Distinctive exercise-induced inflammatory response and exerkine induction in skeletal muscle of people with type 2 diabetes

Abstract

Mechanistic insights into the molecular events by which exercise enhances the skeletal muscle phenotype are lacking, particularly in the context of type 2 diabetes. Here, we unravel a fundamental role for exercise-responsive cytokines (exerkines) on skeletal muscle development and growth in individuals with normal glucose tolerance or type 2 diabetes. Acute exercise triggered an inflammatory response in skeletal muscle, concomitant with an infiltration of immune cells. These exercise effects were potentiated in type 2 diabetes. In response to contraction or hypoxia, cytokines were mainly produced by endothelial cells and macrophages. The chemokine CXCL12 was induced by hypoxia in endothelial cells, as well as by conditioned medium from contracted myotubes in macrophages. We found that CXCL12 was associated with skeletal muscle remodeling after exercise and differentiation of cultured muscle. Collectively, acute aerobic exercise mounts a noncanonical inflammatory response, with an atypical production of exerkines, which is potentiated in type 2 diabetes.

Fig. 1.. Exercise triggers waves of transcription associated with inflammation and metabolic processes.

(A) Gene expression changes in skeletal muscle after exercise were collected from the MetaMEx database and processed as described in Materials and Methods. Heatmap of the logFC of genes regulated at each time point. The top 100 genes ranked on FDR are presented. (B and C) Peak time for each significant gene was calculated and the log(fold change) of genes with the same peak time averaged. Orange, average of genes peaking 0 to 1 hour after exercise; blue, average of genes peaking 2 to 3 hours after exercise; green, average of genes peaking 4 to 6 hours after exercise; pink, average of genes peaking 24 hours after exercise; and red, average of genes peaking 48 hours after exercise. (D) Gene set enrichment analysis was performed on logFC-ranked genes at each time point. The top 3 gene ontology terms based on FDR for each time point are presented.

Fig. 2.. Enhanced inflammatory response to exercise in skeletal muscle from men with type 2 diabetes.

(A) Heatmap of the logFC of genes significantly regulated by exercise in either normal glucose tolerance (NGT; n = 17) or type 2 diabetes (T2D; n = 20). (B) Number of genes regulated by exercise (FDR < 0.01) and their intersection. Orange, genes annotated with inflammatory processes in gene ontology. (C and D) Number of genes associated with the gene ontology pathways “chemotaxis” and “inflammatory response” increased after exercise (FDR < 0.01). (E) Exercise-responsive gene ontology biological processes calculated with an overrepresentation test. (F and G) Gene set enrichment analysis performed on genes ranked based on log(fold change) interaction effect exercise*T2D at the recovery time point.

Fig. 3.. Elevated exercise-induced infiltration of skeletal muscle by immune cells in individuals with type 2 diabetes.

(A) Overrepresentation analysis using Fisher’s exact test on pro- and anti-inflammatory genes signatures. (B) Signaling pathway impact analysis (SPIA). (C) Representative images of skeletal muscle labeled with CD11b. (D) Quantification of the number of immune cells per myofiber performed with ImageJ. Data are means ± SE and individual data points, n = 3 to 5. Two-way ANOVA (exercise and type 2 diabetes) and pairwise t tests. (E) Macrophage markers and inflammatory pathways analyzed by Western blotting. Representative blots. (F and G) Western blot quantifications. Data are means ± SE and individual data points, n = 8 to 10. Two-way ANOVA (exercise and type 2 diabetes) and pairwise t tests. a.u., arbitrary units.

Fig. 4.. Cytokine cross-talk in skeletal muscle tissue.

