MicroRNAs (miRNAs) are short, noncoding RNAs that influence biological processes by regulating gene expression after transcription. It was recently discovered that miRNAs are released into the circulation (ci-miRNAs) where they are highly stable and can act as intercellular messengers to affect physiological processes. This review provides a comprehensive summary of the studies to date that have investigated the effects of acute exercise and exercise training on ci-miRNAs in humans. Findings indicate that specific ci-miRNAs are altered in response to different protocols of acute and chronic exercise in both healthy and diseased populations. In some cases, altered ci-miRNAs correlate with fitness and health parameters, suggesting causal mechanisms by which ci-miRNAs may facilitate adaptations to exercise training. However, strong data supporting such mechanisms are lacking. Thus, a purpose of this review is to guide future studies by discussing current and novel proposed roles for ci-miRNAs in adaptations to exercise training. In addition, substantial, fundamental gaps in the field need to be addressed. The ultimate goal of this research is that an understanding of the roles of ci-miRNAs in physiological adaptations to exercise training will one day translate to therapeutic interventions.
- circulating microRNA
- intercellular communication
scientists first identified microRNAs (miRNAs or miRs) circulating in human serum/plasma in approximately 2008 (19, 59, 71). These circulating miRNAs (ci-miRNAs) are secreted into the circulation by a variety of cell types, and can be taken up by other cells, where they may regulate numerous physiological processes (16). In this way, they mediate communication between cells and tissues throughout the body. Due to their remarkable stability and ease of access, the promise of ci-miRNAs as useful biomarkers of diseases was recognized immediately (71). Since their discovery, ci-miRNAs have been investigated as biomarkers and moderators of essentially all human diseases (42, 105, 107). Specifically, the roles of ci-miRNAs as biomarkers for cardiovascular diseases (CVD) and skeletal muscle pathologies have been detailed (5, 76). Conversely, miRNAs are crucial to the normal development and health of these systems (46, 61). Considering the well-established beneficial effects of exercise training on the cardiovascular system, skeletal muscle, and body as a whole, it seems logical that ci-miRNAs would be altered in response to exercise, and may be involved in mediating or moderating some or all of the adaptations to exercise training. Indeed, this has very recently become an area of increased research interest, and several studies have shown changes in ci-miRNAs in response to various modes of acute and chronic exercise.
Although recent “-omics” studies (e.g., genomics, proteomics, transcriptomics, metabolomics) have been used to interrogate the pathways underlying physiological training adaptations, results of these studies to date have been modest. In this respect, ci-miRNAs may be important missing links, given their potential to modulate pathways upstream in the “-omic” adaptations to exercise training. Ci-miRNAs hold the promising potential to unlock a greater understanding of the integrative physiology of exercise and individual exercise responses. Therefore, the goals of this review are threefold: 1) present a comprehensive review of studies examining the effect of acute exercise and/or training on ci-miRNA expression in human serum, plasma, or whole blood; 2) highlight potential mechanisms by which ci-miRNAs may induce adaptations with training; and 3) identify current gaps limiting knowledge in the field and suggest future directions for the study of exercise and ci-miRNAs. Although we will address all studies on exercise/physical activity and ci-miRNAs in humans, the primary focus of this review paper will be on aerobic exercise and endurance training, because that has been the major focus of the field thus far.
MiRNAs are noncoding RNAs, ~22 nucleotides in length, that regulate gene expression at the posttranscriptional level. The first miRNA was discovered in the nematode Caenorhabditis elegans in 1993 by Lee, Feinbaum, and Ambros (60), who in simultaneous experiments with Ruvkun’s group (111), reported that the small lin-4 RNA could base pair with a segment of the 3′-untranslated region of another RNA, lin-14, to decrease abundance of the LIN-14 protein. In 1998, Fire and colleagues (32) demonstrated the experimental potential of the RNA interference mechanism when they injected small, double-stranded RNA into C. elegans and observed potent downregulation of specific endogenous mRNA. Since those initial discoveries, the field of miRNA research has expanded immensely. The miRNA pathway is now known to be highly conserved in mammals and vital in biological processes such as cell growth, cell proliferation, motility, tissue differentiation, and embryonic development (35, 63, 89, 113). Displaying their importance, the number of miRNAs within the genome correlates with the complexity of the organism (10). Currently, 2,588 miRNAs have been identified in the human genome. They are expressed in seemingly all cell types and it is predicted that over 60% of protein coding-genes are conserved targets (35).
Genes for miRNAs are encoded in both intergenic and intronic regions typically located in polycistronic transcription units (41). Interestingly, miRNAs encoded within intronic regions often repress their cotranscribed host gene(s) and, in addition, have their own promoters, enabling independent transcription and distinct expression from their host gene(s) (84). For in-depth reviews of canonical miRNA biosynthesis, we direct readers to other excellent reviews (41, 113). Briefly, polymerase II transcribes miRNA genes into a ≥1,000 base pair (bp) initial primary miRNA (pri-miRNA) containing the mature miRNA sequence in a stem-loop structure (113). The pri-miRNA is cleaved within the nucleus by Drosha into a pre-miRNA of ~65 bp, which then associates with exportin-5 to be transported into the cytoplasm. Here, Dicer further cleaves the pre-miRNA near its terminal loop resulting in a 21- to 25-bp double-stranded RNA consisting of a mature miRNA and its complementary strand. The miRNA duplex quickly unwinds, and the mature miRNA is loaded onto an Argonaute protein, which associates with Dicer to form the miRNA-induced silencing complex (miRISC) (15, 95). The mature miRNA then acts as a guide for miRISC by complementarily binding to the 3′-untranslated region of target mRNA transcripts, while the associated “passenger” strand is usually released and degraded (41, 95). The miRISC-targeted mRNA may be perfectly or imperfectly complementary, having implications on the mechanism of silencing that is carried out. Target recognition and binding due to complementarity are determined primarily by nucleotides 2 to 8 on the 5′ end of the mature miRNA (28). Imperfect binding is advantageous in that an individual miRNA can target numerous mRNAs. In fact, individual miRNAs can regulate up to 200 mRNA transcripts, whereas conversely, it is possible for multiple miRNAs to share the same mRNA target (28). After binding, miRISC partially or completely suppresses gene expression via induction of translational repression, mRNA deadenylation, mRNA degradation, or sequestration in processing bodies (P-bodies) (50). Recent evidence also indicates that miRNA-mediated upregulation of transcription can occur in certain cases (102). Lastly, miRNAs often exhibit tissue specific expression and functions, which are occasionally contradictory. For instance, in liver and intestinal cancers, miR-26 acts as a tumor suppressor, whereas it is an oncogene in brain cancer (49, 55, 117). Thus, miRNAs act in an intricate and coordinated manner to fine-tune protein expression, although knowledge about the mechanisms governing their expression and regulation is incomplete.
