The effect of endurance exercise on circulating microvesicle dynamics and their impact on surrounding endothelial cells is unclear. Here we tested the hypothesis that exercise intensity modulates the time course of platelet (PMV) and endothelial-derived (EMV) microvesicle appearance in the circulation through hemodynamic and biochemical-related mechanisms, and that microvesicles formed during exercise would stimulate endothelial angiogenesis in vitro. Nine healthy young men had venous blood samples taken before, during, and throughout the recovery period after 1 h of moderate [46 ± 2% maximal oxygen uptake (V̇o2max)] or heavy (67 ± 2% V̇o2max) intensity semirecumbent cycling and a time-matched resting control trial. In vitro experiments were performed by incubating endothelial cells with rest and exercise-derived microvesicles to examine their effects on cell angiogenic capacities. PMVs (CD41+) increased from baseline only during heavy exercise (from 21 ± 1 × 103 to 55 ± 8 × 103 and 48 ± 6 × 103 PMV/μl at 30 and 60 min, respectively; P < 0.05), returning to baseline early in postexercise recovery (P > 0.05), whereas EMVs (CD62E+) were unchanged (P > 0.05). PMVs were related to brachial artery shear rate (r2 = 0.43) and plasma norepinephrine concentrations (r2 = 0.21) during exercise (P < 0.05). Exercise-derived microvesicles enhanced endothelial proliferation, migration, and tubule formation compared with rest microvesicles (P < 0.05). These results demonstrate substantial increases in circulating PMVs during heavy exercise and that exercise-derived microvesicles stimulate human endothelial cells by enhancing angiogenesis and proliferation. This involvement of microvesicles may be considered a novel mechanism through which exercise mediates vascular healing and adaptation.
- platelet microvesicles
- shear stress
- human umbilical vein endothelial cell
NEW & NOTEWORTHY
Increases in intravascular platelet microvesicle concentration occur during exercise, but this depends on exercise intensity, and correlates with elevations in vascular shear stress and plasma norepinephrine concentration. Circulating microvesicles isolated from exercising humans display proangiogenic potential on cultured endothelial cells. Thus, it is possible that microvesicles are involved in vascular responses to exercise.
intraluminal shear stress is a key modulator of endothelial nitric oxide synthase (eNOS) activity and expression in arteries (45, 61) and has been identified as a major stimulus leading to the exercise-induced improvement in vascular function (4, 13, 52, 53), but endothelial adaptations also take place at sites not directly exposed to increased shear forces, and acute changes in circulating factors are believed to be involved in such adaptations (43). However, the mechanisms by which exercise bring about endothelial adaptations not directly dependent on shear stress are not fully understood and may relate to a variety of still unidentified factors.
Cell-derived microvesicles have been identified as potential biomarkers linked to obesity (19), dyslipidemia (1), and coronary artery disease (3, 25). In addition, increased concentrations of several microvesicle varieties are found in the blood postexercise (11, 31, 34, 48, 49). As small membrane-derived vesicles (diameter <1 μm), their phenotype depends on their cell of origin and stimulus of formation (29, 47). Platelet-derived microvesicles (PMVs) are the most abundant circulating microvesicle population (24, 48, 49), and in vitro stimulation of platelets with thrombin (7), norepinephrine (54), IL-6 (42), and shear stress (39, 44) induce shedding of PMVs, whereas high shear stress downregulates the shedding of endothelial microvesicles (EMVs) from cultured endothelial cells (56).
Plasma PMV appearance occurs during recovery from maximal (11) and submaximal (34, 36, 48, 49) exercise protocols, whereas a less pronounced increase or completely absent alteration in plasma EMV (31, 34, 48, 49), erythrocyte (11), and monocyte microvesicles (11, 48, 49) has been observed. The dynamics of PMV during exercise, however, is unknown, and the impact of exercise intensity on circulating microvesicle appearance has not been explored. Several intravascular factors known to stimulate in vitro PMV and EMV production (29, 39, 44, 54) are modulated by exercise intensity. Because vascular shear stress is expected to increase with workrate, and elevated shear forces are strong agonists stimulating ex vivo microvesicle blebbing from platelets, we anticipated that PMVs would increase during exercise according to exercise intensity, whereas EMV release would be restricted by elevated shear stress (56).
Beyond being simple inert plasma membrane fragments, microvesicles are now recognized as bioactive blood constituents capable of interacting with diverse cell populations, including the vascular endothelium (47). Certain EMVs isolated from cell culture exhibit paracrine effects, stimulating inflammation, increasing superoxide anion production while decreasing nitric oxide bioavailability in recipient endothelial cells (8, 14). Depressed endothelial function has also been reported in arteries exposed to circulating microvesicles from cardiac patients and preeclamptic women, but not from apparently healthy donors (6, 55). In contrast, PMVs may protect endothelial cells from apoptosis (30) and stimulate angiogenesis in vitro (7, 30) and in vivo (7). The physiological relevance of microvesicles produced with exercise upon the vascular endothelium, however, has not been explored, but based on previous research we postulated that microvesicles formed during exercise bouts could be involved in vascular adaptations, leading to a proliferative and angiogenic endothelial phenotype.