(A) Differential response in skeletal muscle from men with T2D versus NGT for exercise-responsive cytokines (FDR < 0.05). Area under the curve of mRNA to estimate the induction of genes compared with their relative baseline. Individual data for each cytokine in all conditions are available in fig. S2A. (B) Cytokines regulated by either EPS in primary human myotubes (green), hypoxia in macrophages (orange), or hypoxia in endothelial cells (blue). Cytokines induced in human THP1 macrophages in response to conditioned media from electrical pulse–stimulated myotubes (red). No cytokines were induced in human primary myotubes exposed to hypoxia (black). Several cytokines were not induced under any condition (side list). Individual data for each cytokine in all conditions are available in figs. S2 and S3. (C) Publicly available RNA sequencing data from bulk skeletal muscle tissue and peripheral blood leukocyte (PBL), as well as primary myotube and monocyte-derived macrophages differentiated in vitro. (D to G) CXCL12/SDF-1 measurement in plasma and skeletal muscle tissue lysate of men with T2D (n = 6) versus healthy individuals (n = 7). Alpha and beta isoforms of CXCL12 were quantified using enzyme-linked immunosorbent assay (ELISA). Dotted lines represent the detection threshold of the assays. Data are mean ± SE and individual datapoints, n = 6 to 7, two-way ANOVA (exercise and T2D) and pairwise t tests.

Fig. 5.. CXCL12 is induced by hypoxia in skeletal muscle tissue but not in primary myotube cultures.

(A) Publicly available data (GSE81286) for cytokine mRNA expression in skeletal muscle from mice exposed to hypoxia (8% oxygen for up to 6 hours). Data are means ± SE. (B) Publicly available data (GSE95244) for cytokine mRNA expression in skeletal muscle from mice in which HIF1α was activated by an injection of an HIF prolyl hydroxylase inhibitor. Data are area under the curve of the time courses presented in fig. S3C. *P < 0.05 at least one time point after injection of the prolyl hydroxylase inhibitor. (C) Primary human skeletal muscle cells treated for 4 hours with the HIF1A stabilizers DMOG (100 μM) or IOX2 (200 μM). Box-and-whisper plots from five independent donors. *P < 0.05 paired t test versus respective vehicles. (D) CXCL12 mRNA expression in human skeletal muscle. Biopsies were collected before and after an acute bout of exercise in men with prediabetes versus metabolically healthy controls. Half of the participants performed exercise under hypoxia. Box-and-whisper plots from seven to eight independent donors. (E) HIF mRNA in skeletal muscle biopsies from men with type 2 diabetes (n = 20) versus healthy individuals (n = 18) from RNA sequencing data presented in Fig. 2. Data are means ± SE. (F) HIF1A protein abundance in skeletal muscle biopsies from men with type 2 diabetes (n = 7) versus healthy individuals (n = 7). Data are means ± SE.

Fig. 6.. CXCL12 alters skeletal muscle cell differentiation.

(A and B) Gene set enrichment analysis of genes correlated with CXCL12 and CXCL16 in the RNA sequencing of skeletal muscle biopsies. The pathway “muscle cell migration” was positively enriched, while “skeletal muscle contraction” was negatively enriched. (C) Inhibition of forskolin-induced cAMP production by CXCL12 or CXCL16 (100 ng/ml) measured in C2C12 myotubes as described in Materials and Methods. Box-and-whisper plots from five independent experiments. *P < 0.05. (D) C2C12 myoblasts were exposed to CXCL12 or CXCL16 (100 ng/ml) every other day for 5 days during the differentiation process as described in Materials and Methods. mRNA expression of genes markers of myoblasts or myotubes was measured by qPCR. Data are means ± SE, n = 5 independent experiments, repeated-measures ANOVA with Tukey’s posttest comparing cytokines to vehicle, *P < 0.05. (E to G) Primary human skeletal muscle cells from individuals with type 2 diabetes (T2D) or NGT were exposed to CXCL12a/SDF-1α or CXCL12b/SDF-1β (100 ng/ml) during the differentiation process as described in Materials and Methods. mRNA expression of gene markers of myoblasts or myotubes were measured by qPCR. Data are means ± SE, n = 4/6 cultures from independent donors (NGT/T2D).