining
HUVEC‐seeded slides were cut with a glass cutter after the incubation
time with different media under laminar flow and used for staining. For
static culture, 50 000 HUVECs were placed in each well of an 8 well
ibidi slide that was coated with 300 µL of 50 µg mL^−1 collagen. The
cells were then incubated overnight before being washed with PBS and
exposed to different media for 72 h. Following this, the cells were
washed with 300 µl PBS and fixed with 300 µL of 1% warm
paraformaldehyde (PFA) for 15 min at room temperature. The cells were
then washed again with PBS and permeabilized by incubating for 10 min
in 300 µL of 0.1% Triton X‐100. The cells were subsequently washed with
PBS and blocked with blocking buffer (#MB‐070, Rockland) for 30 min.
300 µL medium containing antibodies against actin (#P1951, Sigma) and
von Willebrand Factor (vWF) (2 µL per well) (#AHP062F, AbD Serotec) was
used to stain the cells for 40 min. Excess antibodies were removed by
washing the cells with the same blocking buffer; after which the cells
were incubated for 10 min with 300 µL of 1µg mL^−1 Hoechst 33342
(#62249, Thermo Fisher) to stain the nucleus. HCT116 cells were used as
negative control. Finally, the cells were observed under a fluorescence
microscope (Axio Observer Z1 epifluorescence microscope, Zeiss,
Oberkochen, Germany).
2.5. Gene Expression
Total RNA was isolated using the Direct‐zolTM RNA MiniPrep Kit (#R2052,
Zymo Research). The concentration of isolated RNA was quantified by
NanoDrop™ (Thermo Fisher Scientific). Equal amounts of RNA were
transcribed using the High‐Capacity cDNA Reverse Transcription Kit
(#4368813, Thermo Fisher Scientific) in the presence of an RNase
inhibitor (#10777‐019, Invitrogen) according to the manufacturer's
instructions. qPCR was performed using a 5xHotFirePol EvaGreen qPCR Mix
(#08‐24‐00020, Solis BioDyne) and a total volume of 20 µL. The primer
sequences for each transcript are detailed in Table [90]2 . For each
primer pair, an annealing temperature of 60 °C was used (except NOS3
with 62 °C annealing temperature). The PCR was performed in a CFX96
touch™Real‐Time PCR detection system (BioRad). Data were normalized to
the beta‐actin housekeeping gene (ACTB).
Table 2.
Primer sequences used for qPCR (10 µm stock).
Gene Accession number Primer forward sequence Primer reverse sequence
ACTB [91]NM_001101.3 TGC GTG ACA TTA AGG AGA AG GTC AGG CAG CTC GTA GCT
CT
NOS3 NM_00603.4 AACCCCAAGACCTACGTGC CATGGTAACATCGCCGCAGA
ICAM1 [92]NM_000201.3 TGA CCG TGA ATG TGCTCT CC TCC CTT TTT GGG CCT GTT
GT
KLF2 [93]NM_016270.2 AGACCACGATCCTCCTTGAC AAGGCATCACAAGCCTCGAT
MMP2 [94]NM_001302510.1 CGTCGCCCATCATCAAGTTC GAAGGTGTTCAGGTATTGCACTG
DUSP1/MKP‐1 [95]NM_004417.4 GGCCATTGACTTCATAGACTCCATC
ACTCAAAGGCCTCGTCCAGC
HMOX1/HO‐1 [96]NM_002133.3 GTGCCACCAAGTTCAAGCAG GCAACTCCTCAAAGAGCTGGA
VEGFA [97]NM_001171623.1 CGCTTACTCTCACCTGCTTCTG GGTCAACCACTCACACACACAC
TSC22D3/GILZ [98]NM_004089.4 CATGTGGTTTCCGTTAAGCTGG
AGGATCTCCACCTCCTCTCTC
KLF4 [99]NM_001314052.2 TGCTCCCATCTTTCTCCACG TCCCGCCAGCGGTTATTC
NQO1 [100]NM_000903.3 CTTGTGATATTCCAGTTCCCCC GGCAGCGTAAGTGTAAGCAA
CYP1A1 [101]NM_000499.5 CATCCCCCACAGCACAACAA TACAAAGACACAACGCCCCT
CYP1B1 [102]NM_000104.4 TCCTCCTCTTCACCAGGTATCC TGGTCACCCATACAAGGCAG
PODXL [103]NM_005397.4 CCAACAAGCTCGGGACATGA TAACCGATGACGGTAGGGTG
NHERF2 [104]NM_001130012.3 GACCGGCTCATTGAGGTGAA CGAAGCCGCTTGAAGTGTTC
ADAMTS1 [105]NM_006988.5 CACAGCCCATGAATTAGGCCA ATTGACGCCATCATGTGGGA
PI16 [106]NM_153370.3 CTGACAAGCCTAGCGTCGTG GCTGACCTCTTCACCCTTTG
CMKLR1 [107]NM_004072.3 GAGGGGGATCTTGAATGAACAA GAGGCTGTTGGGGAGACTT
IGFBP5 [108]NM_000599.4 ACAAGAGAAAGCAGTGCAAACC CGTCAACGTACTCCATGCCT
CCN3 [109]NM_002514.4 GGCCTTACCCTTGCAGCTTAC TGCTGTCCACTCTGTGGTCT
APOLD1 [110]NM_001130415.2 CGCGGGGACAGAGATGTAAC GCCTCTCCATTCCCTTTCCAA
[111]Open in a new tab
2.6. Sex Determination of HUVECs
HUVECs were lysed after mixing with 1 µL of Proteinase K (#03115836001,
Roche), 5 µL of 10x Taq Buffer (#[112]E00007, Genscript), and 44 µL of
water (#A7398, AppliChem) to a total volume of 50 µL. The mixture was
then incubated in a heating block set to 55 °C for 60 min at 1500 rpm,
followed by 95 °C for 15 min. qPCR was performed as previously
described. The primer sequences are detailed in Table [113]3 .
Table 3.
Primer sequences used for HUVEC sex determination (10 µm stock).
Gene Accession number Primer forward sequence Primer reverse sequence
RPS4Y1 [114]NM_001008.4 TTTGCTCATGATTTTGGCACTGT TCCACAAAAGAATGCCGTCCT
RPS4X [115]NM_001007.5 CAGTGATTAAGTTCTCAGGCAGG CTTAACAGGGCAGAGGGGTC
[116]Open in a new tab
2.7. EV Isolation
To prepare EV‐depleted FCS, 30% FCS‐containing medium was
ultracentrifuged at 100 000 g for 18 h at 4 °C, followed by collecting
half of the supernatant and filtering through a 0.2 µm stericup filter
(Merck Millipore, Germany). The flow‐through was used to prepare 2%
EV‐depleted medium. For each biological replicate, 3 individual female
HUVECs were mixed when thawing the cryo tube from −80 °C and let grow
until confluency. For static culture, cells were seeded into three T75
flasks with 10^6 cells per flask. The next day, old medium was removed
and cells were incubated in 25 mL 2% EV‐depleted FCS medium (Promocell)
for 48 h. For flow condition, HUVECs in three T75 flasks were
trypsinised and injected (using a luer‐lock syringe) into a
collagen‐coated hollow fiber cartridge according to the protocol. Cells
were let to attach overnight with the 100 mL^−1 complete medium flowing
through ECS with degree 5 on the duet pump. The next day, the medium in
the reservoir bottle was refreshed with complete medium and the
direction of flow was connected through the fibers on the cells. The
flow was set to 5 overnight. The next day the medium was replaced with
fresh medium, and the flow was increased from 5 to 25 degree gradually
from morning to afternoon. The cells were incubated for 48 h under
laminar flow (20 dynes cm^−2). After the incubation time, conditioned
media were collected and centrifuged for 10 min at 300 g at 4 °C to
remove remaining cells and debris. The supernatant was subjected for 30
min to 10 000 g at 4 °C to remove larger particles. EVs were isolated
by ultracentrifuging for 4 h at 100 000 g at 4 °C using a 45Ti rotor
(Beckman). Due to limitations in EV purification methods, such as
sample loss, sample dilution and re‐concentration, the EV pellets were
not further purified in this work.