Therefore, we aimed to 1) characterize the impact of modulating exercise intensity on the time course of PMV and EMV concentrations during exercise and recovery; 2) examine putative relationships between shear stress and microvesicle formation during exercise in vivo; and 3) investigate the potential impact of circulating microvesicles produced during exercise on angiogenesis, proliferation, and migration of human endothelial cells. We hypothesized that PMVs, but not EMVs, would increase during exercise with intensity, and remain elevated up to 1 h postexercise; that the increase in PMVs would be associated with shear stress during exercise; and that exercise-derived circulating microvesicles would stimulate endothelial cell proliferation, migration, and angiogenesis in vitro to a greater extent than those isolated at rest.
Written informed consent was obtained from all participants before taking part in the experiments. Human and cell culture experimental procedures were approved by Brunel University London Research Ethics Committee (RE55-12 and RE39-13), and experiments were performed in accordance with the Declaration of Helsinki.
Nine healthy young participants [25 ± 1 yr, 1.79 ± 0.03 m, 80.5 ± 4 kg, maximal oxygen uptake (V̇o2max) 3.3 ± 0.2 l/min] completed all five visits of this study. Volunteers were young lean male nonsmokers, who were free from cardiorespiratory and metabolic diseases, and mostly involved in recreational physical activities (self-reported). Microvesicles used in the cell culture experiments were obtained from samples collected from a subset of this group (n = 6; 24 ± 1 yr, 1.79 ± 0.04 m, 82.5 ± 5.9 kg, V̇o2max 3.3 ± 0.3 l/min).
Two experiments were performed to 1) elucidate the time course of microvesicle appearance in the circulation in response to two intensities of dynamic exercise, while gaining insights about the potential mechanisms involved with this response in vivo; and 2) investigate the impact of circulating microvesicles produced during exercise on endothelial proliferation, chemotaxis, and angiogenesis in vitro.
All participants attended the laboratory for two preliminary visits, followed by three experimental trials during which exercise or resting control experiments were performed in a thermoneutral environment. Participants were familiarized with the experimental procedures, the semirecumbent cycle ergometer (Angio, Lode, Netherlands), and the maximal incremental exercise test at the first visit to minimize possible learning effects. After at least 48 h of recovery, participants had their oxygen uptake (V̇o2), CO2 output (V̇co2), and derived variables determined using a ramp incremental test with continuous collection of expired gases using open-circuit spirometry (Quark B2, Cosmed, Italy). Briefly, the test started with participants cycling at 80 revolutions/min at a power output of 25 watts with a ramp slope of 25 W/min. They cycled continuously until the limit of exercise tolerance, and a confirmatory step test to exhaustion was performed after 5 min of recovery at a relative intensity 5% above their ramp peak power output (46). Data were analyzed using 10-s moving average windows, and V̇o2max was determined as the highest V̇o2 obtained during the incremental test. The first ventilatory threshold (VT1) was determined using a two-criteria confirmation method, comprising the first deflection point in the V̇o2-V̇co2 curve, and the point when the ventilatory equivalent of O2 increased without a concomitant rise in the respective CO2 equivalent (9).
The last three visits were the experimental trials. Participants arrived at the laboratory in the morning, ∼1.5 h after a light breakfast, and rested for ∼30 min during instrumentation. The first measurement took place ∼20 min after cannulation. During the resting control trial (CON) participants remained in a semirecumbent position throughout the protocol (240 min) while blood samples were obtained at baseline and 30, 60, 80, 100, 120, 150, 180, and 240 min of the protocol. Hemodynamic measurements were obtained at similar time points. Exercise trials were performed in random order. For the moderate-intensity exercise trial (MI), 60 min of cycling at 80% of the workrate relative to VT1 (i.e., 46 ± 2% of V̇o2max) was performed followed by 180 min of rest (recovery period), whereas during the heavy exercise intensity trial (HI) participants cycled at VT1 plus 30% of the difference between the workrate relative to their individual VT1 and V̇o2max (i.e., 67 ± 2% of their V̇o2max). The delta method was selected for the HI protocol because it reduces the variability of interindividual physiological responses (35), and the classification of exercise domains as moderate and heavy intensity was based on previous work (60). The three experimental trials were separated by at least 1 wk to reduce potential cumulative effects of blood sampling.
The second set of experiments investigated the impact of exercise-derived microvesicles on endothelial cells. Culture studies were performed by incubating human umbilical vein endothelial cells (HUVECs) with circulating microvesicles obtained from a subset of the previous experimental group at rest (rMVs) or during heavy exercise (exMVs). Endothelial cell capacity to migrate, proliferate, repair a disturbed monolayer, and form tubule-like structures in vitro was assessed. Experiments were performed by supplementing the experimental media with microvesicles. FBS or VEGF was used as positive control, and the supernatant of microvesicle pellets (microvesicle-free plasma) was used as an internal control.
Catheter placement and blood sample storage.
