Brain-Derived Extracellular Vesicles Induce Vasoconstriction and Reduce Cerebral Blood Flow in Mice

Abstract

Traumatic brain injury (TBI) impairs cerebrovascular autoregulation and reduces cerebral blood flow (CBF), leading to ischemic secondary injuries. We have shown that injured brains release brain-derived extracellular vesicles (BDEVs) into circulation, where they cause a systemic hypercoagulable state that rapidly turns into consumptive coagulopathy. The BDEVs induce endothelial injury and permeability, leading to the hypothesis that they contribute to TBI-induced cerebrovascular dysregulation. In a study designed to test this hypothesis, we detected circulating BDEVs in C57BL/6J mice subjected to severe TBI, reaching peak levels of 3 × 10⁴/μL at 3 h post-injury (71.2 ± 21.5% of total annexin V-binding EVs). We further showed in an adaptive transfer model that 41.7 ± 5.8% of non-injured mice died within 6 h after being infused with 3 × 10⁴/μL of BDEVs. The BDEVs transmigrated through the vessel walls, induced rapid vasoconstriction by inducing calcium influx in vascular smooth muscle cells, and reduced CBF by 93.8 ± 5.6% within 30 min after infusion. The CBF suppression was persistent in mice that eventually died, but it recovered quickly in surviving mice. It was prevented by the calcium channel blocker nimodipine. When being separated, neither protein nor phospholipid components from the lethal number of BDEVs induced vasoconstriction, reduced CBF, and caused death. These results demonstrate a novel vasoconstrictive activity of BDEVs that depends on the structure of BDEVs and contributes to TBI-induced disseminated cerebral ischemia and sudden death.

Introduction

Extracellular vesicle (EV) is a collective term for subcellular fragments (membrane fragments and intracellular organelles) that are released from cells undergoing apoptosis or active microvesiculation, as well as exosomes that are secreted from cells.^1,2 The EVs have diverse morphologies, structures, and biological activities^3–5 and have been increasingly recognized as a unique class of biological mediators that carry out the function of parental cells while also engaging in distinctive activities of their own. These EV-associated activities are mediated by molecules expressed on these EVs and the cargo molecules they carry.^6–8

We have shown that traumatically injured brains release brain-derived EVs (BDEVs) that disrupt the blood–brain barrier (BBB) and are released into circulation. These BDEVs promote intravascular coagulation through tissue factor and anionic phospholipids such as phosphatidylserine (PS) expressed on membrane vesicles and cardiolipin (CL) on extracellular mitochondria, causing a systemic hypercoagulable state that rapidly turns into consumptive coagulopathy.^9,10 This consumptive coagulopathy has been widely reported in patients with traumatic brain injury (TBI) and is closely associated with poor outcomes among these patients.^11–14 This TBI-induced coagulopathy (TBI-IC) can also be induced in non-injured mice infused with BDEVs.^9,15

A TBI exerts a physical impact on a small part of the brain, but secondary injures that affect the whole brain and disseminate systemically develop in patients with severe TBI. Ischemia is the major cause of these secondary injuries, which are attributable to the reduction of cerebral blood flow (CBF) caused by cerebral edema and the resultant high intracranial pressure (ICP).

There is extensive clinical and laboratory evidence, however, that TBI impairs cerebrovascular autoregulation, which maintains steady CBF over a wide range of perfusion pressures during homeostasis.¹⁶ For example, the core of a cortical contusion often shows the rapid but transient lack of perfusion.¹⁷ Tissue histology^18–20 and continuous monitoring of peri-contusional CBF²¹ find vasoconstriction, compression of microvasculature by swollen perivascular glial cells, and microvascular thrombosis.

The CBF reduction is not confined to the site of injury and affects other regions of the brain as well.^17,22,23 Cerebral vasospasm also develops in patients with subarachnoid hemorrhage (SAH),²⁴ which is reported in 39%–65% of TBI cases,²⁵ and has been independently associated with poor clinical outcomes.^25,26 The vasospasm is believed to be caused by oxygenated hemoglobin from the lysed erythrocytes in the cerebral spinal fluid,²⁷ but erythrocyte EVs have also been implicated in causing endotheliopathy and vascular dysregulation.^9,15,28,29

Although the biological activities of exosome secreted from live cells have been extensively reported, EVs from injured cells are far less understood. Here we report results from a study designed to investigate the role of BDEVs in the development of vascular dysregulation and the suppression of CBF in mouse models and in vitro experiments.