The fairly recent discovery of miRNAs in human biofluids, including serum, plasma, saliva, sweat, cerebrospinal fluid, urine, and milk, has revealed new possibilities for miRNA research (16, 71). Such ci-miRNAs are actively or passively secreted into the bloodstream where they circulate in association with extracellular vesicles (EVs) (microvesicles, exosomes, or apoptotic bodies), proteins (e.g., Argonaute), or high-density lipoproteins (HDLs). Ci-miRNAs can then prompt downstream effects upon uptake by target recipient cells by regulating translation of complementary mRNAs (11, 45, 85, 118, 120). Only select miRNAs within cells are released into the circulation, and although miRNAs are released passively due to cell damage or senescence (e.g., miR-1 following myocardial infarction) in some cases (13), evidence supports the deliberate packaging and release of miRNAs in response to stimuli as a means of intercellular communication (45). The intercellular transport of miRNAs and subsequent functional regulation of gene expression in recipient cells is now a well-supported mechanism of cell-to-cell communication involving a variety of cell types and transport methods (11, 45, 85, 118, 120). However, the precise mechanisms underlying miRNA secretion/packaging and uptake by recipient cells are not known.
MiRNAs with aberrant expression in the circulation are reportedly associated with at least 70 diseases, and ci-miRNAs have been extensively investigated as useful disease biomarkers given their ease of acquisition, high stability, and convenient amplification (31, 42, 43, 58, 63). The potential use of ci-miRNA-based therapeutic interventions is also compelling. Long-term suppression of the hepatitis C virus by intravenous infusion of a complementary antagonist to miR-122 has been demonstrated in chimpanzees (58). This and other miRNA-based therapies are already in preclinical development and clinical testing for diseases such as coronary heart failure and myocardial infarction.
Although the majority of studies have focused on differentially expressed ci-miRNAs in disease settings, the effects of other physiological alterations have also been investigated, albeit to a much lesser extent. Exercise is one such stressor that induces a well-characterized response, and if repeated (training), results in beneficial adaptations within numerous physiological systems. Yet, the study of ci-miRNAs in the context of acute exercise and exercise training has lagged, with only a recent upsurge in interest. To date, there have been 30 published studies written in English on the effects of acute exercise, training, or fitness level on ci-miRNA levels (Table 1). These papers were found through a comprehensive search of the PubMed database and by reviewing references of included studies. Of the studies performed thus far, most have investigated responses to acute exercise, with most of them focused on aerobic exercise. Still, results from these few investigations are promising. Ci-miRNAs altered with exercise have been proposed as potential biomarkers with clinical applications (38), and are proposed to participate in mechanisms governing the acute response to exercise and adaptations to exercise training. The studies reviewed in the following sections are presented according to both exercise modality and (dis)agreements in results, which facilitates a discussion of proposed roles for ci-miRNAs in the physiological responses to exercise. Additionally, the results of investigations on the two groups of most studied ci-miRNAs that are enriched in striated muscle and the endothelium are concisely summarized in Table 2 and Table 3, respectively.
Ci-miRNAs in the Response to Acute Exercise
The first investigation of the effect of exercise on ci-miRNAs came in 2011 from Baggish and colleagues (7). They examined the effect of a cycle ergometer maximal oxygen consumption (V̇o2max) test on 12 ci-miRNAs in young, moderately trained men. Immediately after exercise, ci-miRs -146a, -222, -21, and -221 increased in plasma whereas the others remained unchanged. These four miRNAs are highly enriched in the endothelium compared with other cell types and were chosen a priori on the basis of their roles in angiogenesis and/or inflammation. Along with ci-miRs -126 and -210, this group of endothelial-derived ci-miRNAs is now among the best described in regards to the response to exercise (Table 3). Interestingly, absolute ci-miR-146a level immediately after exercise exhibited a significant linear correlation with absolute V̇o2max (r = 0.63), prompting the investigators to suggest its potential as a biomarker for cardiorespiratory fitness and peak exercise capacity, although this has not been verified.
More recently, Backes et al. (6) similarly assessed the effect of an incremental, maximal cycle ergometer test in young men and women. They compared the ci-miRNA changes in whole blood of elite endurance athletes with that of moderately active controls. Using a microarray of 1,205 miRNAs, they identified three possible targets, but found no significant differences upon validation using quantitative RT-PCR (qRT-PCR). Discrepancies between these two studies may have resulted from the sample type used (whole blood vs. plasma), small sample sizes/gender differences, and/or time of sampling. RNA isolated from whole blood contains intracellular miRNA from circulating cells and therefore cannot be compared directly with RNA in plasma or serum. Backes et al. (6) included six men and six women in each of their two study groups, but they did not report any sex-based comparisons. Ci-miRNAs were shown to be associated with age, gender, and body mass index in a population-based cohort study, and there are known hormonal and genetic differences in miRNA regulation between sexes (3, 93). Moreover, it was shown in another study that ci-miRNAs varying at rest between high V̇o2max and low V̇o2max groups were different depending on gender (14). Ci-miR-21 was higher in the low V̇o2max vs. the high V̇o2max group only in men (14). Finally, Backes et al. (6) sampled blood 30 min after cessation of exercise, and Baggish et al. (7) found significant changes immediately after exercise, but showed that all ci-miRNAs were back to baseline after 1 h.