Venous blood samples were taken during the CON, MI, and HI trials. Upon arrival, participants rested in the supine position in a quiet room, and an 18-gauge cannula (BD Venflon; Becton-Dickson) was inserted in a superficial antecubital vein of the arm. A 0.9% NaCl solution (BD PosiFlush; Becton-Dickson) was flushed through the cannula to maintain patency following each blood draw. Samples were obtained without venous stasis, and the first 3 ml of blood were discarded. Collected blood was immediately mixed in tubes containing 0.129 mol/l sodium citrate (Sigma-Aldrich) or 8 mg of EDTA (Sarstedt). Hct, Hb, and lactate were determined in whole blood samples. Platelet-rich plasma was obtained by centrifugation at 3,000 g at 4°C for 10 min. EDTA samples were separated into aliquots in tubes and stored at −80°C. Sodium citrate plasma underwent a second centrifugation step at 15,000 g at 4°C for 10 min to obtain platelet-poor plasma (PPP). Aliquots were stored at −80°C for microvesicle quantification or further cell culture experiments.
Systemic and limb hemodynamics.
Arterial blood pressure and heart rate (HR) were recorded continuously using a Finometer Pro device (Finapress Medical Systems) throughout the three experimental trials. Participant's stroke volume (SV) was determined from apical four-chamber echocardiography (phased-array probe, Vivid 7; GE). Measurements were obtained with participants in the semidecubitus position on their left side, which was accomplished by tilting the recumbent cycle ergometer. End diastolic and end systolic left ventricular areas were determined using EchoPac software (version 112; GE), and volumes were calculated using the single-plane Simpson's method (33). At least three heart cycles were selected for analysis. Cardiac output (Q̇) was calculated as SV multiplied by the corresponding HR. Q̇ and SV were estimated from the Finometer Pro in situations where the echocardiographic recordings could not be obtained or were of low quality. Q̇ of five participants obtained during diverse resting and exercise conditions displayed a moderate agreement between echocardiographic and Modelflow methods, with the mean difference of −0.5 l/min not being different from zero (single-sample t-test P > 0.05; 95% confidence intervals: 4.1 to −5.2), and no proportional bias (P > 0.05).
Upper and lower limb hemodynamics were assessed by vascular ultrasonography with a multifrequency linear array transducer (Vivid 7 ultrasound; GE) to examine the relationship between hemodynamic changes and circulating microvesicle concentrations. The left brachial artery was assessed at rest and during leg exercise. Blood flow was calculated as the product of Vmean × π × (D2/4) × 60; and vascular shear rate (SR) was calculated as 4 × Vmean/D, where Vmean refers to weighted time-averaged mean Doppler velocity, and D refers to vessel diameter (4, 53). The common femoral artery was assessed at baseline and throughout recovery, but it could not be assessed during exercise due to the dynamic nature of cycling. Hence, leg blood flow during exercise was estimated assuming two-leg blood flow = exercise Q̇ − (resting Q̇ + exercise two-arm blood flow).
PPP samples were thawed at room temperature, and Fc receptors were blocked with Human TruStain FcX (BioLegend) for 10 min. Samples were incubated in the dark for 25 min with PE-conjugated anti-human CD62E [E-selectin (BioLegend) at 18 μg/ml] and PE/Cy5 anti-human CD41 [integrin-α2b (BioLegend) at 2.04 μg/ml] fluorescent antibodies. Samples were diluted 1:40 in 0.2 μm filtered PBS and centrifuged for 15 min at 17,960 g and 4°C. The microvesicle pellet was resuspended in PBS. Microvesicle concentration was quantified using an ImageStreamx Mark II imaging flow cytometer (Amnis) using INSPIRE software (Amnis) with ×60 magnification. All events were collected, and sample acquisition was limited to no less than 1,000 events in the expected PMV gate. The a priori choice for PMV gate relied on previous evidence indicating that PMVs are more responsive to an exercise stimulus than EMVs. Single-stained samples were used to create a multistaining compensation matrix.
Data were analyzed off-line with IDEAS software (version 6.1; Amnis) using an adapted microvesicle gating strategy described by others (24). Briefly, after compensation, microvesicles were determined as events below 1 μm diameter using calibration beads (Fluoresbrite; Polysciences) and displayed very low side scatter and moderate mean fluorescence intensity for PE/Cy5 (PMVs) or PE (EMVs) (Fig. 1) (24). A threshold for positive events was set as adapted from the fluorescence minus one procedure.
Plasma norepinephrine and IL-6 quantification.
Plasma norepinephrine and IL-6 concentrations were measured in EDTA samples by commercial ELISA kits [Noradrenalin ELISA (IBL); Human IL-6 Quantikine (R&D Systems)]. Samples were analyzed in duplicate according to the manufacturer's instructions. The mean coefficient of variation between duplicates was 9.7% for norepinephrine and 6% for IL-6.
Whole blood hematocrit, hemoglobin, lactate, and plasma volume corrections.
Blood (1 ml) was immediately separated to determine Hct, Hb, and whole blood lactate concentrations. Hct was determined by the packed cell volume method after standard centrifugation of sodium-heparinized capillary tubes (microhematocrit tubes, HaematoSpin 1400 centrifuge; Hawksley). Blood Hb concentration was obtained by photometric analysis (HemoCue Hb 201+ System; HemoCue), and lactate concentration was determined using a Biosen C-line analyzer (EKF Diagnostics) after daily calibration. Plasma volume shifts were calculated as described previously (17).
Cell Culture Experiments
Culture of human umbilical vein endothelial cells.