2.8. Nanoparticle Tracking Analysis
Particle size distribution and yield of EV preparations were analyzed
by nanoparticle tracking analyzer (NTA, LM‐10, Malvern, UK).
Preparations of EVs were diluted in 0.22 µm filtered PBS before the
analysis. A 500 µLl diluted EV sample was introduced into a green
laser‐illuminated chamber to maintain vesicle concentration within the
range of 20–120 particles/frame, and a high‐sensitivity video with
camera level 13–15 was captured; three videos of 30 s length were
recorded and processed by the NanoSight 3.1 software.
2.9. Cryo‐TEM Imaging
Cryogenic transmission electron microscopy (cryo‐TEM) was performed on
EV pellets after ultracentrifugation. Three to four microliters of the
sample were dropped onto a holey carbon grid (type S147‐4, Plano,
Wetzlar, Germany) and plotted for 2 s before plunging into liquid
ethane at T = −165 °C using a Gatan (Pleasanton, CA, USA) CP3 cryo
plunger. The sample was transferred under liquid nitrogen to a Gatan
model 914 cryo‐TEM sample holder and analyzed at −173 °C by low‐dose
TEM bright‐field imaging using a JEOL (Tokyo, Japan) JEM‐2100 LaB6 at
200 kV accelerating voltage. Images with 1024 × 1024 pixels were
acquired using a Gatan Orius SC1000 CCD camera at 2 s binning and 4 s
imaging time.
2.10. Western Blot
The EV pellets were lysed with Laemmli lysis buffer (50 mm Tris‐HCl, 1%
SDS, 10% glycerol, and 0.004% bromophenol blue). HUVECs were also
harvested in the same lysis buffer containing 1% protease inhibitors.
Samples were boiled for 9 min in 95 °C before loading to the gel. The
presence of EV markers was studied by loading equal volumes of samples
subjected to 10% sodium dodecyl sulfate‐polyacrylamide gel
electrophoresis (SDS‐PAGE) for 20 min at 90 V. Then the voltage was
increased to 110 V for another 45 min. Proteins were transferred to
polyvinylidene difluoride (PVDF) membrane (#88518, ThermoFisher), under
250 mA for 75 min in 4 °C. Following 1 h incubation in blocking buffer
(#MB144 070, Rockland) membranes were probed with primary antibodies
for CD9 (1:1000, #MA1‐80307, Thermofischer) and CD63 (1:1000, #sc‐5275,
Santa Cruz) overnight at 4 °C. Membranes were washed three times with
PBS‐0.05% Tween 20 and incubated in the dark with IRDye 800 CW goat
anti‐mouse (1:10 000, Li‐COR Biosciences) for 1 h. The blots were then
washed three times for 5 min. Bound antibody was visualized by scanning
the membrane with an Odyssey Infrared Imaging System (Li‐COR
Biosciences) in 800 nm channel. All blots were cut in order to detect
several proteins on the same blot.
2.11. Zeta Potential
The surface charge of isolated EVs was measured in triplicates for each
batch by DLS using the Zetasizer nano‐ZS (Malvern instruments,
Malvern). All samples were diluted 1:500 in 0.22 µm filtered PBS before
measurements.
2.12. RNA Sequencing
2.12.1. RNA Library Preparation
The library was prepared from static and flow EVs and their parental
HUVECs, each in three biological replicates, while each biological
replicate was a mix of three individual female donors (EVs from this
preparation were used for proteomics as well). RNAs from EVs and cells
were isolated using the miRNeasy Serum/Plasma kit (#217184, Qiagen) and
Direct‐zol^TM RNA MiniPrep Kit (#R2052, Zymo Research) respectively,
according to the manufacturer's protocols. RNA concentration was
quantified by Nanodrop spectrometer (ThermoFisher Scientific, USA) at
260 nm. Small RNA libraries were prepared according to the MGIEasy
small RNA library preparation kit (#1000005269, China). The final small
RNA libraries were sequenced by MGI Tech (China).
Libraries for RNA‐Seq were prepared with the MGIEasy rRNA depletion kit
and MGIEasy Universal Library Prep Set (MGI Tech, Shenzhen, China)
according to manufacturer's protocols. Sequencing was performed on an
DNBSEQ‐G400RS instrument by the Sequencing Unit of the Core Facility
Molecular Single Cell and Particle Analysis of Saarland University
using the 100 bp paired end sequencing strategy.
2.12.2. MiRNA Processing
Fastq sequencing files were analyzed using the miRMaster 2.0 pipeline
with default parameters as previously described^[ [117]^34 ^] and using
miRbase as reference (release 22.1). As an output, miRMaster generated
a list with the expression of all mapped miRNAs. We used our in‐house
sncRNA pipeline to normalize to rpmmm (reads per million mapped to
miRNAs) and filter miRNA based on a raw count detection of at least 5
in > 30% in each group. The normalized count matrices were used to
create PCA plots and hierarchical clusterings. Differential expressions
were calculated based on t‐tests on the normalized values with multiple
testing correction using Benjamini‐Hochberg with a threshold of false
discovery rate (FDR) < 0.05 and absolute fold‐change ≥ 2.0. Potential
miRNA targets were identified using TargetScan (Release 8.012).^[
[118]^35 ^]
2.12.3. mRNA Processing
The mRNA module from snakePipes^[ [119]^36 ^] was used for processing
paired‐end fastq files: STAR^[ [120]^37 ^] was used to align to GRChm38
p6 at the gene level, followed by RNA quantification using
FeatureCount.^[ [121]^38 ^] FastQC^[ [122]^39 ^] along with multiQC^[
[123]^40 ^] was used for quality checking. The raw count matrix was
transformed using Deseq2's^[ [124]^41 ^] variance stabilizing
transformation and the resulting matrix was used to create PCA plots
and gene expression clustering. Fold changes for differential
expression was calculated on the raw counts using Deseq2 and a
Benjamini‐Hochberg correction and a false‐discovery rate of 0.05 was
applied. RPKM values subjected to unsupervised k‐means clustering using
iDEP.96. Pathway enrichment analysis was performed using ShinyGO
0.82.^[ [125]^42 ^]
2.12.4. Integrative Analysis
MiRNA‐mRNA target pairs were obtained from TargetScan Release 8.012.^[
[126]^35 ^] Only the pairs with a weighted context++ score about the
75th percentile was kept. Pearson's correlation was calculated using
rpmmm values for miRNA and rpkm values for mRNA for matched samples.
