We predict that RNA level regulation is as diverse and powerful as protein level regulation when considering physiological adaptation. Non-coding RNA molecules, such as miRNAs (microRNAs), have emerged as a powerful mechanism for post-transcriptional regulation of mRNA. In an effort to define the role of miRNA in human skeletal-muscle biology, we have initiated profiling of muscle RNA before and after endurance exercise training. The robust molecular phenotype of muscle is established using unbiased analysis strategies of the raw data, reflecting the statistical power of gene ontology and network analysis. We can thus determine the structural features of the skeletal-muscle transcriptome, identify discrete networks activated by training and utilize bioinformatics predictions to establish the interaction between non-coding RNA modulation and Affymetrix expression profiles.
- endurance exercise
- non-coding RNA
- systems biology
Regular physical training improves insulin-sensitivity and vascular function of skeletal muscle . Lack of cardiorespiratory (aerobic) capacity and low habitual physical activity levels are hallmarks of increased risk for the development of metabolic syndrome and premature death . Although one can modify aerobic capacity through exercise training, humans demonstrate a heterogeneous response in adaptability, with some individuals unable to significantly improve their aerobic capacity at all , while fitness gains may be blunted with aging . Over the past decade, we have utilized genomics technologies (e.g. Affymetrix oligonucleotide arrays) to examine the molecular characteristics of skeletal muscle. We provided an indication that gene expression parallels physiological adaptability in a genome-wide analysis of human skeletal-muscles response to aerobic training [5,6]. The approach of studying gene-by-gene canonical signalling following acute exercise [7–9] has proved inadequate for predicting the regulation of longer-term alterations in skeletal-muscle phenotype. Mutation or knockout of candidate proteins (identified previously from canonical pathway analysis studies), such as AMPK (AMP-activated protein kinase) and CaMKIV (Ca2+/calmodulin-dependent protein kinase IV) signalling, have failed to prevent muscle adaptation following contraction . Thus previously accepted regulators identified by these traditional approaches may not reliably identify mechanisms directly responsible for the chronic adaptation of human skeletal muscle.
Taking a systems biology approach to studying skeletal-muscle physiology
More recently, transcriptomics are used to study acute exercise responses in human skeletal muscle  and they provide an unbiased approach to discover links between acute endurance exercise and long-term changes following endurance exercise training . In the present review, we present a cross-comparison of the biological response to acute intense aerobic cycling for 75 min, as presented by Mahoney et al. , and our coding-genome-wide assessment of 6 weeks of endurance cycle training at 70% of peak aerobic capacity (four times per week) . Direct comparison of microarray data generated with more than one technology platform and from different laboratories is challenging, but possible , and relies on the manner in which raw data are deposited . Ideally, an unbiased analysis is carried out prior to detailed examination of gene lists using e.g. gene network analysis or gene ontologies. This approach has the following benefits: first, microarrays provide high coverage at the expense of both precision and sensitivity, so while individual genes in a pathway may not be detected in a given experiment (false negatives), networks or families of genes (ontologies) will be reliably represented. Determining which families or networks are activated by the physiological intervention is therefore an ideal starting point. Secondly, individual genes (be it at the sequence, RNA or protein level) cannot adequately represent a physiological process; genes act in networks and thus emphasis should be on discovering new processes rather than changes in individual genes. Finally, there is limited evidence that the most regulated transcript represents one with the greatest biological importance and such arguments should probably be avoided.
Mahoney et al.  deposited the raw data from the four healthy male subjects they profiled, before and after a single bout of endurance exercise and presented 73 genes that are differentially expressed [FDR (false discovery rate) <5%] 3 h post-exercise. We used the list of the 73 sequence IDs and mapped 59 of them reliably to official gene symbols in the IPA (Ingenuity Pathway Analysis) database. This allowed us to determine what statistically enriched biological processes and gene networks this gene list represents, also ensuring accurate annotations. While metabolism, oxidant stress and electrolyte transport ‘functions’ were manually identified by the authors , an unbiased analysis identified regulation of metabolism, transcription factor binding and cation binding as being statistically enriched within the data set (FDR <1%), i.e. our analysis overlapped with the authors’ observations. Interestingly, IPA identified only three substantial gene networks in the acute endurance exercise transcriptome. Of the over-represented canonical pathways within these networks, only PPAR (peroxisome-proliferator-activated receptor) and interferon signalling and G1/S cell cycle check point approached significance. The top IPA network is presented in Figure 1(A), where most members of this ‘gene-expression/cell cycle’ network are up-regulated (red), and demonstrates where regulated nodes strongly interconnect with kinase and VEGF (vascular endothelial growth factor) signalling pathways identified from a traditional gene-by-gene study [14,15]. This agreement between the conventional and unbiased analysis approaches strongly validates the use of the unbiased analysis approach when investigating large-scale gene array data sets. This is in turn has benefits, in that investigators’ attention will be quantitatively drawn towards genuinely new observations.
