Men and women display distinct extracellular vesicle biomarker signatures in response to military operational stress

Abstract

Extracellular vesicles (EVs) are mediators of physiological changes that occur during physical exertion. This study examined the effects of physical exertion with and without sleep and caloric restriction on EV size, concentration, and surface proteins in men and women. Twenty participants (10 men) completed a 5-day simulated military operational stress protocol with daily physical exertion. Blood was drawn before and immediately after exertion at baseline (D1) and following 48-h of sleep and caloric restriction (D3). EV size and concentration were assessed using nanoparticle tracking analysis. EVs were identified with markers associated with exosomes (CD63), microvesicles (VAMP3), apoptotic bodies (THSD1), and skeletal muscle-derived EVs (SGCA) and quantified using imaging flow cytometry. Interactive and main effects of sex, day, and time on EVs were assessed using three-way ANOVAs. EV concentration declined pre to postexertion in women on D1 and D3 but was stable in men. EV size increased from pre to postexertion and from D1 to D3 in men and women. Physical exertion following sleep and caloric restriction increased CD63+ EV concentration, proportion of total EVs, and CD63 surface protein expression regardless of sex. The proportion of SGCA+ EVs increased in men and women following exertion and from D1 to D3 but was higher in women than in men. No differences were observed in VAMP3+ and THSD1+ EVs. This study identified sexually dimorphic EV profiles in response to various stressors. Further investigations are necessary to determine if dimorphic EV responses affect health and performance outcomes during stress.

NEW & NOTEWORTHY Sex is understudied in EV research, and most studies limit EV analysis to single stress conditions such as exercise. Multistress conditions consisting of physical exertion and sleep and caloric restriction are common in real-world settings. We demonstrate that physical exertion results in sex-specific EV signatures and that EV profiles vary according to single versus multistress conditions. Our data highlight important biological and ecological characteristics that should be considered in EV research.

INTRODUCTION

Multicellular organisms coordinate an array of physiological functions through a complex network of communication systems that promote near and distant tissue cross talk (1). Circulating hormones have been the focus of endocrine research and biomarker discovery for decades, but, more recently, extracellular vesicles (EV) have emerged as a novel mechanism through which cells communicate in an autocrine, paracrine, and endocrine manner (2). EVs are now being recognized as mediators of biological function, such as improving insulin sensitivity (3), browning of white adipose tissue (4), and increasing angiogenesis (5), through delivery of their cargo to targeted cells. Mostly studied in the context of pathological states such as cancer (6, 7), cardiovascular disease (8, 9), and neurodegenerative disease (10, 11), EVs have provided novel insights into disease diagnosis, progression, and prognosis, with high sensitivity and specificity, earning the nomenclature as a “liquid biopsy” (12). However, comparatively less is known about changes in EV characteristics in the context of stress physiology. Investigation of EVs under physiologically relevant stress related conditions has the potential to provide important advancements in our ability to objectively monitor and quantify physiological stress and identify targets for future therapeutic applications (13, 14).

For decades, military research has provided important insights into stress physiology with far-reaching implications dating back to the early seminal work of Ancel Keys and colleagues, who showed the negative impact of starvation and prolonged physical work on the health and physical fitness of men (15, 16). Since then, researchers have followed the trail blazed by Dr. Keys by detailing the impact of stress during military operations on Soldier readiness and performance (16–20). Specifically, Soldiers undergoing intense and/or prolonged military operations experience sharp declines in physical performance, including measures of strength, power, speed, and endurance, which may take days or weeks to recover (17, 21–23). Deterioration of physical capacity is often preceded by or in conjunction with declining anabolic status, compromised immune status, and increased activation of the hypothalamic–pituitary–adrenal axis (19, 20, 24, 25). As such, military operational stress, which consists of the combination of rigorous physical and cognitive demands, sleep restriction/disruption, undernutrition, and/or environmental extremes, serves as a useful model for studying human stress physiology and provides good ecological validity with applications that extend beyond the military to other populations such as emergency personnel, shift workers, and athletes to name a few.

Circulating hormones (e.g., cortisol, insulin-like growth factor-I, testosterone) have conventionally been used to objectively monitor and quantify stress in military research. More recently, EVs have gained notoriety as biomarkers of pathology and health (2, 12, 14). EVs are a heterogeneous population of nano-sized particles that can be classified by their biogenesis into exosomes (endosomal origin), microvesicles (ectosomal origin), and apoptotic bodies (ectosomal origin) (2). EVs can be further differentiated based on biophysical characteristics, such as size, density, and morphology. EVs are characterized by a lipid bilayer membrane, which makes them stable in the extracellular environment, and are enriched with lipids, proteins, and nucleic acids according to the state of their parent cell during EV biogenesis (2). This makes them information-rich packages that are analogous to a “biological signature.” Once released into the extracellular space, EVs can be collected and isolated from virtually any biological fluid and analyzed to decipher both cell-specific and systemic messages from and across a range of tissues. Their ubiquitous presence in biological fluids, stability in the extracellular environment, and robust cargoes makes them appealing targets for biomarker research.

Insights into the physiological functions of EVs may be observed in their features, including size, shape, morphology, and cargo (2). For example, size and shape may suggest a particular class of EV, with larger (50–5,000+ nm) and more irregular particles being indicative of apoptotic bodies and medium (100–1,000 nm) or smaller (30–150 nm) and more circular vesicles associated with microvesicles and exosomes, respectively (26). Similarly, EV surface markers may be related to their origin or biogenesis pathway with markers such as CD63, vesicle-associated membrane-3 (VAMP3), and thrombospondin-1 (THSD1) present in exosomes, microvesicles, and apoptotic bodies, respectively (27). α-Sarcoglycan (SGCA) has also been used as a marker for skeletal muscle-derived EVs, which is useful for studying EVs in the context of exercise and muscle activation (28–30).