P‐values were adjusted with the Benjamini‐Hochberg correction with an
FDR threshold 0.05.
2.13. Proteomics
EVs from three independent preparations were analyzed. 88 micrograms of
EV protein were precipitated by trichloroacetic acid (TCA)
precipitation with an end concentration of 20% TCA. Samples were washed
thrice with acetone. After a final centrifugation of 15 min in a
SeedVac Plus concentrator (Savant, Thermo Fisher, Waltham, USA),
samples were resuspended in 2x Laemmli buffer (4% SDS, 20% glycerol,
120 mm Tris‐HCl (pH 6.8), 0.02% bromophenol blue in Millipore water)
and denatured at 95 °C for 5 min. Proteins were separated on NuPAGE®
4%–12% gradient gels (ThermoFisher Scientific, Karlsruhe, Germany)
until the bromophenol dye front reached the center of the gel. Proteins
were fixed in the presence of 10% acetic acid /40% ethanol and
visualized with colloidal Coomassie stain (10% (v/v) phosphoric acid,
10% (w/v) ammonium sulfate, 20% (v/v) methanol, and 0.12% (w/v)
Coomassie G‐250). Six gel pieces were cut/ cell lysate, washed,
reduced, carbamidomethylated, and trypsin digested as described before
(Fecher‐Trost et al. 2013). After extraction, 6 µl of tryptic peptides
were analyzed by data‐dependent nano‐LC‐ESI‐HR‐MS/MS analysis using the
instrument setup: Ultimate 3000 RSLC nano system equipped with an
Ultimate3000 RS autosampler and Nanospray Flex NG ion source coupled to
an Orbitrap Eclipse Tribrid mass spectrometer (Thermo Scientific,
Germany). Peptides were separated with a gradient generated with buffer
A (water and 0.1% formic acid) and buffer B (90% acetonitrile and 0.%
formic acid) at a flow rate of 300 nL min^−1: 0–5 min 4% B, 5–80 min to
31% B, 80–95 min to 50% B, 95–100 min to 90% B, 100–105 min hold 90% B,
105–106 min to 4% B and 106–120 min to 4% B. Peptides were trapped on a
C18 trap column (75 µm × 2 cm, Acclaim PepMap100C18, 3 µm,) and
separated on a reverse phase column (nano viper Acclaim PepMap
capillary column, C18; 2 µm; 75 µm × 50 cm,). The effluent was sprayed
into the mass spectrometer using a coated emitter (PicoTipEmitter, 30
µm, New Objective, Woburn, MA, USA, ionization energy: 2.4 keV). MS^[
[127]^1 ^] peptide spectra were acquired using the Orbitrap analyzer (R
= 120k, RF lens = 30% m/z = 375‐1500, MaxIT: auto, profile data,
intensity threshold of 10^4). Dynamic exclusion of the 10 most abundant
peptides was performed for 60 s. MS^[ [128]^2 ^] spectra were collected
in the linear ion trap (isolation mode: quadrupole, isolation window:
1.2, activation: HCD, HCD collision energy: 30%, scan rate: fast, data
type: centroid).
Peptides and fragments were analyzed using the MASCOT algorithm and TF
Proteome Discoverer (PD) 1.4 software (ThermoFisher, Waltham, USA).
Therefore, peptides were matched to tandem mass spectra by Mascot
version 2.4.0 by searching of a SwissProt database (2021_05, number of
protein sequences for all taxonomies: 564.638, for taxonomy human:
20.397). Peptides were analysed with the following mass tolerances:
peptide tolerance: 10 ppm, fragment tolerance: 0.7 D. The workflow
included tryptic digest and up to two missed cleavage sites. Cysteine
carbamidomethylation was set as a fixed modification and deamidation of
asparagine and glutamine, acetylation of lysine and N‐term and
oxidation of methionine were set as variable modifications. The PD
output files were loaded in the software Scaffold (5, Proteome
SoftwareInc., Portland, OR, USA). The identification of two unique
peptides per protein was set as the minimum for protein identification.
2.14. Statistical Analysis
GraphPad Prism 9 software (GraphPad, USA) was used for data analysis.
Shapiro‐Wilk test was performed to analyze the data distribution. For
normally distributed data, means of two groups were compared with
Student's t‐test. For group analysis, one‐way analysis of variance
(ANOVA) followed by Dunnett's post hoc test was applied to compare
every mean with the mean of control group. All data are presented as
mean ± SD, and p < 0.05 was considered significant. ^* p < 0.05, ^**
p < 0.01, ^*** p < 0.001. Schematic illustration were made using
BioRender.com.
3. Results
3.1. Finding the Optimum Medium for EV Isolation
The experiments involving different media were conducted at various
times (chronologically), with some options being introduced during
later phases of the study. Consequently, not all experiments in this
section included all media. Table [129]4 describes the options and the
experiments in which they were investigated.
Table 4.
Media options and experiments.
Options Morphology Endothelial characteristics (vWF) Gene expression EV
production RNA yield
10% FCS medium (Co)
* ✓
* ✓
* ✓
* ✓
x
2% EV‐depleted FCS
* ✓
* ✓
* ✓
* ✓
* ✓
10% EV‐depleted FCS
* ✓
* ✓
* ✓
* ✓
* ✓
Endopan medium
* ✓
* ✓
* ✓
* ✓
x
EGM™ BulletKit™ (Lonza)
* ✓
x x
* ✓
x
ITS‐supplemented medium
* ✓
x x x x
FCS‐free medium
* ✓
x x x x
[130]Open in a new tab
3.1.1. Cell Morphology
The experiment on morphology began by culturing HUVECs under static
conditions with the hypothesis that if the cells remained stable in
static culture first, then they could be examined under flow. HUVECs
were subjected to various media for 72 h, revealing normal morphology
in 10% and 2% EV‐depleted FCS medium, Endopan medium, and Lonza medium
(Figure [131]3 ). However, when grown in ITS‐containing medium and
serum‐free medium, some cells were found to be partly detached.
Consequently, the first four media were selected to be tested under
flow conditions, revealing normal elongation of the cells in the
direction of the flow for both EV‐depleted FCS media and Endopan
medium, while cells grown in Lonza medium detached under flow.
Figure 3.
Figure 3
[132]Open in a new tab
Morphology of HUVECs after 72 h culture in different media under static
and 20 dynes cm‐2 flow conditions using the parallel flow chamber.
Scale bar = 100 µm. Cells were a mix of two HUVEC donors with unknown
sex, conducted in two independent experiments, each including one
technical replicate.
3.1.2. Von Willebrand Factor
To make sure HUVECs keep their endothelial characteristics, we
investigated the presence of von Willebrand Factor as an endothelial
marker after incubation with different media under static and flow
culture conditions (Figure [133]4 ). The immunofluorescent staining
detected the presence of vWF in HUVECs cultured in complete (Co),
Endopan, 10%, and 2% EV‐depleted FCS medium in both culture conditions.