Van Craenenbroeck and colleagues (103) compared the response of 12 ci-miRNAs after a symptom-limited maximal cycling test in patients with ambulatory chronic kidney disease (CKD) with the response of healthy controls. Immediately after exercise, ci-miR-150 increased in all subjects, whereas ci-miR-146a decreased only in patients with CKD. Additionally, the same research group recently showed a negative correlation of ci-miRs -146a, -150, and -210 with V̇o2peak in patients with CKD, which was lost when corrected for arterial stiffness, as measured by carotid-femoral pulse wave velocity (104). Interestingly, this negative correlation of ci-miR-146a with aerobic capacity is opposite to the positive correlation previously found in young, healthy men (7). Both miR-146a and miR-150 also correlated with arterial stiffness in patients with CKD and have been implicated in numerous mechanisms of CVD development and disease (104). Further work is necessary to determine their use as biomarkers or therapeutic targets in this population. MiR-150 is also involved in physiological left-ventricular hypertrophy (66), raising the possible role for ci-miR-150 in this beneficial adaptation to training.
Conversely, patients with heart failure have exhibited no change in ci-miRs -146a or -150 immediately following a symptom-limited maximal test, although ci-miRs -21, -378, and -940 all increased (116). Changes in these ci-miRNAs did not correlate with V̇o2max, markers of inflammation, or muscle damage. The fact that cardiac muscle-enriched miR-940 increased in circulation while other cardiac and skeletal muscle-enriched miRNAs and markers of muscle damage and inflammation did not, implies distinct mechanisms of miRNA release from muscle into the circulation with exercise. This is supported in a report by Banzet et al. (9) comparing the response to downhill backward walking with that of uphill walking in young, recreationally active, healthy men. For this study, subjects performed 30 min of both types of treadmill walking at a 25% grade while wearing a weighted backpack, and changes in eight muscle-related ci-miRNAs were assessed for up to 72 h afterward. The only change with uphill exercise was an increase in ci-miRs -181b and -214 immediately after. On the other hand, there was no change immediately following downhill exercise, although ci-miR-1 was increased 2 h after exercise; and ci-miRs -1, -133a, -133b, and -208b were increased 6 h after exercise. Although the four ci-miRNAs that increased with downhill running were postulated to be released as a result of skeletal muscle damage, miR-181b and miR-214 are not muscle specific, but are present in many tissues. For example, they are upregulated in muscle in response to hypoxia (56, 91), and miR-181b is upregulated in neutrophils and peripheral blood mononuclear cells following cycling exercise (82, 83). Although the cellular origins of these ci-miRNAs are unknown in this study, it seems likely that they are actively as opposed to passively released, which may be the case with those released after the muscle-damaging downhill exercise, for example.
Notably, Guescini et al. (40) have shown that at rest, muscles secrete EVs containing miRNA into the circulation. They then found increased ci-miR-181a-5p in EVs isolated from plasma of healthy men after a 40-min treadmill run at 80% V̇o2max. Ci-miRs -1, -133b, -206, and -499 were also present in muscle-derived EVs, all of which exhibited a positive correlation between expression and relative V̇o2max. A failure of these other common muscle-specific ci-miRNAs to increase, though, indicates a mechanism of selective miRNA packaging and release in EVs from skeletal muscle in response to exercise. The mechanisms underlying this process, the ultimate destination and purpose of these EV-encapsulated ci-miRNAs, and whether they may explain adaptations to exercise, are all relevant questions that are yet to be investigated.
In another investigation of muscle-enriched ci-miRNAs, Aoi and colleagues (4) had healthy men undergo cycling at 70% V̇o2max for 60 min. Out of seven ci-miRNAs, only ci-miR-486 showed a response, decreasing immediately after and returning to baseline 3 h after exercise. There was a negative correlation between the fold-change in ci-miR-486 with exercise and relative V̇o2max (r = 0.58). To explain this, the authors suggested the possibility of increased uptake of ci-miR-486 by contracting skeletal muscle, which may then affect energy metabolism. They did not investigate ci-miRs -181b or -214, but the lack of a response in ci-miRs -1, -133a, and -133b implies that they are not released by muscle contraction alone, and supports the hypothesis that increased secretion of these ci-miRNAs may be due to muscle damage. More recently, ci-miR-486 was also shown to decrease in whole blood of young, recreationally inactive men immediately following a maximal treadmill test (26). Furthermore, in a larger sample including both endurance-trained and inactive men, resting levels of ci-miR-486 correlated positively with V̇o2max (r = 0.20) and negatively with resting heart rate (r = −0.31), signifying it may play a positive role in cardiovascular fitness (26). Ci-miRs -1 and -133a additionally decreased following the max test, and resting levels of ci-miR-1 correlated with V̇o2max (r = 0.25) (26).
Nielsen and colleagues (77) had healthy, trained men cycle for 60 min at 65% of maximal power output. Eight ci-miRNAs were downregulated immediately after, including -146a. Following 1 and 3 h of rest after the exercise, seven and four different ci-miRNAs were upregulated, respectively. The fact that there was no overlap of differentially expressed ci-miRNAs at any time point suggests distinct time-sensitive mechanisms of release. The investigators performed an initial global screening of 752 ci-miRNAs and a validation of 188 of these targets, which also raises the possibility of false positives. Because most exercise and ci-miRNA studies examine only a small number of targets chosen a priori, the results of this study are not necessarily in disagreement with previous results, but they should be verified in another cohort. Cui et al. (21) likewise found a reduction in serum ci-miRNAs immediately after exercise in an investigation of seven a priori chosen targets. Ci-miRs -1, -133a, -133b, -122, and -16 all decreased following two Wingate tests separated by a 4-min period of rest. In addition, ci-miR-133b correlated positively with peak power of the first Wingate (r = 0.712), whereas ci-miR-122 correlated positively with the ratio of peak power of sprint one with that of sprint two (r = 0.665). What role these ci-miRNAs may play in anaerobic capacity deserves attention. The same investigators more recently compared a session of repeated bouts of sprinting [high-intensity interval exercise (HIIE)] with a bout of vigorous-intensity continuous exercise (VICE) in endurance-trained men, using a global screening of ci-miRNAs (22). Immediately following both exercise bouts, an upregulation of 12 ci-miRNAs occurred, including ci-miRs -1, -133a, and -133b. The only difference was a lower ci-miR-1 level following HIIE compared with VICE, suggesting the sprint intervals may induce more or different muscle damage than the sustained endurance exercise. Kilian et al. (53) performed a similar comparison, although it was in adolescent, competitive male cyclists, and the researchers isolated ci-miRNAs directly from capillary blood obtained from the ear. Both 20 min into and 60 min after VICE, ci-miR-126 was increased. At the 30-min postexercise time point, ci-miR-16 was increased due to VICE, whereas ci-miR-21 was decreased due to HIIE. The decrease in ci-miR-21 with HIIE and increase in ci-miR-126 with VICE may reflect specificity of exercise adaptations, because both of these miRNAs are positive regulators of the proangiogenic protein vascular endothelial growth factor (VEGF) and are upregulated in endothelial cells by sustained laminar shear stress (74, 110).