HUVECs were cultivated in culture flasks (T75; Sarstedt) in growth medium (medium 199 supplemented with 20% FBS, 15 μg/ml endothelial cell growth supplement, 5 U/ml of heparin, 100 U/ml-100 μg/ml of penicillin-streptomycin, 0.6 mg/ml pyruvate, and 20 mM HEPES) at 37°C in 5% CO2, with half-medium changes every 48–72 h. Cells were removed from wells with trypsin when monolayers became 60–80% confluent for passaging or experiments. Experiments were performed with cells between passage 3 and 6.
Microvesicle experimental medium preparation.
On the day of the experiments, PPP vials were thawed at room temperature, diluted 1:1 in PBS, and isolated by centrifugation for 1 h at 17,500 g at 4°C as adapted from others (6, 55). Experimental media were prepared by suspending the microvesicle pellet to its original concentration in FBS-free growth medium for migration and wound-healing assays, or to double the original microvesicle concentration in medium 199 supplemented with penicillin-streptomycin and heparin for proliferation or tubule formation assays (diluted to the original concentration in subsequent steps of those corresponding assays). Thus, microvesicles were administered at their physiological concentrations determined in plasma samples at baseline [e.g., 24.1 × 103 PMV/μl (25.9, 20.0) and 12.4 × 103 EMV/μl (14.3, 10.6), median (96% confidence intervals), n = 5/assay] or during exercise [e.g., 49.9 × 103 PMV/μl (63.7, 41.8) and 12.7 × 103 EMV/μl (14.3, 10.5), median (96% confidence intervals), n = 5/assay]. The plasma supernatant obtained after centrifugation was used to make separate aliquots of microvesicle-free experimental media. Results from a pilot study confirmed very low concentrations of microvesicles in these plasma supernatant samples (i.e., 9 ± 2% positive microvesicle events compared with the microvesicle pellet suspension; n = 4). Hence, microvesicle-free plasma experimental medium was used as an internal control in the current experiments.
Proliferation of HUVECs was quantified using the thiazol blue tetrazolium bromide (MTT) assay in 0.5% gelatin-precoated 96-well plates. A total of 50 μl of the cell suspension (2,000 cells) was seeded in each well followed by the addition of 50 μl of experimental medium. Experimental medium with 20% FBS served as positive controls. After 48 h each well was incubated with the MTT solution for a further 4 h. All but 30 μl was removed, and the wells were incubated for 10 min in DMSO before absorbance analyses were obtained at 540 nm using a microplate reader (ELx808; BioTek Instruments). The average of quadruplicates was calculated and expressed as a percentage of the negative control condition.
Migration of HUVECs toward microvesicle-containing wells was determined using a modified 48-well Boyden chamber (AP48; Neuro Probe). Experimental medium (rMVs, exMVs, or microvesicle-free supernatant) was loaded in lower-chamber wells to form a slight meniscus, and then a precoated gelatin polycarbonate filter was positioned over the lower chamber. Wells loaded with 10% FBS served as positive controls. The upper chamber was assembled and loaded with 50 μl of a suspension of serum-starved cells in microvesicle-free medium (25,000 cells) and incubated for 4 h. Adherent cells on the underside of the filter were fixed with methanol and stained with giemsa. Three images of each well acquired at random locations were obtained with an Axioskop 2 microscope (Zeiss) at ×20 objective magnification, and the number of migrated cells was counted with ImageJ software (version 1.48; National Institutes of Health). The average of triplicates was expressed as a percentage of positive controls.
Scratch wound-healing assay.
The scratch wound-healing assay was performed to investigate the impact of microvesicles on HUVECs undergoing the repair process. Cells were seeded on gelatin-precoated 96-well plates and incubated at 37°C in 5% CO2 until cells became almost confluent. On the experimental day and after 4 h of serum starvation, a scratch was made by disturbing the cell monolayer with a 200-μl pipette tip. All medium was aspirated and replaced with experimental medium. Medium supplemented with 20% FBS served as positive control. Images were obtained at baseline and after 3.5 h of incubation using an inverted microscope (Axiovert 200M; Zeiss) at ×10 objective magnification. The cell-free area was measured using ImageJ software (version 1.48; National Institutes of Health), and the wound closure was calculated by subtracting the cell-free area obtained postincubation from the baseline measurement. The average of quadruplicates for each condition was expressed as a percentage of the wound closure observed in the negative control condition.
Tubule formation assay.
The angiogenic potential of microvesicles was assessed by a tubule formation assay. Briefly, 96-well plates were coated with basement membrane protein gel (Geltrex; Thermo Fisher Scientific) on ice at 78 μl/cm2. The gel was gently spread with combitip inserts, and the plate was incubated at 37°C to allow the gel to solidify (20). Serum-starved cells were suspended in experimental medium 199 (with penicillin-streptomycin and 0.5% FBS), and 50 μl of cell suspension (10,000 cells) were incubated in Geltrex-precoated wells with 50 μl of experimental medium. VEGF (50 ng/ml) was used as a positive control. Wells were imaged after 24 h of incubation using an AxioVert microscope (Zeiss) at ×5 objective magnification. The number of tubule-like structures and branching points of three fields of view per well were averaged using ImageJ software (version 1.48; National Institutes of Health), and the average of three wells per condition was expressed as a percentage of the results observed in the negative control condition.