Figure 4.
Figure 4
[134]Open in a new tab
Fluorescence microscopy images of HUVECs cultured under static and
laminar flow conditions (20 dynes cm^−2, using the parallel flow
chamber) after 72 h. HCT116 cells were used as negative control. Blue:
Hoechst, red: Actin, green: von Willebrand factor. Scale bar = 50 µm.
Cells were mix of HUVEC donors, conducted in two experiments, including
two technical replicates.
Quantification of immunofluorescence signal intensities across multiple
male and one female donor revealed stable vWF protein expression under
both static and flow conditions. In addition, transcriptomic analysis
confirmed that vWF mRNA levels remained unchanged under shear stress
(data not shown).
3.1.3. Gene Expression
Additional investigations were conducted using qPCR to evaluate the
impact of various media on HUVECs on the expression of genes known to
be altered upon laminar flow. This aimed to identify a medium that
exhibits the least deviation in gene expression compared to the
complete medium. Since laminar flow modulates the expression of
adhesion molecules and anti‐inflammatory factors,^[ [135]^30 , [136]^43
^] the expression of relevant genes was examined in HUVECs cultured
under laminar flow relative to static cultures. The data indicate a
shift in gene expression that closely resembles the control condition
when using a medium containing 2% EV‐depleted FCS medium (Figure [137]5
).
Figure 5.
Figure 5
[138]Open in a new tab
Gene expression of HUVECs incubated with different media under laminar
flow conditions using the parallel flow chamber (20 dynes cm^−2) for 72
h. Data are normalised to static culture as control (dashed line), and
shown as mean ± SD. Cells were a mix of two HUVEC donors with unknown
sex. Dots show biological replicates, and each dot is the average of
three technical replicates. Means of two groups were compared with
Student's t‐test. For group analysis, one‐way analysis of variance
(ANOVA) followed by Dunnett's post hoc test was applied to compare
every mean with the mean of control group. # shows significant
differences between groups. * indicates significant differences
compared to the control (Co, indicated with the dashed line). p < 0.05
is considered significant. ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001.
3.1.4. RNA Yield
Up until this point, the initial flow culture experiments were
conducted using a parallel flow chamber. However, for large‐scale EV
isolation, we needed to use a hollow fiber cartridge to culture HUVECs
under flow. Before proceeding with large‐scale EV collection, we first
needed to optimize the hollow fiber system. Given that the hollow fiber
cartridge functions as a closed system, we aimed to ensure cell
stability following the incubation period in the 2% EV‐depleted FCS
medium that was suggested to be suitable based on microscopic analyses
and qPCR data. Attempts to image cells adhered to the fibers using
scanning electron microscopy (SEM) were unsuccessful due to limitations
in accessing the fibers. Consequently, our alternative approach
involved assessing the RNA concentration of the cells. We hypothesized
that if the cells remained adherent throughout the incubation period,
it should be possible to isolate RNA in a concentration within an
acceptable range relative to the initial cell seeding number. The RNA
was less concentrated when incubated longer (72 h) in the low serum
medium (2% EV‐depleted FCS medium), while the RNA extracted after
shorter (48 h) incubation in the same medium had a higher concentration
(n = 1, cells were a mix of 4 female HUVEC donors) (Table [139] 5 ).
Table 5.
RNA concentration of HUVECs.
Medium Flow incubation time h RNA C. ng µL^−1
2% EV‐depleted FCS medium 72 19.5
2% EV‐depleted FCS medium 48 67.5
[140]Open in a new tab
3.1.5. RNA‐Seq
Since the 2% EV‐depleted medium suggested to be suitable, we conducted
RNA‐Seq with HUVECs under static and flow conditions using the hollow
fiber cartridge. Principal component analysis (PCA) revealed distinct
clustering patterns driven by culture conditions (Figure [141]6A). This
separation was further confirmed by hierarchical clustering of the 2000
most variable genes, which grouped the samples according to their
respective culture conditions (Figure [142]6B).
Figure 6.
Figure 6
[143]Open in a new tab
Transcriptomic profiling of female HUVECs under static and flow
conditions using the hollow fiber cartridge (n = 3). A) Principal
component analysis (PCA) of gene expression data. B) Heatmap of the
2000 most variable genes, clustered using hierarchical clustering
(Euclidean distance, average linkage). Expression levels are displayed
as log10‐transformed RPKM values, with genes centered by subtracting
the mean expression across samples. C) Volcano plot of differentially
expressed genes (DEGs) between flow and static conditions. Red and blue
dots represent significantly upregulated (759) and downregulated (608)
genes under flow conditions, respectively (adjusted p‐value < 0.05,
fold change > 2). D–G) Pathway enrichment analysis. Selected pathways
are shown; see Table [144]S1 (Supporting Information) for the full
list. D,E) KEGG pathway enrichment analysis of upregulated D) and
downregulated E) DEGs under flow conditions. F,G) GO term enrichment
analysis of upregulated F) and downregulated G) DEGs under flow
conditions.
Differential gene expression analysis identified 1,367 DEGs (adjusted
p‐value < 0.05, fold change > 2), with 759 genes upregulated and 608
downregulated under flow conditions (Figure [145]6C; Table [146]S1,
Supporting Information). Pathway enrichment analysis showed that
upregulated genes under flow were mainly involved in
mechanotransduction, metabolism, and cellular signaling, with a
significant enrichment in the KEGG “Fluid Shear Stress and
Atherosclerosis” pathway, confirming that the cells exhibit a
well‐characterized endothelial response to flow (Figure [147]6D; Figure
[148]S1, Supporting Information). Conversely, downregulated genes were
associated with cell cycle regulation, DNA replication, and repair
processes, indicating a shift toward a quiescent endothelial state
(Figure [149]6E).
GO term analysis further supported these findings, with upregulated
genes linked to extracellular matrix remodeling, adhesion, and
migration, while downregulated genes were enriched in cell division and
chromatin organization, reflecting the reduced proliferative activity
under shear stress (Figures [150]6F,G).
To further explore gene expression patterns, we analyzed the 2000 most
variable genes using k‐means clustering (Figure [151]S2 and Table
[152]S1, Supporting Information), identifying four distinct gene
clusters (A–D): Cluster A showed strong upregulation, Cluster B
moderate upregulation, Cluster C moderate downregulation, and Cluster D
strong downregulation under flow conditions. The functional pathways
associated with Clusters A, B, and D closely mirrored those identified
in the DEG analysis, reinforcing the observed mechanotransduction and
quiescence‐associated signatures. Interestingly, Cluster C was enriched
for pathways related to RNA processing, translation, and ribosome
function, suggesting a broader suppression of biosynthetic activity.
qPCR analysis confirmed the upregulation of shear‐responsive markers in
both the parallel flow chamber and hollow fiber cartridge systems,
demonstrating consistent gene expression patterns under physiological
laminar flow conditions (Figure [153]7 ).
Figure 7.
Figure 7
[154]Open in a new tab
qPCR validation of selected marker genes for laminar shear stress in
female HUVECs. Values are shown as x‐fold of static controls (n = 3).