Extracellular vesicles have primarily been studied under isolated stress conditions, and few data exist comparing men and women. Specifically, acute exercise (mostly aerobic) and, to a lesser extent, nutrient manipulation have been two of the more commonly tested physiological conditions (28–35). In studies comprised of men only or those that include both sexes but lack a comparison of men and women, aerobic exercise has been shown to cause an acute rise in circulating EV concentration followed by return to baseline within minutes to hours postexercise (28, 35–37). Studies that directly compare men and women have reported an effect of sex whereby EV concentration is unaltered in men but decreases in women following exercise (30, 33). Previous data also indicate an effect of exercise intensity on EV release (33, 38). Wilhelm et al. (38) observed a more than twofold increase in platelet-derived EVs following 1 h of high-intensity cycling exercise (67 ± 2% V̇o_2max) compared with control, whereas there was no difference between moderate intensity exercise (46 ± 2% V̇o_2max) and control. In a study by Shill et al. (33), there was an 18% decline in endothelial-derived EVs immediately following continuous exercise (65% V̇o_2max) compared with baseline but no difference from baseline following energy-matched high-intensity interval exercise. EV mean size is unchanged with exercise (39). EV subpopulations, as defined by specific surface protein markers, are altered by exercise as well. Specifically, EVs positive for CD63, CD81, and CD9—tetraspanins commonly associated with exosomes—have been observed to increase with exercise (39). Others have also reported a significant increase in SGCA+ EVs following moderate to vigorous intensity aerobic exercise (29, 30). A similar rise and fall of EV concentration has been noted following food intake (31, 34, 36, 40), and there are some data to suggest that exercise and nutrient intake may interact (34). However, the interaction of multistress environments common in real-world settings on EV characteristics remains largely unexamined.

Using a simulated military operational stress model, the purpose of this study was to compare the effects of two stress conditions—physical exertion only versus physical exertion plus combined sleep and caloric restriction—on EV size, concentration, and surface proteins according to sex. Based on previous findings, we hypothesized that 1) EV concentration would decrease following physical exertion in women only (30, 33), 2) EV size would remain stable in men and women across stress conditions (39), and 3) EV surface markers would increase from pre to postexertion (39). In addition, we explored whether there would be differences in EV surface markers in response to physical exertion during sleep and caloric restriction compared with physical exertion alone, and whether the responses would vary by sex. The results from this study will elucidate the interactive effects of multistressor conditions common in real-world settings and delineate EV features that differ by sex for consideration in future studies.

METHODS

Data presented here are a subset of a larger prospective cohort study (US Department of Defense Award No. W81XWH-17-2-0070) from which some data have been published previously (41, 42). The University of Pittsburgh Institutional Review Board approved all study procedures. Research activities were compliant with the US Army Medical Research and Development Command Human Research Protection Office and in accordance with the Declaration of Helsinki. This study adhered to the requirements of the U.S. Federal Policy for the Protection of Human Subjects (45 CFR, Part 46).

Subjects were recruited via electronic correspondence, flyers, and in-person briefings. Written informed consent was provided before participation. Subjects were free to withdraw at any time. Male and female service members between 18 and 41 yr old were eligible for the study. Subjects were physically fit and had to meet their service-specific physical fitness standards to qualify for the study. Individuals were excluded if they were pregnant or had moderate to severe obstructive sleep apnea as determined by an apnea hypopnea index of ≥ 15, measured on the first night of the study.

Experimental Procedures

This study used a repeated measure, prospective cohort experimental design and was conducted in a controlled, laboratory setting. The protocol consisted of 5 days of simulated military operational stress characterized by daily physical and cognitive exertion and simulated marksmanship (Fig. 1). The details of the parent study have been previously reported (41, 42). In short, subjects arrived on day 1 to provide written informed consent and complete additional screening procedures before participation. On day 0, subjects completed physiological testing and were familiarized to all tasks. Physiological testing included a V̇o_2peak test using the Bruce protocol (43) on a Woodway treadmill (Woodway; Waukesha, WI), vertical and broad jump measurements, and a body composition assessment measured via air displacement plethysmography (BOD POD; Cosmed, Concord, CA). Maximum strength of the lower limbs was assessed via bilateral isometric knee extensions using an S-type load cell sampling at 2,000 Hz (SSM-AJ-500, Interface Inc., Scottsdale, AZ). Subjects were seated with ankle-, knee-, and hip joint angle fixed at 90° and performed a series of four 3–5 s maximal voluntary contractions (MVC) with at least 1 min of rest between sets. Verbal encouragement was provided and trials displaying excessive postural deviations were excluded from analysis. MVCs were averaged between legs, and only the best of the four trials was retained for analysis. Individual caloric needs were determined based on total daily energy expenditure estimated from air displacement plethysmography. On day 1 (baseline), subjects were afforded 100% of their individual estimated caloric needs and 8 h of sleep opportunity on night 1 (2300–0700 h). On days 2 (stress onset) and 3 (peak stress), caloric intake was restricted to 50% of estimated caloric needs, and subjects were only allowed two 2-h blocks of interrupted sleep from 0100–0300 and 0500–0700 h on nights 2 and 3. Sleep was restored to 8 h of uninterrupted sleep on night 4 (recovery). Day 3 was defined as “peak stress” based on 1) the accumulation of 48 h of sleep and caloric restriction and 2) decreased physical and cognitive performance as reported in previously published data from the same study (41, 42). Meals were standardized and consumed each day at ∼0830, 1530, and 2000 h, except on days 2 and 3 when the third meal was not provided during caloric restriction. All study requirements were completed and subjects were released before the dinner meal on day 4. Water was consumed ad libitum, and caffeine and alcohol were prohibited during the entire study period. Sleep was monitored using standard polysomnography.