MiR-126 is a well characterized endothelial-enriched miRNA that induces angiogenesis by inhibiting SPRED1 and PIK3R2, resulting in upregulation of VEGF (33). da Silva Jr. et al. (24) evaluated the change in whole blood ci-miR-126 and other circulating angiogenic factors after a maximal treadmill walking test in patients with intermittent claudication. Furthermore, the effect of the antioxidant drug N-acetylcysteine (NAC) was assessed. They first found that NAC supplementation improved redox balance 1 h after supplementation and following the exercise bout. However, expression of angiogenic factors ci-miR-126, VEGF, and eNOS increased in circulation 30 min after exercise in the placebo session, but did not change in the test following NAC supplementation. Also, PIK3R2 mRNA, the antiangiogenic target of miR-126, increased in circulation after exercise only in the NAC session. These results were contrary to the authors’ hypotheses and led them to conclude that miR-126 signaling is redox-sensitive in patients with intermittent claudication. This role of exercise-induced oxidative stress in stimulating angiogenic ci-miR-126 release from the endothelium is intriguing and remains to be elucidated. Whether this process is also true in healthy individuals, or is a result of peripheral arterial disease, should also be determined.
Ci-miR-126 has been shown to increase in healthy, middle-aged individuals immediately after a maximal cycle ergometer test (101). In addition, Uhlemann et al. (101) measured changes in ci-miR-126 and -133 in response to three other modes of exercise. Ci-miR-126 increased 30 min into cycling at 70% ventilatory threshold, and remained elevated throughout the entire 4-h bout. Following a marathon, both ci-miR-126 and -133 were elevated, whereas only ci-miR-133 was upregulated after resistance exercise with added eccentric load. This lends support for ci-miR-133 as a marker of skeletal muscle damage. The authors argued that the increase in ci-miR-126 is evidence of endothelial damage/lysis during aerobic exercise, although they did not provide any evidence to support this claim. In their study, ci-miR-126 was increased after just 30 min of cycling at 70% ventilatory threshold, whereas a previous paper showed no evidence of endothelial damage, determined by number of circulating endothelial cells, until 2 h into cycling at the same intensity (72, 101). Together, these data do not point toward endotheli al damage as the cause of increased ci-miRNA, indicating that other mechanisms are responsible for the secretion of endothelial ci-miR-126. In addition to the potential role of redox balance discussed above, laminar shear stress is known to upregulate endothelial cell expression of miRs -126 and -21 (74, 110), as well as secretion of miRs -143, -145, and -150 in vitro (45, 51). Moreover, Jae et al. (51) recently showed that shear stress-responsive transcription factor KLF2 induces selective packaging and export of miRNAs from endothelial cells that is independent of intracellular levels of upregulation. Finally, miR-126 is released from circulating angiogenic CD34+ cells (73), so their contribution to circulating levels cannot be discounted.
Long-distance running is a well characterized exercise stimulus in relation to ci-miRNA response. In 2014, Baggish et al. (8) examined the changes in seven ci-miRNAs in healthy men following the Boston Marathon. Immediately after the run, ci-miRs -1, -126, -133a, -134, -146a, -208a, and -499-5p were all increased, and returned to baseline by 24 h after the run. Meanwhile, creatine phosphokinase, troponin I, NH2-terminal prohormone of brain natriuretic peptide (NT-proBNP), and high-sensitivity C-reactive protein (hsCRP) as markers of skeletal muscle damage, cardiomyocyte stress, cardiomyocyte damage, and inflammation respectively, all remained elevated 24 h after the marathon. This adds credence to miRNA-specific mechanisms of secretion and uptake from circulation, even in the case of muscle-damaging exercise. That same group of researchers (70) also found increases in the skeletal muscle-derived ci-miRs -1, -133a, -134, and -206 in individuals with hypercholesterolemia after the Boston Marathon. Again, all ci-miRs returned to baseline after 24 h, although ci-miR-499-5p was increased at 24 h only in those runners who used statins. These data were supplemented with the ci-miRNA response of contracting mouse C2C12 myotubes exposed to statins, as an in vitro model of exercise and statin exposure. In conjunction with the human data, ci-miRs -1, -133a, -134, and -206 were increased in the culture media following contraction and regardless of statin exposure. An increase in miR-499-5p secretion from the myotubes was observed only in response to a combination of statin exposure and contraction. Statins are known to cause skeletal muscle dysfunction/damage when combined with exercise (68), and this study identifies ci-miR-499-5p as a potential biomarker for this phenomenon. Interestingly, statins have also been shown to attenuate the normal increase in skeletal muscle mitochondrial content with aerobic exercise training (69), whereas intracellular miR-499 has been shown to regulate expression of mitochondrial proteins (119). Thus, future studies should investigate the role that secretion of miR-499-5p plays in statin-induced muscular dysfunction and whether it is taken up by nonactive muscle or other tissues to cause a systemic response. Contrarily, statins are known to elicit beneficial vascular adaptations (106). In one study of statin use, plasma levels of ci-miR-122 and -370 were higher in patients with hyperlipidemia than in healthy controls, but were lower in statin users than nonusers (36). It is unclear whether ci-miRs -499-5p, -122, -370, or other vascular-related miRNAs are involved in statin-induced vascular adaptations, and also, what interaction exercise might have.
