The Effects of Regular Exercise on Circulating Cardiovascular-related MicroRNAs

The purpose of the present study was to examine the effects of regular exercise on the abundance of targeted circulating microRNAs (miRNAs). The present analysis examined 20 previously sedentary adults from the HERITAGE Family Study who completed 20 weeks of endurance exercise training. The expression of 53 miRNAs related to cardiovascular disease were measured in serum collected at baseline and post-training by performing RT-qPCR on the Human Cardiovascular Disease miRNA array (Qiagen, Germany). The effect of regular exercise on circulating miRNAs was assessed using paired t-tests of baseline and post-training expression levels. A false discovery rate threshold of 5% was used to determine significance. Regular exercise resulted in significantly decreased mean serum expression of nine miRNAs (miR-486-5p, let-7b-5p, miR-29c-3p, let-7e-5p, miR-93-5p, miR-7-5p, miR-25-3p, miR-92a-3p, and miR-29b-3p; fold change range: 0.64–83, p = 0.0002–0.01) and increased mean expression of five miRNAs (miR-142-3p, miR-221-3p, miR-126-3p, miR-146a-5p, and miR-27b-3p; fold change range: 1.41–3.60, p = 0.001–0.006). Enrichment analysis found that these 14 miRNAs target genes related to over 345 different biological pathways. These results provide further evidence of the effects of regular exercise on the circulating miRNA profile.

miRNAs is somewhat limited in that many studies have examined fewer than 10 targeted miRNAs, with little overlap of miRNAs between studies 19 . Furthermore, little is known about how exercise affects miRNAs related to CVD. Therefore, the purpose of the current study was to explore the potential effects of regular exercise on a relatively large panel of miRNAs known to be associated with CVD. Specifically, the Human Cardiovascular Disease miRNA array (Qiagen, Germany) was examined in 20 individuals from the HERITAGE Family Study. This array includes 84 miRNAs that are differentially regulated during CVD progression and have been associated with CVD in multiple studies involving a variety of CVD models and a variety of miRNA profiling methodologies [20][21][22][23][24][25][26][27][28] . We hypothesized that regular exercise would significantly alter the circulating miRNAs related to CVD.