Methods

Mouse models

Three mouse models were used to study the impact of BDEVs on vascular regulation and CBF: fluid percussion injury (FPI; Supplementary Methods and Results), controlled cortical impact (CCI), and adaptive transfer. We have extensively used the FPI model to study TBI-induced coagulopathy.^9,30

In addition, we also subjected mice to injury induced by CCI³¹ to validate data obtained from FPI mice. Briefly, an anesthetized adult male C57BL/6J mouse (12–16 weeks and 22–25 g; Jackson Laboratory, Bar Harbor, ME) was affixed to a mouse stereotactic frame (Harvard Apparatus) to expose the skull through a midline incision. A 2-mm hole was drilled through the skull, with the dual matter intact. The CCI was applied at a speed of 5 m/sec, depth of 1.5 mm, and exposure time of 150 msec. The skin incision was closed under sterile conditions. Vital signs were monitored during and immediately after injury. The mouse was examined for Evans blue extravasation, CBF, and the levels of BDEVs in peripheral blood samples.

For the adaptive transfer model, non-injured mice were infused with BDEVs made from brains subjected to freeze-thawing injury (Supplementary Methods and Results).^9,10,32 Briefly, adult male C57BL/6J mice (12–16 weeks and 22–25 g; Jackson Laboratory) were infused with BDEVs through the tail vein⁹ and monitored for outcomes discussed in the following sections. Control mice received an equal volume of the supernatant from the brain homogenates centrifuged at 100,000 × g for 60 min (twice) at 4°C.

Blood pressure (BP) was measured using a tail-cuff device (CODA; Kent Scientific Co, Torrington, CT) 30 min after BDEV infusion. In a subset of mice, nimodipine (Nimotop, Bayer) was given intravenously at 1 mg/kg 5 min before BDEV infusion. Nimodipine is a dihydropyridine calcium channel blocker that was originally indicated for hypertension but is now used primarily for relieving the cerebral vasospasm of patients with SAH,³³ including traumatic SAH.³⁴

For TBI experiments, we chose to study male mice because TBI occurs predominantly with men. The mice were randomized to be subjected to FPI/CCI or sham surgery and for the adaptive transfer model, non-injured mice were randomized for infusion with BDEVs or the supernatant of brain homogenates, which was obtained through two consecutive ultracentrifugation cycles, each at 100,000 × g for 60 min at 4°C.

Laser speckle contrast analysis (LASCA)

We used non-invasive LASCA to measure the effects of FPI and BDEVs on cerebral cortical blood flow in real time.³⁰ In brief, an anesthetized mouse was fixed on a temperature-controlled surgery platform with the head constrained (Supplementary Figure S1). The hair was removed from the frontoparietal region to expose the scalp, which was then exposed to a laser beam from the PeriCam PSI HR System (Perimed AB, Järfälla, Sweden). After allowing the baseline CBF to stabilize for 5 min, the mouse was subjected to either FPI or CCI or infused with 3 × 10⁷ of BDEVs. The CBF was monitored continuously, and the data were quantified at the time period of interest (TOI).

Endothelial injury, permeability, and cerebral edema

We used four complementary methods to measure BDEV-induced endothelial injury and permeability. First, the Evans blue-dye extravasation test measured the vascular leakage of mice subjected to FPI or infused with BDEVs.¹⁵ Briefly, a mouse received 50 μL of 2% Evans blue (Sigma-Aldrich, St. Louis, MO) through the tail vein and 60 min later was exposed to FPI or infused with BDEVs. The mouse was euthanized 30 min later and perfused with 50 mL of warm phosphate buffered saline (PBS) through cardiac puncture to remove the dye and blood cells from the vasculature. The brain was dissected and its topical and coronal views were photographed. It was then snap-frozen in liquid nitrogen, homogenized in formamide (1:20 w/v), and incubated at 60°C overnight. The brain homogenate was centrifuged at 16,000 × g for 30 min. Evans blue in the supernatant was quantified at OD620 nm in a SpectraMax M5 microplate reader (Molecular Devices, Sunnyvale, CA).

Second, we also used a transwell system to measure the transendothelial migration of BDEVs in vitro (Supplementary Methods and Results).^9,30 Third, we measured brain water content to detect vascular leakage and cerebral edema (Supplementary Methods and Results).³⁰ Finally, tissue histology was performed to detect microvascular bleeding and fibrin deposition.

