Insulin- and exercise-induced phosphoproteomics of human skeletal muscle identify REPS1 as a regulator of muscle glucose uptake

Abstract

Skeletal muscle glucose uptake, essential for metabolic health, is regulated by both insulin and exercise. Using phosphoproteomics, we analyze skeletal muscle from healthy individuals following acute exercise or insulin stimulation, generating a valuable dataset. We identify 71 phosphosites on 55 proteins regulated by both stimuli in the same direction, suggesting a convergence of exercise and insulin signaling pathways. Among these, the vesicle-associated protein, REPS1, is highly phosphorylated at Ser709 in response to both stimuli. We identify p90 ribosomal S6 kinase (RSK) to be a key upstream kinase of REPS1 S709 phosphorylation and that the RSK-REPS1 signaling axis is involved in insulin-stimulated glucose uptake. Insulin-induced REPS1 Ser709 phosphorylation is closely linked to muscle and whole-body insulin sensitivity and is impaired in insulin-resistant mice and humans. These findings highlight REPS1 as a convergence point for insulin and exercise signaling, presenting a potential therapeutic target for treating individuals with insulin resistance.

Keywords: REPS1; RSK; exercise; glucose metabolism; insulin; phosphoproteomics; skeletal muscle signaling.

Graphical abstract

Figure 1

Phosphoproteomic signature of insulin and exercise signaling in human skeletal muscle (A) Study design: eight healthy men underwent, in randomized order, an acute bout of high-intensity cycling exercise (study day 1) and a hyperinsulinemic euglycemic clamp (study day 2). Skeletal muscle biopsies were obtained before and immediately after each intervention. (B and C) Glycogen content (B) and leg glucose balance during clamp (C). (D) Western blot confirmation of insulin and exercise-induced signaling. (E) Phosphoproteomics workflow with 2x TMT 11-plex labeling, phosphopeptide enrichment, and LC-MS/MS analysis. (F and G) Principal-component analysis of insulin and exercise signaling responses. (H–I) Volcano plots of insulin and exercise signaling (x axis = logFC, y axis = −log10 (p value)). A linear model (limma) was used to test for difference in phosphosite abundance between conditions with a false discovery rate set to 5%. Mean +/− standard deviation is represented in (C). N = 8 individuals are presented in (A)–(D). N = 5 individuals are presented in (E)–(I). Two-sided paired t test was used to test for mean differences in (B). ∗∗p < 0.01, ∗∗∗p < 0.001.

Figure 2

Shared and distinct features of insulin and exercise signaling Predicted insulin- and exercise-activated kinases from the kinase library (adjusted p value < 0.05) (A and B). Two-sided Fisher’s exact test of phosphosite functionality based on the PhosphoSitePlus database (C). Heatmap of individual fold-change response of significant phosphosites to exercise and insulin stimulation across the five subjects (D). Selected phosphosites regulated in opposite or the same direction (E). AlphaFold-predicted structure of CLASP2 with highlighted serine residues phosphorylated or dephosphorylated by both insulin and exercise (F). Sequence window of phosphosites upregulated by insulin and exercise (G). A two-sided Fisher’s exact test was used to test for residue overrepresentation. Kinase enrichment analysis of phosphosites upregulated by both insulin and exercise (H). Individual fold changes of p-REPS1 S709 in response to insulin and exercise (I). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Figure 3

The insulin and exercise-responsive protein, REPS1, is a critical regulator of skeletal muscle glucose uptake Highly conserved REPS1 sequence window around the serine 709 site (A). Western blot validation of REPS1 phosphorylation in human skeletal muscle before (basal) and after (insulin) a 2-h hyperinsulinemic euglycemic clamp (B) and at rest and after 10 min high-intensity cycling exercise (C). Pearson’s correlation analysis of steady-state leg glucose uptake and delta (post-pre clamp) REPS1 S709 phosphorylation (D). In vivo³H-2-deoxyglucose uptake during rest or in situ contractions of tibialis anterior (TA) muscle with saline or 15 mU insulin stimulation (E). Western blot analysis of REPS1 phosphorylation during rest or in situ contractions of TA muscle with saline or 15 mU insulin stimulation (F). Pearson’s correlation analysis of insulin-stimulated glucose uptake into TA muscle and REPS1 S709 phosphorylation (G). Representative western blot from in situ contraction experiment (H). Western blot validation of siRNA-mediated knockdown of Reps1 in C2C12 myotubes (I). Insulin-stimulated ³H-2-deoxy-glucose uptake in C2C12 myotubes transfected with control or Reps1 siRNA (J). Data in (B) and (C) were analyzed with a two-sided paired-sample t test (N = 8). Data in (E) and (F) were analyzed by a two-way ANOVA with repeated measures and Sidak multiple comparison test (n = 7–10). Data in (I) were analyzed by a two-sided two-sample t test. Data in (J) were analyzed by two-away ANOVA with Tukey’s multiple comparisons test (n = 6) (J). In paired analyses, bars represent the mean. In unpaired analyses, the mean +/− standard error of the mean is displayed. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Figure 4