3.1.6. EV Surface Markers
Since 48 h culture under laminar flow resulted in higher amounts of
isolated RNA, we proceeded with large‐scale EV isolation from flow
cultures and performed western blot analysis with EVs isolated from
conditioned media of cells used for RNA yield analysis to investigate
whether lower incubation time would affect EV markers as well. Western
blot analysis showed that CD63 and CD9 EV markers are detectable in
samples after 48 h under static and flow cultures when 2% EV‐depleted
FCS medium is used (Figure [155]8 ). These data on EVs obtained from
cell culture supernatants after 48 h were in line with surface markers
on EVs obtained after 72 h of culture. Interestingly, CD63 was not
detectable in serum‐rich media. Given that our data confirmed the
suitability of 2% EV‐depleted medium, we did not further investigate
this observation (Figure [156]S3, Supporting Information).
Figure 8.
Figure 8
[157]Open in a new tab
HUVEC EV marker analysis of static and flow cultures after 48 h. Mix of
4 female HUVEC donors was cultured in 2% EV‐depleted FCS medium (under
static, and under 20 dynes cm^−2 laminar flow using the hollow fiber
cartridge for 48 h). Presence of EV markers (CD63, CD9) was examined
using western blot. 30 µL (30 µg protein) of EVs were loaded into each
pocket (n = 1).
Taken together, we decided to culture the cells for 48 h in 2%
EV‐depleted FCS medium under static and laminar flow conditions for
further EV sample collection and analysis.
3.2. HUVEC EV Isolation and Characterization Obtained from Static and Laminar
Flow Cultures
3.2.1. EV Characterization
Having identified the optimal medium for EV isolation suitable for both
static and flow conditions, we proceeded with the main EV sample
collection of both EV types with three biological replicates (while
each biological replicate was a mix of three individual female donors),
and their characterization. EVs were isolated by ultracentrifugation
from cell culture supernatants from HUVECs cultured in 2% EV‐depleted
FCS medium under static and laminar flow conditions (20 dynes cm^−2)
for 48 h. The concentration of EVs was determined using NTA, revealing
an average of 2.44 × 10^12 ± 0.71 × 10^12 particles per milliliter for
static EVs and 2.29 × 10^12 ± 0.54 × 10^12 particles per milliliter for
flow EVs. Furthermore, NTA showed 129 ± 3 and 134 ± 9 nm for the mode
size of static and flow EVs, respectively (Figure [158]9A,B). The
morphology of the EVs was then verified through cryo‐TEM, which
confirmed their spherical structure for both EV types
(Figure [159]9C,D). The zeta potential of the vesicles was negative,
averaging from −10.9 ± 1.12 mV for static EVs to −10.2 ± 0.77 mV for
flow EVs (Figure [160]9E). The average protein concentration was
significantly higher in static EVs (Figure [161]9F) (Table [162] 6 ).
Figure 9.
Figure 9
[163]Open in a new tab
HUVEC EV characterization isolated from static and flow cultures using
the hollow fiber cartridge. EVs were isolated using UC from HUVECs
cultured in 2% EV‐depleted FCS medium under static and laminar flow
conditions (20 dynes cm^−2) for 48 h. A,B) Representative size
distribution of particles by NanoSight particle tracking analysis of
static and flow EVs, respectively. C,D) Representative cryo‐TEM images
of static and flow EVs, respectively, scale bar = 200 nm. E) Zeta
potential of the vesicles (n = three biological replicates, each
replicate is a mix of three HUVEC female donors). F) Protein
concentration of isolated EVs was assessed by BCA assay (n = six
biological replicates, each replicate was a mix of three HUVEC female
donors). Statistical differences were analyzed by Student's t‐test. ^*
p < 0.05.
Table 6.
HUVEC EV characterization isolated from static and flow cultures from
three individual EV isolations each measured in triplicates.
Static EVs Flow EVs
Particle c.
(particles per milliliter)
2.44 × 10^12 ± 0.71 × 10^12 2.29 × 10^12 ± 0.54 × 10^12
Size
(nm)
129 ± 3 134 ± 9
Zeta potential
(mv)
−10.9 ± 1.12 −10.2 ± 0.77
Protein c.
(mg mL^−1)
4.93 ± 1.8 3.1 ± 0.87
[164]Open in a new tab
3.2.2. miRNA‐Seq Analysis of Cells and EVs
To investigate the impact of shear stress on miRNA expression, we
performed miRNA sequencing on HUVECs and their released EVs under
static and flow conditions. Principal component analysis (PCA) revealed
a clear separation between static and flow cells, while EVs clustered
separately from cells but did not exhibit distinct grouping based on
culture conditions (Figure [165] 10A). Hierarchical clustering of the
1,000 most variable miRNAs confirmed this pattern (Figure [166]10B).
Figure 10.
Figure 10
[167]Open in a new tab
miRNA sequencing analysis of HUVECs and EVs under static and flow
conditions using the hollow fiber cartridge (n = 3, female donors). A)
Principal component analysis (PCA) of miRNA expression data from cells
and EVs. B) Heatmap of the 1,000 most variable miRNAs, clustered using
hierarchical clustering (Euclidean distance, average linkage). SC:
static cells, FC: flow cells, SEV: static EVs, FEV: flow EVs. Numbers
indicate biological replicates. Values represent log10 rpmmm values. C)
Volcano plot of differentially expressed miRNAs in cells (adjusted
p‐value < 0.05, fold change > 2). D,E) Inverse correlation between
differentially expressed miRNAs and target mRNAs in selected pathways.
Colors indicate condition: violet for static, orange for flow. F,G) GO
term enrichment analysis of differentially expressed mRNAs with inverse
miRNA regulation.
Differential expression analysis identified 36 miRNAs significantly
regulated by flow in cells (fold change > 2, adjusted p‐value < 0.05),
with 20 miRNAs upregulated and 16 downregulated (Figure [168]10C;
Table [169]S2, Supporting Information). However, in EVs, no significant
differences were observed between static and flow conditions (Table
[170]S2, Supporting Information), suggesting that flow‐induced miRNA
changes occur primarily within cells and are not reflected in the EV
cargo.
Correlation analysis revealed 409 miRNA‐mRNA pairs with negative
correlation and 388 with positive correlation. Among the negatively
correlated interactions, 144 were associated with genes upregulated
under flow conditions, while 96 involved downregulated genes (Table
[171]S2, Supporting Information). Flow‐downregulated miRNAs correlated
with mRNAs involved in the KEGG “Fluid Shear Stress and
Atherosclerosis” pathway, while flow‐upregulated miRNAs correlated with
genes linked to cell cycle regulation. Examples of such inverse
interactions are shown in Figure [172]10D,E, while the complete dataset
is provided in Table [173]S2 (Supporting Information).
As KEGG pathway analysis yielded limited results for predicted miRNA
targets, we used GO Biological Process (GOBP) enrichment analysis to
highlight functional categories. Upregulated mRNAs (with corresponding
downregulated miRNAs under flow) were associated with vascular
remodelling and endothelial function, including blood vessel
morphogenesis, cell migration, and anatomical structure morphogenesis
(Figure [174]10F). In contrast, downregulated mRNAs (with upregulated
miRNAs under flow) were enriched in processes related to cell cycle
progression, cytoskeleton organization, and spindle assembly
(Figure [175]10G).
