Exercise intensity and training alter the innate immune cell type and chromosomal origins of circulating cell-free DNA in humans

Abstract

Exercising regularly promotes health, but these benefits are complicated by acute inflammation induced by exercise. A potential source of inflammation is cell-free DNA (cfDNA), yet the cellular origins, molecular causes, and immune system interactions of exercise-induced cfDNA are unclear. To study these, 10 healthy individuals were randomized to a 12-wk exercise program of either high-intensity tactical training (HITT) or traditional moderate-intensity training (TRAD). Blood plasma was collected pre- and postexercise at weeks 0 and 12 and after 4 wk of detraining upon program completion. Whole-genome enzymatic methylation sequencing (EM-seq) with cell-type proportion deconvolution was applied to cfDNA obtained from the 50 plasma samples and paired to concentration measurements for 90 circulating cytokines. Acute exercise increased the release of cfDNA from neutrophils, dendritic cells (DCs), and macrophages proportional to exercise intensity. Exercise training reduced cfDNA released in HITT participants but not TRAD and from DCs and macrophages but not neutrophils. For most participants, training lowered mitochondrial cfDNA at rest, even after detraining. Using a sequencing analysis approach we developed, we concluded that rapid ETosis, a process of cell death where cells release DNA extracellular traps, was the likely source of cfDNA, demonstrated by enrichment of nuclear DNA. Further, several cytokines were induced by acute exercise, such as IL-6, IL-10, and IL-16, and training attenuated the induction of only IL-6 and IL-17F. Cytokine levels were not associated with cfDNA induction, suggesting that these cytokines are not the main cause of exercise-induced cfDNA. Overall, exercise intensity and training modulated cfDNA release and cytokine responses, contributing to the anti-inflammatory effects of regular exercise.

Keywords: DNA methylation; ETosis; circulating cell-free DNA; exercise biology; extracellular traps.

Fig. 1.

Acute exercise induces cfDNA in plasma more in HITT than TRAD. (A) Experimental overview of cfDNA collection pre-exercise (pre), postexercise (post), and after 4 wk of detraining (rest) from week 0 (W0) to week 16 (W16). (B) Concentration of cfDNA from samples collected from participants of exercise training programs Traditional (TRAD), and HITT. Lines connect measurements taken from the same individual. Exercise increases cfDNA in blood plasma (P-value < 0.0001) and the increase is intensified with higher intensity exercise (P-value = 0.017) and slightly weaker after 12 wk of exercise (P-value = 0.04). Linear mixed-effects modeling used for statistical testing.

Fig. 2.

Acute exercise alters cfDNA methylation composition. PCA of EM-seq from plasma samples of participants for exercise training programs Traditional (TRAD), and HITT, pre-exercise (pre) and postexercise (post) for week 0 and week 16 (W0 and W16) and after 4 wk of detraining (rest).

Fig. 3.

Exercise intensity influences the magnitude of cfDNA release from neutrophils, DCs, and macrophages. (A) Volcano plot of postexercise (post) compared to pre-exercise (pre). (B) Volcano plot of postexercise HITT program vs. postexercise Traditional program (TRAD). (C–E) Comparing TRAD to HITT for absolute abundance of cfDNA estimated to be derived from neutrophil, DCs, and macrophages using EM-seq and CelFie with linear mixed-effects modeling.

Fig. 4.

Exercise training diminishes the magnitude of cfDNA released from DCs and macrophages, but not neutrophils. (A) Volcano plot of week 12 (W12) postexercise (post) compared to week 0 (W0) postexercise (post). (B–D) Comparing week 12 exercise to week 0 exercise for absolute abundance of cfDNA estimated to be derived from neutrophil, DCs, and macrophages using EM-seq and CelFie with linear mixed-effects modeling.

Fig. 5.

Exercise at weeks 0 and 12 immediately induces proinflammatory, promigratory, and anti-inflammatory cytokines across exercise participants. (A) Volcano plot for average fold change immediately after exercise vs. before exercise for 90 soluble proteins measured via Luminex across weeks 0 (W0) and 12 (W12). (B and C) W12 postexercise (post) compared to W0 postexercise (post) for IL-6 and IL-17F measured via log2 mean fluorescent intensity (MFI). Linear mixed-effects modeling used for statistical testing. Horizontal fine lines connecting dots indicate samples from the same participant. (D) Heatmap (Left) showing pairwise correlation of exercise-induced changes across both W0 and W12 for soluble proteins significant in panel (A). Scatterplots (Right) of exercise-induced changes across participants (each point is one participant) for significant fold change correlations with Benjamini–Hochberg adj. P-value < 0.05. (E) Scatterplot of correlation of fold change in postexercise vs. pre-exercise for week 0 (light red) and week 12 (dark red) in plasma total cfDNA concentration to fold change in plasma IL-6 MFI, Spearman correlation test P-value = 0.984. Inset scatterplot of correlation of plasma total cfDNA concentration to plasma IL-6 MFI, Spearman correlation test P-value = 2.74 × 10⁻⁶.

Fig. 6.

Training reduces mitochondrial cfDNA at rest while acute exercise enriches chromosomal and NuMT sequences in an exercise intensity-associated manner. Circles represent participants in the traditional intensity exercise training program (TRAD), while triangles represent participants in the HITT program. (A) Concentration of mitochondrial cfDNA for pre-exercise (pre) for weeks 0 (W0) and 12 (W12), and rest at week 16 (W16). (B) Nuclear chromosomal enrichment, defined as nuclear DNA: mitochondrial (mito) DNA read coverage for pre-exercise and postexercise and rest for weeks 0, 12, and 16. (C) Fold change in chromosomal enrichment for W0 and W12 for TRAD and HITT. Paired Student t test for average HITT chromosomal enrichment fold change vs. average TRAD, P = 0.0019. (D) Methylated DNA mapping to mitochondrial genome as percent of all reads mapping to the mitochondrial genome for pre-exercise (pre) and postexercise (post), averaged for weeks 0 and 12 (paired Student t test, P = 0.0024). (E) Fold change (post/pre) of NuMT enrichment upon exercise compared to fold change (post/pre) in chromosomal enrichment for exercise programs TRAD and HITT, averaged for weeks 0 and 12. Pearson correlation P = 0.0025, r = 0.84.