Serum extracellular vesicle miR-203a-3p content is associated with skeletal muscle mass and protein turnover during disuse atrophy and regrowth

Abstract

Small noncoding microRNAs (miRNAs) are important regulators of skeletal muscle size, and circulating miRNAs within extracellular vesicles (EVs) may contribute to atrophy and its associated systemic effects. The purpose of this study was to understand how muscle atrophy and regrowth alter in vivo serum EV miRNA content. We also associated changes in serum EV miRNA with protein synthesis, protein degradation, and miRNA within muscle, kidney, and liver. We subjected adult (10 mo) F344/BN rats to three conditions: weight bearing (WB), hindlimb suspension (HS) for 7 days to induce muscle atrophy, and HS for 7 days followed by 7 days of reloading (HSR). Microarray analysis of EV miRNA content showed that the overall changes in serum EV miRNA were predicted to target major anabolic, catabolic, and mechanosensitive pathways. MiR-203a-3p was the only miRNA demonstrating substantial differences in HS EVs compared with WB. There was a limited association of EV miRNA content to the corresponding miRNA content within the muscle, kidney, or liver. Stepwise linear regression demonstrated that EV miR-203a-3p was correlated with muscle mass and muscle protein synthesis and degradation across all conditions. Finally, EV miR-203a-3p expression was significantly decreased in human subjects who underwent unilateral lower limb suspension (ULLS) to induce muscle atrophy. Altogether, we show that serum EV miR-203a-3p expression is related to skeletal muscle protein turnover and atrophy. We suggest that serum EV miR-203a-3p content may be a useful biomarker and future work should investigate whether serum EV miR-203a-3p content is mechanistically linked to protein synthesis and degradation.

INTRODUCTION

Skeletal muscle atrophy is a risk factor for all-cause morbidity and mortality in both aging and critical illness (38, 49, 50). As such, it has been suggested that skeletal muscle is an important organ for maintenance of overall health and homeostasis of other tissues via interorgan communication (28). That is, skeletal muscle releases factors that can be systemically delivered to other tissues/organs and impart biologically relevant adaptations or changes in organ function (26, 30, 43, 48, 51, 56). During times of muscle contraction or disuse, this communication paradigm may change and impact systemic function. Growing evidence suggests that extracellular vesicles (EVs) are an important component of this intercellular communication across a variety of circumstances. Despite a recent growth in the appreciation of EVs in mediating cross talk and the clinical importance for understanding muscle atrophy, there is a paucity of information about the role of EVs during muscle atrophy.

EVs are a heterogeneous population of small lipid membrane-bound vesicles that are released from virtually all cell types (6). The most commonly discussed EVs are exosomes (30–150 nm) and microvesicles (100 nm–1 μm), although apoptotic bodies (>1 μm) are now gaining support as vesicles that are not merely a by-product of apoptosis but can serve regulatory roles (5, 7). Exosomes are formed within the endosomal pathway and released from cells when multivesicular bodies (MVBs) merge with the plasma membrane and release the exosomes into the extracellular space (6). Microvesicles are formed from outward blebbing of the plasma membrane and subsequent scission and release of the plasma membrane bleb. There is still no consensus on specific markers of specific EV subtypes, however, so the classifications of vesicles for most studies are based primarily on physical characteristics such as size (i.e., <100 nm), density (i.e., low, middle, high), or biochemical composition (i.e., CD63+) (58). The overlap in the characteristics of the various subtypes of vesicles has led to the use of the term “extracellular vesicles” to collectively describe them (58). The release of EVs into the extracellular space facilitates intercellular/intertissue communication that is important for orchestrating systemic physiological outcomes. EVs are loaded with bioactive cargo such as proteins, lipids, RNA, and microRNA (miRNA) that impact the physiological function of cells to which they are delivered. A plethora of in vitro studies (reviewed in 54) have shown how EVs are involved in skeletal muscle cell function, but in vivo examinations are lacking, especially in the context of disuse atrophy.

Small noncoding microRNAs (miRNAs) are often carried through the systemic circulation in EVs and are an important component in the regulation of atrophy (24, 37). Circulating miRNAs (potentially within EVs) may be a mechanism contributing to skeletal muscle atrophy and its associated systemic effects (reviewed in 2). It has been shown that atrophying muscle cells lower internal miR-23a levels by selectively loading miR-23a into EVs, thus releasing the inhibition of expression of MAFbx and MuRF1, two critical ubiquitin ligases in muscle atrophy (21). In addition, there is evidence of muscle-kidney cross talk via EVs that export miR-23a, miR-27a, and miR-29a from the muscle to the kidneys, thereby modulating renal fibrosis in both a diabetic and a chronic kidney disease mouse model (62, 69). A muscle-liver communication paradigm has also been proposed via EVs that are liberated from muscle during exercise (65). Altogether, available data suggest that EV-mediated delivery of muscle-derived miRNAs may impact distant organ homeostasis under a variety of physiological conditions.

Finally, there is interest in finding noninvasive biomarkers that can provide diagnostic and prognostic value during pathophysiological states, such as during muscle disuse and atrophy. A study by Margolis et al. (34) demonstrated an inverse relationship between circulating muscle-specific miRNAs (myomiRs) and whole body rates of protein synthesis in human subjects. Numerous other studies have now shown that circulating miRNAs may indeed be useful biomarkers for understanding muscle atrophy diagnosis and muscle disease progression (23, 35, 36). The potential for investigating EVs as biomarkers with a mechanistic contribution to atrophy prompted the work presented here, with the aim of uncovering the relationship between serum EV miRNA and skeletal muscle atrophy in vivo. We hypothesized that there is a relationship between serum EV miRNA, muscle miRNA, and indexes of atrophy and regrowth in skeletal muscle.

METHODS

Animals and experimental procedures.