Figure 1.

Overview of the 5-day simulated military operational stress protocol. Dashed lines indicate days/nights with caloric and sleep restriction. A single black box indicates uninterrupted sleep from 2300 to 0700 h. Two small black boxes indicate interrupted sleep consisting of two 2-h blocks from 0100 to 0300 h and 0500 to 0700 h. *Subjects were released before dinner on the final day.

Each day, subjects completed a physical testing battery designed to mimic common military occupational tasks and was completed in an obstacle course format. The physical exertion protocol consisted of nine consecutive events, with minimal rest between each event. Events were conducted in the following order: unloaded and loaded (16-kg vest) vertical jumps, 2-min water can carry (20 kg in each hand), fire and movement course, 20-m casualty drag (91 kg), unloaded 300-m shuttle run, loaded (16-kg vest) 300-m shuttle run, unloaded and loaded (16-kg vest) vertical jumps, 4-mile loaded (16 kg) ruck march, and unloaded and loaded (16-kg vest) vertical jumps. Standardized rest periods were administered for 3 min before the unloaded and loaded shuttle runs and 10 min before the ruck march. The physical exertion protocol began at ∼1200 h each day and took ∼90 min to complete. Subject ratings of perceived exertion were recorded by a test administrator before and after each event using a 6–20 Borg scale (44), with 6 indicating little to no effort and 20 indicating maximal effort. The 6–20 Borg scale is a reliable indicator of overall effort and fatigue and was designed to correspond with heart rate (e.g., RPE of 15 corresponds to a heart rate of 150 beats/min) (44, 45).

Blood Processing and Storage

Blood was drawn from an upper arm vein before and immediately after the physical exertion protocol using a 21-gauge or 23-gauge butterfly needle (Becton, Dickinson and Company Vacutainer, Franklin Lakes, NJ). Ethylenediaminetetraacetic acid (EDTA)-lined tubes were immediately placed on ice, and serum tubes were allowed to clot for 30 min at ambient temperature. Within 60 min of blood draw, samples were centrifuged at 1,500 g at 4°C for 15 min. Platelet-poor plasma and serum supernatants were aliquoted and stored at −80°C for later analysis.

Blood Assays

Serum myoglobin, creatine kinase, estrogen, and progesterone concentrations were measured using enzyme-linked immunoassays according to each manufacturer’s protocol. Myoglobin (Alpco 25-MYOHU-E01; Salem, NH) assays had a range of 25–1,000 ng/mL and sensitivity of 5 ng/mL. Creatine kinase MM (LSBio LS-F20706-1; Seattle, WA) assays had a range of 1.563–100 ng/mL and sensitivity of 0.94 ng/mL. Estradiol (Alpco 11-ESTHU-E01; Salem, NH) assays had a range of 20–3,200 pg/mL and sensitivity of 10 pg/mL. Progesterone (Alpco 11-PROHU-E01; Salem, NH) assays had a range of 0.3–60 ng/mL and sensitivity of 0.1 ng/mL. Each analyte was measured in duplicate with coefficients of variation of ≤ 10%. Myoglobin and creatine kinase were measured from blood drawn pre and postexertion and days 1 and 3. Estrogen and progesterone were measured from morning blood drawn in a fasted state on day 0.

Extracellular Vesicle Analysis

We used the MIFlowCyt-EV framework as a guide for reporting preanalytical factors, sample preparation, assay controls, and data acquisition to promote experimental reproducibility (46).

Size exclusion chromatography.

Extracellular vesicles were isolated from plasma using size exclusion chromatography (SEC) (qEV 70 nm Original; IZON, Medford, ME) according to the manufacturer’s protocol. Thawed plasma samples were centrifuged at 1,500 g at 4°C for 10 min using a FiberLite F21-48x1.5/2.0 fixed angle rotor (Thermo Fisher Scientific, Waltham, MA) to remove any cells or large particles. Following stored conditions, SEC columns were flushed with 10 mL of 0.22-μm filtered phosphate buffer solution (PBS) before loading with 450 μL of plasma. The first 3 mL of eluate was discarded, and the subsequent 1.5 mL was collected as the EV fraction. This process was repeated for each sample. SEC columns were used no more than five times per column and were flushed with 15 mL of 0.22-μm filtered PBS between samples.

Nanoparticle tracking analysis.

Extracellular vesicle size and concentration (particles/mL) were determined using nanoparticle tracking analysis (NanoSight NS300, Malvern Panalytical Ltd, Malvern, UK) equipped with a 532 nm (green) laser. Samples were analyzed by a single, trained user. Ten microliters from each EV sample were diluted 1:100 in type 1 EV-free water and infused into the flow cell using a syringe pump (Harvard Apparatus 98–4730). Three 45-s videos were recorded for each sample, with the camera level set to 14. The flow cell was flushed with 1 mL of type 1 water between each sample. All samples were batch analyzed using computer software (NTA 3.4, build 3.4.003). Samples from each within-subject time point were assessed on the same day. The remaining isolated EV sample was divided into 150-μL aliquots and stored at −80°C for later analysis.

Staining.