Death, autopsy, and tissue histology

Mice subjected to TBI and those infused with BDEVs were monitored for survival. Autopsy was performed on mice that died during the monitoring period, and the brains, hearts, lungs, and intestines were collected, washed extensively with PBS, and fixed in 5% paraformaldehyde.^9,10 Blood was also collected using 0.38% sodium citrate as an anticoagulant (final concentration). Surviving mice were euthanized under anesthesia 12 h after BDEV infusion to collect blood and organs. The organs were fixed in 5% paraformaldehyde, sectioned, and stained with hematoxylin and eosin (H&E) to detect microvascular bleeding, tissue infarction, and occlusive thrombosis or with phosphotungstic acid hematoxylin (PTAH) to detect intravascular fibrin deposition, which is a hallmark of intravascular coagulation.⁹

Flow cytometry

Flow cytometry (LSR II, Becton Dickinson and Co, San Jose, CA) was used to quantify plasma levels of BDEVs from neurons and glial cells and EVs from endothelial cells (eEVs) and platelets (pEVs) of TBI and control mice.⁹ These EVs were identified first by their sizes (<1 mm) using 0.5, 0.9, and 3 mm standard microbeads (Biocytex, Marseille, France), and then by their expression of anionic phospholipids by APC-conjugated annexin V (ANV⁺, eBioscience) together with one of the following antibodies: a monoclonal fluorescein isothiocyanate (FITC)-anti-NSE antibody (neuron specific enolase; Abcam, Cambridge, MA) for neuronal EVs, a polyclonal anti-glial fibrillary acidic protein (GFAP) antibody (eBioscience, San Diego, CA) followed by a PE-anti-rabbit IgG (eBioscience) for glial EVs, an eFluor 450-anti-mouse CD144 antibody (Invitrogen, Carlsbad, CA) for endothelial EVs (eEVs), and an eFluor 450-anti-mouse CD41a antibody (Invitrogen) for platelet-derived EVs (pEVs).

Isotype-specific IgGs were tested as negative controls. Sphero AccuCount beads were used to quantify EVs. All buffers were filtered with a 0.1 mm filter (EMD Millipore, Burlington, MA) to reduce small-particle contaminations.

Protein and phospholipid extraction from BDEVs

The known numbers of BDEVs were lysed in a 2% sodium dodecyl sulfate (SDS) buffer for 30 min at 4°C. The SDS was then removed using an SDS-Out™ kit (ThermoFisher Scientific, Waltham, MA). The protein extract was centrifuged at 13,000 × g for 20 min at 4°C to collect the supernatant, which was tested as BDEV-derived proteins. Phospholipids were extracted from BDEVs and profiled using mass spectrometry (Avanti Polar Lipids, Inc., Alabaster, AL). The phospholipids thus identified were mixed proportionally according to mass spectrometry data.

The phospholipid mix was sonicated at 60 mH for 10 min to promote the formation of lipid micelles immediately before testing as BDEV-derived phospholipids. The protein and phospholipid components of BDEVs were tested separately at amounts equivalent to 10 times the IC50 number of BDEVs.

BDEV-induced vasoconstriction

A 2-mm segment was dissected from the carotid artery of a C57BL/6J male mouse immediately after euthanasia and mounted on two wires of the DMT 610 M multi myograph system (Danish Myo Technology, Hinnerup, Denmark). The segment was incubated with 5 mL of the physiological salt solution (PSS), which contained 130 mM NaCl, 4.7 mM KCl, 1.17 mM KH₂PO₄, 1.17 mM MgSO₄.7H₂O, 14.9 mM NaHCO₃, 5.5 mM glucose, 0.026 mM ethylenediaminetetraacetic acid (EDTA), and 1.6 mM CaCl₂ (pH 7.4) for 30 min at 37°C to set the baseline (2 mN).

The PSS was then replaced with the oxygenated (95% O₂) potassium-enriched PSS (KPSS), which contained 74.7 mM NaCl, 60 mM KCl, 1.18 mM KH₂PO₄, 1.17 mM MgSO₄.7H₂O, 14.9 mM NaHCO₃, 5.5 mM glucose, 0.026 mM EDTA, and 1.6 mM CaCl₂ (pH 7.4) to trigger vasoconstriction to ensure the contractility of the arterial segment. After washing with PSS to reset the baseline, the segment was incubated with BDEVs and monitored continuously for 60 min at 37°C. Changes in vascular wall tension were recorded by the PowerLab system (AD Instruments, Sydney, Australia).

Hopping probe ion conductance microscopy (HPICM)

The BDEV-induced morphological changes of cultured human umbilical vein endothelial cells (HUVECs) and rat brain smooth muscle cells (RBSMC; ATCC, Manassas, VA) were monitored in real-time using HPICM (Ionscope Ltd, Cambridge, UK).^35,36 The Axon MultiClamp 700B amplifier (Molecular Devices, San Jose, CA) supplied a DC voltage of +200 mV to generate an electrical current between the nanopipette electrode and bath electrode. After baseline scanning, cells were treated with BDEVs for 2 min at 37°C and scanned for up to 60 min under an inverted TiU microscope (Nikon Corporation, Japan). Topographical data were continuously acquired and linearly interpolated into images with ScanIC Image Viewer (version 1.0; Ionscope Ltd, Cambridge, UK).