List of Materials
Medium 199 (catalog no. SH30024.01; Hyclone), modified medium 199 (phenol red, l-glutamine, and sodium hydrogen carbonate free; catalog no. M3769), MTT (catalog no. M5655), and VEGF (catalog no. RP-75746) were from Fisher Scientific. Geltrex (catalog no. A1413202) and FBS (catalog no. 10270-106) were from Life Technologies. Fc receptor blocking solution (Human Trustain FcX; catalog no. 422302), PE anti-human CD62E (catalog no. 322606), and PE/Cy5 antihuman CD41 antibodies (catalog no. 303708) were from BioLegend, and combitips used for coating plates for angiogenesis assay were from Eppendorf (0.5 ml, catalog no. 12674587).
All data are presented as means ± SE. For the first study, dependent variables over time and conditions were compared by two-way repeated-measures ANOVA. If a significant F ratio was observed for time × condition interactions, the Dunnett's test for multiple comparisons was performed to compare within-condition results vs. baseline values. Between-condition comparisons across time were performed with repeated-measure ANOVA with Bonferroni correction. A within-subject repeated-measures multiple regression was performed to examine the relationship between circulating microvesicles, hemodynamics, and biochemical variables (5). In vitro experiments were compared with one-way repeated-measures ANOVA, and differences between conditions were identified by the least significant difference when a significant F ratio was observed.
Due to technical issues, data of one individual had to be acquired using a different ultrasound system (ProSound SSD5500; Aloka) during the exercise trials, with no data recorded on the control day. Due to the completely random nature of the missing data, a mean substitution treatment was applied for the missing data (i.e., CON trial), instead of the listwise deletion procedure. In addition, Hb results could not be collected in the MI protocol for one participant because of technical limitations, so a similar mean substitution data treatment was used.
General statistical analyses were performed using statistical software (SPSS version 20; IBM), with Dunnett's test derived from GraphPad Prism software (version 5.03; GraphPad Software), and the within-subject repeated-measures multiple regression calculated with SigmaPlot (version 13; Systat Software), always setting the significance level at α < 0.05 for all analyses.
Limb and Systemic Hemodynamics
Estimated leg blood flow increased during cycling as a function of exercise intensity (P < 0.05), and measured leg blood flow remained elevated at 5 min of recovery with more pronounced values after HI exercise (P < 0.05) (Fig. 2C). Accordingly, early recovery femoral artery mean SR was higher than baseline, with greater values after heavy cycling compared with moderate exercise (P < 0.05) (Table 1). This blood flow and SR response occurred mostly through a rise in blood velocity to the lower limbs (P < 0.05), with arterial vasodilation (increases in conduit vessel diameter) playing a smaller role (Table 1).
During exercise, the brachial artery dilated and mean blood velocity increased (P < 0.05) (Table 1), with a resultant rise in arm blood flow during exercise and into early recovery (P < 0.05) (Fig. 2D). The higher blood flow to upper limbs during heavy compared with moderate exercise was mostly driven by changes in blood velocity, so mean brachial artery SR was augmented during cycling as a function of exercise intensity (P < 0.05), and remained elevated at 5 min of heavy exercise recovery (P < 0.05), returning to resting values by 20 min.
Cardiac output was elevated throughout cycling in proportion with exercise intensity (P < 0.05) (Fig. 2A) as a result of increases in HR and SV, with corresponding adjustments in left ventricular end-diastolic and end-systolic volume (Table 2). In the first 5 min of recovery after HI exercise, Q̇ remained higher than baseline and CON (P < 0.05), whereas it returned to values similar to rest following the MI protocol (P > 0.05). Mean arterial pressure increased during exercise, reaching higher values during the HI compared with the MI protocol at 40 min (P < 0.05) but quickly returned to baseline at the end of exercise. Mean arterial pressure at 5 min of recovery, however, was lower following the HI compared with MI protocol, reflecting a tendency for postexercise hypotension after prolonged heavy exercise (Fig. 2B).
Circulating Platelet and Endothelial-Derived Microvesicles at Rest and During Exercise
PMV and EMV concentrations were stable throughout the 4 h of resting control and did not change with moderate exercise (P > 0.05). Similarly, EMV concentrations were unaltered by heavy exercise (Fig. 3B). Yet, PMV concentrations increased more than twofold by 30 and 60 min of heavy exercise (P < 0.05; Fig. 3A) after which they decreased. However, a second rise above baseline was evident at 60 min of recovery (P < 0.05). The PMV area under the time-concentration curve was larger during HI compared with CON and MI trials (154 ± 16 vs. 89 ± 6 and 108 ± 9 × 103 PMV × h/μl, respectively) (P < 0.05). There were no differences in the EMV area under the curve between CON (46 ± 3 × 103 EMV × h/μl), MI (46 ± 1 × 103 EMV × h/μl), and HI (48 ± 3 × 103 EMV × h/μl) trials. Changes in plasma volume did not affect the increases in PMV during and after heavy exercise, and thus the aforementioned results represent uncorrected values.