Isolated EVs (suspended in PBS) were fixed for 10 min using equal parts (140 μL each) of sample and 4% paraformaldehyde. Fixed samples were then centrifuged for 30 min at 16,000 g at 4°C before removing 140 μL of supernatant and subsequently adding 140 μL of blocking buffer (3% bovine serum albumin, 0.1% Triton-X), then incubating on a rocker for 1 h at ambient temperature. Following the 1-h incubation, samples were centrifuged for 30 min at 16,000 g at 4°C and 140 μL of supernatant was removed. Samples were then stained with the following antibodies: 0.5 μL anti-human CD63 Alex Fluor 700 (1:280, NBP2-42225AF700, Novus Biologicals, CO), 0.5 μL anti-human vesicle-associated membrane-3 (VAMP3) Alex Fluor 405 (1:280, NBP1-97948AF405, Novus Biologicals, CO), 1.4 μL anti-human thrombospondin-1 (THSD1) Alex Fluor 594 (1:100, FAB5178T-100UG, Novus Biologicals, CO), and 0.35 μL anti-human α-sarcoglycan (SGCA) FITC (1:400 dilution, orb29665, Biorbyt, MO). Stained samples incubated overnight at 4°C. The following day, samples were centrifuged for 30 min at 16,000 g at 4°C before removing 60 μL and adding 20 μL for a final volume of 100 μL.

Single-stained compensation controls (100 μL UltraComp eBeads Plus; Invitrogen) were used to correct for fluorescence carryover between channels. Fluorescence minus one (FMO) controls were used to establish gates for each spectral channel. Compensation and FMO controls followed the same staining protocol as outlined above with the exception that antibody volumes were half that of the volume used in samples. In a subset of samples, buffer only, buffer plus reagent, and isotype controls were prepared using the same procedures as sample preparation. There were no differences in scatter or fluorescent signals between these controls and samples. Therefore, buffer only, buffer plus reagent, and isotype controls were excluded from the remaining data acquisition. Single-stained compensation and FMO controls were used for all sample acquisition and analysis.

Imaging flow cytometry.

Imaging flow cytometry (ImageStreamX Mark II, EMD, Millipore Sigma, Seattle, WA) combines conventional flow cytometry capabilities, including forward and side scatter detection, with up to 10 fluorescent channels and two channels of high-resolution microscopy of up to ×60 magnification in a high-throughput manner (47). Data were collected using INSPIRE software with the following settings: normal gain mode, objective ×60, slow speed, high sensitivity, 7 μm core size, auto-focus, and auto-centering. Data were collected for 3 min for all EV samples, and 2,000 events were collected for compensation and FMO controls. SpeedBeads were gated out during data collection by plotting a histogram for channel 6 (side scatter) and collecting events less than the high side scatter peak indicative of SpeedBeads (i.e., <1e + 5 channel 6 intensity x-axis) (Supplemental Fig. S1; all Supplemental Material is available at https://doi.org/10.6084/m9.figshare.16632595). The same gate was applied to all samples. Lasers were set to maximum voltage as follows: 405 nm 175 mW, 488 nm 200 mW, 561 nm 200 mW, 642 nm 150 mW, and SSC 70 mW. Bright-field images were collected using channels 1 and 9. Samples from the same subject point were stained and analyzed on imaging flow cytometry on the same day to control for potential day-to-day variability.

Following data collection in INSPIRE, samples were analyzed using the IDEAS 6.2 software. Positive events in each EV subpopulation were manually gated using the following procedures (Supplemental Fig. S1). Scatterplots were made by plotting each stained fluorescence marker (y-axis) against its neighboring channel (x-axis). Events with a positive intensity above background were gated for each fluorescence channel. The file was then saved as a template, and gates were adjusted using each FMO control. Samples were batch processed using the template containing the final adjusted gates so that all samples run on the same day were gated in the same manner. Features exported for analysis included objects per mL, percentage of total events, and intensity of channels 2 (SGCA), 4 (THSD1), 7 (VAMP3), and 11 (CD63). Intensity is the sum of fluorescence within the defined pixel region for each EV after correction for background pixel values. Intensity was used to measure protein expression (e.g., SGCA, THSD1, VAMP3, CD63) and estimate fluorescence activity.

Statistical Analysis

A sample size of 20 (10 men) was determined a priori to be adequate to detect differences in EV characteristics based on previously reported CD62E+ EV concentration data (33) and assuming an α of 0.05, β of 0.8, and effect size of 0.5. Data were tested for assumptions before inferential statistical analysis. Normality was assessed using Shapiro–Wilk, and Levene’s test was used to determine homogeneity of variance. Data violating assumptions were transformed using natural log, square root, or reciprocal transformations for analysis. Assumptions for statistical analysis were met for all outcomes using either raw or transformed data. Raw data are presented for reader interpretability. Independent variables included sex (men, women), time (pre, postexertion), and day (day 1, or baseline; day 3, or peak stress). Primary outcomes included: 1) total EV mean size, 2) total EV mean concentration, 3) objects per mL for each subpopulation (i.e., CD63+, THSD1+, VAMP3+, SGCA+ EVs), 4) percentage of each subpopulation as a proportion of total EVs, and 5) sum, mean, and median fluorescence intensity within each subpopulation of EVs. Secondary outcomes included sum, mean, and median fluorescence intensity of CD63, THSD1, and VAMP3 within the SGCA+ EV subpopulation as a measure changes in EV subtype tetraspanin expression within the skeletal muscle-derived EVs. Changes in intensity were measured to assess changes in overall tetraspanin expression within each subpopulation, and as an estimate of tetraspanin surface density when normalized to particle count. A separate three-way analysis of variance was used to examine interactions or main effects of sex, time, or day on each of the primary outcomes followed by Bonferroni adjusted pairwise comparisons when appropriate. Baseline subject characteristics were analyzed using independent samples t tests. Partial eta squared (${\mathrm{\eta }}_p^2$) and Cohen’s d effect sizes were calculated for analysis of variance and t tests, respectively. Significance was considered P < 0.05 (two-tailed). Statistical analyses were performed using SPSS, version 27 (IBM, Armonk, NY).

Steps to Ensure Rigor

Several steps were taken to ensure rigor and reduce bias for data collection and analysis. Samples were randomly assigned a unique code by a third party not involved in data acquisition or analysis. Assessors performing ELISA assays, EV isolation, NTA, immunofluorescence staining, and imaging flow cytometry were blinded to group allocation and time point. Samples from the same subject were analyzed on the same day, and an equal number of men and women were analyzed at the same time.

