MicroRNAs (miRNAs) are short (~21 nucleotide-long) endogenous non-coding RNAs that mediate mRNA degradation and translational repression (Filipowicz et al., 2008; Huntzinger and Izaurralde, 2011). In humans, miRNAs have been established as vital regulator of almost every physiological process including development, aging and disease (Alvarez-Garcia and Miska, 2005; Sayed and Abdellatif, 2011; Jung and Suh, 2014). Following the identification of muscle-specific miRNAs such as miR-1 and miR-133 and their role in skeletal muscle development, plasticity, and regeneration (McCarthy and Esser, 2007; Simionescu-Bankston and Kumar, 2016), miRNAs have only recently been suggested as markers for individual exercise response (Zacharewicz et al., 2013; Flowers et al., 2015; Polakovičová et al., 2016). In addition, the discovery of inducible circulating miRNAs in human plasma in combination with high-sensitive detection methods have led to the concept of novel minimally invasive approaches to monitor the physiological response to different exercise regimes. Circulating plasma miRNAs are commonly preserved by association with either RNA-binding proteins or small membranous vesicles [i.e., exosomes, microvesicles (MVs)] (Arroyo et al., 2011). In this conformation, miRNAs are involved in inter-cell communication by miRNA-rich MVs shed from the plasma membrane into the extracellular environment to regulate targets in recipient cells (Zernecke et al., 2009). Significant amounts of MVs are commonly released from endothelial cells (ECs) into the plasma and thus ECs are an abundant source of miRNA-rich MVs (Deregibus et al., 2007). One of the most frequent and EC-specific miRNAs is miR-126 which is processed from intron 7 of the EGF-like domain-containing protein 7 (EGFL7) gene (Wang et al., 2008). The precursor miR-126 molecule gives rise to two mature strands, miR-126-3p and miR-126-5p (Wang et al., 2008; Zernecke et al., 2009). It has been shown that endothelial microparticles contain and transfer miR-126 as a highly bioactive molecule with considerable activity to promote EC migration and proliferation (Wang et al., 2008; Zernecke et al., 2009). As such, miR-126 mimics have already been successfully used to limit atherosclerosis and increase plaque stability in atherosclerotic mouse models (Zernecke et al., 2009), which mark miR-126 as a functional biomarker for anti-atherosclerotic training interventions in humans.

It was initially assumed that only the miRNA-3p strand with weaker 5′ thermostability was actively involved in the regulation of cellular processes, while the miRNA-5p strand would be degraded without further usage (Poissonnier et al., 2014). However, a recent study by Poissonnier and coworkers provided evidence that endothelial miR-126-5p is functional and regulates leucocyte adhesion and transmigration (Poissonnier et al., 2014). In addition, the group of Weber and colleagues reported that miR-126-5p promotes endothelial proliferation and also limits atherosclerosis independent of the -3p molecule (Schober et al., 2014). With respect to these results, which suggest a strong and specific atheroprotective capacity of miR-126, the current study focused on the effect of physical exercise on circulating miR-126 levels. It is well-known that physical activity has widespread health benefits and is an important factor in primary and secondary prevention of cardiovascular disease (Myers et al., 2002; Haskell et al., 2007). High-intensity interval training (HIIT) has been identified to be efficient for increasing health-related fitness in general (Burgomaster et al., 2008; Costigan et al., 2015; Milanović et al., 2015) and in lifestyle-induced chronic diseases such as coronary artery disease, heart failure, diabetes mellitus, hypertension, obesity, and metabolic syndrome (Weston et al., 2014). However, exercise-induced miRNA-dependent processes triggering these protective effects remain only partly understood. Therefore, the aim of this study was to explore the dose-response relationship of HIIT on miR-126-3p and miR-126-5p plasma levels in humans. We hypothesized that progressive HIIT (proHIIT) would result in increased miRNA-126 levels compared to HIIT performed at constant intensity.

In this 4-week interventional study, we analyzed the dose-response relationship of HIIT on circulating anti-atherosclerotic miR-126-3p/-5p levels in young healthy moderately trained females and males. Our main findings are: (1) A single low- or high-intensity running exercise bout at baseline did not elevate miR-126 levels, instead miR-126 levels were slightly reduced; (2) proHIIT led to increased miR-126-3p and -5p exercise levels while no elevated miR-126 levels were observed in the HIIT group; (3) the LIT group also presented increased exercise miR-126-3p levels after the intervention; and (4) both mature miR-126 forms, -3p and -5p can be elevated by regular running exercise.