Calcium influx imaging

The RBSMCs were cultured in a Dulbecco's Modified Eagle Medium (DMEM) medium (ThermoFisher Scientific) containing 10% fetal bovine serum until confluent. They were then labeled with the fluorescent calcium-binding dye Fluo-3 (Thermo-Fisher Scientific) for 10 min at 37°C. After being washed with PBS, the cells were treated with BDEVs for 10 min at 37°C, counter-stained with the DNA-binding dye Hoechst (Beyotime, Shanghai, China), and reviewed under a laser-scanning confocal microscope (Olympus, Japan).

Study design and statistical analysis

To minimize the confounding influence of blood draws and other technical manipulations on the outcome assessments of the TBI mice, blood collection and analyses, histopathology assessments, measurements of the CBF, and evaluation of vascular permeability were performed on separate groups of mice and analyzed as independent variables. To reduce experimental and analytical biases, experimental results were analyzed by a designated biostatistician who was blinded to the experimental conditions.

Power analyses were performed for individual experiments. Categorical (frequency) and continuous variables were expressed as the percentage and mean ± standard error of the mean (SEM), respectively. They were analyzed using a paired t test, one-way analysis of variance (ANOVA) or repeated measures ANOVA for Bonferroni post hoc comparisons, as indicated in the specific datasets. Cumulative survival was analyzed with a Kaplan-Meier plot. The analyses were performed using SYSTAT software SigmaPlot (version 11.2; San Jose, CA). A p value of less than 0.05 was considered statistically significant.

Results

An impact force of 1.9 ± 0.2 atm resulted in an overall mortality rate of 41.7 ± 5.8%, and all deaths occurred within the first 6 h of injury (Fig. 1A). Vascular leakage, measured by Evans blue extravasation, was most severe at the site of injury but was also detected in a much larger area of the ipsilateral hemisphere (Fig. 1B) as well as in the contralateral hemisphere of the brain, but at 22.2 ± 6.8% of levels in the ipsilateral hemisphere (Fig. 1C). For validation, we also subjected mice to injury induced by CCI, which caused the most severe leakage at the injured hemisphere than the contralateral hemisphere (Supplementary Fig. S2A). Pulmonary vascular leakage was also detected by Evans blue extravasation (Fig. 1D) and tissue histology (Fig. 1E).

FIG. 1.

Fluid percussion injury (FPI)-induced vascular leakage and suppressed cerebral blood flow (CBF). (A) Mortality of mice subjected to FPI at 1.9 atm (n = 36; Kaplan-Meier survival analysis). (B) Traumatic brain injury (TBI)-induced vascular leakage measured by Evans blue extravasation. Images are representative of 36 mice (circle: area with direct impact). (C) Evans blue extravasation was quantified in the ipsilateral and contralateral hemisphere to the injury by optical density at 620 nm (n = 36; one-way analysis of variance [ANOVA]). Vascular leakage was also measured in the lungs of mice with TBI by Evans blue extravasation (D; n = 36; one-way ANOVA) and by the tissue histology of hematoxylin and eosin stains (E), which show enlarged perivascular space (interstitial edema, arrowhead) and microvascular bleeds (arrows; representative of 36 mice; bar = 20 μm). (F) Representative images CBF measured by laser speckle. (G) Dynamic changes of annexin V (ANV⁺)-bound brain-derived extracellular vesicles (BDEVs) from neurons (NSE⁺/ANV⁺) and glial cells (GFAP⁺/ANV⁺; n = 36; one-way ANOVA; *p < 0.05 between FPI and sham mice, **p < 0.05, time-dependent changes compared with the baseline). Color image is available online.

Brain water content, which measures vascular leakage and tissue edema, was significantly increased not only in the ipsilateral hemisphere but also in the contralateral hemisphere, albeit at a lower level (Supplementary Fig. S3). Laser speckle detected reduced cortical CBF of the whole brain of mice subjected to FPI (Fig. 1F) and those to CCI (Supplementary Fig. S2B). The BDEVs from neurons (NSE⁺) and glial cells (GFAP⁺) that expressed anionic phospholipids were detected within 15 min after injury and reached peak levels at 3 h post-FPI (Fig. 1G) when they accounted for 71.2 ± 21.5% of total annexin V-binding EVs, before declining and being surpassed by platelet-derived (CD41a⁺) and endothelial cell-derived (CD144⁺) EVs (Supplementary Fig. S4).

Peak levels of circulating BDEV were closely associated with brain water content (Supplementary Fig. S5). Similarly, BDEVs were detected in peripheral blood samples from mice subjected to CCI (Supplementary Fig. S2C). These results demonstrate that BDEVs played a critical role in TBI-induced global CBF reduction and pulmonary endotheliopathy.