All procedures were approved by the University of Kentucky’s Institutional Animal Care and Use Committee. Male Brown Norway/F344 rats at 10 mo of age (National Institute on Aging, Bethesda, MD) were used in this study. Rats were randomly assigned into three groups: weight-bearing conditions (WB), hindlimb suspension for 7 days (HS), and hindlimb suspended for 7 days and subsequently reloaded and allowed to move around the cage freely for 7 days (HSR). Rats were allowed free access to food and water at all times and were housed on a 12:12-h light-dark cycle. Hindlimb suspension was performed as previously described (40). Briefly, a tail device containing a hook was attached with gauze and cyanoacrylate glue while the animals were anesthetized with isoflurane (2% by inhalation). After the animal regained consciousness, the tail device was connected via a thin cable to a pulley sliding on a vertically adjustable stainless steel bar running longitudinally above a high-sided cage. Rats were raised progressively over a 6-h period to allow them to adjust to the suspension. The system was designed in such a way that the rats could not rest their hindlimbs against any side of the cage but could move around the cage on their front limbs and could reach water and food easily. Cages were randomly placed in the room, and room temperature was 27°C. Rats in the HSR group were released from the tail suspension device after 7 days of unloading, and they were allowed to maintain normal ambulation for 7 days (reloading). The study timeline for the rats is depicted in Fig. 1.

Fig. 1.

Study timeline for rats. HS, hindlimb suspension; HSR, hindlimb suspension followed by reloading; WB, weight bearing.

Deuterium oxide labeling and determination of protein synthesis and degradation rates.

For determination of protein synthesis rates, rats received a bolus of (99%) deuterium oxide (D₂O) (Sigma-Aldrich, St. Louis, MO) before the start of the experiment, equivalent to 5% of the body water pool, followed by drinking water enriched 8% with D₂O for the remainder of the experimental period. Specifically, the HS rats received the bolus of D₂O 2 days before the 7-day suspension period. The WB rats were euthanized 9 days after receiving the bolus and beginning D₂O drinking water to match the labeling period of the HS group. Rats in the HSR group received the same bolus of D₂O 2 days before the start of reloading followed by 8% D₂O-enriched drinking water until euthanasia for a total of 9 days.

Protein synthesis was determined according to our established methods (27, 40, 41). Skeletal muscle tissue was homogenized 1:10 in isolation buffer (in mM: 100 KCl, 40 Tris·HCl, 10 Tris Base, 5 MgCl₂, 1 EDTA, 1 ATP, pH = 7.5) with phosphatase and protease inhibitors (Halt; Thermo Fisher Scientific) with a bead homogenizer (Next Advance Inc., Averill Park, NY). After homogenization, subcellular fractions were isolated via differential centrifugation (9, 10, 39). Protein fractions were derivatized for analysis of deuterium enrichment in alanine with an Agilent 7890A GC coupled to an Agilent 5975C MS (9, 10, 39). To determine the precursor pool enrichment, serum samples were prepared for analysis on GC-MS (9, 10, 39).

Calculations.

Disuse atrophy, such as that caused by hindlimb suspension, results in a rapid loss of muscle mass as well as the protein pool of measurement, creating a non-steady-state condition. Therefore, we used our previously published equations that account for non-steady-state conditions (27, 40, 42). The mass of protein at time t, P(t), obeys the differential equation

$$\frac{\text{d}P}{\text{d}t}=K_{\text{syn}}-K_{\text{deg}}P\left(t\right)$$ (1)

where K_syn is the synthesis rate, with dimensions of mass over time, and K_deg is the degradation constant, with dimensions of inverse time. From this relationship:

$$K_{\text{syn}}=K_{\text{deg}}P{}_{\text{eq}}$$ (2)

Blood and tissue collection.

At the end of the experimental period, rats were anesthetized with pentobarbital sodium and blood was collected through a cardiac puncture; gastrocnemius muscles were then dissected, weighed, and frozen. A subset of rats (WB n = 5, HS n = 5, HSR n = 8) also had the liver and kidneys harvested for analysis. Muscles and organs were frozen in liquid nitrogen and stored at −80°C for later biochemical analyses. Serum was isolated by allowing blood to clot at room temperature for 30 min before centrifugation for 10 min at 2,000 g and 4°C. The serum supernatant was collected and stored at −80°C until analysis.

Mean fiber cross-sectional area.

Mean fiber cross-sectional area (CSA) was determined as previously described (40). Briefly, gastrocnemius cross sections (8 μm) were rehydrated in PBS and incubated in dystrophin antibody (1:50; Vector Laboratories, Burlingame, CA) for 1 h at room temperature and overnight at 4°C. Secondary antibody (1:200, directly conjugated Texas Red goat anti-mouse; Rockland Immunochemicals, Gilbertsville, PA) was applied, and sections were coverslipped. Images were captured with a Zeiss AxioImager MI upright fluorescent microscope (Zeiss, Gottingen, Germany). Mean fiber CSA was measured with the automated software program MyoVision (64).

Serum EV isolation and nanoparticle tracking analysis.

EVs were isolated from 500 µL of serum with ExoQuick Exosome Precipitation Solution [System Biosciences (SBI), Palo Alto, CA]. The serum was first centrifuged at 3,000 g for 15 min to remove cells and debris, and then the supernatant was collected and filtered through a 0.22-µm low-binding PVDF filter (Millex-GV; Millipore, Tullagreen, Ireland). Approximately 240 µL of ExoQuick was added to the sample and incubated at 4°C overnight. The ExoQuick-serum mixture was then centrifuged at 1,500 g for 30 min to pellet the EVs. The supernatant was removed, and the EV pellet was reconstituted in 300 µL of PBS. Validation of EV isolation from serum was made by measuring particle size with a ZetaView nanoparticle tracking analyzer (Particle Metrix, Grayslake, IL) and via Western blotting.

EV miRNA isolation and analysis.