Applying the multiple array comparison strategy
When the acute endurance exercise network, described above, is directly compared with the functional networks derived from the ∼500 genes modulated 24 h following the last training session of 6 weeks of endurance training (n=8 healthy male subjects), little direct connection is observed. Figure 1(B) represents the acute significant pathway showing expression values from our endurance training study . MYO1B (myosin 1B) and PKA (protein kinase A) are up-regulated (red) with training, but are unchanged or down-regulated acutely (blue, Figure 1A). The expression of PDHK4 (pyruvate dehydrogenase kinase 4) varies substantially with time, is down-regulated by training, whereas PDHK4 is up-regulated by acute exercise  and/or up-regulated with time . Thus direct comparison between ‘acute’ and ‘chronic’ transcriptomes does not identify the link between acute exercise responses and the chronic phenotype shift associated with endurance exercise training. To address this issue, one should identify the significant networks that define skeletal-muscle phenotype post-endurance training (Figures 2A–2C) and then establish the canonical regulators. It is worth mentioning that we generated our array data set solely from those that demonstrated evidence of physiological adaptation (increased aerobic capacity). Acute studies, by their nature, cannot prove that acute perturbations lead to physiological adaptation in the subjects studied; thus comparison is still complex. Sixteen interconnected networks were generated using IPA from our exercise training expression changes. The top three significant networks reflected integrin- (Figure 2A), IGF-1 (insulin-like growth factor-1)- (Figure 2B) and VEGF-related signalling networks (Figure 2C), and all significantly affected networks are summarized in Figure 2(D). Re-analysis of our original data confirmed our previous unpublished interpretation that endurance training-mediated remodelling of skeletal-muscle phenotype demonstrates significant biological overlap with cancer biology. Thus cross-comparison of physiological tissue remodelling with tumour-related growth may provide insight into novel cancer targets.
Junk DNA rules
RNA level regulation is probably as diverse, and powerful, as protein level regulation when considering physiological adaptation . Non-coding RNA molecules such as miRNAs (microRNAs) (originating from sequence previously referred to as junk DNA) have emerged as a powerful mechanism for post-transcriptional regulation of mRNA . In the coming year, we will present an updated genome-wide profiling of human muscle tissue, analysed following chronic training and interrogated relative to physiological adaptation and combined with large-scale miRNA profiling (to provide insight into post-transcriptional regulation). We will present the physiologically important gene networks required for successful adaptation to exercise training in humans, as changes will be related to the extent of physiological adaptation. A central network of genes was identified in humans which track with improved aerobic capacity and modulated miRNAs. Finally, we will demonstrate that TGF-β (transforming growth factor-β)/integrin-, TNFα (tumour necrosis factor α)- and IL-6 (interleukin-6)-regulated gene networks are prominent gene expression networks demarcating successful adaptation to aerobic exercise training in humans.
Exercise: A Focus Topic at Life Sciences 2007, held at SECC Glasgow, U.K., 9–12 July 2007. Edited by C. Downes (Dundee, U.K.), P. Greenhaff (Nottingham, U.K.) and P. Taylor (Dundee, U.K.).
Abbreviations: CaMK, Ca2+/calmodulin-dependent protein kinase; FDR, false discovery rate; IGF, insulin-like growth factor; miRNA, microRNA; MYO1B, myosin 1B; PDHK4, pyruvate dehydrogenase kinase 4; PKA, protein kinase A (or cAMP-dependent protein kinase); PPAR, peroxisome-proliferator-activated receptor; TGF-β, transforming growth factor-β; VEGF, vascular endothelial growth factor
- © The Authors Journal compilation © 2007 Biochemical Society