BDEVs induced death by transient causes

The specific effects of BDEVs are difficult to study in mice with TBI because (1) EVs found in these mice are highly heterogeneous, including those from erythrocytes, platelets, and leukocytes, the levels of which were significantly reduced in EVs made from isolated brains because the mice were perfused extensively with PBS to remove blood from the vasculature; (2) purification steps could change the activity of BDEVs; and (3) the confounding influence of trauma on other organs is difficult to stratify.

We addressed these issues by studying BDEVs in an adaptive transfer model using BDEVs made from isolated brains.^9,15,37 The BDEVs generated using this method contained 76.8 ± 16.2% of BDEVs from neurons and glial cells, similar to the peak levels of BDEVs found in FPI mice. They also resembled BDEVs found in FPI mice in sizes, PS expression, and procoagulant activity (Supplementary Fig. S6).

We infused BDEVs into non-injured C57BL/6J mice (3 × 10⁷/mL as found in FPI mice at 3 h; each mouse has ∼1 mL of circulating blood). Within 1–5 min of infusion, 76% of mice commenced circular movements, suggesting hemiparesis, which occurred randomly in both directions and at similar frequencies (Supplementary Video S1), followed by rapid tachypnea, and 46.6 ± 5.8% of mice died within 6 h of infusion. This mortality rate was statistically indistinguishable from that of FPI mice (Fig. 2A). In contrast, hemiparesis and death did not develop in mice infused with the supernatant. This BDEV-induced death was dose-dependent, reaching 100% in mice receiving a single infusion of 5 × 10⁷/mL BDEVs (Fig. 2B), with an IC50 at 3 × 10⁷/mL, which was also the peak level of BDEVs in FPI mice.

FIG. 2.

Brain-derived extracellular vesicles (BDEVs) caused rapid death. (A) Mortality rates of mice subjected to fluid percussion injury (FPI) and those infused with BDEV at the number found in TBI mice (n = 27; Kaplan-Meier survival analysis). (B) BDEV-induced death is dose-dependent (n = 150; 30 at each dose, Kaplan-Meier survival analysis). Hematoxylin and eosin stain shows subarachnoid and intracerebral hemorrhage of the brain (C) and interstitial edema and perivascular microbleeding in the lung (E), but minimal changes in the heart (G) of BDEV-infused mice. These changes were not detected in organs from mice infused with the supernatant (D, F, H; bar = 20 μm). (I) Brain water content of mice at the baseline (white bar) and after receiving either BDEVs or the supernatent (n = 27; one-way analysis of variant [ANOVA]). Phosphotungstic acid hematoxylin (PTAH) stain detects fibrin deposition to the vasculature of the brain (J), the lungs (L; arrowhead), and the heart (N) of mice receiving BDEVs, but not those receiving the supernatant (K, M, O; bar = 40 μm for images C, D, and G to N, bar = 50 μm for images E and F). (P) Plasma fibrinogen levels in the three groups of mice (n = 27; one-way ANOVA). Histology images presented in this figure are representative of experiments from 126 mice (50% received BDEVs). Color image is available online.

The mortality decreased from 46.6% to 7.1% in mice infused with BDEVs together with 400 mg/kg of lactadherin (milk fat globule-EGF factor 8 protein; Supplementary Fig. S7), which is an apoptotic cell-scavenging factor³⁸ that we have shown to remove anionic phospholipid-expressing EVs from circulation through phagocytosis in the liver.¹⁵ Blood cell counts measured 15 min after infusion were comparable between mice receiving BDEVs and those receiving the supernatant, except for mild to moderate thrombocytopenia in the BDEV-infused mice (Supplementary Table S1). No death occurred in mice infused with either glial cells from the T98G glioblastoma cell line³⁹ or neurons from the SH-SY5Y neuroblastoma cell line (Supplementary Fig. S8), which were tested as parental cells of glial cell- and neuron-derived EVs, respectively.

Autopsies conducted immediately after death found no apparent cause of death. Tissue histology showed widespread microvascular leakage, defined by the accumulation of red blood cells in enlarged perivascular spaces of the brain, lungs (Fig. 2C, 2E), and intestines (Supplementary Fig. S9) but minimally in the hearts (Fig. 2G) of BDEV-infused mice. These changes were not detected in mice infused with the supernatant (Fig. 2D, 2F, 2H). The histological evidence of cerebral vascular leakage was consistent with increased brain water content (Fig. 2I).