In 2015, de Gonzalo-Calvo et al. (25) measured an array of 106 ci-miRNAs involved in inflammation and four muscle-specific ci-miRNAs in nine trained men after both a 10-km run and a marathon race. Immediately after the 10-km run, only ci-miR-150 was increased. After the marathon, a total of 12 inflammatory ci-miRNAs were upregulated, not including ci-miR-150, and all returned to baseline following 24 h of rest. Inflammatory markers IL-6, IL-8, IL-10, and C-reactive protein also increased immediately after the marathon, but not the 10-km run. This study provides evidence for a dose-dependent response of inflammatory ci-miRNAs to exercise and suggests that they play a role in exercise-induced inflammation. Still, it cannot be determined whether these ci-miRNAs were a cause or merely a response to the inflammation, as mentioned by the authors (25). Ci-miR-150 also correlated positively with increased leukocyte and neutrophil (as mediators of inflammation) counts immediately after the 10-km run, suggesting them as possible sources. Surprisingly, there was no change in the four muscle-specific ci-miRs (-1, -133a, 133b, or -206) after either race. Increases in skeletal and heart secreted ci-miRNAs immediately after both a marathon and half marathon have been shown in other studies of endurance-trained men (39, 75). Although the study by Gomes et al. (39) included only five subjects, it found an increase in ci-miRs -1, -133a, and -206 immediately after a half marathon. Uniquely, only the study by Mooren et al. (75) reported a sustained elevation in ci-miRs -1, -133a, and -206 after 24 h of rest. The reasons for this discrepancy are unknown, given that the subjects were healthy, middle-aged men who were exercise trained, the same characteristics as those used in the studies by Baggish et al. (8), de Gonzalo-Calvo et al. (25), and Min et al. (70). In addition, a positive correlation was reported for the increases in ci-miRs -1, 133a, and -206 with relative V̇o2max and running speed at lactate threshold (75). Ci-miR-133a was also correlated to thickness of the intraventricular septum. Relatedly, Clauss et al. (20) found a negative correlation between ci-miR-1 and -133a levels immediately after a marathon and left atrial diameter. This correlation was true for left atrial diameter both immediately and 24 h after the marathon only in “elite” marathon runners. This result indicates that training intensity and/or volume may alter the ci-miRNA response to a single exhaustive bout of exercise. Furthermore, the potential roles of ci-miRs -1 and -133a in endurance training-induced atrial remodeling should be explored.
Currently, only three studies have examined the effect of resistance exercise on ci-miRNAs. As mentioned, Uhlemann et al. (101) found increased ci-miR-133 after 3 sets, 15 repetitions of lat pulldown, leg press, and butterfly with added eccentric load. Sawada and colleagues (92) found no immediate response to 5 sets, 10 repetitions of bench press and leg press at 70% of 1 repetition maximum. There was a delayed increase in ci-miR-149* 1 day after, and a decrease in -146a and -221 three days after the exercise. What role the endothelial-enriched ci-miRs -146a and -221 play in adaptations to resistance training and whether they regulate muscle hypertrophy remains to be determined. Moreover, changes in ci-miR-21 correlated with adrenaline and noradrenaline, whereas changes in ci-miR-221 correlated with insulin-like growth factor-1 and testosterone (92). Although Uhlemann et al. (101) examined only two ci-miRNAs selected a priori, Sawada et al. (92) performed a global screening of miRNAs in circulation.
While the previous two studies of resistance exercise were performed on relatively young participants, Margolis et al. (65) compared the response to 3 sets, 10 repetitions of bilateral leg extension and leg press in young (age, 22 ± 1 yr) and older (age 74 ± 2 yr) men. Out of 90 serum miRNAs assessed using qRT-PCR, there were no changes immediately after exercise, though nine were significantly increased 6 h after exercise only in young men. By combining these data with changes in skeletal muscle mRNA expression from biopsies (vastus lateralis) of the same participants, presented in a separate article (87), the investigators performed an Ingenuity miRNA target filter analysis (65). Positive correlations were observed for six of the altered ci-miRNAs (miR-19a-3p, miR-19b-3p, miR-20a-5p, miR-26b-5p, miR-143-3p, and miR-195-5p) with p-AktSer473 and p-S6K1Thr389, phosphorylated proteins upstream and downstream of mTORC1 that are important in the anabolic/hypertrophic response to resistance exercise (65). Interestingly, several of the miRNAs that increased with exercise are part of the miR-17~92 cluster transcribed from the same primary transcript. These miRNAs overlap in function and target phosphatase and tension homolog (PTEN), a positive regulator of Akt-mTOR signaling (114). The lack of responses observed for these ci-miRNAs that may facilitate hypertrophy are consistent with the blunted response to resistance exercise that is characteristic of older individuals. Thus, failure of these ci-miRNAs to become upregulated may partially explain the resistance to anabolic stimuli with age through inadequate regulation of important gene targets. Although causal mechanisms cannot be determined, those results provide some of the strongest evidence that ci-miRNAs are involved in training adaptations by corroborating ci-miRNA and tissue gene expression data using integrative analytic techniques. The authors (65) suggested that these findings support ci-miRNAs as predictive markers of age-associated changes in body composition and metabolic health, and as potential future biomarkers for adaptive responses to resistance exercise. Still, the ci-miRNA response to acute resistance exercise is a largely unexplored area. Several questions stand unanswered, such as the effects of gender, age, exercise intensity/volume, and resistance training status.
Effects of Exercise Training on ci-miRNAs
In addition to the investigations of acute exercise responses discussed above, several studies have examined the effects of different exercise training protocols. Aoi et al. (4) found a similar ci-miRNA response to 4 wk of moderate-intensity cycling training for 3 days/wk, as that found for a 60-min acute bout of cycling. Before training, out of seven muscle-enriched ci-miRNAs, only ci-miR-486 decreased immediately after exercise. Resting levels of ci-miR-486 were then decreased after training, and this change correlated positively with serum insulin level (r = 0.43) (4). miR-486 is known to target PTEN, which is a negative regulator of the PI3k/AKT pathway downstream of insulin signaling. Thus, during acute exercise, ci-miR-486 may be taken up by the muscle, where it stimulates glucose uptake by suppressing PTEN. More muscle uptake of miR-486 during sustained endurance exercise may therefore translate to lower exercise capacity, because individuals with higher aerobic capacity will rely more heavily on lipid utilization (47). Furthermore, a reduction in ci-miR-486 at rest may reflect higher concentration in skeletal muscle, reduced PTEN, and enhanced insulin signaling as an adaptation to endurance training. Interestingly, in a cohort including inactive, young and older men, fat mass and blood glucose concentration explained 52% of the variance in serum miR-486 level (r = 0.72) (65), providing evidence that ci-miR-486 may be involved in pathways underlying metabolic health. These are attractive hypotheses that merit more extensive investigations.