While a number of studies investigated the effects of acute or prolonged exercise on miRNA abundance in tissue or plasma, only few studies have addressed changes in miRNA levels in response to specific training interventions (for review see Flowers et al., 2015; Polakovičová et al., 2016). Among other miRNAs regulated in response to physical activity, miR-126 is of particular interest as it exerts beneficial effects on vascular integrity, angiogenesis, and atherosclerotic plaque stability (Wang et al., 2008; Zernecke et al., 2009). So far, some studies focused on the effect of prolonged physical activity on miR-126 levels in rodents. The group of Oliveira reported an intensity-dependent increase of cardiac miR-126 levels after a 10-week swimming protocol imposed on female Wistar rats (Da Silva et al., 2012). The authors also documented increased capillary/fiber ratio in the heart, suggesting a relationship between the induced angiogenesis and miRNA-126 expression. By contrast, they did not observe significant changes of skeletal muscle miR-126 in male Wistar rats after the 10-week swimming protocol (Fernandes et al., 2012). However, they detected normalization of reduced skeletal muscle miRNA-126 levels in spontaneously hypertensive rats (SHR) in response to the protocol, which was paralleled by revascularization and a reduction of blood pressure (Fernandes et al., 2012). In male ApoE null C57BL/6 J mice, miR-126 levels were reported to be increased in vascular tissue after a 12-week treadmill protocol (Wu et al., 2014). In humans, Uhlemann et al. (2014) reported plasma concentrations of miR-126 after different single exercise bouts. The authors analyzed miR-126 levels after cycling in healthy individuals using a symptom-limited test and in well-trained men after cycling for 4 h at 70% of the individual anaerobic threshold as well as in trained runners after a marathon race and in trained subjects after a single resistance training. They found increased miR-126 levels in all conditions except for the resistance training, an observation in line with another report on miR-126 upregulation during prolonged aerobic exercise (Baggish et al., 2014). The group of Hilberg analyzed competitive adolescent male cyclists performing a single high-volume or high-intensity interval session reporting that the single acute high-intensity exercise did not increase miR-126 levels (Kilian et al., 2016). Taken together, these data indicate that prolonged aerobic exercise can induce elevated miR-126 levels. So far, an investigation on the effects of short intensive running exercises bouts on miR-126 is missing from the literature and most studies have not differentiated their investigations toward the two mature miR-126 forms, miR-126-3p and miR-126-5p, most likely because the anti-atherosclerotic potential of the -5p form has only recently been revealed. The group of Weber and colleagues reported that miR-126-5p affects different cellular target molecules compared to miR-126-3p and suppresses the notch1 inhibitor delta-like 1 homolog (Dlk1) inducing atheroprotective effects by promoting endothelial proliferation (Schober et al., 2014). Thus, it seems conceivable that both mature miR-126 forms act synergistically in endothelial gene expression regulation exerting cardioprotective effects.

Our data indicate that a single exercise bout in moderately trained individuals is not sufficient to induce elevated levels of miR-126 as a direct response. Instead, at the first exercise session, miR-126-3p and -5p levels were slightly reduced immediately after the exercise in all groups. This observation is partly in line with a report on moderately trained men by Radom-Aizik et al. (2010) who also observed a subsequent reduction of miR-126 levels in response to 30 min of cycle ergometry. In our study, we observed a distinct difference between the HIIT and proHIIT group after the intervention. After 4-weeks of training, the proHIIT group presented increased exercise miR-126 levels during and directly after the exercise while the HIIT group still presented slightly lowered miR-126-3p levels. Notably, exercise intensity at follow-up was not different to baseline or in comparison to the HIIT group. Since we observed a significant difference in miR-126 levels between the proHIIT and HIIT group, a dose-response effect of HIIT on miR-126 levels can be suggested.

A comparison of resting miR-126 levels between the two HIIT groups did not suggested that any of the training programs induced a permanent increase of miR-126-3p levels. Notably, when comparing the resting levels of the two HIIT groups after the training intervention we noticed a significant difference of miR-126-3p resting levels before the intervention in that the resting levels of the designated proHIIT group were significantly lower. We investigated different parameters that could explain this difference (including baseline exercise capacity, sex, age, height, activity level, medication) but detected no obvious reason for the different levels. However, it does not seem conceivable that this might have affected the diverse training response detected between HIIT and proHIIT participants as the stimulus in the proHIIT group appeared strong enough to induce higher miR-126 levels despite the lower resting levels.