Intravascular fibrin deposition, which indicates intravascular coagulation, was also detected in these organs, but not in those infused with the supernatant (Fig. 2J–2O). This is consistent with a significant reduction of plasma fibrinogen in BDEV-infused mice (Fig. 2P), suggesting that the mice developed consumptive coagulopathy. We did not detect massive cardiac or cerebral infarction, occlusive arterial thrombosis in large vessels, pulmonary thromboembolism, or cerebellar tentorium herniation, which are common causes of sudden death, in the autopsy or tissue histology of all 196 mice examined.

Death was also not caused by anaphylactic attack caused by BDEV infusion because (1) eosinophil counts (Supplementary Table S1) and IgE (Supplementary Fig. S10) did not increase in BDEV-infused mice and (2) epinephrine, which is the most common treatment for severe anaphylactic attack, did not reverse the reduction of CBF or prevent death in mice infused with BDEVs (Supplementary Fig. S11). These results demonstrate that BDEVs at levels comparable to those found in FPI mice caused death by transient causes.

BDEVs rapidly reduced CBF

A common cause of sudden death by transient causes is severe cerebral and cardiac vasoconstriction, which deprives blood supply to these vital organs; we therefore measured cortical CBF in real-time using LASCA. At the IC50 dose of 3 × 10⁷/mouse, BDEVs reduced CBF by 93.8 ± 5.6% within 15 min of infusion (Fig. 3), significantly faster than the FPI-induced suppression of CBF, which was most severe at 3 h post-injury (Fig. 1F). The CBF reduction persisted in mice that eventually died (Fig. 3A and Supplementary Video S2) but quickly recovered in those that survived (Fig. 3B and Supplementary Video S3) and was not observed in mice infused with the supernatant (Fig. 3C and Supplementary Video S4).

FIG. 3.

Profiles of brain-derived extracellular vesicles (BDEV)-induced reduction of cerebral blood flow (CBF). Representative images at three time points (top panel), representative tracing of speckle signal (middle panel), and summary from multiple mice (bottom panel) for BDEV-infused mice that eventually died (A), those that survived (B), and mice infused with the supernatant (C). The bottom panels summarize data from 36 mice in each group (one-way analysis of variant [ANOVA]). (D) The mortality (D) and CBF (E) of mice after receiving 3 × 10⁷/mouse of BDEVs (black dot), the supernatant (white dot), BDEV-derived proteins (black triangle), and BDEV-derived phospholipids (white triangle; n = 32; one-way ANOVA; *p < 0.001 vs. BDEVs). Both protein and phospholipids were extracted from 10 times the IC50 number of BDEVs. Color image is available online.

Proteins extracted from 10 times the IC50 numbers of BDEVs did not induce death (Fig. 3D) or reduce CBF (Fig. 3E). Phospholipids extracted from BDEVs were profiled by mass spectrometry (Table 1). Phospholipids found in BDEVs were then mixed proportionally based on their mass spectrometry profiles and sonicated to promote the formation of lipid micelles. The phospholipid micelles from 10 times the IC50 numbers of BDEVs were infused into non-injured mice but did not cause death (Fig. 3D) or reduce CBF (Fig. 3E). These results suggest that (1) BDEVs, but not their protein or phospholipid components suppressed CBF and (2) the prolonged reduction of CBF predicts death.

Table 1.

Phospholipid Compositions of Brain-Derived Extracellular Vesicles^*

Phospholipids (including lyso-form)	Amount (μg/mg proteins)	% of total
Phosphatidylcholine (PC)	8.0	41.27
Phosphatidylethanolamine (PE)	5.6	29.41
Phosphatidylserine (PS)	2.9	15.31
Phosphatidylinositol (PI)	1.8	9.56
Sphingomyelin (SM)	0.8	4.19
Phosphatidylglycerol (PG)	0.1	0.27

^* Brain-derived extracellular vesicles were made and pooled from 11 mouse brains.

BDEVs rapidly induced cerebral vasoconstriction

The rapid and transient reduction of CBF suggests that BDEVs induced vasoconstriction instead of vasoocclusive thrombosis. We found that BDEVs increased the wall tension of the carotid artery in a dose-dependent manner in vitro (Fig. 4A) and induced transient hypertension in vivo (Fig. 4B), whereas their protein and lipid components did not (Fig. 4C). The vasoconstriction was blocked by the calcium channel blocker nimodipine (Fig. 4D), which also prevented CBF reduction (Fig. 4E) and significantly improved the survival of mice infused with BDEVs (Fig. 4F). The hypertension was not from the volume effect because hypertension did not develop in mice infused with an equal volume of the supernatant.

FIG. 4.

