ECSS Paris 2023: OP-PN14
INTRODUCTION: Proteins can be fractionated in skeletal muscle into e.g. “myofibrillar” and “sarcoplasmic” fractions, but these are not pure and distinct profiles (1). Alternatively, high-end proteomics has been used to examine responses of individual proteins and pathways to exercise. It can also be used to study protein profiles based on cellular location where their molecular function is known to take place (i.e. myofibrillar, ribosomal, nuclear, enzyme etc. proteins). A study design including resistance training (RT), detraining (DT), and retraining (reRT) (2) enables investigating short- and long-term effects of RT and a potential “proteomic memory” of RT. Therefore, the present study aims to investigate the effects of RT, DT, and reRT on different protein profiles based on their cellular location. METHODS: Thirty untrained participants were divided into training-detraining-retraining (TDR, n = 17) and control (n = 13) groups. TDR underwent 10 weeks of RT, 10 weeks of DT, and 10 weeks of reRT, while the control group remained non-trained for 10 weeks. Biopsies were collected from vastus lateralis in weeks 0, 10, 20 and 30. High-end diaPASEF on the mass spectrometry of the timsTOF Pro 2 platform was used for proteomics analysis. In total, > 3000 unique proteins were quantified. Of these, 50 unique proteins were classified based on the literature as myofibrillar proteins (37 sarcomeric and 13 sarcomere-associated proteins). Other quantified protein profiles were classified using Panther and Gene Ontology Cellular Component databases: mitochondrial (714 proteins), ribosomal (120), and nuclear (992) proteins. Statistics were conducted using ANOVA and Tukey´s post hoc test. Statistical significance was set at p<0.05. RESULTS: The relative abundance of sarcomeric and sarcomere-associated proteins displayed no significant changes between the timepoints, suggesting these protein profiles change at the same rate as muscle proteins on average during RT, DT, and ReRT. Instead, the relative abundance of ribosomal, mitochondrial, and enzyme protein profiles increased during 10 weeks of RT (p<0.05) and compared to controls, and there was a trend for increases in nuclear proteins (p=0.06). Enzymes and mitochondrial proteins returned to baseline after DT, whereas nuclear and ribosomal proteins remained elevated after DT (p<0.05). CONCLUSION: Herein we show different temporal changes in protein profiles after resistance training depending on their classified cellular location. Our results suggest relative and persisting expansion of nuclear and ribosomal protein profiles after resistance training and detraining. This may indicate proteomic memory of resistance training-induced hypertrophy. References 1. Roberts et al., Aging, 2024 2. Halonen et al., Scandinavian journal of medicine and science in sport, 2024.
Read CV Eeli HalonenECSS Paris 2023: OP-PN14
INTRODUCTION: Physical exercise can modify circulating microRNA (c-miRNA) profiles in vein and skeletal muscle in a stimulus-specific manner [1]. Along with the posttranscriptional regulation of miRNAs, this makes them potential functional biomarkers in sports [2]. However, the specific secretory and target tissues during exercise remain unclear. Therefore, the aim of this study was to assess the c-miRNA profile in femoral artery and vein influenced by exercise and its potential physiological and performance implications. This approach enables to isolate the specific influence of the muscle environment on the c-miRNA profile in vivo. METHODS: Nine healthy trained men performed a maximal aerobic test where VO2max and Wpeak were measured. Blood samples were obtained from femoral artery and vein immediately before and after the test. Plasma obtention was followed by miRNA extraction and massive sequencing. Only those c-miRNAs with at least 30 RPM in 3 or more subjects and in one of the two time points or vessels were considered. Principal Component Analysis followed by Principal Component Regression was used to relate c-miRNAs with performance parameters. Functional analysis was performed by a multiscale in silico analysis of the most enriched protein-protein interaction networks. RESULTS: In basal state, 56% (23 miRNAs) of the c-miRNAs were present in both vessels. Of them, 83% showed a trend to a lower expression level in vein. Consistently, a higher number of c-miRNAs were unique in artery vs vein (16 vs 2), highlighting a possible