Total RNA was isolated from EVs with the commercially available miRCURY RNA Isolation Kit (Exiqon, Woburn, MA) according to the manufacturer’s instructions. Sample miRNA abundance was quantified with a small RNA kit on an Agilent Bioanalyzer (Agilent, Santa Clara, CA). We submitted 50 ng of miRNA from WB EV miRNA (n = 3) and HS EV miRNA (n = 3) to the University of Kentucky Genomics Core Laboratory for miRNA microarray analysis. Samples were hybridized to the Thermofisher/Affymetrix miRNA 4.0 array. MetaboAnalyst 4.0 (66) was used to perform detailed statistical analyses of the microarray data. Predictions of Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways altered by changes in serum EV miRNA expression during HS were made with DIANA miRpath v3.0 (61).

Reverse transcription of miRNA was performed with 10 ng of total RNA with the miRCURY LNA RT kit (Qiagen). RT-qPCR was performed with the miRCURY LNA SYBR Green PCR kit (Qiagen) combined with the appropriate miRCURY LNA primer sets for the miRNA of interest (Qiagen). The expression of each miRNA was normalized to the expression of UniSp6 with the −∆C_T method (where C_T is threshold cycle) (55). UniSp6 is an exogenous spike-in that resembles miRNAs in structure but lacks close sequence similarities to known miRNAs, provided by the miRCURY LNA RT kit (Qiagen).

EV protein isolation and Western blotting.

Protein was isolated from EVs with Pierce RIPA lysis buffer (Thermo Scientific, Rockford, IL) with a protease and phosphatase inhibitor cocktail (Halt protease inhibitor cocktail; Thermo Scientific, Rockford, IL) added. Protein concentration was determined with the Pierce BCA protein assay kit (Thermo Scientific, Rockford, IL). Samples were then prepared in Laemmli buffer at a concentration of 1 µg/µL and boiled at 95°C for 5 min. After 5 µg of sample was loaded, proteins were separated by SDS-PAGE with 4–15% TGX gels (Criterion; Bio-Rad) by first stacking for 10 min at 100 V and then running for ~40 min at 200 V and room temperature. Samples were then transferred for 60 min at 100 V on ice onto a nitrocellulose membrane (Bio-Rad) in 20% methanol Tris-glycine buffer. The Revert Total Protein Stain Kit (Li-Cor Biosciences, Lincoln, NE) was used to stain total protein, and the membrane was imaged to verify transfer efficiency and loading. Membranes were blocked in 5% nonfat dry milk in Tris-buffered saline-Tween (TBS-T, 0.1% Tween 20) for 1 h at room temperature. The membranes were then incubated overnight at 4°C with primary antibody [anti-CD63, catalog no. EXOAB-CD63A-1; System Biosciences (SBI), Palo Alto, CA] at a 1:1,000 dilution in 5% nonfat dry milk in TBS-T. The membrane was then washed 3 × 5 min in TBS-T before a 1-h incubation with goat anti-rabbit secondary antibodies [1:10,000 dilution; System Biosciences (SBI), Palo Alto, CA]. Blots were developed with enhanced chemiluminescence (Clarity Western ECL substrate; Bio-Rad), imaged, and quantified with ImageJ (National Institutes of Health).

Human subjects and experimental procedures.

We recruited six healthy subjects (5 women, 1 man) between the ages of 18 and 30 yr with a body mass index (BMI) < 26 kg/m². Exclusion criteria included subjects who had preexisting health conditions, current musculoskeletal injuries, previous surgeries involving lower extremities, and pregnancy. Subjects were randomly assigned to one of two study groups: unilateral lower limb suspension (ULLS; n = 3) or unilateral lower limb suspension and massage, an intervention used as a mechanotherapy (27, 40) (ULLS+M; n = 3) (Table 1). ULLS was used to induce skeletal muscle atrophy in a single leg as previously described (17). Briefly, a knee immobilizer was fitted and secured to the left leg, thereby immobilizing the lower limb. Crutches were fitted, and subjects ambulated with the crutches in a non-weight-bearing manner for the 7-day duration of the study. Subjects in the ULLS+M group had a controlled massage-like mechanotherapy applied to the immobilized vastus lateralis (VL) muscle for 30 min on the first, third, fifth, and final day of the intervention. Muscle biopsies of the VL were obtained before the intervention and on the final day of the intervention (day 7) with a Bergstrom needle as previously described (13). Briefly, ~100 mg of each biopsy was mounted with fibers perpendicular to a cork with tragacanth gum. The sample was then snap-frozen in liquid nitrogen-cooled isopentane and stored at –80°C until analysis of fiber CSA. A separate portion of the muscle biopsy was also snap-frozen and stored at –80°C until the analysis of miRNA. All study procedures were approved by the University of Kentucky’s Institutional Review Board, and written informed consent was obtained from all subjects before initiating participation in the study. The study timeline for the human subjects is depicted in Fig. 2.

Table 1.

Human subject characteristics

	Sex	Age, yr	BMI, kg/m2
ULLS	F	19	24.1
	F	23	22.5
	F	20	26.3
ULLS+M	F	25	27.0
	M	26	23.8
	F	33	26.6

BMI, body mass index; F, female; M, male; ULLS, unilateral lower limb suspension; ULLS+M, unilateral lower limb suspension and massage.

Fig. 2.

Study timeline for human subjects. ULLS, unilateral lower limb suspension; ULLS+M, unilateral lower limb suspension and massage.

Statistics.

Muscle atrophy outcome measures of CSA, mass, K_syn, and K_deg were visualized with histograms and Q-Q plots, and the Shapiro–Wilk normality test was used to determine whether they were normally distributed. All three variables were not normally distributed. In addition, the miRNA data were assessed for normality in a similar manner, and it was also determined that many of the miRNAs were not normally distributed. Since many of the outcome measures were not normally distributed, the nonparametric version of analysis of variance, Kruskal–Wallis, and independent t tests, Dunn, were used to determine differences across groups. The Bonferroni method corrected for multiple post hoc comparisons. Significance was assumed at P < 0.05.