Discussion
This exploratory study found that regular exercise significantly altered the circulating expression levels of 14 miRNAs related to CVD in 20 previously sedentary, but otherwise healthy adults from the HERITAGE Family Study. The 14 miRNAs altered with exercise potentially contribute to the regulation of thousands of genes and are associated with hundreds of different biological pathways. Thus, the potential downstream biological effects of altered miRNA levels with regular exercise are enormous.
Little is known about the potential mechanism(s) by which exercise could alter circulating miRNA expression. MiRNA expression can certainly be regulated and altered, but the regulation of miRNAs is complex and not fully understood 29 . One potential mechanism could be an augmented degradation of the miRNAs in circulation with exercise. However, miRNAs are transported within vesicles or within protein complexes that prevent degradation 1,2 , thus this seems unlikely. Another potential mechanism could be the promotion of selective uptake of these miRNAs by skeletal muscle with exercise, as several studies have demonstrated the ability of target cells, such as skeletal muscle, to take up circulating miRNAs 1,2,30 . Similarly, it has been proposed that exercise may cause the release of certain miRNAs from skeletal muscle, either through muscle damage or selective release in response to acute or regular exercise 31,32 . Thus, one plausible explanation is that the increased or decreased expression of these 14 miRNAs in circulation with regular exercise may be due in part to increased uptake of miRNAs by skeletal muscle (9 of 14 decreased with regular exercise) and the converse increased release of certain miRNAs by skeletal muscle (5 of 14 increased with regular exercise). Baggish et al 33 . found that regular exercise resulted in elevated levels of circulating miR-222, miR-21, and miR-221. Interestingly, expression levels of these same three miRNAs were also elevated following acute exercise, indicating a potential additive effect of repeated acute bouts, as with training, that serves to increase circulating expression of certain miRNAs 33 .
Previous studies have examined the response to exercise of some of the same miRNAs included in this study. We found that regular exercise significantly decreased the expression of circulating miR-486-5p. Similarly, a study www.nature.com/scientificreports www.nature.com/scientificreports/ of 11 men found a single cycling exercise session and four weeks of regular exercise both significantly decreased circulating levels of miR-486-5p 31 . A 90-day period of rowing training in 10 males increased circulating miR-221-3p and miR-146a-5p expression 33 , the same responses found with cycle ergometer training in the current study. Circulating miR-92a-3p and miR-29b-3p expression levels were both significantly decreased in seven males following 12 weeks of regular exercise on cycle ergometers 34 . Both of these miRNAs also significantly decreased with training in the present analysis. The agreement with the current literature on the effects of regular exercise on these select miRNAs is encouraging and strengthens the existing evidence of a significant effect of exercise on the circulating miRNA profile. However, many of the 14 circulating miRNAs significantly altered with regular exercise in the current study have not been previously examined.    www.nature.com/scientificreports www.nature.com/scientificreports/ Our study benefitted from the use of a miRNA PCR array that included 84 miRNAs known to be associated with CVD. Additionally, the entire assay from RNA extraction to miRNA quantification was done using a fully robotic automated platform (QIAcube workstation) minimizing any potential pipetting errors. The study also benefitted from a fully standardized and supervised exercise program, thus all included subjects received the same relative dose of exercise and fully adhered to the program. Further research is also needed to determine the mechanisms by which exercise alters circulating miRNA expression and the association of changes in the circulating miRNA profile with changes in clinical outcomes. Importantly, the present study must be seen as a small, exploratory study. The same research questions should be investigated in greater detail with a much larger sample size and an untargeted approach to accurately and comprehensively examine the effects of regular exercise on the circulating miRNA profile. This study was also limited by a lack of technical replicates in sample analysis. Finally, further research in the field of circulating miRNAs is needed to create clear standards for data handling. Specifically, the difficulty in differentiating between missing data due to a true lack of expression in the sample and missing data due to various experimental issues is a major limitation in the field. Current conventions suggest ignoring Ct values above detection threshold limits, however these 'missing' data may have physiological relevance as the appearance or disappearance of a miRNA in response to an intervention is potentially even more meaningful than changes in expression. Similarly, further research is needed to determine an ideal normalization strategy for circulating miRNAs, as different normalization methodologies may lead to different conclusions and impact reproducibility and comparability of studies.
In summary, regular exercise may alter the circulating miRNA profile of select CVD related miRNAs and these alterations may have physiological relevance given the number of genes regulated by and diverse roles of these miRNAs. The current study provides further support for the effects of regular exercise on circulating miRNAs miR-486-5p, miR-146a-5p, miR-92a-3p, miR-29b-3p, and miR-221-3p and new evidence for other miRNAs. The existing evidence suggests that circulating miRNAs should be thoroughly investigated as biomarkers of regular exercise response. Further research is needed on the effects of exercise on the global circulating miRNA profile to potentially identify novel miRNAs associated with exercise and confirm the existing findings.

Materials and Methods
The HERITAGE Family Study was designed to examine the role of genetic factors on cardiometabolic responses to endurance regular exercise. The HERITAGE cohort is composed of 481 whites (232 men and 249 women) from 99 families and 250 blacks (88 men and 162 women) from 105 family units that completed a 20-week endurance exercise program at one of four clinical centers (Indiana, Minnesota, Québec, Texas). For the present analysis, 20 non-related white subjects (10 males, 10 females) from the Québec center that completed the 20-week exercise program were selected as part of an ongoing ancillary study. The study design, training protocol, and measurements have been described in detail elsewhere 35 . Briefly, the training program consisted of three weekly exercise sessions on cycle ergometers starting at the heart rate associated with 55% VO 2 max and progressing to the heart rate associated with 75% baseline VO 2 max for the final six weeks of the study. The study protocol was approved by the Institutional Review Boards at each of the five participating centers of the HERITAGE Family Study consortium (Indiana University, Laval University, University of Minnesota, Texas A&M University, and Washington University at St. Louis). Written informed consent was obtained from each study participant. All research was performed in accordance with the Declaration of Helsinki.
Exercise tests. Sub-maximal and maximal exercise tests on a stationary cycle ergometer were administered prior to training and additional tests were given following the completion of training 35 . Maximal cycle ergometer tests were performed twice on separate days at least 48 hours apart. VO 2 max was determined by the average of the two tests if the values were within 5% of each other, or the higher of the two values if the measures differed by more than 5% 36 .
Blood collection. Blood samples were obtained in the morning after a 12 hour fast, both at baseline and again 24 hours after the last training session. Serum was collected using standard methods 36 .