Together, these results indicate that shear stress modulates
endothelial miRNA expression, with key miRNAs potentially regulating
pathways involved in vascular adaptation and cell cycle control, while
EV‐associated miRNA cargo remains unchanged under flow conditions.
3.2.3. Enrichment of Specific miRNAs in EVs Compared to Parental Cells
To identify miRNAs selectively enriched in EVs, we compared miRNA
expression levels between EVs and their parental HUVECs under static
and flow conditions. A total of 48 miRNAs were more abundant in EVs
than in cells under at least one condition (static, flow, or both)
(Table [176]S2, Supporting Information). These miRNAs were visualized
in a heatmap (Figure [177]11A).
Figure 11.
Figure 11
[178]Open in a new tab
Enrichment of specific miRNAs in EVs compared to parental cells (n = 3,
female donors). A) Heatmap of miRNAs enriched in EVs compared to their
parental cells under static or flow conditions. Displayed are 48 miRNAs
that were more abundant in EVs in at least one condition. Values
represent log[10] rpmmm values. B,C) Fold enrichment analysis of miRNAs
in EVs compared to cells, plotted against statistical significance
(p‐value), for static B) and flow C) conditions. Only miRNAs with fold
change > 2 and p < 0.05 are shown. miRNAs with p < 0.01 are highlighted
in a darker shade. The most highly abundant miRNAs in EVs (miR‐451a,
miR‐122‐5p, miR‐9‐5p, see D) are marked in red. D) Scatter plot of
miRNA abundance and enrichment in EVs compared to cells. log₂ fold
change (EVs vs. cells) is plotted against rpmmm values for both static
and flow conditions. Only miRNAs with fold change > 2 in both
conditions are shown. The three most abundant miRNAs in EVs (miR‐451a,
miR‐122‐5p, and miR‐9‐5p) are highlighted. E) KEGG pathway enrichment
analysis of predicted target genes of the three most abundant
EV‐miRNAs.
Fold enrichment analysis confirmed these findings by identifying miRNAs
significantly enriched in EVs under static (Figure [179]11B) and flow
(Figure [180]11C) conditions. While several miRNAs showed strong fold
enrichment in EVs compared to their parental cells, miR‐451a,
miR‐122‐5p, and miR‐9‐5p stood out due to their exceptionally high
absolute abundance in EVs (Figure [181]11D). These miRNAs were not
necessarily the most differentially enriched compared to cells, but
their high presence suggests preferential loading into EVs.
To further characterize the potential impact of these highly abundant
miRNAs, we performed KEGG pathway enrichment analysis for their
predicted target genes. The analysis identified pathways involved in
signal transduction, cell adhesion, immune response, and metabolism
(Figure [182]11E; Table [183]S2, Supporting Information). Given that
EVs can be taken up by different recipient cells, the functional
consequences of these miRNAs may vary depending on the target cell
type. For instance, in endothelial cells, these EV‐miRNAs could
modulate vascular signaling and barrier integrity, while in immune
cells, they may influence inflammatory pathways. Similarly, in
metabolic tissues, they could affect energy homeostasis and glucose
metabolism. These findings suggest that EV‐associated miRNAs have the
potential to fine‐tune diverse cellular processes depending on the
microenvironment and recipient cell type.
3.2.4. Proteomics Analysis of Static and Flow EVs
A total of 3268 proteins were detected including 664 proteins unique in
static EVs and 520 proteins in flow EVs with 2084 proteins common in
both types (Figure [184]12A; Table [185]S3, Supporting Information).
PCA revealed a distinct separation between static and flow EVs, with
biological replicates within each category demonstrating similarity
(Figure [186]12B). Fold changes were calculated using the unique
spectrum counts of flow EVs/static EVs. Figure [187]12C illustrates the
differentially expressed proteins in a volcano plot, i.e. log[2] fold
change was plotted against −log[10] p‐value. Negative log[2] fold
change values represent proteins more abundant in static EVs, whereas
positive values represent abundant proteins in flow EVs. Cellular
component analysis showed that the significantly enriched proteins in
both EV types are annotated with exosomal and cytosolic spaces (Figure
[188]S4, Supporting Information). Interestingly, within the significant
gene ontology (GO) cellular component terms of flow EV proteins,
mitochondrial origin was also observed. Next, we analyzed the
biological processes that these significant proteins are associated
with. Figure [189]12D shows that the proteins significantly enriched in
flow EVs play a role in localization, transport, and respiration.
Biological processes associated with enriched proteins in static EVs
are shown in Figure [190]12E, suggesting a role in cellular metabolism,
and translation. Based on the mitochondrial origin of flow EV‐enriched
proteins as indicated by cellular component terms, and considering
their involvement in cellular respiration, proton transport, and energy
processes, we conducted a detailed analysis of these proteins.
Specifically, we examined their abundance and presence in static EVs as
well. Figure [191]12F illustrates the unique spectrum counts of
mitochondrial proteins significantly present in flow EVs, alongside
their counts in static EVs. Notably, it demonstrates either an absence
or reduced presence of mitochondrial proteins in static EVs. Gene
ontology analysis also showed involvement of the mitochondrial proteins
in biological processes, such as respiration, oxidative
phosphorylation, and ion transport (Figure [192]12G).
Figure 12.
Proteomics data of static and flow EVs. A) Number of detected proteins.
B) PCA shows a clear distinction between static and flow EVs (n = 3).
C) Volcano plot representing the differential enrichment between the
two EV types. Log[2] fold change (1.5) is plotted against −log[10]
p‐value (0.05). Top 20 gene ontology (GO) biological processes for
proteins significantly enriched in D) flow EVs and E) static EVs
according to the STRING database. F) Abundant mitochondrial proteins in
static and flow EVs and their distribution. Exclusive unique spectrum
count raw data are shown for all three independent preparations per
condition (S: static EVs, F: flow EVs). G) Top 20 gene ontology (GO)
biological processes for mitochondrial proteins significantly enriched
in flow EVs according to the STRING database. N = three biological
replicates, each replicate is a mix of three HUVEC female donors.
graphic file with name SMTD-9-2401841-g009.jpg
graphic file with name SMTD-9-2401841-g004.jpg
graphic file with name SMTD-9-2401841-g002.jpg
[193]Open in a new tab
Exclusive, unique spectrum count raw data of a series of EV marker
proteins^[ [194]^1 , [195]^44 , [196]^45 ^] are shown in Figure [197]S5
(Supporting Information) for the independent preparations per
condition. Overall, the EV‐specific protein distribution was quite
similar in both conditions. Only milk fat globule‐epidermal growth
factor‐factor 8 (MFG‐E8) was highly expressed in static EVs compared to
the low expression in flow EVs.
4. Discussion
In the first step, it was necessary to find an approach to prevent
FCS‐derived EV contaminants.^[ [198]^29 ^] Although some protocols
simply proceed with serum‐free medium for EV isolation from human cell
lines,^[ [199]^46 , [200]^47 ^] the utilization of primary endothelial
cells in this work, which were intended to be cultured under flow
conditions, prevented us from removing FCS from our setting. During the
primary setup experiments, these cells were observed to be detached
when grown in FCS‐free medium under static culture conditions, leading
us to conclude that they would not maintain adherence under the
mechanical force of shear flow in serum‐free medium. Consequently, our
approach involved an effort to deplete EVs from FCS, aiming to address
this critical aspect of our experimental setup.