Brain-derived extracellular vesicles (BDEVs) induced vasoconstriction. (A) Left panel: a representative image of BDEV-induced constriction of a carotid arterial segment at 37°C (BDEVs at 3 × 10⁷/mL); and right panel: the effect of BDEVs is dose-dependent (n = 20/dose; one-way analysis of variance [ANOVA]; *p < 0.05 vs. baseline). (B) Blood pressure measured 30 min after mice were infused with 3 × 10⁷/mouse of BDEVs or the supernatant (n = 26; one-way ANOVA). (C) The wall tension of carotid arterial segments treated with different fractions of BDEVs (n = 28; one-way ANOVA; *p < 0.01 vs. BDEV-treated). (D) BDEV-induced vasoconstriction was signficantly reduced by nimodipine (100 mg/mL; n = 28; one-way ANOVA). (E) Cerebral blood flow (CBF) of mice infused with 3 × 10⁷/mouse of BDEVs in the presence or absence of 1 mg/kg of nimodipine. Control mice received the supernatant (left: a representative CBF tracing, right: a summary of 18 experiments (one-way ANOVA). (F) Kaplan-Meier survival analysis of C57BL/6J mice infused with 3 × 10⁷/mouse of BDEVs alone or with 1 mg/kg of nimodipine. Control mice received an equal volume of the supernatant (n = 60). Color image is available online.

The finding that hypertension developed in mice after BDEV infusion further indicates that they did not die of anaphylactic attack, which often results in hypotension. These data demonstrate that BDEVs, but not their protein or phospholipid components, induce rapid vasoconstriction in a calcium-dependent fashion.

BDEVs disrupted endothelial integrity

The results presented in Figures 3 and 4 suggest that BDEVs interacted with vascular smooth muscle cells covered by the endothelium. Consistent with this notion, HPICM detected intercellular gaps, which became larger over time in BDEV-stimulated HUVECs (Fig. 5A–5C). The formation of these intercellular gaps was caused by cell contraction, as indicated by the increase of cell heights (Fig. 5D). These gaps allowed BDEVs to transmigrate through the endothelial barrier, as detected in vitro using a transwell system (Fig. 5E) and in mice infused with BDEVs (Fig. 5F).

FIG. 5.

Brain-derived extracellular vesicles (BDEVs) disrupted endothelial barrier and induced contraction of vascular smooth muscle cells. Representative hopping probe ion conductance microscopy HPICM images show the appearance and time-dependent enlargement of the intercellular gaps between cultured human umbilical vein endothelial cells (HUVECs) (arrowheads) after treatment with 3 × 10⁷/ml of BDEVs (A–C; bar = 25 μm). (D) The quantitative summary from 30 independent experiments showing the increase in cell heights (one-way analysis of variance [ANOVA]). (E) Fluorescent dye-labeled BDEVs or EVs collected from fluid percussion injury (FPI) mice transmigrated through HUVECs (n = 18; one-way ANOVA). (F) Evans blue extravasation from the lungs of mice infused with 3 × 10⁷/mouse of BDEVs or EVs from FPI mice (n = 18; one-way ANOVA). Representative images of BDEV-induced calcium influx of rat brain smooth muscle cells (RBSMCs) (G; bar = 25 μm for all images), which was blocked by nimodipine (H; 100 mg/mL; n = 18; one-way ANOVA). Representative HPICM time-lapse images show the intercellular gaps (arrowheads) of cultured RBSMC treated with 3 × 10⁷/mL of BDEVs (J–L; Bar = 25 μm) and heights of RBSMC treated with the supernatant and BDEVs (M; n = 30; one-way ANOVA). Color image is available online.

Consistent with the effect of nimodipine, we detected calcium influx of RBSMCs stimulated with BDEVs in a dose-dependent manner (Fig. 5G, 5H), which was blocked by nimodipine (Fig. 5I). These RBSMCs formed smaller and less frequent intercellular gaps compared with endothelial cells (Fig. 5J–5L) but also contracted on BDEV treatment (Fig. 5M). These results suggest that (1) BDEVs disrupted endothelial integrity and transmigrated through the endothelial barrier and (2) BDEVs induced calcium influx and the contraction of vascular smooth muscle cells, resulting in vasoconstriction.

Discussion

We have uncovered a novel mechanism of BDEV-induced vasoconstriction that reduces CBF and contributes to TBI-induced coagulopathy and death. This conclusion is supported by several lines of evidence. First, there was a time-dependent increase of circulating BDEVs in FPI mice that correlates with levels of cerebral edema. Second, BDEVs drastically reduced CBF of non-injured mice by triggering the calcium-dependent signaling in vascular smooth muscle cells. Third, the EV-scavenging factor lactadherin improved the survival of non-injured mice infused with BDEVs.¹⁵

The finding that CBF reduction was transient and recovered rapidly in survival mice but persisted in dying mice reaffirms the importance of maintaining sufficient CBF during acute TBI and offers additional perspectives as well. First, BDEVs impair cerebrovascular autoregulation,¹⁶ as was also indicated by the failure of epinephrine to increase CBF in BDEV-infused mice. This cerebrovascular dysregulation has been extensively reported in patients with TBI and is associated with poor outcomes,^40,41 but its underlying cause remains poorly understood. Our results demonstrate, for the first time, a role for BDEVs in TBI-induced cerebrovascular dysregulation.