Plasma volume was reduced by between 10 and 13% during exercise (P < 0.05) but returned to resting values early in recovery after moderate exercise (P > 0.05) (Table 3), with the change in plasma volume still evident 20 min after heavy exercise (P < 0.05). Because changes in plasma volume did not affect the overall microvesicle results, data of the remaining blood-derived variables are also presented without correction. Plasma norepinephrine concentration rose from baseline levels during heavy cycling (P < 0.05), with higher values observed throughout the heavy exercise period compared with resting control and at 30 min of exercise compared with moderate cycling (P < 0.05) (Table 3). Norepinephrine concentration returned to baseline within the 1st h of recovery after heavy exercise, remaining stable until the end of recovery (P > 0.05). No changes from baseline were observed during resting control or moderate exercise (P > 0.05). By the end of 1 h of heavy cycling, IL-6 concentration increased compared with moderate exercise and resting control (P < 0.05) (Table 3), and values remained elevated above baseline throughout recovery. IL-6 concentrations also increased from baseline in the CON and MI trials during the 3rd and 4th h of the protocol (P < 0.05) so that at the end of the recovery period no differences were observed between protocols.
Relationship Between Microvesicles, Shear Rate, and Blood-Derived Variables In Vivo
Modest yet significant correlations were found between PMVs, vascular SR, and norepinephrine concentrations when all time points were investigated (P < 0.05, Fig. 4, A–C). Exploratory analysis revealed that PMV concentrations moderately correlate with brachial artery SR (r2 = 0.43; P < 0.05; Fig. 4E), estimated femoral artery SR (r2 = 0.48; P < 0.05; Fig. 4F), and, to a lesser degree, norepinephrine levels (r2 = 0.21; P < 0.05; Fig. 4G) before and during exercise but not during baseline and recovery time points only (P > 0.05). PMVs displayed no significant correlation to IL-6 within these conditions (P > 0.05).
Circulating Microvesicles Increase Endothelial Cell Proliferation and Migration
HUVECs proliferated almost 50% quicker than the negative control condition when treated with rMVs, and proliferation almost doubled from control with exMVs (Fig. 5A), with smaller effects observed with microvesicle-free supernatant conditions. The exMV treatment increased HUVEC proliferation compared with the rMV and both the resting and exercise supernatants (P < 0.05). rMVs also increased HUVEC proliferation compared with resting supernatant, but no more than the exercise supernatant condition (P > 0.05). No differences in cell proliferation were observed between resting and exercise supernatants (P > 0.05). Both microvesicle and supernatant conditions stimulated HUVEC migration, as depicted in Fig. 5B. The microvesicle treatments, however, induced substantially more endothelial migration compared with their respective supernatants (P < 0.05). A greater number of cells migrated toward wells loaded with exMVs compared with rMVs (P < 0.05), but no difference was observed between the two microvesicle-free supernatant conditions.
Microvesicles Improve Endothelial Scratch Wound Healing
HUVECs treated with supernatants displayed an almost twofold increase in the rate of repair compared with negative control wells, but no difference was observed between resting and exercise supernatants (P > 0.05, Fig. 5C). Both microvesicle conditions enhanced the closure rate compared with the supernatant conditions, with rMVs stimulating wound healing to a similar extent to that observed in the FBS positive control wells (i.e., nearly 3-fold quicker than the negative control), and this process was even more rapid in the presence of exMVs (P < 0.05).
Exercise Microvesicles Stimulate Angiogenesis In Vitro
Treatment with exMVs induced greater formation of tubule-like structures compared with all other conditions (P < 0.05, Fig. 6A), with no differences observed among rMVs and supernatant treatments (P > 0.05). The number of branching points was also increased in endothelial cells treated with exMVs compared with those incubated with microvesicle-free supernatant (P < 0.05, Fig. 6B), although exMV treatment did not increase the number of branching points compared with rMVs (P > 0.5).
This study investigated the time course of plasma microvesicle appearance in the venous circulation during and after prolonged cycling to gain insights into the relationships between exercise intensity and potential physiological stimuli inducing microvesicle formation in vivo and to establish the physiological relevance of intravascular microvesicles produced during exercise on human endothelial cells. We have shown, for the first time, that the increase in circulating PMV concentrations occurs not only after, but also during, exercise within the heavy-intensity exercise domain, with a second peak observed at 1 h of recovery suggesting a biphasic response. In addition, exercise within the moderate exercise domain did not elicit a response. The PMV dynamics during exercise moderately correlate with hemodynamic measurements, suggesting that augmented vascular shear stress may be involved in PMV formation in exercising humans. Moreover, we have taken the first step in understanding the effect of microvesicles produced during exercise on the vascular endothelium by demonstrating that exMVs display a stimulatory effect in cultured endothelial cells, revealing a potential link between intravascular microvesicles and vascular adaptation with exercise.