RSK is an upstream kinase of REPS1 S709 and is associated with vesicle-sorting proteins in skeletal muscle (A) Insulin-stimulated (100 nM, 10 min) glucose uptake in C2C12 myotubes preincubated (20 min) with DMSO or 10 μM of the RSK inhibitor, BI-D1870. (B) Quantified western blot analysis of signaling as in (A). (C) Representative western blot analysis for signaling as in (B). (D) Insulin-stimulated (100 nM, 10 min) GLUT4 translocation in L6 myotubes preincubated (20 min) with DMSO or 10 μM BI-D1870. (E) Basal and insulin-stimulated (300 μIU/mL, 30 min) glucose uptake in ex vivo-incubated mouse soleus muscle preincubated (60 min) with DMSO or 10 μM BI-D1870. (F and G) Quantified western blot analysis of signaling as in (E). (H) Representative images of western blot analysis for signaling in ex vivo insulin-stimulated muscles. (I) Rested and contraction-stimulated glucose uptake in ex vivo-incubated mouse extensor digitorum longus (EDL) muscle preincubated (60 min) with DMSO or 10 μM BI-D1870. (J and K) Quantified western blot analysis of signaling as in (I). (L) Representative images of western blot analysis for signaling in ex vivo contraction-stimulated muscles. (M) FLAG pull-down in C2C12 myotubes transduced with either FLAG-Reps1 or FLAG-CTR . (N) Differentially enriched proteins highlighted in blue (two-sample t test, FDR <5%). Illustration of signaling pathways affected and the proposed role of REPS1 in vesicle trafficking. The illustration was made in BioRender. ∗symbol refers to significance of inhibitor vs. DMSO treatment. # symbol refers to significance of insulin/contraction. ∗∗∗/###p < 0.001, ∗∗/##p < 0.01, ∗/#p < 0.05. Data in (A), (B), (C), and (D) were analyzed by two-way ANOVA with Tukey’s multiple comparisons test. Data in (E)–(G) and (I)–(K) were analyzed by a two-way ANOVA with repeated measures with Šidák’s multiple comparisons test (n = 7–8). In paired analyses, bars represent the mean. In unpaired analyses, the mean +/− standard error of the mean is displayed.

Figure 5

Insulin-induced REPS1 S709 phosphorylation in vivo is impaired in multiple models of insulin resistance Sixteen C57BL6 male mice (7 weeks old) were fed a low-fat diet (LFD; n = 8) or high-fat diet (HFD; n = 8) for 16 weeks. At the day of termination, mice were injected with either saline or 1 U/kg insulin, and tissues were collected after 15 min (n = 4) (A). Body weight of mice after 16 weeks of diet intervention and blood glucose (mM) delta (post-pre) values in response to saline/insulin injection (B). Corresponding western blot analysis of insulin signaling in epidydimal white adipose tissue (eWAT), quadriceps muscle, and liver tissue from same animals (C). BMI and glucose-infusion rate (GIR) in response to a hyperinsulinemic euglycemic clamp (4 h) of 10 normal glucose-tolerant (NGT) and 10 patients with T2D (D). Quantified western blot of REPS1 S709 phosphorylation pre- and post-insulin stimulation (E). Association of GIR and delta REPS1 S709 phosphorylation in skeletal muscle (F). Data in (B) and (C) were analyzed by a two-way ANOVA and Tukey’s multiple comparisons test. Data in (E) were analyzed by a two-way ANOVA with repeated measures and Sidak multiple comparisons test. Data in (F) were analyzed by ANCOVA analysis followed by individual Pearson correlation. In paired analyses, bars represent the mean. In unpaired analyses, the mean +/− standard error of the mean is displayed. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Figure 6

REPS1 variants’ associations with complex traits, mRNA expression, and splicing Lower segment: representation of the phenotypes associated with REPS1 variants using data from NHGRI-EBI GWAS Catalog and Open Targets Genetics. In total, we identified 6 independent signals (r² < 0.1) in the REPS1 region: 4 were located in REPS1 introns and 2 in an intergenic region between REPS1 and ABRACL. Each signal is represented by their lead variant. Upper segment: representation of tissues and cell types with significant changes in REPS1 expression quantitative trait locus (eQTL) or alternative splicing (sQTL) as reported in Open Targets Genetics. We report all tissues and cell types with significant eQTL or sQTL associations for any of the 6 independent REPS1 lead variants. See Tables S2A–S2C for more information on the associations for each lead variant.