Shelke et al. compared the centrifugation of FCS for a short (1.5 h)
and a long period (18 h) to test the efficiency of these two EV
depletion protocols. They found that 18 h centrifugation reduced
FCS‐derived EV RNA content by 95%; however, it does not completely
eliminate EV contaminants from FCS.^[ [201]^48 ^] Later, a study on the
effects of serum dilution on the depletion efficiency suggested that
the amount of RNA in the EV‐depleted supernatant was reduced in diluted
FCS compared to non‐diluted condition, and thus recommended to dilute
the FCS to 30% prior to EV depletion.^[ [202]^49 ^] Therefore, in this
study, a medium containing 30% FCS was ultracentrifuged and then
utilized to formulate the primary culture medium for EV production
during the incubation period.
One study on the impact of different media on EV production has
previously reported that EVs produced from N2a mouse neuroblastoma
cells in Opti‐MEM (reduced‐serum medium) were greater in quantity than
EVs produced in DMEM‐containing serum.^[ [203]^50 ^] Later, the same
group attempted to identify specific media components affecting EV
production. They found higher levels of EV surface markers (CD9, CD63,
and CD81) from HEK293T cells cultured in serum‐free Opti‐MEM compared
to serum‐including conditions. Interestingly, a CD81 + EV population
was not detectable by western blot analysis when complete medium was
used to harvest EVs.^[ [204]^51 ^] Also comparing the enrichment levels
of genes comprising a certain gene ontology term between the different
media conditions, in which cells were cultured for EV production, Bost
et al. found that the sphingolipid and ceramide pathways influencing
exosome production,^[ [205]^52 ^] were upregulated in the Opti‐MEM
samples compared to the serum‐containing media. CD63 functions in
ESCRT‐independent vesicle formation,^[ [206]^53 ^] and
ESCRT‐independent exosome formation relies on ceramide generation by
neutral sphingomyelinase.^[ [207]^54 ^] This could explain the presence
of CD63 marker in HUVEC‐derived EVs when low serum amount was employed.
In this work, in addition to static culture condition, we also
characterized vesicles isolated from HUVECs subjected to laminar flow
trying to simulate the physiological conditions. We utilized two
systems to model laminar shear flow. Our data demonstrated strong
similarity in gene expression between the two approaches showing
upregulation of flow‐induced genes including KLF2—a key master
regulator of the shear stress response that governs the expression of
≈70% of flow‐responsive genes.^[ [208]^55 ^] These findings validate
the effectiveness of both systems in accurately simulating
physiological shear conditions. Several fluid shear stress models have
been used in the literature. Parallel‐plate flow chambers like the one
we used for set up experiments allow the cell layer to be observed with
a microscope.^[ [209]^17 ^] Cone‐and‐plate systems are used to analyze
the shear responses of cells to flow independent of hydrostatic
pressure.^[ [210]^56 ^] The orbital shaker method is able to generate a
larger disturbed flow.^[ [211]^57 ^] In recent years, microfluidic
systems have been often, allowing the creation of constant or active
shear flow with external equipment, like pumps, which dynamically
adjust fluid shear stress by altering the inlet flow.^[ [212]^58 ,
[213]^59 , [214]^60 ^] However, the choice of a specific model depends
on the downstream analysis requirements. Here we used a hollow fiber
cartridge system^[ [215]^33 ^] that allowed for larger‐scale cell
cultivation compared to other commercially available in vitro settings.
This made it possible to isolate EVs from a large volume of conditioned
medium required for downstream processing; therefore, reducing the
number of batches needed for multiple analysis and improving the
consistency of the data generated.
Commonly used EV isolation methods including ultracentrifugation,
density gradient centrifugation, size exclusion chromatography, and
polymer‐based precipitation, vary in EV yield, the depletion of protein
contaminants, labour‐intensity, and cost of the procedure. Utilizing a
combination of two or more methods has the potential to enhance the
removal of protein contaminants; however, it comes at the cost of
reducing the overall number of EVs.^[ [216]^61 ^] Therefore, the choice
of EV isolation method used should depend on the amount of starting
material together with the downstream application. Although commercial
EV separation kits have been used to isolate EVs from HUVECs,^[
[217]^62 , [218]^63 ^] differential centrifugation has been the most
widely used method,^[ [219]^62 , [220]^64 , [221]^65 , [222]^66 ,
[223]^67 , [224]^68 , [225]^69 ^] In our research, we isolated vesicles
from the culture medium using ultracentrifugation, without additional
purification steps. This decision was due to the noticeable sample loss
observed during trial runs of size exclusion chromatography to purify
the isolated EVs (Figure [226]S6, Supporting Information).^[ [227]^70
^]
Definitive characterization of biogenesis‐based EV subtypes is
challenging, as there are no universal molecular markers for ectosomes
(also known as microvesicle or microparticle; refers to EVs originating
from the cell surface), exosomes (refers to EVs originating from
internal compartments of the cell, released via MVBs), or other EV
subtypes.^[ [228]^71 ^] In our work, we examined a series of EV protein
markers based on previous reports,^[ [229]^1 , [230]^44 , [231]^45 ^]
irrespective of the biogenesis routes. A genome‐wide association study
for coronary artery disease involving over a million participants
identified MFG‐E8 as one of the risk variants and genes associated with
cardiovascular diseases,^[ [232]^72 ^] positioning it as a potential
prognostic biomarker for vascular diseases.^[ [233]^73 ^] An in vivo
study on endothelial–vascular smooth muscle cell (VSMC) interactions in
mice further highlighted the role of MFG‐E8 in driving the
pro‐inflammatory phenotypic shift of VSMCs.^[ [234]^74 ^] Dysregulated
EC‐VSMC communication was shown to potentially contribute to the
development of atherosclerosis.^[ [235]^75 ^] Among the more abundant
proteins in flow EVs, we observed a variety of mitochondrial proteins
that were either absent or less prominent in static EVs. This
observation aligns with previous reports documenting the presence of
mitochondrial proteins in EVs from mouse embryonic fibroblasts and
monocyte‐derived dendritic cells.^[ [236]^76 , [237]^77 ^] Vascular
endothelial cells sense shear stress generated by flowing blood and
transmit this information into the cell interior.^[ [238]^78 ^]
Previous data have shown a role of mitochondria in the EC
mechanotransduction of fluid shear stress.^[ [239]^79 , [240]^80 ^] A
recent study suggests that changes in the magnitude and pattern of
fluid shear stress alter the mitochondrial content, shape, and
intracellular distribution in different vessel regions of a mouse model
in vivo and in primary mouse aortic endothelial cells in vitro.^[
[241]^81 ^] It has been shown that unidirectional flow induces an
elevation of oxidative phosphorylation‐dependent ATP generation.^[
[242]^82 , [243]^83 , [244]^84 ^] On the other hand, exposing HUVECs to
laminar flow (20 dynes cm^−2) for 24 h decreases glycolysis pathway.^[
[245]^85 ^] In line with these findings, we saw an increase in ATP
synthase subunits (ATP5MF, ATP5F1A, ATP5F1B, ATP5ME, ATP5PB) and other
respiratory chain members (CYC1 (Cytochrome c1. heme protein), UQCRC1
(Cytochrome b‐c1 complex subunit 1), and COX4I1 (Cytochrome c oxidase
subunit 4 isoform 1)) in flow EVs. Furthermore, PDP1 (Pyruvate
dehydrogenase phosphatase 1), a mediator of glycolysis pathway,^[
[246]^86 ^] was not detected in any of the flow EV replicates.