Time Course of Microvesicle Appearance with Moderate and Heavy Exercise
Earlier attempts to characterize PMV dynamics to exercise generally agree that platelets introduce microvesicles in the circulation during the recovery period, with peak plasma concentrations occurring somewhere within 1 h of the session (11, 36, 48, 49). However, little attention has been placed on microvesicle temporal kinetics, and no study has explored the PMV dynamics during exercise. Our results show for the first time that elevations in PMV concentrations are not limited to the postexercise recovery period, since elevations occurred as early as 30 min into heavy exercise. The concentrations of PMVs increased approximately twofold from baseline during heavy exercise in the present study, which tended to be higher than the 20- and 40-min postexercise values but similar to values after 1 h of recovery. The present 60-min recovery data agree with findings of previous studies using traditional flow cytometry in which two- to threefold increases in PMV concentration were observed after 1 h of recovery (11, 48, 49). This increase in PMVs seemed to be related to exercise intensity rather than being a function of the total work done, since PMV concentrations did not increase during moderate exercise even though a greater amount of mechanical work was accomplished by the end of this condition compared with 30 min of heavy cycling. Moreover, the fact that PMVs increased from baseline and then remained stable from 30 to 60 min during exercise probably indicates a rapid microvesicle turnover that should be confirmed using tracer techniques. Several processes may account for PMV removal, as suggested by in vitro experiments. Specifically, evidence for endothelial and renal epithelial cell uptake exists (10, 50). Infused endothelial progenitor cell-derived microvesicles were localized in renal endothelial and tubular cells of rats during recovery from renal injury, suggesting the kidney as a microvesicle extractor organ within these conditions (10). Besides the fact that microvesicles, including PMVs and EMVs, have been identified in the urine of healthy humans (57), it is still difficult to assess the fate of microvesicles with exercise from these experiments, even more so when one considers that renal blood flow (and thus microvesicle delivery) is reduced during intense exercise (51).
The second rise in PMV concentration observed 1 h after the heavy-intensity bout is intriguing while still in agreement with authors who note peak PMV values between the end of exercise and 1 (36, 48, 49) or 2 (11, 48) h of recovery. Our biphasic response is a function of the improved temporal resolution of our sampling protocol since previous work did not measure the PMV response during early exercise recovery. Our observations imply a mechanism unrelated to shear rate induces platelet activation after 1 h of exercise and causes an accumulation of PMVs in the circulation comparable to the exercise response. This secondary PMV peak might be related to late platelet recruitment, since exercise may lead not only to platelet activation (12, 15) but also to acutely increase platelet concentrations after exercise sessions (15, 27) recruited from various platelet pools, including the spleen and bone marrow (18), that then serve as a source of new microvesicles.
EMV concentrations remained similar to baseline across all experimental protocols, implying that the two submaximal exercise stimuli had little impact on endothelial vesiculation. Differing from PMVs, microvesicles derived from endothelial cells have already been studied during exercise, with no changes observed during moderate-intensity cycling (40). Inconsistent results exist regarding EMV kinetics during recovery, with researchers reporting increases (31, 34, 48), no change (11, 23, 40, 49), or even decreases (59) after exercise. For instance, Kirk et al. (31) reported elevated plasma CD105+ and CD106+ EMVs 1.5 h after supramaximal interval cycling. Endoglin (CD105) expression is known to increase in endothelial cells undergoing angiogenesis (41), whereas the expression of vascular adhesion molecule-1 (CD106) is upregulated in the endothelium under inflammatory stimulation (62), which indicates that the strenuous protocol used by Kirk et al. (31) may have caused endothelial activation. The fact that CD62E+ EMVs were elevated in men in a previous study (34) but not in the current experiment is intriguing and may be related to their slightly higher intensity protocol (∼70% V̇o2max on an upright cycle ergometer). In agreement, results from our laboratory indicate that EMVs expressing the same markers increase during strenuous exercise conditions (i.e., 80% of peak power output under heat stress, E. N. Wilhelm, J. González-Alonso, S. T. Chiesa, S. J. Trangmar, and M. Rakobowchuk; unpublished observations), suggesting that higher exercise intensities may be required to induce endothelial vesiculation.
Relationship Between Platelet Microvesicles and Putative Agonists
The physiological mechanisms eliciting PMV formation during exercise have not been elucidated, and based on previous in vitro studies we thought that shear stress might be a potential candidate (39, 44). Hence, a novel finding of this study was the correlation between PMV concentrations and vascular SRs during exercise coupled with a weaker relationship with norepinephrine in these conditions. The latter relationship also seems logical, since sympathetic nervous system activity has been proposed to play a role in hemostasis and catecholamines have been reported to activate platelets (58), with norepinephrine reported to stimulate PMV formation in vitro (54). The major role of shear stress on PMV formation in vivo still has to be fully elucidated in experiments where shear stress is the single variable manipulated, for example, by independently increasing blood flow with the infusion of vasodilators, but the positive associations observed in the current study provides some support for shear stress as a candidate mediator of PMV formation during exercise, with norepinephrine playing a synergistic role within these experimental conditions. The fact that PMVs did not increase markedly during moderate exercise despite substantial increases in shear is interesting and indicates a shear threshold may be required to stimulate PMV formation. Additional mechanisms may involve the purinergic activation of platelets, since increased concentrations of intravascular nucleotides during heavy exercise stimulate platelet activation ex vivo (63), but the actual link between circulating PMVs and nucleotides should be determined.