Baggish et al. (7) compared the acute exercise response before and after 90 days of rowing training. The training stimuli consisted of daily sessions at low intensities but long duration (1 to 3 h). First, they showed that resting levels of ci-miRs -146a, -222, -21, -221, and -20a were increased following training. ci-miRs -146a and -222 were further increased after acute exercise in the trained state, though there was no increase in ci-miR-21 or ci-miR-221 as was observed with acute exercise before training (7). Conversely, following intense cycling training 5 days/wk for 12 wk, Nielsen et al. (77) found that 11 ci-miRNAs were downregulated, including -21 and -133a, whereas ci-miR-103 and ci-miR-107 were upregulated. They also observed no change in ci-miRs -146a or -221 (77). Only ci-miR-133a was affected by both training and acute exercise, also increasing 3 h after 60 min of cycling performed before training. On the other hand, Clauss et al. (20) found no effect of 10 wk of endurance training on five ci-miRNAs involved in atrial remodeling. Thus, there is no strong consensus on the ci-miRNA response to exercise training in healthy individuals, and more studies are needed. All training protocols used in studies thus far have varied substantially and the crossover of ci-miRNAs studied has been minimal. Furthermore, no study has included healthy women to investigate possible sex differences in training-induced ci-miRNA responses.
All other training studies have been conducted on unhealthy or diseased individuals. Van Craenenbroeck et al. (103) examined ci-miRNA concentrations in patients with CKD following a 12-wk home-based cycling program, both at rest and after a symptom-limited maximal cycling test. There were no changes in resting ci-miRNA concentrations after training. The patients exhibited a decrease in ci-miR-210 immediately after exercise in the trained state, the degree of which correlated with change in V̇o2peak with training (rho = −0.236). However, there was no change in ci-miRs -150 and -146, as was observed with the maximal test before training. As discussed above, such differences indicate that alterations in these ci-miRNAs may facilitate some of the cardiovascular adaptations to training.
In overweight/obese men and women, a 16-wk diet and resistance-training program resulted in an increase in ci-miRs -221-3p and -223-3p (80). This response was not different between high and low responders in weight loss, but ci-miR-140 was higher in the low-responding group than the high-weight-loss group at the end of the intervention. FNDC5 is a target of ci-miR-140 that is processed into the circulating myokine IRISIN, which is secreted into the circulation with exercise and is responsible for stimulating thermogenic processes in adipose tissue (12). The authors proposed a mechanism by which increased ci-miR-140 expression may explain the low-weight-loss response of some individuals to training/diet, through downregulating FNDC5 and attenuating potential IRISIN-induced increases in energy expenditure (80). If this mechanism were elucidated, the use of an antagonist miRNA (antagomir) to miR-140, in conjunction with exercise/dieting might then be a useful method to enhance weight loss in these individuals. Beyond that, whether miRs -221 and -223 respond primarily to resistance training or reduced caloric intake in overweight/obese individuals is unclear, because the individuals in this study received both interventions.
At baseline, older patients with prediabetes exhibited higher ci-miR-192 and -193 levels than either diabetic or healthy controls (81). After a 16-wk aerobic and resistance training program, individuals with prediabetes had reduced ci-miR-192 and -193b compared with levels observed in healthy controls, who did not exhibit differences with training. These results were also replicated in glucose-intolerant mice that underwent exercise training and caloric restriction, indicating that ci-miRs -192 and -193b play some role in the shift from healthy to unhealthy glucose metabolism, and vice versa, with exercise training. The reductions in ci-miRs -192 and -193b cannot be attributed to the effects of exercise training per se, however, because the prediabetic subjects in this study were given diet recommendations in combination with the training program. Lastly, in a study of patients with impaired glucose tolerance/fasting glucose or diabetes, both groups had higher ci-miR-126 levels after a 6-mo exercise training and diet treatment (62). Unfortunately, the exercise training protocol in this study was not described. In sum, there is vast evidence that ci-miRNA alterations facilitate the progression of diseases. Likewise, the ability of exercise training to prevent and mitigate diseases is well characterized, so it seems likely that ci-miRNAs may mediate such processes. The few studies detailed here lend early support to this notion, although more mechanistic approaches are obviously needed, as well as time-course and longer term, well-controlled training studies.
Correlations of Fitness and Activity Level with ci-miRNAs
Cross-sectional comparisons of individuals based on fitness characteristics were performed in four studies. Significant findings have predominantly been on ci-miRNAs enriched in or specific to the endothelium. As mentioned earlier, Bye et al. (14) compared high- and low-V̇o2max groups containing both men and women. Ci-miRs -210 and -222 were higher in the low- vs. high-V̇o2max group, and ci-miR-21 was higher only in men with low V̇o2max. Negative correlations of ci-miR-210 (r = −0.35) and ci-miR-21 (r = −0.20) to relative V̇o2max were also observed. The investigators recognized that miR-210 is a regulator of the response to hypoxia (48), including moderating hypoxia-induced angiogenesis (29). The extent to which higher levels of ci-miR-210 and ci-miR-21 may play a role in individuals with low V̇o2max has yet been elucidated.