A stepwise linear regression with stepwise selection was utilized to determine which miRNAs were associated with atrophy measures. To determine which miRNAs were included in the stepwise analysis, simple linear regressions were run for each miRNA and outcome measure. Only mRNAs with a P < 0.15 were selected for the stepwise analysis. Rats that were missing any miRNA were excluded, yielding 30 rats for the stepwise analysis. Nonparametric ANOVA, simple linear regression, and stepwise regression analyses were completed in R 3.6.1.

A dependent Student’s t test was used to examine changes in VL mean myofiber CSA and serum EV miR-203a-3p expression with ULLS across all six subjects. Significance was assumed at P < 0.05.

RESULTS

Muscle mass and protein turnover response to hindlimb suspension and reloading.

Hindlimb suspension caused significant muscle atrophy as exemplified by reduced gastrocnemius muscle mass and mean myofiber CSA in HS rats compared with WB (P = 0.00018 and P = 0.00009, respectively) (Fig. 3, A and B). In addition, myofibrillar protein synthesis rate (K_syn, mg/day) was ~60% lower (P = 0.0018) and protein degradation (K_deg, 1/t) was 100% higher (P = 0.0008) in HS muscle compared with WB (Fig. 3, C and D). Gastrocnemius mass in the HSR rats was significantly higher compared with HS (P = 0.013) but also significantly lower than WB (P = 0.009) (Fig. 3A). The mean myofiber CSA of HSR muscle was significantly higher than HS muscle (P = 0.0009) but still exhibited only a partial recovery of CSA, as it was significantly lower than WB muscle (P = 0.00014) (Fig. 3B). Myofibrillar protein synthesis was significantly higher and protein degradation was significantly lower in HSR compared with HS (Fig. 3, C and D) (P = 0.00014 and P = 0.0006, respectively) but recovered to levels not different from WB.

Fig. 3.

Response to hind limb suspension and reloading. A: gastrocnemius muscle mass. B: gastrocnemius mean fiber cross-sectional area (CSA). C: myofibrillar protein synthesis [synthesis rate (K_syn)]. D: myofibrillar protein degradation [degradation constant (K_deg)]. HS, hindlimb suspension; HSR, hindlimb suspension followed by reloading; WB, weight bearing. Data presented as means ± SE. *Significantly different compared with WB (P < 0.05); †significantly different compared with HS (P < 0.05).

Isolated serum EVs.

Nanoparticle tracking analysis (NTA; ZetaView) showed that the isolated serum EVs had a mode size of 92 ± 1 nm across all groups, verifying successful isolation of small EVs (Fig. 4A). Although these preparations were enriched for exosomes (30–150 nm), it is likely that small microvesicles (100–1,000 nm) were also captured in these preparations, which is why it is more appropriate to call these preparations EVs and not exosomes. This nomenclature follows the recommendations of the most recent position statement from the International Society for Extracellular Vesicles (58). Furthermore, Western blotting analysis demonstrated that although the ExoQuick preparation captured nonvesicular protein, which is a known drawback of this method (58), EV samples were enriched for CD63+ EVs (Fig. 4, B and C).

Fig. 4.

Characterization of extracellular vesicles (EVs). A: nanoparticle tracking analysis of EVs isolated from rat serum [n = 9 total; n = 3 from weight bearing (WB), hindlimb suspension (HS), and hindlimb suspension followed by reloading (HSR)] shows enrichment of small EVs with a peak of 92-nm diameter. B: representative Western blot image from serum and EV samples. C: quantification of CD63 blot shown in B normalized to the total protein abundance and displayed relative to serum values. a.u., Arbitrary unit.

EV miRNA.

Principal component analysis (PCA) of the miRNA microarray data demonstrated some overlap between WB and HS samples (Fig. 5A). There were 52 mature miRNAs and 20 pre-miRNAs that were significantly different (P < 0.05, unadjusted) in HS compared with WB (Fig. 5B and Table 2). As shown in Fig. 5B, there were a vast number of miRNAs that exhibited small but statistically significant changes. When false discovery rate (FDR) was applied to the data set, there were no significant changes in EV miRNA in HS compared with WB. Interestingly, miR-203a-3p and miR-30a-3p were the only mature miRNAs that exhibited a greater than twofold difference with HS (Fig. 5B and Table 2).

Fig. 5.

MicroRNA (miRNA) microarray analysis of serum extracellular vesicles (EVs) shows limited significant differences in hindlimb suspension (HS) compared with weight bearing (WB). A: principal component analysis of serum EV miRNA microarray results. PC, principal component. B: volcano plot of serum EV miRNA microarray results comparing HS with WB. Gray dots, not significantly different miRNA; purple dots, significantly different miRNA (P < 0.05); green dots, significantly different miRNA (P < 0.05) with a >2-fold difference when comparing HS with WB. FC, fold change.

Table 2.