Effect of Circulating Microvesicles on Cultured Endothelial Cells
Evidence from in vitro studies outside the context of exercise suggests that microvesicles may be involved in vascular adaptations. Endurance training induces angiogenesis in active tissues in animal (26) and human (22, 28) models, which may be stimulated through many mechanisms, including hypoxia (16) and local vascular shear stress (21), with VEGF suggested as the major, albeit not the only, growth factor implicated (22, 26, 28). Because high-intensity training enhances skeletal muscle capillarization (28), and acutely elevates the concentrations of circulating PMVs in humans (as demonstrated in the present study), it seems plausible that microvesicles could play a role in exercise-induced angiogenesis. A unique finding of this study was that exMVs increased the number of endothelial tubule-like formations compared with rMVs and their respective microvesicle-free supernatants, suggesting that microvesicles produced during exercise have proangiogenic potential. The pathways through which exMVs bring about their endothelial effects are unknown but may relate to the delivery of VEGF (2, 7) and biologically active lipids (30) to recipient endothelial cells. Furthermore, microvesicles obtained from thrombin-stimulated platelets have been shown to display angiogenic potential in vitro (7, 30), and in vivo (7), which may indicate that PMV could be the microvesicle population stimulating cultured endothelial cells in the present experiments.
Angiogenesis depends on orchestrated endothelial cell migratory and mitogenic events, which in turn are stimulated by growth factors released where new capillaries will infiltrate (22, 28). By staining the proliferation-related protein Ki-67, Jensen et al. (28) demonstrated that the increased skeletal muscle capillarization observed after training is indeed the result of migration and proliferation of endothelial cells to form new capillaries. Our data indicate that microvesicles may be involved in this intricate process, since exMVs enhanced both endothelial migration and proliferation in individual assays. Endothelial scratch wound healing (a complex process highly dependent on cell migratory and proliferative capacities) was also elevated with exMV treatment, supporting the findings of individual assays. Previous work outside the context of exercise demonstrates that PMVs induce a dose-dependent proliferation and migration of endothelial cells (30), and may substantiate endothelial wound repair through direct and indirect stimulatory mechanisms (30, 37). It is worth noting that exercise increased the concentration of PMVs in the intravascular space, which might seem counterproductive since chemotactic stimulation on the endothelial apical side would stimulate migration in the vessel lumen. PMVs, however, have been reported to adhere to proteins of the extracellular matrix, serving as a binding site for platelets on the vessel wall at regions of extracellular matrix exposure (38), and platelets have been shown to transmigrate to the subendothelial space under certain conditions (32) where they could deliver MVs to the basal surface of endothelial cells, but the existence of similar mechanisms during exercise is unknown.
The current series of studies do have some limitations. It is important to note that correlations do not necessarily represent causation, and the present data only provide indirect evidence of shear stress as a potential stimulus for PMV formation during exercise. Further studies that manipulate shear stress while sampling arterial and venous microvesicles across exercising limbs should be performed to directly evaluate the relevance of shear stress in PMV release with exercise, as well as to address the source of microvesicles (e.g., organs or exercising limbs). Current methods for microvesicle isolation do not enable separation of specific populations from human plasma without ex vivo stimulation, and, even though it seems logical that PMVs were the primary mediators of our findings, we cannot rule out the possibility that other microvesicle populations may be involved. Furthermore, it remains unknown whether the influence of exMVs on endothelial cells resulted simply from their greater concentration or distinct intrinsic characteristics (e.g., different cargo). Finally, it is important to keep in mind that observations taken from static cell culture experiments are not necessarily transferable to other cell lines and to complex whole organism models.
In conclusion, submaximal exercise is a potent physiological stimulus that triggers PMV formation during and 1 h postexercise, with no impact on EMVs. This phenomenon, however, is exercise intensity dependent and requires a substantial stimulus. The rise in PMVs during exercise was coupled with changes in vascular shear stress, and plasma norepinephrine concentration, with both variables explaining part of microvesicle dynamics during exercise. Moreover, human circulating microvesicles produced during exercise increased the angiogenic potential in cultured endothelial cells, which is supported by their mitogenic and chemotactic enhancements. Together, this set of integrative physiology experiments in healthy young humans characterized the physiological time course of PMV and EMV appearance during and after submaximal exercise and provides the first evidence that exercise-derived microvesicles play an important biological role within the human vasculature, suggesting a novel mechanism that may help us to further understand how exercise mediates vascular adaptation.
M. Rakobowchuk received a Physiological Society Research Grant. E. N. Wilhelm was supported by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (Bolsista CAPES, Proc. No. BEX 134/13-6; CAPES Foundation, Ministry of Education of Brazil, Brasília, DF 70040-020, Brazil).
All authors declare no conflict of interests in relation to this work.
E.N.W., J.G.-A., and M.R. conception and design of research; E.N.W. and M.R. performed experiments; E.N.W. and C.P. analyzed data; E.N.W., J.G.-A., and M.R. interpreted results of experiments; E.N.W. and M.R. prepared figures; E.N.W., J.G.-A., C.P., and M.R. drafted manuscript; E.N.W., J.G.-A., C.P., and M.R. edited and revised manuscript; E.N.W., J.G.-A., C.P., and M.R. approved final version of manuscript.
We thank all of the participants for taking part in this study, Dr. Adrian Chester from Imperial College at the Heart Science Centre Harefield Hospital for providing the endothelial cell isolated from human umbilical cords, and Dr. Kameljit Kalsi for technical assistance with the cell work at the start of the project.
- Copyright © 2016 the American Physiological Society