Identification of a resistance-exercise-specific signalling pathway that drives skeletal muscle growth

Abstract

Endurance and resistance exercise lead to distinct functional adaptations: the former increases aerobic capacity and the latter increases muscle mass. However, the signalling pathways that drive these adaptations are not well understood. Here we identify phosphorylation events that are differentially regulated by endurance and resistance exercise. Using a model of unilateral exercise in male participants and deep phosphoproteomic analyses, we find that a prolonged activation of a signalling pathway involving MKK3b/6, p38, MK2 and mTORC1 occurs specifically in response to resistance exercise. Follow-up studies in both male and female participants reveal that the resistance-exercise-induced activation of MKK3b is highly correlated with the induction of protein synthesis (R = 0.87). Additionally, we show that in mice, genetic activation of MKK3b is sufficient to induce signalling through p38, MK2 and mTORC1, along with an increase in protein synthesis and muscle fibre size. Overall, we identify core components of a signalling pathway that drives the growth-promoting effects of resistance exercise.

Extended Figure 1.. Gene ontology enrichment in phosphopeptide clusters 1 and 2

1D enrichment analysis of gene ontology (GO) terms when using the membership score of the phosphopeptides for “cluster 1” (a), or “cluster 2” (b) as defined in Figure 2. All GO terms were assigned a rank-based score between −1 and 1, with a negative score indicating under-representation of the term and a positive score indicating over-representation. Redundant GO terms were then removed with REVIGO. In the graphs, the score for each term was plotted against the −Log₁ of its respective q-value. The displayed dots indicate GO terms with a q-value of < 0.05. GO terms of interest are highlighted in each graph, and a full list of the outcomes is provided in the source data table.

Extended Figure 2.. Reproducibly of the endurance and resistance exercise-induced changes in the phosphorylation of MAPKAPK substrates

Heatmaps of the mean exercise-induced change in the phosphorylation of the known and/or predicted substrates of the MAPKAPK’s that were identified in both the current study and the phosphopeptide dataset of Blazev et al., 2022 . Also shown is the number of participants (n) that the mean values were obtained from in each dataset.

Extended Figure 3.. Quantitative results from the western blots in Figure 4

Biopsies from the experimental interventions described in Figure 4c were subjected to western blot analysis as shown in Figure 4d. a-k, For each participant, the phospho to total protein ratio (P/T) for the indicated signaling event in each biopsy was determined and then expressed relative to the mean value observed in the pre-exercise biopsies. Values in the graphs are presented as the group mean ± SEM, the number of samples per group is indicated at the bottom of the bars in the graphs. The data was analyzed with one-way mixed ANOVA. The q-value for each statistically significant pairwise comparison is annotated in the graphs with a * being used when P < 0.0001.

Extended Figure 4.. Long-term adaptations in the mouse model of endurance exercise

Mice were subjected to 13 weeks of training with treadmill running (TR) or a mock (control) paradigm. The average weekly (a) body weight, and (b) workload per training session, as well as (c) the number of times the rear of the animal was touched during each training session. Individual data points are displayed with hollow symbols and the weekly means for each group are displayed with solid symbols. n = 10 per group for a-c. d-r, After 13 weeks of training, the mice were subjected to measurements of (d) grip strength, and (e) tibia length (TL). The mass of the (f) individual epididymal (Epi.) fat pads, (g) interscapular brown adipose tissue (iBAT), (h) adrenal glands, and (i) heart were measured and normalized to TL. j, The mass of individual muscles (MM) including the gastrocnemius (GAST), plantaris (PLT), soleus (SOL), flexor digitorum longus (FDL), pectoralis major (PEC), triceps brachii lateral head (Tri-Lat), triceps brachii long head (Tri-Long), and the forearm flexor complex (FF) were all normalized to TL and expressed relative to the mean value observed in the control group. k, Mid-belly cross-sections of the FDL muscles were subjected to immunohistochemistry (IHC) for laminin and fiber type identification (i.e., Type I, IIA, IIX, or IIB), scale bars = 500 μm. The entire cross-section was used to determine (l), the average cross-sectional area (CSA) of the different fiber types, and (m) the proportion of the fibers that were represented by each fiber type. n, Mid-belly cross-sections of the FDL muscles were subjected to IHC for laminin and CD31 to identify capillaries, scale bars = 25 μm. o, The entire cross-section was used to determine the average number of capillaries per fiber. p-r, FDL muscles were subjected to western blot analysis for (q) members of the five OXPHOS complexes (i.e., CI - CV), and (r) other mitochondrial (mito.) proteins. For each sample, the individual protein content was normalized to the total amount of protein loaded on the gel and then expressed relative to the mean of the control group. Values in the graphs are presented as the group mean ± SEM, for d-r the number of samples per group is indicated at the bottom of the bars in the graphs. The data were analyzed with two-way repeated measures (RM) ANOVA (a), one-way RM ANOVA (b,c), paired t-tests (d-i, and o), or two-way ANOVA (j, l, m, q, r). ■ Significantly different from week 1, P < 0.05. The specific P-values for all other statistically significant pairwise comparisons are annotated in the graphs.