Cellular culture conditions are not only reflected in exosomal proteins
but also in miRNA contents. Intracellular miRNA expression profiles of
ECs adapt to diverse flow patterns and impact endothelial biology.^[
[247]^87 , [248]^88 , [249]^89 , [250]^90 ^] Therefore, we hypothesized
that the miRNA content of ECs is also regulated by shear stress. To
test this, we performed miRNA sequencing with EVs and cells. A recent
study compared extracellular vesicles (EVs) from HUVECs under static
and laminar flow conditions using a parallel plate flow chamber with a
shear stress of 15 dyn cm^− ^2 for 8 h.^[ [251]^91 ^] In addition to
technical differences between this study (employing 10% EV‐depleted
FBS) and ours, such as the choice of flow model and flow duration for
EV isolation, the authors observed differences in the miRNA profiles of
EVs from the two conditions. To validate their flow model, they noted
changes in inflammatory gene expression; however, the factors they
studied were not among the well‐established flow‐induced
transcriptional patterns.^[ [252]^16 ^] Although we did not observe
significant changes in the miRNA content between EVs from static and
flow conditions, we did find an increased abundance of specific miRNAs
in the EVs compared to the parental cells. MicroRNA‐122‐5p, has been
implicated in various cardiovascular diseases. Studies have shown that
miR‐122‐5p is upregulated in patients with both stable and unstable
coronary artery disease, suggesting its potential role as a biomarker
for plaque instability.^[ [253]^92 ^] MiR‐122 has been shown to
regulate cardiovascular inflammation, autophagy, apoptosis, oxidative
stress and functions as a risk biomarker of cardiovascular diseases,^[
[254]^93 ^] while endothelium‐targeted inhibition of miR‐122 improved
vascular endothelial function in high‐fat diet‐fed mice.^[ [255]^94 ^]
Circulating miR‐451a has been reported as potential marker of coronary
artery aneurysmal disease,^[ [256]^95 ^] and its upregulation could
stimulate HUVECs proliferation and apoptosis by directly targeting
macrophage migration inhibitory factor (MIF), suggesting miRNA‐451a
contribution in regulating atherosclerosis.^[ [257]^96 ^]
Endothelial‐derived miR‐9 plays a key role in the pathogenesis of
diabetic cardiomyopathy and regulates the production of ECM proteins
and inflammatory molecules in human cardiac microvascular endothelial
cells. Using an EC‐specific miR‐9 transgenic model, it was further
demonstrated that EC‐derived miR‐9 regulates cardiac fibrosis.^[
[258]^97 ^] Inducing miR‐9 mimics in HUVECs enhanced cell proliferation
and angiogenesis while simultaneously reducing apoptosis and
inflammation. These effects were mediated through the regulation of the
MAPK/ERK and PI3K/AKT/mTOR pathways, supporting our findings on the
predicted targets of EV‐enriched miRNAs.^[ [259]^98 ^]
Our analysis revealed a correlation between miRNA and mRNA levels in
cells. Differentially expressed miRNAs are associated with the
regulation of their predicted mRNA targets, suggesting that miRNAs
influence endothelial gene expression under different culture
conditions.
However, we found no correlation between cellular mRNA regulation and
EV protein content. As illustrated in Figure [260]S7 (Supporting
Information), the upregulation or downregulation of mRNAs in cells does
not correspond to the same pattern in EV proteins, indicating that
additional factors beyond mRNA expression regulate EV protein content.
Similarly, there was no correlation between EV miRNA content and EV
protein composition. EV miRNA profiles remained largely unchanged
between flow and static conditions, suggesting that the differences in
EV protein composition are not driven by miRNA‐mediated regulation.
Given that intracellular miRNA/mRNA expression correlates but does not
translate to EV protein content, we propose that the observed
variations in EV protein composition are influenced by selective
packaging mechanisms that regulate protein incorporation into EVs,^[
[261]^99 , [262]^100 , [263]^101 ^] biomechanical stress‐induced
changes in cellular signaling affecting EV biogenesis or secretion,^[
[264]^102 , [265]^103 , [266]^104 ^] and post‐translational
modifications^[ [267]^105 , [268]^106 , [269]^107 , [270]^108 ^] or
differential protein stability independent of transcriptional control.
In summary, we developed an in vitro model for culturing endothelial
cells under physiologically relevant conditions, enabling large‐scale
EV isolation. We tested various culture media to identify one that
minimizes FCS‐derived contaminants while maintaining the stability of
primary HUVECs under flow conditions. Our detailed proteome and
transcriptome analysis suggested that EV content differs between static
and laminar flow cultures, potentially affecting EV‐mediated
communication. In this study, we observed significant differences in
the expression of specific miRNAs and proteins in EVs produced under
varying culture conditions. While these findings provide valuable
insights into the impact of culture conditions on EC‐EV composition,
the underlying mechanisms driving these changes remain to be fully
elucidated.
It is well‐documented that HUVECs exhibit sex‐specific differences in
response to mechanical stimulation, gene expression, and other
functional characteristics, which could influence EV production,
composition, and quality. One limitation of our study is the exclusive
use of female HUVECs for the core EV isolation and omics workflows.
While this approach was chosen to minimize biological variability and
ensure reproducibility, we acknowledge that sex‐specific differences in
endothelial function and shear responsiveness have been documented
(e.g., in proliferation, apoptosis sensitivity, and transcriptional
profiles).^[ [271]^109 , [272]^110 , [273]^111 , [274]^112 ^] Although
these differences are often quantitative rather than qualitative in
nature, future studies should incorporate matched male and female
donors to systematically investigate sex‐dependent aspects of EV
release and content. The method we have established provides a
foundation for conducting sex‐specific studies, as well as
investigations into drug effects, different shear stress conditions
(e.g., intensity, oscillatory flow), the impact of hormones, and other
physiological factors. These future directions will contribute to a
deeper understanding of the biological and functional heterogeneity of
EVs across various experimental conditions.
Conflict of Interest
The authors declare no conflict of interest.
Supporting information
Supporting Information
[275]SMTD-9-2401841-s001.docx^ (2.9MB, docx)
Supplemental Table 1
[276]SMTD-9-2401841-s003.xlsx^ (5.2MB, xlsx)
Supplemental Table 2
[277]SMTD-9-2401841-s002.xlsx^ (957.4KB, xlsx)
Supplemental Table 3
[278]SMTD-9-2401841-s004.xlsx^ (279.2KB, xlsx)
Acknowledgements