RESULTS

Subject Characteristics

Subject characteristics are presented in Table 1. Women were shorter, had lower body mass, higher body fat percentage, and lower cardiorespiratory fitness compared with men (P < 0.05 each); ages were similar across groups (P > 0.05). Mean relative rate of oxygen consumption was considered “excellent” for men and women according to American College of Sports Medicine criteria (48).

Table 1.

Subject characteristics

	Men	Women
n	10	10
Age, yr	25.6 ± 5.8 (19–35)	27.1 ± 5.9 (20–37)
Height, cm	178.3 ± 7.2 (166–188)	168.0 ± 7.7 (152–178)*
Body mass, kg	81.4 ± 7.8 (64–95)	70.8 ± 8.1 (59–87)*
Body fat, %	18.8 ± 4.2 (11–24)	28.2 ± 6.7 (18–41)*
V̇o2peak, mL·kg·min−1	50.9 ± 5.3 (43–59)	39.0 ± 5.2 (32–47)*
Average KE MVC (N)#	1,141.1 ± 373.0 (704–1,738)	900.1 ± 234.5 (530–1,247)

Data are presented as means ± SD (range). *Indicates a significant difference from men at P < 0.05, according to independent samples t tests. #Sample size for men (n = 9) and women (n = 8). KE, knee extension; MVC, maximum voluntary contraction.

Menstrual cycle and contraception use were self-reported; one female subject had missing data. Five women reported using contraceptive therapy and four women were not currently using or had not used any form of hormonal contraception within the past 6 mo. Of those not using any contraceptive therapy, three experienced regular menstrual cycles and one reported averaging 5 wk between periods. Means ± standard deviation for estrogen and progesterone were 298.1 ± 264.9 ng/mL and 5.9 ± 6.0 pg/mL, respectively. Individual menstrual cycle and contraception use information as well as corresponding sex hormone data for female subjects are presented in Supplemental Table S1 (see https://doi.org/10.6084/m9.figshare.16632598) and Supplemental Fig. S2.

Mean caloric intake was 42% lower on restricted days (average of days 2 and 3) than on baseline (P < 0.001; 95% CI: –1,085.4, –817.4 kcal). This was slightly less than the targeted 50% restriction due to ad libitum underconsumption at baseline. Total sleep was reduced by 49% on restricted versus unrestricted nights (P < 0.001; 95% CI: –228.2, –215.2 min). The severity of caloric and sleep restriction did not differ between men and women (sex × day: P = 0.481 and 0.618, respectively), and total sleep and caloric intake were similar among sexes (main effect of sex: P = 0.338 and 0.055, respectively).

Qualitative and Quantitative Measures of Physical Effort

To demonstrate adequate physical stress during exertion, subjects were asked to report their rating of perceived exertion before and after each event during the physical exertion protocol using a Borg scale (6–20). We also measured conventional markers of acute (pre to postexertion) and delayed (day 1–day 3) muscle damage, including myoglobin and creatine kinase, respectively. Subjects reported a significant increase in physical effort before and after each event during the ∼90 min physical exertion protocol, with no differences according to sex (main effect of time: P < 0.05 for all events) (Supplemental Fig. S3). In agreement with these subjective data, reciprocal transformed serum myoglobin increased by 169% from pre to postexertion (main effect of time: P < 0.001, ${\mathrm{\eta }}_p^2$ = 0.830) (Supplemental Fig. S4A). Men also had a significantly higher myoglobin concentration than women when averaged by time or day (main effect of sex: P = 0.006, ${\mathrm{\eta }}_p^2$ = 0.351). Reciprocal transformed serum creatine kinase increased 12% from day 1 to day 3, nearing statistical significance and demonstrating the cumulative stress on skeletal muscle from baseline to peak stress (P = 0.115, ${\mathrm{\eta }}_p^2$ = 0.132) (Supplemental Fig. S4B).

Extracellular Vesicle Size and Concentration

We began our analysis of EVs by examining their features using nanoparticle tracking analysis. We observed that physical exertion alone (day 1) and physical exertion plus 48 h of 50% sleep and caloric restriction (day 3) had a significant impact on EV concentration and size (Fig. 2). There was a significant interaction of sex and time on natural log transformed EV concentration (sex × time: P = 0.010, ${\mathrm{\eta }}_p^2$ = 0.316). Pairwise comparisons of pre and postexertion at each level of sex revealed a significant decline in women on day 1 (−37%; P = 0.014; Cohen’s d = 0.965) and day 3 (−38%; P = 0.003, Cohen’s d = 1.287), whereas there was no change in men (Fig. 2A). Natural log transformed EV concentration also declined 29% from day 1 to day 3 with no sex-specific differences (main effect of day: P = 0.001, ${\mathrm{\eta }}_p^2$ = 0.447) (Fig. 2A). Regardless of sex, EV mean size increased 13% from pre to postexertion and 8% from day 1 to day 3 (main effect of time: P < 0.001, ${\mathrm{\eta }}_p^2$ = 0.466; main effect of day: P = 0.002, ${\mathrm{\eta }}_p^2$ = 0.433) (Fig. 2B).

Figure 2.

Extracellular vesicle (EV) mean concentration (objects per mL; A) and size (B) before and after physical exertion (time) on day 1 (single stress) and day 3 (multistress; day) in men and women (sex). EV concentration data were natural log transformed for analysis; raw data were used for EV size analysis. All data are presented as raw means ± SD. †Main effect of time where EV mean size (B) was greater following exertion (B). ‡Main effect of day where EV mean concentration was greater on day 1 (A) and EV mean size was greater on day 3 (B). #Sex-by-time interaction where EV mean concentration (A) declined after exertion in women only. Significance was set at P < 0.05. nm, nanometer.