Wardle et al. (109) compared elite, competitive athletes who were endurance-trained with those who were strength-trained, and with nontrained controls. Ci-miRs -21, -221, -222, and -146a were all higher in the endurance-trained group than the strength-trained group, with the control group tending to fall in between. These four ci-miRNAs each correlated with several performance-related fitness variables. Overall, these data indicate that fitness-related ci-miRNAs may be regulated in opposite directions based on aerobic/endurance vs. strength training. This coincides with previous data showing that all or some of these ci-miRNAs increase following acute endurance exercise and training (7, 8), and decrease 3 days after a bout of resistance exercise (92). Nonetheless, the fact that the athletes in the study by Wardle et al. (109) were competitive elite athletes raises the possibility of underlying genetic differences between the groups. Indeed, subjects were chosen from a larger pool of athletes because they had the largest relative V̇o2max or best performance in strength/power tests. Thus, the observed results attributed to training modes may be confounded by genetic predispositions. On the other hand, Denham and Prestes (26) compared nonelite, endurance-trained men and women with inactive, healthy controls using relatively large sample sizes (n = 67 and n = 61, respectively). They assessed resting levels of five skeletal or cardiac muscle-enriched ci-miRNAs in whole blood. Ci-miRs -1, -486, and -494 were higher in endurance athletes than in controls, and both ci-miRs -1 (r = 0.25) and -486 (r = 0.20) correlated positively with V̇o2max. The finding of higher ci-miR-486 in the endurance-trained individuals is seemingly contradictory to the finding of decreased serum miR-486 expression after 4 wk of endurance training that was observed by Aoi and colleagues (4). This may indicate time-dependent changes, because all endurance athletes had been exercising for at least 1 yr, or may simply be a result of sampling type discrepancies.
Finally, Zhou et al. (121) grouped older men and women with metabolic syndrome into quartiles on the basis of habitual physical activity. The group with the lowest metabolic equivalent (MET) hours per week had higher ci-miR-126 and -130a than the most highly active group. Higher levels of ci-miR-126 were associated with an increased risk for metabolic syndrome. The association of higher physical activity with decreased risk for metabolic syndrome was also abolished when adjusted for ci-miR-126 level. Thus, ci-miR-126 may partially mediate the decreased risk of metabolic syndrome due to regular physical activity, though the mechanism is unknown.
Future Directions of Study
In addition to the roles of ci-miRNAs in exercise physiology suggested thus far, we have identified some novel areas for future study that may reveal potential mechanisms of action for ci-miRNAs. Shear stress is well characterized as the primary mechanism responsible for endothelial and vascular adaptations to exercise training (27, 99), because it beneficially modulates endothelial cell phenotypes through mechanotransduction-mediated effects (23). Intriguingly, though, adaptations may not be directly dependent on shear stress, because similar adaptations are also observed in vasculature-perfusing skeletal muscles and organs that are inactive during exercise (79). Because these vessels are not exposed to a significant degree of increased shear stress, this suggests that other systemic mechanisms such as other hemodynamic or circulating factors, may also be in effect (79). The relationship between shear stress and vascular-related ci-miRNAs in vivo and in relation to exercise has not been investigated, although in vitro, miRNA expression and secretion from endothelial cells is determined by the amount of shear stress (45, 110). Furthermore, endothelial cells secrete miRNA-containing microvesicles, which may be incorporated into vascular smooth muscle cells or other endothelial cells (1, 45). Endothelial cells have been shown to secrete exosomes containing miRs -143 and -145 in response to shear stress, which subsequently confer atheroprotective effects by regulating gene expression in smooth muscle cells (45). Lastly, endothelial cells transfer miR-126 to other endothelial cells through microparticles in vivo, which promotes repair of vascular injury by stimulating migration and proliferation (52). The inflammatory protein TNF-α can further modulate the miRNA “cargo” of released microparticles to induce either proatherogenic or antiatherogenic effects in recipient endothelial cells (1). The idea that shear stress and/or other factors experienced by endothelium of active skeletal muscle vasculature during exercise may induce the secretion of ci-miRNAs to then be taken up by endothelial cells of inactive skeletal muscle vasculature, as a means of inducing systemic endothelial adaptations (Fig. 1), has not been explored.
Besides ci-miRNAs derived from the endothelium itself, ci-miRNAs released from circulating cell types may contribute to endothelial adaptations with exercise and training. Platelet-derived microvesicles are increased in response to intense cycling exercise, are related to brachial and femoral artery shear rate, and induce proangiogenic effects when applied to cultured endothelial cells (112). Monocytes secrete microvesicles that contain miRNA, including miR-150, which stimulates and enhances migration in recipient endothelial cells in vitro (120). Proangiogenic exosomes highly enriched in miRs -126 and -130a are secreted by CD34+ stem cells and are also taken up by endothelial cells (90). Shear stress has been shown to alter miRNA expression in similar CD34+ endothelial progenitor cells (18). MiRNAs are also transported to endothelial cells by HDL (97). HDL isolated from patients with chronic heart failure and cultured with endothelial cells reduced endothelial expression of proangiogenic miRs -126, -21, and -222 (86). After 15 wk of aerobic exercise training, this effect of HDL was attenuated, suggesting their composition/cargo was altered. Whether HDL-encapsulated ci-miRNAs were altered was not assessed. If this were the case, it would raise the attractive prospect of isolating HDL from patients with CVD, altering their miRNA content, and reintroducing them as a means to enhance endothelial function and combat disease. Other circulating EVs such as exosomes are also candidates for such a treatment (57).
Although we focused here on endothelial adaptations, the roles of ci-miRNAs likely extend to training adaptations observed in other tissues (e.g., heart, skeletal muscle, adipose tissue, nervous system, etc.), as suggested by some of the studies reviewed above. A key to elucidating these roles will be combining changes in ci-miRNA with corresponding intracellular miRNA and mRNA data. Only one study has recently done this (65), combining ci-miRNA data with previously published gene expression data from skeletal muscle biopsies (87). Currently, these studies remain difficult because the origin(s) and destination(s) of ci-miRNAs are largely unknown, although some miRNAs are highly enriched in distinct tissues. Additionally, the assessment of predicted downstream target mRNA and protein levels, as well as pre-miRNA levels and components of the miRNA biogenesis pathway, will be necessary for a thorough understanding of the ci-miRNA response to exercise. Although investigations of most tissues will require in vitro and animal models, the use of muscle biopsies can shed light on these pathways within skeletal muscle of humans. As a notable example, Russell et al. (88) took biopsies following both a 60-min bout of cycling and 10 days of cycling training. Drosha, Dicer, and Exportin-5, as well as miRs -1, -133a, and -133b were upregulated in skeletal muscle 3 h after acute exercise, whereas miRNAs known to be involved in muscle wasting and disease were downregulated, indicating that miRNAs can be regulated independently of the primary components of biogenesis. They further observed that some of the predicted targets of affected miRNAs exhibited altered mRNA or protein expression and miRNA levels correlated with the levels of their predicted protein targets (88). It would be interesting to supplement this type of study with ci-miRNA data, although the time course of ci-miRNA packaging, release, uptake, and translational regulation within target cells make it a more complicated endeavor. These processes, as well as the origin(s) and destination(s) of specific ci-miRNAs may first need to be characterized.