Significantly altered serum EV miRNAs and pre-miRNAs with HS

Transcript	Log2(FC)	P Value	Transcript	Log2(FC)	P Value
mir-215	−0.29	0.0002	miR-218a-5p	0.46	0.0217
miR-203a-3p	−1.66	0.0005	mir-27b	0.24	0.0223
mir-203a	0.21	0.0014	miR-34c-5p	−0.43	0.0239
miR-3558-5p	0.41	0.0016	mir-671	0.43	0.0249
miR-328b-3p	−0.25	0.0018	miR-22-3p	−0.22	0.0266
miR-511-3p	−0.26	0.0026	miR-582-3p	−0.56	0.0274
miR-223-3p	−0.55	0.0029	miR-361-5p	−0.18	0.0283
mir-466c	−0.50	0.0034	miR-150-3p	0.27	0.0286
miR-742-3p	−0.18	0.0036	mir-6215	−0.50	0.0287
mir-3085	0.67	0.0050	miR-500-5p	0.23	0.0292
miR-3562	0.12	0.0062	mir-876	0.43	0.0305
miR-877	0.28	0.0064	miR-3550	−0.48	0.0313
mir-151	−0.30	0.0077	miR-20a-5p	−0.23	0.0342
let-7a-1-3p	−0.45	0.0079	miR-29a-3p	−0.25	0.0342
let-7c-2-3p	−0.45	0.0079	miR-152-3p	−0.28	0.0348
mir-17-1	−0.31	0.0079	mir-16	−0.18	0.0358
miR-99a-3p	0.44	0.0084	miR-541-3p	−0.32	0.0378
miR-192-5p	−0.26	0.0084	miR-101b-3p	−0.78	0.0381
miR-30a-3p	−1.21	0.0094	miR-221-3p	−0.53	0.0387
miR-652-3p	−0.45	0.0095	miR-24-2-5p	−0.37	0.0388
miR-375-5p	0.29	0.0098	miR-23b-5p	0.39	0.0396
mir-204	−0.39	0.0105	mir-383	−0.33	0.0396
miR-26a-5p	−0.22	0.0111	mir-20b	−0.21	0.0396
miR-34b-3p	0.54	0.0123	miR-802-3p	0.22	0.0398
miR-194-5p	−0.22	0.0125	miR-378a-3p	−0.19	0.0403
miR-181d-3p	0.71	0.0130	miR-31a-5p	−0.75	0.0404
miR-675-5p	−0.75	0.0138	mir-551b	−0.11	0.0424
mir-219b	−0.46	0.0141	miR-378b	−0.22	0.0428
miR-325-5p	0.35	0.0153	miR-433-5p	0.27	0.0432
mir-134	0.34	0.0155	miR-135b-5p	0.35	0.0432
miR-146a-5p	−0.37	0.0167	mir-29b-1	−0.22	0.0437
miR-3577	0.49	0.0169	miR-151-5p	−0.60	0.0461
mir-361	−0.28	0.0170	miR-222-5p	0.51	0.0464
miR-27b-3p	−0.23	0.0190	miR-92a-3p	−0.17	0.0470
mir-181a-1	0.32	0.0193	miR-18a-5p	−0.46	0.0481
miR-3594-5p	0.18	0.0199	miR-203b-3p	0.44	0.0493

MicroRNAs (miRNAs) in bold were significantly altered (P < 0.05) and greater than 2-fold difference compared with weight bearing (WB). EV, extracellular vesicle; FC, fold change; HS, hindlimb suspension.

To get a better understanding of how the differences in serum EV miRNA profiles may impact overall function, we uploaded the list of mature miRNAs that were significantly different (Table 2) to the DIANA miRpath v3.0 online software and generated KEGG pathways for the genes whose mRNA would be predicted to be targeted by the altered miRNAs (Table 3). There were multiple pathways related to muscle response to mechanical stimulus (11, 14), metabolic demand (25), and, ultimately, muscle growth (12, 18, 47) that were predicted to be affected by the differences in serum EV miRNA content between HS and WB.

Table 3.

Top KEGG pathways predicted to be targeted by altered miRNA with HS

KEGG Pathway	P Value	No. of Genes	No. of miRNAs
Pathways in cancer	0.00626	57	34
MAPK signaling pathway	0.00046	47	27
Proteoglycans in cancer	1.60E−05	42	26
Focal adhesion	0.00173	40	29
Rap1 signaling pathway	0.00156	37	29
Transcriptional misregulation in cancer	1.60E−05	35	26
Endocytosis	0.00626	34	23
FoxO signaling pathway	0.00127	31	23
MicroRNAs in cancer	0.0075	29	25
Ubiquitin mediated proteolysis	0.00727	28	24
Estrogen signaling pathway	0.00483	20	14
Prostate cancer	0.00996	19	21
Renal cell carcinoma	0.00011	18	20
Gap junction	0.00029	18	15
TGF-β signaling pathway	0.00271	18	13
mTOR signaling pathway	0.00447	16	15
Pancreatic secretion	0.00703	16	18
ECM-receptor interaction	1.60E−05	15	10
Bile secretion	0.00156	15	14
Glioma	0.00173	14	14
Aldosterone-regulated sodium reabsorption	0.00727	12	12
Circadian rhythm	0.00017	10	12
2-Oxocarboxylic acid metabolism	1.60E−05	6	6
Valine, leucine and isoleucine biosynthesis	0.00078	2	3

ECM, extracellular matrix; HS, hindlimb suspension; KEGG, Kyoto Encyclopedia of Genes and Genomes; miRNA, microRNA; mTOR, mammalian target of rapamycin; TGF, transforming growth factor.

Follow-up RT-qPCR analysis was performed on miR-203a-3p and miR-30a-3p since these two miRNAs were significantly different (P < 0.05) and had a greater than twofold change in HS compared with WB. RT-qPCR confirmed lower EV miR-203a-3p in HS compared with WB (P = 0.003) and a trend for lower miR-203a-3p in HS compared with HSR (P = 0.066) (Fig. 6A). However, miR-30a-3p was only trending lower in HS compared with WB (P = 0.09), with no difference compared with HSR (Fig. 6A). We then examined miR-23, miR-26, miR-27, and miR-29 since they have been linked to skeletal muscle atrophy and hypertrophy (8, 20, 21, 29, 33, 62, 69) and exhibited changes in the miRNA microarray data set. RT-qPCR analysis demonstrated no significant differences in these miRNAs within EVs across groups, although the miRNAs tended to be lower in HS compared with WB (Fig. 6A). Interestingly, the muscle-specific miR-1 was not detectable in serum EVs under any of the conditions in rats.

Fig. 6.

RT-qPCR validation and follow-up of microRNA (miRNA) microarray: miR-203a-3p, miR-30a-3p, and selected atrophy-related miRNA measurements in serum extracellular vesicles (EVs; A), gastrocnemius muscle (B), kidney (C), and liver (D). HS, hindlimb suspension; HSR, hindlimb suspension followed by reloading; WB, weight bearing. miRNA expression values shown were normalized to UniSp6 with the −∆C_T method (where C_T is threshold cycle; 55) and then normalized to WB values. Data presented as means ± SE. ^aSignificantly different (P < 0.05) compared with WB; ^bsignificantly different (P < 0.05) compared with HSR. N.D., not detectable.