Extended Figure 5.. Long-term adaptations in the mouse model of resistance exercise

Flexor digitorum longus (FDL) muscles were collected from mice that had completed 13 weeks of training with weighted pulling (WP) or an unweighted (control) paradigm as previously reported by Zhu et al. 2021 . a, Mid-belly cross-sections of the FDL muscles were subjected to immunohistochemistry for laminin and CD31 to identify capillaries, scale bars = 25 μm. b, The entire cross-section was used to determine the average number of capillaries per fiber. c-e, FDL muscles were subjected to western blot analysis for (d) members of the five OXPHOS complexes (i.e., CI - CV), and (d) other mitochondrial (mito.) proteins. For each sample, the individual protein content was normalized to the total amount of protein loaded on the gel and then expressed relative to the mean of the control group. Values in the graphs are presented as the group mean ± SEM, the number of samples per group is indicated at the bottom of the bars in the graphs. The data were analyzed with two-sided paired t-tests (b), or two-way ANOVA (d, e).

Extended Figure 6.. A rapid and robust activation of signaling through MKK3/4/6, p38, and MK2 occurs specifically in response to resistance exercise in mice

a, Schematic of how C57BL6 mice were subjected to endurance exercise with treadmill running (TR), resistance exercise with weight pulling (WP), or their respective mock-trained (control) conditions. b, FDL muscles from the mice were collected immediately after the last training bout and subjected to western blot analysis for the phospho (P) and total (T) levels of the indicated proteins. Long isoform of MK2 (L), short isoform of MK2 (S). c, For each sample, the phospho to total protein ratio (P/T) for each signaling event was determined and expressed relative to the mean value observed in the treadmill control group. Values in the graphs are presented as the group mean ± SEM, the number of samples per group is indicated at the bottom of the bars in the graphs. The data were analyzed with two-way ANOVA or a Student’s t-test. ■ Significant difference between the TR control and TR trained group when the planned comparison was analyzed with a Student’s t-test, P < 0.05. The P-value for each statistically significant pairwise comparison is annotated in the graphs with a * being used when P < 0.0001.

Extended Figure 7.. Quantitative results from the western blots in Figure 7

FDL muscles from the experimental conditions described in Figure 7a were subjected to western blot analysis as shown in Figure 7e. a-k, For each sample, the phospho to total protein ratio (P/T) for each signaling event was determined and then expressed relative to the mean value observed in the treadmill (TR) control group. Values in the graphs are presented as the group mean ± SEM, the number of samples per group is indicated at the bottom of the bars in the graphs. The data were analyzed with two-way ANOVA. The P-value for each statistically significant pairwise comparison is annotated in the graphs with a * being used when P < 0.0001.

Extended Figure 8.. Quantitative results from the western blots in Figure 8

TA muscles were subjected to western blot analysis as described in Figure 8b. a-i, For each sample, the total (T) protein level, or the phospho to total protein ratio (P/T) for the indicated signaling event, was determined and then expressed relative to the mean value observed in the LacZ control group. Values in the graphs are presented as the group mean ± SEM, n = 4 per group unless otherwise indicated at the bottom of the bars in the graphs. The data were analyzed with one-way ANOVA. The P-value for each statistically significant pairwise comparison is annotated in the graphs with a * being used when P < 0.0001.

Extended Figure 9.. Genetic activation of MKK3b induces hypertrophy through a mechanism that is only partially dependent on mTORC1.