Currently, there is no standardized protocol for the study of ci-miRNAs. Methods differ in sample type and timing, RNA extraction, normalization, and quantification. As mentioned, the time course for miRNA appearance and clearance from circulation has yet to be characterized. In addition, plasma and serum differ in ci-miRNA content. Either serum or plasma have been shown to exhibit a higher concentration of ci-miRNA, possibly as a result of the coagulation process or remnant cellular components, respectively (37, 67, 98, 108). Previously, the sample type selected for study has apparently been chosen arbitrarily, and the degree to which this accounts for differences between studies is unknown. Hemolysis is another possible confounder of ci-miRNA concentration, because many miRNAs are present in erythrocytes (67). Hemolysis is only occasionally accounted for, though Nielsen et al. (77) have proposed a protocol for hemolyzation quality control. Furthermore, there is currently no known invariant “housekeeping” miRNA in circulation, and methods for normalization vary. The most popular method is to introduce a known amount of a synthetic spike-in miRNA that is not expressed in humans during the RNA isolation process. This miRNA can then be quantified following isolation and used to normalize samples based on efficiency of extraction. Other methods include normalizing based on the mean expression of all miRNAs in an array, to the least variant ci-miRNA found between groups in an array, or to a priori chosen ci-miRNA(s) that usually show low variation between samples (64). A universally accepted, gold standard method of normalization would be useful to increase reproducibility and confidence in the legitimacy of results. Strategies such as using algorithms to choose the most stable ci-miRNA or combining the use of endogenous and exogenous controls have been proposed (64, 96). Lastly, both the extraction kit and qRT-PCR system used may affect accuracy and reproducibility of results (96, 98). Thus, it is important to consider the methods used in studies of ci-miRNAs because they may play a significant role in data interpretation and may provide an explanation for some apparent discrepancies between results.
To date, available review papers summarizing studies of ci-miRNA responses to acute exercise and exercise training have either not been exclusively focused on ci-miRNAs or the effects of exercise, or were published before other more-recent ci-miRNA and exercise studies. For instance, the most recent reviews published in 2015 include 5 (2), 6 (115), 8 (38), and 10 (34) primary articles on exercise/training and ci-miRNAs, whereas we have identified and presented 30 such peer-reviewed papers. A previous review paper (38) highlighted the potential of ci-miRNAs as biomarkers in exercise and training. In addition to that, we have discussed possible mechanisms of action and proposed areas of future study. There is a need for more studies describing ci-miRNA responses to different protocols of exercise and training in diverse populations. Most studies performed to date involved men, but sample sizes were small. Of those that involved women, only one reported sex-based comparisons (14). In light of known ci-miRNA gender differences, more studies of women are needed. Additionally, most studies have examined a small number of ci-miRNAs chosen a priori. This is understandable because of the large number of miRNAs and added cost of high-throughput methods, but it may be limiting our current knowledge. Of those studies that have examined large numbers of ci-miRNAs, some use samples from the same subjects for follow-up qRT-PCR validation, whereas others did not validate their array findings. This step is important due to the increased sensitivity and specificity of qRT-PCR and the fact that results of initial microarray and screening results often fail to be reproduced (17). Findings from these studies should be replicated in other, larger cohorts.
As of now, only one study has investigated the effects of aerobic training in completely sedentary/untrained men (4). More studies are needed that compare young, healthy, sedentary and trained subjects, either cross-sectionally or in longitudinal training studies, to more completely understand the effects of exercise training. Meanwhile, no study has explored the effect of a resistance-training program on ci-miRNAs in healthy individuals. Conversely, studying the effect of training cessation in well-trained individuals may be a useful paradigm to shed light on the roles of ci-miRNAs in adaptations to long-term training. The intensity of exercise and the amount of muscle mass that must be recruited to significantly alter miRNA levels in circulation is also unclear, as well as whether a systemic response to exercise is necessary to upregulate ci-miRNAs, or if localized responses are sufficient. A primary gap in knowledge limiting the progress of the ci-miRNA field as a whole is the unknown cellular origin of ci-miRNAs. Whereas extensive work has been completed on miRNAs enriched in, or specific to, certain cell types, the sources of many ci-miRNAs are speculative or unknown. Likewise, the significance of the distinct modes by which ci-miRNAs travel in circulation (associated with proteins, vesicles, or HDL) is largely unknown, especially in the context of exercise. The different processes of packaging, secretion, and uptake/clearance may be important determinants of ci-miRNA effects.
In conclusion, miRNAs appear vital to the acute exercise response and adaptations to training, although most data are still correlative (30, 54). Future research should move beyond simply describing exercise-induced alterations in ci-miRNA and should instead focus on identifying the mechanisms and pathways by which ci-miRNAs elicit responses to acute exercise and adaptations to exercise training. The biological roles of ci-miRNAs in these processes remain to be elucidated, but once they are understood they may offer future opportunities for developing therapeutic interventions (38, 78, 94).
No conflicts of interest, financial or otherwise, are declared by the authors.
R.M.S., D.D.S, S.M.R., and J.M.H. conceived and designed research; R.M.S. interpreted results of experiments; R.M.S. prepared figures; R.M.S. drafted manuscript; R.M.S., D.D.S, S.M.R., and J.M.H. edited and revised manuscript and figures; R.M.S., D.D.S, S.M.R., and J.M.H. approved final version of manuscript.
- Copyright © 2017 the American Physiological Society