Muscle, kidney, and liver miRNA.

To better understand how changes in serum EV miRNA are related to muscle miRNA, we measured the same miRNAs within the gastrocnemius muscle. miR-203a-3p was not significantly different in HS muscle compared with WB but was significantly higher compared with HSR (Fig. 3B; P = 0.047). The miRNAs 23a-3p, 26a-3p, 27a-3p, and 29a-3p were elevated in HS muscle compared with both WB and HSR. There was no difference in muscle miR-30a-3p or miR-1-3p between WB, HS, and HSR (Fig. 6B).

Studies have established a potential cross talk between muscle and kidneys that is mediated by EVs (69, 70); thus we also measured the miRNAs in the kidney to see whether kidney miRNAs were altered or related to muscle levels (Fig. 6C). There were no differences in miR-203a-3p, miR-30a-3p, miR-26a-5p, and miR-27a-3p, while, as expected, muscle-specific miR-1 was undetectable. miR-29a-3p was significantly lower in HSR compared with WB (P = 0.019).

We measured the miRNAs in liver since there has been recent work claiming that exercise-induced release of EVs, which may be enriched for muscle-derived EVs, is preferentially delivered to the liver (65). Thus, under scenarios that induce muscle atrophy we postulated that there may be a relationship between the liver, muscle, and EVs. We found that miR-203a-3p, miR-23a-3p, miR-26a-5p, miR-27a-3p, and miR-29a-3p were higher in HS compared with HSR (Fig. 6D). There were no differences in miR-30a-3p, and, as expected, miR-1 was undetectable.

Relationship between EV, muscle, kidney, and liver miRNA.

We further investigated the relationship between miRNA across EVs and organs via correlational analysis (Fig. 7). In general, the direction of change of the miRNAs that we measured was correlated within EVs, muscle, and liver across all rats. The direction of change of the miRNAs within the kidney was more moderately correlated but still showed patterns similar to each other. There was little relationship between serum EV miRNA and the corresponding miRNA within the muscle, kidney, and liver (i.e., EV miR-23a-3p vs. muscle miR-23a-3p). The only miRNA within serum EVs that showed a relationship to any organ was EV miR-203a-3p, which was negatively correlated with liver miR-203a-3p (R = −0.54, P < 0.05). Interestingly, there were significant correlations between muscle and kidney miRNAs 27a-3p (R = 0.64, P < 0.05) and 30a-3p (R = −0.60, P < 0.05). Muscle miR-23a-3p was also associated with liver miR-23a-3p (R = 0.53, P < 0.05).

Fig. 7.

Correlations of microRNA (miRNA) expression within serum extracellular vesicles (EVs), muscle, kidney, and liver. Values within the matrix represent correlation coefficients (R). Values that are highlighted represent significant (P < 0.05) correlations.

Relationship between muscle size, protein synthesis, protein degradation, and miRNA.

We examined the relationship of EV miRNA and muscle miRNA with muscle size and protein synthesis by stepwise regression analysis (Table 4). We found that miR-203a-3p was the only EV miRNA that associated with three of the outcomes examined, namely, gastrocnemius muscle mass, K_syn, and K_deg. Interestingly, muscle miR-29a-3p was inversely associated with muscle mass, mean myofiber CSA, K_syn, and K_deg.

Table 4.

Results from stepwise regression

	Mass	CSA	Ksyn	Kdeg
Slope ± SE
EV miR-203a-3p	153.57 ± 70.42*		8.40 ± 2.91**	−0.005 ± 0.002**
Muscle miR-23a-3p		−1,252.33 ± 583.02*
Muscle miR-26a-5p				0.004 ± 0.003
Muscle miR-27a-3p		1,091.52 ± 492.41*		−0.006 ± 0.004
Muscle miR-29a-3p	−129.91 ± 93.69	−918.99 ± 357.78*	−13.073 ± 3.87**	0.012 ± 0.004**
Intercept	1,835.51 ± 139.75***	4,335.85 ± 360.74***	38.18 ± 5.77***	0.013 ± 0.004**
Observations	30	30	30	30
R2	0.25	0.39	0.50	0.63
Adjusted R2	0.19	0.32	0.46	0.57
F statistic	4.49* (df = 2; 27)	5.62** (df = 3; 26)	13.53*** (df = 2; 27)	10.72*** (df = 4; 25)

CSA, cross-sectional area; EV, extracellular vesicle; K_deg, degradation constant; K_syn, synthesis rate.

^*P < 0.05;

^**P < 0.01;

^***P < 0.001.

EV miR-203a-3p with muscle atrophy in humans.

ULLS caused a significant decrease in mean myofiber CSA of the vastus lateralis muscle (−16 ± 3%, P = 0.0023; Fig. 8A) in human subjects. Similar to the rats, there was a significant decrease in serum EV miR-203a-3p (−35 ± 10%, P = 0.0020; Fig. 8B) with muscle disuse atrophy in the humans.

Fig. 8.

Disuse atrophy in humans lowers serum extracellular vesicle (EV) miR-203-3p expression. A: mean myofiber cross-sectional area (CSA) in control leg and immobilized leg vastus lateralis after unilateral lower limb suspension (ULLS) for 7 days. B: serum EV miR-203a-3p expression before and after ULLS. Black dots represent ULLS subjects, and purple dots represent unilateral lower limb suspension and massage (ULLS+M) subjects. microRNA (miRNA) expression values shown were normalized to UniSp6 with the −∆C_T method (where C_T is threshold cycle; 55) and then normalized to preimmobilization values.

