Jacob L. Barber1, Guoshuai Cai1, Jeremy M. Robbins2, Prashant Rao2, Michael Mi2, Sujoy Ghosh3, Clary Clish4, Dan H. Katz2, Robert E. Gerszten2, Claude Bouchard, FACSM5, Mark A. Sarzynski, FACSM1. 1University of South Carolina, Columbia, SC. 2Beth Israel Deaconess Medical Center, Boston, MA. 3Duke-NUS Medical School, Singapore. 4Broad Institute of Harvard and MIT, Cambridge, MA. 5Pennington Biomedical Research Center, Baton Rouge, LA.

BACKGROUND: Regular exercise is associated with beneficial effects on lipid metabolism, however the molecular mechanisms responsible for these benefits are unclear. METHODS: Circulating proteins (n=4979), metabolites (n=300), and lipids/lipoproteins were measured in 647 Black and White adults from the HERITAGE Family Study at baseline and after 20 weeks of supervised endurance training. The current analysis focused on 7 lipid/lipoprotein traits that significantly changed with training in the overall HERITAGE cohort: HDL- cholesterol (HDL-C), triglycerides (TG), large TG-rich lipoprotein particles (LTRLP), large HDL particles (LHDLP), small LDL particles (SLDLP), and mean TRLP (TRLPz) and LDLP size (LDLPz). The relationship between exercise-induced fold changes in circulating molecules and changes in lipid traits was examined using sparse canonical correlation analysis (R package ‘PMA’), a technique for assessing the joint associations between two sets of variables by creating composite canonical variates. All variables were corrected for age, sex, race, baseline BMI, baseline value, and family membership via linear mixed models. RESULTS: We identified 3 canonical variate pairs of exercise-induced changes in lipid traits and circulating molecules. Molecular variate 1 was positively correlated with changes in TG, LTRLP, SLDLP, and TRLPz (r=0.29-0.57, p<0.0001). Conversely, molecular variate 3 was negatively correlated with changes in TG, LTRLP, SLDLP, and TRLPz (r=-0.30 to -0.39, p<0.0001), and positively correlated with changes in LDLPz (r=0.38, p<0.0001). Molecular variate 2 was negatively correlated with HDL-C (r=-0.46, p<0.0001) and LHDLP (r=-0.22, p<0.0001) changes with training. Molecular loadings on the respective variates were largely distinct. For example, plasma diacylglycerols (DAGs), phosphatidylethanolamines (PEs), and proteins such as WNT5A, QORL1, and APOC-III were positively loaded, while several plasma carnitines were negatively loaded on variate 1. Molecular variate 2 was negatively loaded by proteins such as adiponectin and neuropeptide S and positively loaded by multiple complement system proteins. CONCLUSIONS: We identified groups of circulating molecules whose changes in response to exercise training are associated with changes in the plasma lipid and lipoprotein profile and may provide insights into the mechanisms underlying exercise-induced changes in lipid metabolism.

This document is currently not available here.