Extracellular Vesicle Subpopulation Profiles

Next, we set out to broadly characterize various EV subpopulations by probing EVs with surface protein markers associated with exosomes (CD63), microvesicles (VAMP3), and apoptotic bodies (THSD1), as well as a marker of skeletal muscle-derived EVs (SGCA). Intensity features were exported for each subpopulation and analyzed to determine the effect of each condition on 1) objects per mL, 2) percentage of total EVs, and 3) intensity of each subpopulation. As a secondary aim, we also examined changes in CD63+, VAMP3+, and THSD1+ intensities within the skeletal muscle-derived EV population. Baseline comparisons revealed no differences between men and women in any EV subpopulation outcome (P > 0.05 each).

CD63+ extracellular vesicles.

A significant two-way interaction of time × day was present in CD63+ EV objects per mL (P = 0.033, ${\mathrm{\eta }}_p^2$ = 0.228). Subsequent pairwise comparisons of time at each level of day revealed a 43% increase in CD63+ EVs from pre to postexertion following 48 h of sleep and caloric restriction (day 3), but there were no differences between pre and postexertion measures on day 1 (day 3: P = 0.017, Cohen’s d = 0.583 vs. day 1: P = 0.773, Cohen’s d = 0.065) (Fig. 3A). A similar pattern emerged for the percentage of CD63+ EVs, whereby the proportion of total EVs that were CD63+ increased 76% from pre to postexertion on day 3 but did not differ on day 1 (time × day: P = 0.039, ${\mathrm{\eta }}_p^2$ = 0.216; day 3: P = 0.018, Cohen’s d = 0.579 vs. day 1: P = 0.727, Cohen’s d = 0.079) (Fig. 3B). Although objects per mL and percentage of CD63+ EVs increased pre to postexertion on day 3, reciprocal transformed CD63 median fluorescence intensity normalized to count decreased 55% from pre to postexertion on day 3, with no difference on day 1, suggesting a possible decline in CD63 surface density per particle (time × day: P = 0.028, ${\mathrm{\eta }}_p^2$ = 0.241; day 3: P = 0.010, Cohen’s d = 0.643 vs. day 1: P = 0.646, Cohen’s d = 0.104) (Fig. 3C). There were no differences according to sex (P > 0.05 each).

Figure 3.

CD63+ extracellular vesicle (EV) objects per mL (A), percentage of total EVs (B), and median fluorescence intensity (C) before and after physical exertion (time) on day 1 (single stress) and day 3 (multistress; day) in men and women (sex). Raw data were used for analysis of CD63+ EV objects per mL and percentage of total EVs; reciprocal transformed data were used for median fluorescence intensity. All data are presented as raw means ± standard deviation. @Time-by-day interaction where CD63+ EV concentration (A) and percentage of CD63+ EVs increased (B), but CD63+ EV intensity decreased following exertion on day 3 only (C). Significance was set at P < 0.05.

SGCA+ extracellular vesicles.

Although the objects per mL of SGCA+ EVs did not differ across time or day or between sex, there were significant main effects for each of these factors on the percentage of SGCA+ EVs. Specifically, from pre to postexertion, there was a 20% increase in the percentage of SGCA+ EVs, and a 37% increase from day 1 to day 3 (main effect of time: P = 0.027, ${\mathrm{\eta }}_p^2$ = 0.244; main effect of day: P = 0.001, ${\mathrm{\eta }}_p^2$ = 0.438) (Fig. 4A). When averaged across time and day, the percentage of SGCA+ EVs was 27% higher in women than in men (main effect of sex: P = 0.049, ${\mathrm{\eta }}_p^2$ = 0.198) (Fig. 4A). Despite no difference in SGCA+ EV objects per mL from pre to postexertion, SGCA mean fluorescence intensity increased significantly following physical exertion, suggesting a possible increase in SGCA proteins per EV (main effect of time: P = 0.005, ${\mathrm{\eta }}_p^2$ = 0.361) (Fig. 4B).

Figure 4.

SGCA+ extracellular vesicle (EV) percentage of total EVs (A) and mean fluorescence intensity (B) before and after physical exertion (time) on day 1 (single stress) and day 3 (multistress; day) in men and women (sex). Raw data were used for analyses of each outcome and are presented as means ± SD. $Main effect of sex where percent of SGCA+ EVs was greater in women (A). †Main effect of time where percent of SGCA+ EVs (A) and SGCA+ EV intensity was greater following exertion (B). ‡Main effect of day where percent of SGCA+ EVs was greater on day 3 (A). Significance was set at P < 0.05. SGCA, α-sarcoglycan.

THSD1+ extracellular vesicles.

There were no significant interaction or main effects of sex, time, or day on THSD1+ EV objects per mL, percentage of total EVs, or measure of fluorescence intensity (P > 0.05 for all).

VAMP3+ extracellular vesicles.

Although there was a modest increase in percentage of VAMP3+ EVs gated from pre to postexertion, it failed to meet statistical significance (main effect of time: P = 0.086, ${\mathrm{\eta }}_p^2$ = 0.155). There were no significant differences in VAMP3+ EV objects per mL or measure of fluorescence intensity across time or day or between sex (P > 0.05 for all).

CD63, THSD1, VAMP3 fluorescence intensity within the SGCA+ extracellular vesicle subpopulation.