Second, transient hypertension developed in mice infused with BDEVs; such hypertension is consistently found in patients with TBI⁴² and associated with worsened outcomes.^43–45 The hypertension is believed to be a physiological response to increased intracranial pressure to preserve cerebral perfusion (Cushing reflex) during acute TBI, and our study identified BDEVs also as a causal factor.

Third, CBF reduction occurred rapidly after BDEV infusion, well before death, which occurred between 1–6 h, suggesting that early and persistent CBF suppression predicts death, consistent with clinical observations.^46–48 It is important to note that the reduction of CBF is consistently bilateral whereas the increase in BBB leakage is always more severe at the site of injury. These two presentations could result from either two distinct pathologies involving endothelial cells and smooth muscle cells, respectively, or a dose effect depending on the time and level of BDEV exposure. The latter may be determined by the dose titration of BDEVs in the future.

We further demonstrated that BDEV-treated endothelial cells formed large intercellular gaps, which allowed EVs to transmigrate through the endothelial barrier to activate vascular smooth muscle cells, causing vasoconstriction in a calcium-dependent manner. There are several potential mechanisms for the BDEV-induced vasoconstriction. First, BDEVs induced intravascular coagulation to generate thrombin,⁹ which is associated with cerebral vasospasm,⁴⁹ cerebral edema, and BBB permeability.⁵⁰ Antithrombin III is reported to attenuate the cerebral vasospasm.⁵¹ This is unlikely, however, because BDEVs induced contraction of a carotid artery segment in the absence of coagulation.

More importantly, we found that vasoconstriction, CBF reduction, and death were not observed in mice infused with either protein or phospholipid components of BDEVs, even at much higher concentrations. This finding suggests that BDEVs did not cause vasoconstriction by molecules they carried but rather by their specific structures. The underlying mechanism of this structure-dependent activity remains to be investigated, but it is unlikely from the membrane proteins losing their activity when they are solubilized by SDS because they are routinely expressed and purified as soluble proteins that maintain their biological activities. Further, while we have focused on BDEVs in this study, EVs from endothelial cells, platelets, erythrocytes, and leukocytes likely also play a role in enhancing or modifying the vascular activity of BDEVs.

Conclusion

We have demonstrated that BDEVs cause sudden death in mice by inducing severe vasoconstriction, thus depriving blood flow to the brain and likely to other vital organs as well. The protein and phospholipid components of BDEVs are not lethal on their own, but they become lethal when they are assembled into EVs. This novel cause of vasoconstriction, CBF reduction, and sudden death differs from the pathogen- and toxin-induced death found with most diseases. It provides a new mechanism of TBI-induced sudden death.

Authors' Contributions

Jiwei Wang, Xiaofeng Xie, Yingang Wu, and Yuan Zhou: performed experiments, analyzed data, and wrote the manuscript; Qifeng Li, Ying Li, Xin Xu, Min Wang, Lydia Murdiyarso, Tristan Hilton, and Katie Houck: performed experiments and analyzed data;

Dominic Chung: designed experiments; Jingfei Dong, Min Li, and Jianning Zhang: developed hypotheses, designed experiments, analyzed data, and wrote the manuscript.

Funding Information

This study is supported by the Natural Science Foundation of Tianjin (No.20JCQNJC01270 and No.19JCQNJC0 9500), Natural Science Foundation of Anhui (No.200 8085MH250), Natural Science Foundation of China grants 81720108015 and 81930031 (JZ), 81672399 and 81871919 (ML), 81801234 (YZ) and the National Institutes of Health grants HL152200 and HL154250 (JFD).

Author Disclosure Statement

No competing financial interests exist.

Supplementary Material

Supplementary Methods and Results

Supplementary Figure S1

Supplementary Figure S2

Supplementary Figure S3

Supplementary Figure S4

Supplementary Figure S5

Supplementary Figure S6

Supplementary Figure S7

Supplementary Figure S8

Supplementary Figure S9

Supplementary Figure S10

Supplementary Figure S11

Supplementary Figure S12

Supplementary Table S1

Supplementary Video S1

Supplementary Video S2

Supplementary Video S3

Supplementary Video S4