DISCUSSION

The overall goal of this study was to investigate the relationship between serum miRNAs and muscle atrophy. The main finding of our work is that serum EV miR-203a-3p expression is associated with protein synthesis, protein degradation, and muscle atrophy. This association is a novel finding and recognizes a previously unrecognized circulating marker for muscle protein turnover during disuse atrophy and regrowth. Our results further demonstrate that previously recognized muscle atrophy-related miRNAs may not be drastically altered in serum EVs during prolonged disuse atrophy. As such, changes in miRNAs 23a-3p, 26a-5p, and 27a-3p within the muscle may not contribute to, or be affected by, serum EV miRNAs. However, we did find an association between muscle miR-27a-3p and kidney miR-27a-3p suggesting a muscle-kidney cross talk and specific delivery of miR-27a-3p to the kidney as previously demonstrated by other laboratories (69).

The limited significant differences in serum EV miRNA with 7 days of HS were surprising and suggest either that we missed changes that occurred early in HS or that large shifts in serum EV miRNA expression do not occur with disuse atrophy. However, when we examined predicted pathways that may be altered by the differences in EV miRNAs with HS (miRNAs P < 0.05) we found multiple pathways related to anabolic and catabolic signaling (3, 4, 16, 57), response to mechanical stimulus (11, 14), and metabolic demand (25). In particular, KEGG pathways such as MAPK signaling, focal adhesion, FoxO signaling, ubiquitin-mediated proteolysis, transforming growth factor (TGF)-β signaling, and mammalian target of rapamycin (mTOR) signaling were predicted to be targeted by the differentially expressed EV miRNAs. This demonstrates that although there were few major changes in individual EV miRNAs, the general shifts in EV miRNA expression reflected the alterations in major anabolic, mechanosensitive, and catabolic pathways that typically occur with disuse atrophy.

Microarray analysis revealed that miRNAs 203a-3p and 30a-3p were the only two miRNAs to exhibit a greater than twofold change in serum EV expression with muscle disuse atrophy. However, qPCR validation revealed that only miR-203a-3p was significantly lower in EVs with HS, although there was a trend for lower miR-30a-3p. miR-203a-3p has been shown to promote proliferation of prostate cancer cells (67) and hepatocytes (71) while hindering proliferation of ovarian cancer cells (31), gastric cancer cells, and skeletal muscle cells (32). The impact on cellular proliferation and differentiation is a consequence of miR-203-3p targeting and suppression of IGF-1R (63), MEF2C (32), p63 (68), Rap1A (67), PTEN (71), and Akt/GSK-3β (31). Our data across both rats and humans suggest that EV miR-203a-3p is a good biomarker for representing the catabolic/anabolic state of skeletal muscle. It was the only EV miRNA associated with muscle mass, protein synthesis, and protein degradation. A previous study has shown that elevated circulating miR-203 levels are associated with muscle wasting (44), which is opposite to the positive association of EV miR-203 expression and muscle size shown in this study. Disparity between the studies may be due to the type of atrophy examined: our study was solely focused on disuse atrophy, whereas the study by Okugawa et al. (44) examined cancer-related muscle loss in colorectal cancer patients. It is possible that different relationships exist in circulating miR-203 depending on the paradigm in which muscle loss is induced. It is also important to note that we examined miR-203 in EVs whereas their studies encompassed all circulating miR-203.

miR-23, miR-26, miR-27, and miR-29 have been linked to skeletal muscle atrophy and hypertrophy (8, 20, 21, 29, 33, 62, 69) and exhibited differences with atrophy and regrowth in the muscle but showed no significant changes within EVs. Both miR-23a and miR-27a have been proposed to be modulated by EVs in a diabetic model of muscle wasting in mice (21, 69). Additionally, miR-29 has also been associated with muscle wasting in a model of chronic kidney disease and was also associated with exosome-driven regulation of kidney fibrosis (62). The work by Hudson et al. (21) suggested that miR-23a was reduced in muscle during dexamethasone-induced or diabetes-induced atrophy via expulsion of miR-23a from the muscle within exosomes. This finding contradicts the elevation we see in miR-23a in the gastrocnemius muscle with disuse atrophy and the lack of elevation in miR-23 in serum EVs. However, bed rest or limb immobilization in humans led to elevated miR-23a and miR-26a in skeletal muscle, which matches our results (53). Furthermore, miR-29 has been shown to attenuate atrophy in a mouse model of kidney disease (62), whereas other recent work by Li et al. (29) demonstrated that elevated miR-29 in muscle contributes to multiple types of atrophy. The study from Li et al. (29) agrees with our present data showing that elevated miR-29 in muscle is associated with atrophy. Resistance training-induced hypertrophy has also been associated with a reduction in the abundance of muscle miR-29a and miR-26a in muscle (8). Altogether, our data show no association between atrophy-associated miRNAs within EVs and those within muscle. It also clearly demonstrates that there is significant work left to understand how miRNA within muscle changes with atrophy and how miRNA loading into EVs plays a role.

There is no definitive evidence regarding the contribution of muscle EVs to the total pool of serum EVs. As mentioned above, miR-23a has been shown to be exported out of muscle cells via EVs with dexamethasone treatment and create an inverse relationship between EV and muscle miRNA miR-23a content (21). If we only consider the average differences in serum EV miRNA and muscle miRNA (Fig. 6) in our study, there is generally an inverse relationship between them, and it is tempting to speculate that these changes are related. However, when all data points from each rat are assessed via correlational analysis, we show that the serum EV expression of miR-203, miR-23, miR-26, miR-27, and miR-29 is not reflective of muscle expression values. We observed that the muscle-specific miR-1-3p was not readily detectable in serum EVs, which may also reflect a very small contribution of muscle-derived EVs to the serum or very little regulation of muscle miRNA via loading into EVs. It is also possible that there are abundant muscle-derived EVs circulating in the serum but they contain very small amounts of miR-1 relative to other miRNAs. Finally, it cannot be directly ascertained from our samples whether any serum miRNA changes are due to muscle or contributions from other organs since there are no reliable ways to track in vivo origin of serum EVs. There is much work needed in future studies to better understand these specific inquiries about EVs, their origin when in serum, and muscle atrophy.