We next examined the fluorescence intensities of CD63, THSD1, and VAMP3 within the skeletal muscle-derived EV subpopulation to determine if there were changes in markers associated with EV exosomes (CD63), microvesicles (VAMP3), or apoptotic bodies (THSD1). CD63 sum intensity increased significantly from day 1 to day 3 in SGCA+ EVs (main effect of day: P = 0.038, ${\mathrm{\eta }}_p^2$ = 0.218) (Fig. 5A). CD63 mean intensity was greater in men than in women, although this difference did not reach statistical significance (main effect of sex: P = 0.086, ${\mathrm{\eta }}_p^2$ = 0.680). There were no differences in THSD1 or VAMP3 intensities within the skeletal muscle-derived EV subpopulation (P > 0.05 each) (Fig 5, B and C).

Figure 5.

Sum fluorescence intensities of CD63 (A), THSD1 (B), and VAMP3 (C) within the SGCA+ EV population before and after physical exertion (time) on day 1 (single stress) and day 3 (multistress; day) in men and women (sex). Raw data were used for analyses of each outcome and are presented as means ± SD. ‡Main effect of day where CD63 sum intensity within the SGCA+ EV population was greater on day 3 (A). Significance was set at P < 0.05. EV, extracellular vesicle; SGCA, α-sarcoglycan; THSD1, thrombospondin-1; VAMP3, vesicle-associated membrane-3.

DISCUSSION

In this study, we characterized the effects of sex and two types of stress—physical exertion versus physical exertion plus sleep and caloric restriction—on EV characteristics. As hypothesized, we observed a decline in EV concentration following exertion in women only as well as an increase in SGCA+ and CD63+ EVs (peak stress only) following exertion, whereas there were no differences in VAMP3+ or THSD1+ EVs. Contrary to our hypothesis, we observed an increase in mean EV size following exertion and from baseline to peak stress. Novel findings in this study were that the proportion of SGCA+ EVs was higher in women compared with men regardless of stress condition, and that SGCA+ EVs increased following 48 h of restricted sleep and caloric intake compared with baseline independent of sex. Collectively, these results suggest that sex and stress condition have an impact on EV profiles and should be taken into consideration during experimental design of EV studies.

EV concentration declined ∼38% in women following physical exertion, whereas there was no change in men. Previous studies have described similar sex-specific changes in EV concentration following moderate intensity (60%–65% V̇o_2max) treadmill exercise (30, 33). As noted by Toth et al. (49), vesicle biogenesis differs between men and women at rest and varies according to menstrual cycle phase in women, suggesting that female sex hormones likely account for some of the sex differences noted in this study. However, sex-specific changes in EV concentrations may be in part attributed to coisolates. Although size exclusion chromatography is able to isolate EVs from most plasma proteins and small lipoproteins (50), other molecules with a similar size as EVs, such as low-density lipoproteins (LDL), very low-density lipoproteins (VLDL), and chylomicrons, may still be present as coisolates and exhibit similar sex-specific responses to exercise (51). For example, VLDL concentrations have been reported to decline 30% following exercise in women (52), whereas no change in high-density lipoproteins (HDL) or LDL concentrations was observed in men (28). Thus, changes in similarly sized coisolates with EVs could account for at least some of the sex-specific response patterns noted in EV concentration following physical exertion as measured by nanoparticle tracking analysis.

Although there were no significant differences in THSD1+ and VAMP3+ EVs—markers associated with apoptotic bodies and microvesicles, respectively—CD63+ EVs (exosomes) changed substantially. Specifically, CD63+ EV concentration increased by 43%, proportion of CD63+ EVs increased by 76%, and CD63 median fluorescence intensity decreased by 55% pre to postexertion during sleep and caloric restriction, but there were no differences pre to postexertion at baseline. Collectively, these results suggest that the stress of physical exertion alone may have been insufficient to cause a significant change in CD63+ EVs, whereas the cumulative stress of physical exertion plus sleep and caloric restriction resulted in an increase in EV-mediated cell signaling. Furthermore, the inverse directional change in concentration of CD63+ EVs and median fluorescence intensity suggests a possible decline in the average number of CD63 proteins per particle (i.e., surface density of CD63 protein). Previous studies have reported similar increases in circulating CD63+ EVs (35, 53, 54) and total CD63 protein content (28, 53) following moderate to intense exercise. CD63+ EVs, such as exosomes, are known mediators of metabolism and favorable physiological adaptations such as neurogenesis, immunoregulation, angiogenesis, and myogenesis (26).

Given the physical exertion-based stress conditions in this study, we were particularly interested in EVs from skeletal muscle origin. Therefore, we probed EVs for SGCA, which is highly enriched in skeletal muscle, and has been previously used as a marker to analyze skeletal muscle-derived EVs in circulation (29, 30). The proportion of SGCA+ EVs increased from 8.1 to 9.7% of total EVs from pre to postexertion (main effect of time) and ranged from 6.4 to 9.6% in men and 7.2%–12.8% in women from preexertion at baseline to postexertion during sleep and caloric restriction, respectively. The increase in proportion of SGCA+ EVs suggests augmented signaling originating from skeletal muscles, relative to other cell types, during physical exertion. These data are substantiated by others who have reported a similar increase in SGCA+ EVs following exercise (29, 30). SGCA+ EVs are taken up by a range of tissues (55) and have been shown to carry miRNA that target pathways involved in skeletal muscle differentiation (29, 56–59). The increase in proportion of skeletal muscle-derived EVs was complemented by an increase in SGCA mean fluorescence intensity despite no change in SGCA+ EV concentration, suggesting greater SGCA protein surface density.