role of the muscle environment as a target. After the test, 69% (25 miRNAs) of the c-miRNAs were shared between vessels, with a tendency to higher levels in vein (63%). In this line, exclusive vein c-miRNAs were predominant vs artery (1 vs 10), showing a shift of the muscle environment towards increased secretion after exercise. A significant increase in miR-26b-5p (p-value=0.018, FC=9.07) and miR-21-5p (p-value=0.044, FC=4.04) was found in vein before and after the test, supporting the secretion role. Arterial miR-22-3p, miR-20a-5p, miR-107, miR-106b-5p, miR-103a-3p and let-7g-5p were found to be predictors of VO2max, while post-test arterial levels of miR-486-5p, miR-451a, miR-93-5p, miR-20a-5p, miR-144-3p and miR-15a-5p were related to VO2max. Higher uptake of miR-106b-5p by the muscle environment was correlated with VO2max (p-value=0.021, r=-0.767), highlighting a potential implication of c-miRNA regulation on performance. In silico analysis showed that c-miRNAs correlated with VO2max have a high association at the protein level involved in phosphorylation-mediated processes. CONCLUSION: The specific c-miRNA profiles described in vein and artery suggest a role of secretion and uptake from the muscle environment, influenced by exercise. The association with performance parameters and phosphorylation functions highlights the importance of c-miRNAs in regulating the response to exercise. 1. Margolis et al, J Physiol, 2022 2. Fernandez-Sanjurjo et al, ESSR, 2018
Read CV David Fernández ViveroECSS Paris 2023: OP-PN14
INTRODUCTION: Precision training management requires continuous monitoring of athletes physiological states to optimize performance and prevent overtraining. Traditional blood tests are limited by their invasiveness and complexity. Urinary metabolomics has emerged as a non-invasive and information-rich alternative, offering significant potential in sports physiology. This study uses UHPLC-HRMS-based urinary metabolomics to develop multivariate regression models for predicting key biochemical markers, including hemoglobin (Hb), testosterone (T), cortisol (C), creatine kinase (CK), and blood urea nitrogen (BUN). These models provide a novel tool for non-invasive monitoring and personalized training strategies. METHODS: Urine samples from 90 National Level athletes were analyzed using high-throughput targeted metabolomics, detecting 416 metabolites per sample. Stepwise Multiple Linear Regression (SMLR) was employed to construct predictive models for the five biochemical markers. The models were evaluated using Adjusted R², Mean Absolute Error (MAE), and Mean Relative Error (MRE). Predictive performance was validated through cross-validation and independent validation set testing. Comparative analyses with machine learning methods, including Random Forest (RF) and Neural Networks (NN), confirmed the superior stability, accuracy, and interpretability of SMLR. RESULTS: The models demonstrated high predictive accuracy, with Adjusted R² values of 0.855 (Hb), 0.974 (T), 0.915 (C), 0.896 (CK), and 0.970 (BUN). Cross-validation results showed mean relative errors (MRE) of 14.25% (Hb), 21.16% (T), 6.12% (C), 20.8% (CK), and 27.27% (BUN), indicating robust performance across diverse markers. For example, the hemoglobin prediction model is represented as: ŶHb=122.71 − 0.008 × Indole-3-methyl acetate − 0.003 × Glycolithocholic acid 3-Sulfate + 0.034 × Dethiobiotin+0.003 × Nicotinamide − 0.002 × Riboflavin 5-monophosphate + 0.001 × Arginine − 0.004 × Octanoylcarnitine + 0.001 × X4 Pyridoxic acid − 0.052 × N-Acetylphenylalanine − 0.001 × Stachyose − 0.006 × Methylcysteine + 0.000054 × Cystine − 0.006 × Adenosine monophosphate AMP − 0.001 × Phthalic-acid − 0.005 × Sepiapterin + 0-259 × Glycoursodeoxycholic acid GUDCA + 0.00008 × Uracil − 0.033 × Acetylcholine The models effectively captured metabolic patterns associated with athletes physiological states, offering a non-invasive approach to optimize training and recovery. CONCLUSION: This study is the first to establish urinary metabolomics-based predictive models for multiple biochemical markers, providing a non-invasive and efficient tool for athlete monitoring. Despite a higher MRE for BUN, the model offers valuable insights for optimizing recovery strategies. Future research will expand the sample size and explore advanced machine learning techniques to enhance predictive accuracy and applicability.
Read CV Zhang YangECSS Paris 2023: OP-PN14