RNA extraction.
Total RNA was extracted on the QIAcube workstation from baseline and post-training whole serum samples using miRNeasy Serum/Plasma Advanced Kits (Qiagen, Hilden, Germany). Samples were spiked with a known amount of synthetic C. elegans miR-39-3p (cel-miR-39-3p).  Table 3. Select pathways associated with the 14 microRNAs significantly altered in response to regular exercise and the number of genes per pathway targeted by these microRNAs. Pathways were identified from the exRNA atlas database (http://exrna-atlas.org).

Number of miRNAs Involved
www.nature.com/scientificreports www.nature.com/scientificreports/ MiRNA quantification. MiRNA quantification was performed using reverse transcription real-time polymerase chain reaction (RT-qPCR). Briefly, cDNA was transcribed from the extracted RNA and then aliquoted into the Human Cardiovascular Disease miScript miRNA PCR Array (Qiagen, Hilden, Germany). This array contains primers for 84 miRNA sequences, an individual miRNA primer sequence per well, identified as exhibiting altered expression during cardiovascular disease. The selected miRNAs are listed in Fig. 3. All cDNA steps and PCR setup were performed by the QIAgility instrument (Qiagen, Hilden, Germany) using an automated pipetting protocol. RT-qPCR was performed on the miRNA PCR array in the Rotor-Disc 100 format by the Rotor-Gene Q real-time PCR cycler (Qiagen, Hilden, Germany). Rotor-Gene PCR cycling was performed according to the manufacturer's suggested protocol and conditions. Cycle threshold (Ct) represents the cycle number at which there is an exponential increase in miRNA fluorescence. Individual miRNAs were determined to be detected when Ct values were lower than 33, while miRNAs with a Ct value ≥ 33 were considered not detected. Only miRNAs detected in at least 50% of subjects were retained for analysis, resulting in 53 total miRNAs analyzed.

MiRNA validation.
To validate the miRNA quantification methods, samples from six participants were randomly selected from the study inventory for analysis. Total RNA was isolated and spiked with a known amount of synthetic cel-miR-39-3p and miRNA analyzed as described above. Analysis demonstrated that cel-miR-39-3p exhibited excellent linearity between the target input and measured values with regression coefficients approaching 1 and the qPCR-based extraction efficiency (Efficiency = 10 (−1/slope) −1) was approximately 80% (79.08% ± 11.26%). Additionally, as a positive control for RT-qPCR performed with the Human Cardiovascular Disease miScript miRNA PCR Array, three wells of the array contained cel-miR-39-3p. Using this internal control, the Ct coefficient of variation (CV) values were determined to be 1.14% ± 0.49% intraplate and 0.9% ± 0.62% interplate. Using these automated methods and repeating this miRNA array in triplicate for these six samples, the overall CV for each miRNA was less than 20%.
MiRNA normalization and fold change calculation. Normalized miRNA expression at baseline and post-training are represented by ΔCt. ΔCt values were calculated by subtracting the global geometric mean signal of miRNAs that were commonly expressed in the present study from individual miRNA Ct values as previously described 37,38 . Commonly expressed miRNAs were identified as those with Ct values < 33 in all samples, utilizing the GeneGlobe Data Analysis Center PCR software (Qiagen). ΔΔCt was then calculated by subtracting baseline ΔCt values from post-training ΔCt values. Fold change was calculated as 2 −∆∆Ct .
Statistical analysis. The effects of regular exercise on circulating miRNAs were assessed using paired t-tests of baseline and post-training ΔCt levels. A false discovery rate (FDR) threshold of 5% using the Benjamini-Hochberg procedure was implemented to define statistical significance 39 . All analyses were performed using SAS 9.4 (Cary, NC).
MiRNA target prediction and pathway analysis. Experimentally validated miRNA targets were identified using miRNet (https://www.mirnet.ca), which utilizes data from well-annotated databases: miRTarBase v7.0, TarBase V7.0, and miRecords 40,41 . The connectivity of the 14 miRNAs and their target genes was visualized using miRNet with a minimum network adjustment to simplify the network visualization. Pathway analysis was performed using the extracellular RNA Atlas (http://exrna-atlas.org) 42 . All 14 miRNAs significantly altered with regular exercise were entered into the pathway analysis.