The addition of sleep and caloric restriction (main effect of day) resulted in an increase in the proportion of SGCA+ EVs from 7.5% at baseline to 10.3% at peak stress. Moreover, within the skeletal muscle-derived EV subpopulation, the CD63 intensity was significantly increased from baseline to peak stress, whereas THSD1 and VAMP3 did not change, suggesting that exosomes are the major component of skeletal muscle-released EVs in response to the cumulative stressors imposed in this study. The increased presence of SGCA+ EVs in the circulation suggests augmented cell–cell communication originating from skeletal muscle tissue during combined sleep and caloric restriction. It is well established that skeletal muscle undergoes proteolysis to liberate amino acids for energy production via gluconeogenesis during periods of underfeeding (60). Therefore, it is plausible that skeletal muscle-derived EVs may provide greater contribution to energy metabolism during hypocaloric compared with eucaloric energy states, though future studies are necessary to confirm this potential mechanism as a means to coordinate and share limited energy resources when nutrient intake is suboptimal. The increase in proportion of SGCA+ EVs may also be due to the cumulative effects of daily physical activity. Creatine kinase is a conventional marker of muscle damage that peaks ∼48 h after onset of intense exercise (61). Interestingly, the increase in creatine kinase concentration did not reach statistical significance 48 h after baseline physical exertion in this study (P = 0.115), whereas the proportion of SGCA+ EVs steadily increased from pre to postexertion at baseline through postexertion at peak stress. These findings suggest skeletal muscle-derived EVs may be a more sensitive marker of endocrine signaling following physical exertion.

Finally, we observed a significantly higher proportion of SGCA+ EVs in women compared with men (main effect of sex: 10.0% in women vs. 7.9% in men), which may be explained by sex-specific physiological differences. Prior literature demonstrates that women operate at a higher percentage of heart rate reserve than men when performing the same work and with no difference in ratings of perceived exertion (62–64). These higher relative demands of performing a given task in women compared with men may have accounted for some of the difference in proportion of SGCA+ EVs noted here. In addition, sex differences in muscle fiber type are well-established in the literature with women having a higher proportion of type I versus type II muscle fibers compared with men (65), and EV protein cargo has been shown to vary according to fiber type (66). This could serve as an additional explanation for the higher proportion of SGCA+ EV in women compared with men, although these hypotheses would need to be tested.

Several considerations must be acknowledged when interpreting results from this study. First, although size exclusion chromatography is a reliable and effective method for obtaining highly enriched EVs (67, 68), no method exists for achieving entirely pure samples, and coisolation of similarly size molecules is likely. Second, as previously discussed, although nanoparticle tracking analysis is considered a gold standard method for size and concentration characterization, it is unable to distinguish between EVs and similarly sized particles. In light of these facts regarding the isolation and size/concentration characterization methods used here, we cannot discount the fact that at least some of the variation in EV size and/or concentration may be attributed to changes in coisolates across conditions. Third, although our goal was to broadly characterize different EV subpopulations, including exosomes, microvesicles, and apoptotic bodies, we used a limited panel of fluorescent markers for identifying these populations; therefore, EVs that were present but CD63^–, THSD1^–, or VAMP3^– went undetected. Fourth, although SGCA⁺ EVs in circulation have been suggested to originate from muscle tissue (29), it is possible that some muscle-derived EVs are absent SGCA. Fifth, although we noted interactive effects on EV characteristics between the two stress conditions, we are unable to distinguish between the effect of sleep restriction and energy deficit. Although sleep and caloric restriction often go hand in hand, future studies should tease out the effects of each. Finally, because this study did not include a control group, we cannot exclude the possibility that observed changes from baseline to peak stress were due to time-dependent events.

To our knowledge, this is the first study to examine the combined effects of sleep and caloric restriction in addition to physical exertion on EV profiles in men and women. Although single stress studies provide good internal validity, generalizability to an applied setting is often compromised. Our use of two different stress conditions demonstrates the interactive effects that occur by combining multiple stresses as they occur in real-world settings, providing greater ecological validity. Furthermore, the inclusion of men and women adds to the scant pool of data examining EVs across the sexes and helps distinguish aspects of EV research that should consider sex for analysis and interpretation of results.

Conclusions

Data from this study extend previous findings and add to an exciting and burgeoning field of EV research. We demonstrated sex- and stress-specific changes in EV profile dynamics, which should be considered in the design of future EV studies. These data lay a foundation for future studies to examine changes in EV cargo and determine physiological implications that can lead to the long-term goal of EV therapeutics. Basic science research is showing exceptional advancements using EVs in this manner (69). One can conceptualize similar translational benefits applied in humans, whereby isolated EVs or engineered EVs containing synthetic cargoes may be administered to individuals with the goal of promoting health and optimizing performance. For now, characterization of EVs in humans is a necessary first step for identifying physiological targets of interest.

SUPPLEMENTAL DATA

Supplemental Figs. S1–S4: https://doi.org/10.6084/m9.figshare.16632595.

Supplemental Table S1: https://doi.org/10.6084/m9.figshare.16632598.

GRANTS

This work was supported by the Department of Defense (Award No. W81XWH-17-2-0070) and the Freddie H. Fu, MD Graduate Research Award.

DISCLAIMERS

The authors have no professional relationships with companies or manufacturers who will benefit from the results of the present study. The results of this study do not constitute endorsement of the product by the authors, the Department of Defense, or the US Government. Citations of commercial organizations and trade names in this report do not constitute an official Department of Defense endorsement or approval of the products or services of these organizations. The results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

W.R.C., F.A., and B.C.N. conceived and designed research; W.R.C., M.E.B., A.S., and Z.J.C. performed experiments; W.R.C. and M.L. analyzed data; W.R.C., M.E.B., A.S., and F.A. interpreted results of experiments; W.R.C. prepared figures; W.R.C. drafted manuscript; W.R.C., M.E.B., A.S., Q.M., Z.J.C., M.L., S.D.F., B.J.M., F.F., F.A., and B.C.N. edited and revised manuscript; W.R.C., M.E.B., A.S., Q.M., Z.J.C., M.L., S.D.F., B.J.M., F.F., F.A., and B.C.N. approved final version of manuscript.