a, Schematic describing how electroporation was used to transfect mouse tibialis anterior (TA) muscles. Created in BioRender. Hornberger, T. (2024) BioRender.com/e11s145. Specifically, the TA muscles of male and female C57BL6 mice were co-transfected with plasmid DNA encoding tdTomato and LacZ as a control condition, Rheb as a direct activator of mTORC1, or with constitutively active (c.a.) mutant of MKK3b. Following electroporation, the mice were given daily intraperitoneal (IP) injections of the drug rapamycin (1.5 mg/kg) to inhibit signaling through mTORC1 or the solvent vehicle as a control. b, At 7 days post-transfection, the muscles were collected and subjected to immunohistochemistry for laminin to identify the periphery of the transfected (tdTomato positive) vs. non-transfected (tdTomato negative) fibers, scale bars = 50 μm. c, The cross-sectional area (CSA) of randomly selected fibers were measured, and then the values in the transfected fibers were expressed relative to the mean of the values observed in the non-transfected (control) fibers within each sample (n = 60-155 transfected and non-transfected fibers per sample). Values in the graphs are presented as the group mean ± SEM, the number of samples per group is indicated at the bottom of the bars in the graphs (605-889 fibers per group). The data were analyzed with two-way ANOVA. The P-value for each statistically significant pairwise comparison is annotated in the graphs with a * being used when P < 0.0001.

Extended Figure 10.. The genetic activation of MKK3b prevents immobilization-induced atrophy.

a, Illustration of how electroporation was used to co-transfect the left and right tibialis anterior (TA) muscles of male and female C57BL6 mice with plasmid DNA encoding tdTomato and constitutively active (c.a.) MKK3b or LacZ as a control. Immediately following electroporation, the right hindlimb was subjected to immobilization while the left hindlimb was untouched and used for the control condition. Schematic created in BioRender. Hornberger, T. (2024) BioRender.com/a12z496. b, At 7 days post electroporation, the TA muscles were collected and cross-sections were subjected to immunohistochemistry for laminin to identify the periphery of the transfected (tdTomato positive) and non-transfected (tdTomato negative) fibers, scale bars = 100 μm. The mean cross-sectional area (CSA) of the transfected and non-transfected fibers in each sample was determined from n = 58-130 transfected and non-transfected fibers per sample, and the resulting values for each sample were expressed relative to the mean value obtained in the sex-matched control group (i.e., the tdTomato negative fibers from vehicle-treated muscles that were co-transfected with LacZ). Values in the graphs are presented as the group mean ± SEM, the number of samples per group is indicated at the bottom of the bars in the graphs (669-1269 fibers per group). The data were analyzed with two-way RM ANOVA. The P-value for each statistically significant pairwise comparison is annotated in the graphs with a * being used when P < 0.0001.

Figure 1.. Overview of the phosphoproteomic alterations that occur after a bout of endurance versus resistance exercise in humans

a, Timeline of the experimental interventions and a general description of the analytical procedures that were employed. b, Measurement of the mean rate of myofibrillar protein synthesis during three different 3 hr windows of time including: i) 3 hr pre-exercise (Pre), ii) immediately post (0 hr) to 3 hr post endurance exercise (END), and iii) 0 hr to 3 hr post resistance exercise (RE). Values are group means ± SEM, n = 4 per group. Data were analyzed with one-way repeated measures (RM) ANOVA. The P-value for each statistically significant pairwise comparison is listed. c, Unsupervised hierarchical clustering of the phosphopeptide data from each of the participants. Squares denote particpants that performed RE before END. d, Principal component analysis of the phosphopeptide data from each of the participants. Dots/squares represent the values for each participant and the color of the dot indicates the experimental condition (as described in c). e, UpSet plot illustrating the number of phosphopeptides for each condition that experienced a > 1.5-fold exercise-induced change (Δ) in abundance at an FDR-corrected P-value (q) of < 0.05 when compared with Pre. Icons created in BioRender. Hornberger, T. (2025) https://BioRender.com/7mi8p27.

Figure 2.. Cluster-based identification of phosphorylation events that are specific to endurance and resistance exercise in humans

a, Volcano plots from the phosphoproteomic analyses at 0 hr or 3 hr post endurance (END) or resistance (RE) exercise. b, Volcano plots from the proteomic analyses for the same conditions listed in a. The highlighted proteins experienced a significant alteration in abundance, q ≤ 0.05. c, Soft clustering of the phosphoproteomic data revealed four major clusters of phosphopeptides that were affected by exercise. In a, the phosphopeptides that had a membership score of ≥ 0.5 for cluster 1 are highlighted in pink, while those with a membership score of ≥ 0.5 for cluster 2 are highlighted in blue.