Previous work has shown muscle-specific cross talk to the kidneys in two mouse models of muscle wasting: unilateral ureteral obstruction (UUO) [a model of chronic kidney disease (CKD)] (70) and a streptozotocin (STZ)-induced diabetic model (69). The muscle loss in the CKD model was associated with lower serum EV miR-26a-5p, which is similar to our study; however, their mice had lower expression of muscle miR-26a-5p with atrophy. Interestingly, when miR-26a-5p was overexpressed in the atrophied muscle via intramuscular injection of exosomes containing miR-26, it caused an increase in muscle and serum EV miR-26a-5p combined with an increase in kidney expression of miR-26a-5p. Our data support that muscle atrophy is associated with lower serum EV miR-26 but disuse atrophy in our rats caused an increase in muscle expression of miR-26. Our data did not show a relationship between muscle and kidney expression levels of miR-26 and suggest that the muscle and kidney expression of miR-26-5p are not related. Interestingly, the same research group showed evidence that muscle may also deliver miR-23 and miR-27 to the kidney to combat fibrosis in a STZ-induced diabetic mouse model (69). Specifically, when they elevated miR-23/27 via adeno-associated virus (AAV) injection into the muscle, they saw an increase in kidney miR-23/27 in the kidney via exosomal delivery. In our disuse atrophy model, we observed a significant positive relationship between muscle and kidney expression of miR-27a-3p, which supports the model of EV-mediated delivery of miR-27a-3p from the muscle to the kidney. Altogether, our data provide further support of a relationship between muscle and kidney miRNA cross talk, but further work is needed to verify the specific roles of EVs in this process.

Our data showing that miR-203a-3p may be a good biomarker for muscle atrophy and protein synthesis in rats are novel, so we further investigated whether this finding was clinically relevant in human subjects. The ULLS method had been used extensively to induce atrophy in human subjects (reviewed in 17), and our 7-day ULLS intervention caused significant atrophy of the vastus lateralis muscle in our subjects. Furthermore, the atrophy of a single limb was associated with a significant drop in serum EV miR-203a-3p expression similar to what we observed in the hindlimb suspension model with rats. The decrease in serum EV miR-203a-3p with the more subtle atrophy stimulus in humans suggests that serum EV miR-203a-3p is indeed a sensitive marker of muscle size status. Recently published work has highlighted a large disparity in EV small RNA profiles between rats and humans when EVs were isolated with ExoQuick methods (72). Despite this, we were able to find that miR-203a-3p content is a consistent marker of atrophy and protein turnover across species and lends confidence that our results in rats were not species specific or due to our methodology used for EV isolation.

Our data provide novel and useful information regarding changes in serum EV miRNA and their relationship to muscle atrophy and other organs; however, there are several limitations that must be recognized. First, the field of EV research is rapidly evolving, and techniques for the isolation of EVs have been a growing and debated topic. In particular, there has been a growing appreciation for the need to use labor-intensive density gradient ultracentrifugation and/or size exclusion chromatography techniques to isolate more pure EV samples since serum and plasma samples are saturated with an abundance of non-EV lipids, proteins, and nucleic acids (1, 45, 46). Because of the small sample volume availability in the present study, however, we opted to isolate EVs with the commercially available ExoQuick reagent. This reagent is extremely efficient at isolating a large yield of EVs from serum and plasma samples for miRNA analysis (19, 52, 72), but it is known that with these types of methods many other non-EV nucleic acids, proteins, and lipoproteins are precipitated and may have impacted some of the results (15, 22, 59, 60). Second, our results may not be transferable to other forms of atrophy as there are disparate results on miRNA changes with other forms of atrophy in the literature as discussed above. We believe, though, that the combination of our rat and human data provides adequate evidence to promote further investigation into miR-203a-3p and its role specifically in disuse atrophy. We also acknowledge that we may have missed divergent changes in miRNA occurring at earlier time points in disuse atrophy. Altogether, we believe that despite these limitations we have provided a novel glimpse into the relationship and changes between serum EV miRNA and muscle miRNA with disuse atrophy.

In conclusion, our novel findings demonstrate that serum EV miR-203a-3p expression is a sensitive measure of muscle protein status. Specifically, serum EV miR-203a-3p expression was associated with muscle mass, protein synthesis, and protein degradation across weight bearing, disuse, and reloading conditions. In general, we also show that although the overall changes in the profile of serum EV miRNAs are rather subtle in the hindlimb suspension model of atrophy and reloading, the changes generally reflect alterations in anabolic and catabolic signaling during atrophy and regrowth. Future studies should determine the mechanistic impact and specificity by which changes in serum EV miR-203a-3p expression alter function in both muscle and other organs.

GRANTS

This work was supported by NIH Grant AT-009268 (E.E.D.-V, T.A.B., and B.F.M.). The project described was supported by the NIH National Center for Advancing Translational Sciences through grant number UL1TR001998. The content is solely the responsibility if the authors and does not necessarily represent the official views of the NIH.

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

D.W.V.P. and E.E.D.-V. conceived and designed research; D.W.V.P., I.J.V., M.M.L., and P.P. performed experiments; D.W.V.P., M.M.L., K.L.V.P., P.P., and B.F.M. analyzed data; D.W.V.P., I.J.V., K.L.V.P., B.F.M., T.A.B., and E.E.D.-V. interpreted results of experiments; D.W.V.P. and K.L.V.P. prepared figures; D.W.V.P. drafted manuscript; D.W.V.P., I.J.V., M.M.L., K.L.V.P., B.F.M., T.A.B., and E.E.D.-V. edited and revised manuscript; D.W.V.P., I.J.V., M.M.L., K.L.V.P., P.P., B.F.M., T.A.B., and E.E.D.-V. approved final version of manuscript.