A transcriptional map of the impact of endurance exercise training on skeletal muscle phenotype

Abstract

The molecular pathways that are activated and contribute to physiological remodeling of skeletal muscle in response to endurance exercise have not been fully characterized. We previously reported that ∼800 gene transcripts are regulated following 6 wk of supervised endurance training in young sedentary males, referred to as the training-responsive transcriptome (TRT) (Timmons JA et al. J Appl Physiol 108: 1487-1496, 2010). Here we utilized this database together with data on biological variation in muscle adaptation to aerobic endurance training in both humans and a novel out-bred rodent model to study the potential regulatory molecules that coordinate this complex network of genes. We identified three DNA sequences representing RUNX1, SOX9, and PAX3 transcription factor binding sites as overrepresented in the TRT. In turn, miRNA profiling indicated that several miRNAs targeting RUNX1, SOX9, and PAX3 were downregulated by endurance training. The TRT was then examined by contrasting subjects who demonstrated the least vs. the greatest improvement in aerobic capacity (low vs. high responders), and at least 100 of the 800 TRT genes were differentially regulated, thus suggesting regulation of these genes may be important for improving aerobic capacity. In high responders, proangiogenic and tissue developmental networks emerged as key candidates for coordinating tissue aerobic adaptation. Beyond RNA-level validation there were several DNA variants that associated with maximal aerobic capacity (Vo(₂max)) trainability in the HERITAGE Family Study but these did not pass conservative Bonferroni adjustment. In addition, in a rat model selected across 10 generations for high aerobic training responsiveness, we found that both the TRT and a homologous subset of the human high responder genes were regulated to a greater degree in high responder rodent skeletal muscle. This analysis provides a comprehensive map of the transcriptomic features important for aerobic exercise-induced improvements in maximal oxygen consumption.

Fig. 1.

Differentially expressed microRNAs (miRNAs) in human skeletal muscle following 6 wk of aerobic exercise training. Exiqon locked nucleic acid (LNA) miRNA array data were analyzed using Significance Analysis of Microarrays (SAM) with a false discovery rate (FDR) <15% and fold change (FC) > 1.5. Inset represents miRNA expression validated by qPCR (change from pre: P < 0.01 for all 4 miRNAs). qPCR data are presented as mean percent change from the pretraining expression level ± SE while the array data are plotted as fold change, where <1 equals a reduction in gene expression.

Fig. 2.

Gene Ontology (GO) classes of predicted miRNA gene targets for human miRNAs differentially expressed following 6 wk of aerobic exercise training. The main ontology classes represented were regulation of transcription (∼22% of target genes) and metabolism (∼16% of target genes).

Fig. 3.

Upregulation of a highly significant Ingenuity pathway analysis (IPA) gene network (IPA network score = 50) in response to endurance training in high responders (A), but not in low responders (B). Aerobic training was found to be associated with differential increases (>1.5 FC, dark gray boxes) and decreases (>1.4 FC, light gray boxes) in gene expression mostly in the high responders to aerobic training. Those genes that were upregulated in the low responders were upregulated to a significantly greater extent in high responders (Table 2). The network involves a number of extracellular matrix genes, VEGF signaling and associates with terms such as “cardiovascular development.” This is consistent with the TopGO analysis presented online in Supplementary Fig. S4.

Fig. 4.

Upregulation of a second highly significant IPA gene network (IPA network score = 41) in response to endurance training in high responders (A), but not in low responders (B). Aerobic training was found to be associated with differential increases (>1.5 FC, dark gray boxes) and decreases (>1.4 FC, light gray boxes) in gene expression almost exclusively in the high responders to aerobic training. This network is regulated by NF-kB, TGF-β, IGF2 and insulin, includes a number of major histocompatibility complex (MHC) genes, and is associated with embryonic tissue development according to IPA.

Fig. 5.

Rat genetic model for high and low training response to aerobic exercise training. A: differences in basal level of muscle gene expression in 6th-generation rats that demonstrate a high or a low aerobic training response. Of 26 human “high responder genes” analyzed in soleus muscle of rats selectively bred for high and low adaptation to aerobic exercise training, 13 genes already demonstrated >25% higher expression levels in high responder rats (n = 5, black columns) compared with low responder rats (n = 6, white columns) even in this early stage of the breeding model development. As a whole, the 26 genes were expressed at a higher level in the high responder rats (P < 0.0001). Data are presented as mean arbitrary expression units ± SE. B: the change in treadmill running capacity following 8 wk of aerobic exercise training was significantly different (P < 0.001) for rats bred across 10 generations for low (n = 8) and high (n = 8) response to training. Data are presented as meters gained in maximal aerobic running capacity produced by the training mean ± SE. ***P < 0.001 compared with pretraining; ###P < 0.001 compared with low responder rats.