Figure 3.. Identification of kinases that are reproducibly inferred as being regulated by endurance and/or resistance exercise in humans

a, Heatmap of Z-scores for the KSEAapp inferred alterations in kinase activity at 0 hr or 3 hr post endurance (END) or resistance (RE) exercise when compared to the pre-exercise state. Kinases are listed by their gene names and only kinases expressed in skeletal muscle (see methods) that possessed at least 5 known and/or predicted substrates in the phosphopeptide dataset are listed. The list of kinases was clustered according to their group (e.g., AGC) and then listed in alphabetical order. b, The results from the current study are compared with the results that were derived from identical processing of the phosphopeptide dataset published by Blazev et al., 2022 . c, Heatmaps of the Z-scores for the kinases whose activity was inferred to be significantly altered (q ≤ 0.05) in at least one condition in both the Blazev et al., 2022 dataset and the current dataset. d, Heatmaps of the mean exercise-induced change in the phosphorylation of the known and/or predicted substrates of each kinase listed in (c). Also shown is the number of known and/or predicted substrates for each kinase (m), the number of m that were common to both datasets (m ⁼), and the combined number of distinct m from the two datasets (m ^c).

Figure 4.. Prediction and validation of a signaling pathway that is activated specifically by resistance exercise in humans

a, Multivariable plot with Z-scores, q-values, and the mean differences in the phosphorylation of the known and/or predicted substrates for each kinase that was inferred by KSEAapp to have a significant difference (q ≤ 0.05) in activity at 3 hr post resistance (3 hr RE) vs. 3 hr post endurance (3 hr END) exercise. The list includes the gene name of each kinase as well as its common alias. Also highlighted with a * are kinases that were inferred by KSEAapp to have significant differences in activity at 3 hr RE vs. 3hr END as determined from the phosphopeptides dataset published by Blazev et al., 2022 . b, Manually curated prediction of the interconnectivity and functional outcomes of the most robustly perturbed kinases listed in (a). c, Schematic of the experimental intervention that was used to test the predictions in (b). d, The samples from (c) were subjected to western blot analysis for the phospho (P) and total (T) levels of the indicated proteins. Non-specific band (ns), long isoform of MK2 (L), short isoform of MK2 (S). The quantitative analysis of the western blots are provided in Extended Figure 3. e, Measurement of the mean rate of myofibrillar protein synthesis during three different 3 hr windows of time including: i) 3 hr pre-exercise (Pre), ii) immediately post (0 hr) to 3 hr post END, and iii) 0 hr to 3 hr post RE. Values are group means ± SEM, n = 12 per group. Data was analyzed with one-way repeated measures (RM) ANOVA. * Significant difference between the indicated groups, P < 0.0001. Icons in the figure were created in BioRender. Hornberger, T. (2025) https://BioRender.com/7mi8p27.

Figure 5.. Changes in MKK3b(S218) phosphorylation are highly correlated with the resistance exercise-induced increase in myofibrillar protein synthesis

For each of the signaling events analyzed in Figure 4, linear regression was used to compare each participant’s endurance (END) and resistance exercise (RE) induced change (Δ) in myofibrillar protein synthesis (MyoPS) with the mean ( $\stackrel{-}{x}$ ) Δ of the phospho to total protein ratio (P/T) at 0 hr and 3 hr post-exercise. a, Multivariable plot illustrating the coefficient of determination (R-squared) and P-value of the co-relationship for all comparisons that revealed an R-squared value of >0.1. NS indicates not significant. b, Graph of the co-relationship between the Δ in MyoPS and the Δ in the P/T for MKK3(S218). Individual values for each participant were expressed as a percentage of the value obtained in their respective pre-exercise sample. The solid line represents the line of best fit and the dashed lines represent the 95% confidence intervals.

Figure 6.. Mouse models of endurance and resistance exercise lead to distinct adaptations

a, Male mice were subjected to 13 weeks of endurance exercise with treadmill running (TR) or resistance exercise with weight pulling (WP). Values in each exercise group were expressed relative to their respective mock-trained (control) groups (see Extended Figures 4, 5 and Zhu et al. 2021 ), and then the effects of the different modes of exercise were compared. b, Measurements of grip strength. c-d, The mass of the individual epididymal (Epi.) fat pads (c), and flexor digitorum longus (FDL) muscles (d), after being normalized to tibia length (TL). e-g, The average cross-sectional area (CSA) of the different fiber types (e), the proportion of the fibers that were represented by each fiber type (f), and the average number of capillaries per fiber (g), as assessed in mid-belly cross-sections from FDL muscles. h-i, FDL muscles were subjected to western blot analysis for (h) members of the five OXPHOS complexes (i.e., CI - CV), and (i) other mitochondrial (mito.) proteins. Values in the graphs are presented as the group means ± SEM, the number of samples per group is indicated at the bottom of the bars in the graphs. ● Significantly different from the condition-matched control group as presented in Extended Figures 4, 5 and Zhu et al. 2021. The P-value for each statistically significant comparison between the endurance and resistance conditions are annotated in the graphs with a * being used when P < 0.0001. The data were analyzed with two-sided Student’s t-tests (b-d, and g), or two-way ANOVA (e, f, h, and i).