The physiological model underlying exercise induced mir-126 levels might be characterized by our two main findings. mir-126 levels were not increased by the stimulus of a single exercise bout at baseline but increased quickly in only 4 min of high-intensity running after the training intervention. This rather rapid increase points to a release or secretion mechanism of miR-126 as de novo expression would most likely require more time (Baggish et al., 2011). With respect to the observed high reactivity it is noteworthy that the process of miRNA release into the blood stream especially in response to physical exercise is largely unknown (Chen et al., 2012; Makarova et al., 2014). However, it has been suggested that miRNA-releasing cells possess a sorting mechanism that guides specific miRNAs to enter exosomes resulting in a concentration of selected miRNAs (Zhang et al., 2015). It also seems conceivable that regular exercise training could results in modified (i.e., elevated) basal miRNA expression (Makarova et al., 2014) and thus an increased pool of concentrated miRNAs to be secreted. In particular, the molecular mechanisms involved in mammalian miR-126 expression regulation are still under debate. Pre-miR-126 is located to intron 7 of the EGFL7 gene and promoter studies have revealed different transcription factors involved in its expression regulation including ETS1/2 (Harris et al., 2010). High shear stress has been discussed to be an activator of miR-126 expression as miR-126 is inducible by flow-sensitive Krüppel-like Factor 2 (KLF2) in zebrafish (Hergenreider et al., 2012). However, no flow or KLF2-dependent regulation of miR-126 has been detected in human ECs (Hergenreider et al., 2012). Notably, a connection between the lactate anion and miR-126 levels may be suggested based on the observation that hypoxia targets transcription factor ETS1 (Oikawa et al., 2001), while blockade of the endothelial lactate transporter MCT1 led to inhibition of hypoxia-inducible factor-dependent angiogenesis (Sonveaux et al., 2012). Preliminary in vitro data from our laboratories suggest that low lactate levels of ~1.5 mM have the potential to induce miR-126 expression, while lactate concentrations of 3–4 mM seem to have an inhibitory effect. Notably, blood lactate levels of our LIT group were in the potentially activating range while lactate levels in the HIIT group were quickly exceeding this range. High lactate levels were also observed in the proHIIT group after exercise, however the slope of lactate increase in this group was significantly lower after the training intervention at identical exercise intensities compared to the HIIT group. Future studies will be needed to address the potential mechanism underlying miR-126 expression regulation.

Different pathophysiological conditions have been associated with decreased miR-126 plasma levels including type 2 diabetes mellitus (Zampetaki et al., 2010). Vice versa, increased levels of miR-126 have been associated with a lower rate of cardiovascular events in patients with stable coronary artery disease (Jansen et al., 2014). HIIT, in particular, has been suggested to improve skeletal muscle insulin sensitivity (Eskelinen et al., 2015) and to increase health-related fitness in general (Burgomaster et al., 2008; Costigan et al., 2015; Milanović et al., 2015). While HIIT has also been proposed for patients with lifestyle-induced chronic diseases such as coronary artery disease, heart failure, diabetes mellitus, hypertension, obesity, and metabolic syndrome (Weston et al., 2014), it may be even detrimental for cardiac insulin sensitivity and blood flow capacity as shown for healthy but untrained middle-aged men (Eskelinen et al., 2016). It will therefore be of interest to investigate the benefits and limitations of different HIIT forms (running or cycling, progressive vs. no increment) in the general population and different patient groups in future studies. If cardioprotective effects of HIIT in healthy subjects (primary protection) and patients with different cardiovascular disease manifestations (secondary protection) can be substantiated, miR-126 could be a useful marker to control and optimize individual training interventions.

We conclude that LIT and proHIIT may be performed to increase plasma levels of miR-126 in healthy females and males. Standard HIIT performed without progression may be less effective to increase miR-126 levels. The value of miR-126 as a marker to optimize individual cardioprotective training will be the scope of future studies.

Our results may not be directly translated to other groups or populations as our results were detected in a population of young healthy female and male Caucasians. The presented results are based on the determination of circulating miRNAs from blood. As miR-126 in particular can be highly concentrated in MVs with the potential to shuttle into target cells with high efficiency, future studies are needed to investigate the effect of physical exercise on miRNA concentrations in MVs. Even if the applied novel method for miRNA determination in the field allows for an efficient screening of a larger number of individuals, the sample size of our three training groups might be some limitation and future studies involving larger groups and longer observation periods post-exercise may generate additional insight. Our participants were free in their daily diet. Although training was not controlled, our study involved highly motivated university students with high training compliance (>90%). However, we cannot completely exclude effects of extra non-prescribed training. In addition, the extent to which participants complied with the intensity of the intervention was not measured and therefore quantification of the exercise dose was not performed.

BS designed and coordinated the study, recruited and tested participants, analyzed the data, and drafted the manuscript. KS and JN performed sample preparation and miRNA measurements and analyzed data. LT tested participants and analyzed training and testing data. AK coordinated the study, recruited, and tested participants. ML participated in data analysis. MK, SB, and EB helped finalizing the manuscript. All authors read and approved the final version of the manuscript.