Figure 7.. Mouse models affirm that prolonged activation of signaling through MKK3/4/6, p38, MK2, and mTORC1 occurs specifically in response to resistance exercise

a, Schematic of how male transgenic mice that express a mutated form of the methionyl-tRNA synthetase (MetRS^{L274G +/−}) were subjected to endurance exercise with treadmill running (TR), resistance exercise with weight pulling (WP), or their respective mock-trained (control) conditions. At the end of the last training bout, the mice were injected with azidonorleucine (ANL) which is an azide-bearing analog of methionine that can be incorporated into newly synthesized proteins of mice that possess the MetRS^L274G transgene. The FDL muscles were collected at 3 hr post-training and subjected to the analyses described below. b, The muscles were homogenized and whole lysates were separated into myofibrillar and sarcoplasmic fractions. The different fractions were then subjected to western blot analysis for myofibrillar and sarcoplasmic proteins. c, As illustrated in (a), a DBCO-based ‘click’ reaction was used to label the ANL-containing proteins with a fluorophore (Cy5.5). Fluorescently labeled proteins in the myofibrillar fraction were then subjected to SDS-PAGE and used to visualize the in-gel amount of ANL-labeled proteins as well as the total amount of protein in each sample. d, The ANL-labeled to total protein ratio for each sample was quantified and used as a readout for the rate of myofibrillar protein synthesis. e, The FDL muscles were subjected to western blot analysis for the phospho (P) and total (T) levels of the indicated proteins. Non-specific band (ns), long isoform of MK2 (L), short isoform of MK2 (S). The quantitative analysis of these western blots is provided in Extended Figure 7. Values in the graph are presented as the group mean ± SEM, the number of samples per group is indicated at the bottom of the bars in the graph. Data were analyzed with two-way ANOVA. The P-value for each statistically significant pairwise comparison is annotated in the graph with a * being used when P < 0.0001.

Figure 8.. Genetic activation of MKK3b and MKK6 is sufficient to induce resistance exercise-specific signaling events, protein synthesis, and growth

a, Schematic describing the electroporation procedure that was used to transfect mouse tibialis anterior (TA) muscles. Created in BioRender. Hornberger, T. (2024) BioRender.com/e11s145. b, TA muscles (n = 4 per group) of male C57BL6 mice were transfected with plasmid DNA encoding FLAG-tagged and constitutively active (c.a.) mutants of human MKK3b or MKK6, or with LacZ as a control condition. The muscles were collected at 3 days post-transfection and subjected to western blot analysis for the phospho (P) and total (T) levels of the indicated proteins. Long isoform of MK2 (L), short isoform of MK2 (S). Quantitative analysis of the western blots is provided in Extended Figure 8. c, TA muscles of male MetRS ^+/− mice were co-transfected with tdTomato and c.a. MKK3b, c.a. MKK6, or LacZ as the control condition. At 3 days post-transfection the mice were injected with ANL to label newly synthesized proteins. At 3 hr after the ANL injection, the muscles were collected for histological analysis. Cross-sections were used to visualize the transfected (tdTomato positive) vs. non-transfected (control) fibers, ANL-labeled proteins, and phosphorylated (P) S6(S240_4) as a marker of signaling through mTORC1, scale bars = 50 μm. For each sample, the intensity of the signal for ANL (i.e., protein synthesis) (d), and P-S6(S240_4) (e) were simultaneously measured in randomly selected fibers, and then the values in the transfected fibers were expressed relative to the mean of the values observed in the non-transfected (control) fibers within that sample (n = 60-120 transfected and non-transfected fibers per sample). f,g, Graphs of the co-relationships between protein synthesis and P-S6(S240_4) in the fibers from (d,e) that were transfected with c.a. MMK3b (f), or c.a. MKK6 (g). The data were analyzed with linear regression, dashed lines represent the 95% confidence intervals. h, The cross-sectional area (CSA) of the same fibers analyzed in (d,e) was measured and then the CSA of the transfected fibers was expressed relative to the mean of the non-transfected (control) fibers within that sample. i, TA muscles of male C57BL6 mice were transfected, collected at 7 days post-transfection, and analyzed for fiber CSA as described above (n = 46-140 transfected and non-transfected fibers per sample). Values in the graphs are presented as the group mean ± SEM, the number of samples per group is indicated at the bottom of the bars in the graphs (290-526 fibers per group). The data in (d,e,h,i) were analyzed with one-way ANOVA. The P-value for each statistically significant pairwise comparison is annotated in the graphs with a * being used when P < 0.0001.