2.1 Human Participants

Ten healthy males aged 18–30 years with a body mass index 18.5–27.0 kg/m2 were recruited to participate in this study. Participant characteristics are shown in Fig. 1B. Inclusion criteria were physical inactivity (i.e., inactive in terms of exercise training and job, < 150 min/week moderate-intensity exercise, and no structured physical activity for 6 months prior to recruitment); no cardiopulmonary abnormalities; no injuries; the ability to pass the Exercise and Sport Science Australia (ESSA) pre-exercise screening tool and/or obtain general practitioner clearance to exercise; and the ability to ride a stationary cycle at high intensity. Exclusion criteria were known cardiovascular disease or diabetes mellitus; major or chronic illness that impairs mobility and/or eating/digestion; taking prescription medications (i.e., beta-blockers, anti-arrhythmic drugs, statins, insulin sensitizing drugs, or drugs that increase the risk of bleeding [anticoagulants, antiplatelets, novel oral anticoagulants, nonsteroidal anti-inflammatory drugs, selective norepinephrine reuptake inhibitors, or selective serotonin reuptake inhibitors]; or known bleeding disorders (i.e., hemophilia A [factor VIII deficiency], von Willebrand disease, or other rare factor deficiencies including I, II, V, VII, X, XI, XII, and XIII).

Fig. 1

Preliminary testing and randomized crossover trial design, participant baseline characteristics, and plasma lactate and glucose responses to HIIT and MICT. As detailed in the overall study schematic (A), participants first underwent preliminary testing and dietary control prior to each experimental HIIT or MICT trial day. Participants arrived at the laboratory following overnight fasting for baseline measurements, a dual-energy X-ray absorptiometry (DXA) body composition scan and resting metabolic rate (RMR) testing. Each participant then completed an incremental fitness test to volitional fatigue on a cycle ergometer to determine peak oxygen uptake (\(\dot}_} }\)) and maximal aerobic power (MAP) to calculate the work rate for the subsequent two workload (67.9 ± 10.2 kJ) and total duration (10 min) matched HIIT and MICT exercise trials. Participants’ food and fluid intake for all meals and snacks was recorded over a 3-day period using a mobile phone application and analyzed by an accredited research dietician. A standardized dinner was consumed by each participant the evening prior to each exercise trial, with no caffeine or alcohol consumed 20 or 24 h prior, respectively. In a randomized crossover design, participants were randomly assigned their first exercise trial (i.e., HIIT or MICT) prior to commencing trial days and did not perform any exercise in the 72 h prior to each trial day. HIIT and MICT exercise trials were separated by at least a 10-day recovery period. On each experimental day, participants reported to the laboratory following overnight fasting, and vastus lateralis skeletal muscle biopsies and venous blood samples were collected pre-exercise (0 min), mid-exercise (5 min), and immediately post-exercise (10 min). Participant characteristics are listed in (B). Plasma (C) lactate and (D) glucose concentrations across the acute HIIT and MICT exercise bouts were determined using a YSI Analyzer. No interaction effect was observed for plasma lactate in (C); P = 0.0573. Heart rate (E) and rating of perceived exhaustion (F; RPE; Borg RPE scale out of 20) were recorded at 1 min (i.e., following completion of the first HIIT “on” interval), 5 min, and 10 min during HIIT or MICT trials. Data are presented as mean ± SD; two-way ANOVA with repeated measurements, Tukey’s test for multiple comparisons; **P < 0.01 versus 0 min (or 1 min in E and F); ***P < 0.001 versus 0 min (or 1 min in E and F); ****P < 0.0001 versus 0 min (or 1 min in E and F); #P < 0.05 versus 5 min; n = 10 for each exercise intensity and timepoint

2.2 Participant Baseline Measurements, Exercise Testing, and Familiarization

Participants arrived at the laboratory (Melbourne, Australia) following a 10–12 h overnight fast for dual-energy X-ray absorptiometry (DXA)-based body composition analysis (Lunar iDXA; GE HealthCare, Chicago, IL, USA). Upon arrival each participant’s height and body mass were recorded, bladder was voided, and any metal jewelry or clothing items containing metal were removed prior to DXA scanning. Next, resting metabolic rate (RMR) testing was performed using a calibrated TrueOne 2400 (Parvo Medics, Sandy, UT, USA) with expired air collected for a total of 25 min, including a 10 min baseline measurement and 15 min data collection. Following RMR, resting blood pressure and heart rate (HR) were recorded in a seated position.

Following baseline measurements, each participant completed an incremental fitness test to volitional fatigue on an electronically braked cycle ergometer (Lode Excalibur Sport; Lode, Groningen, the Netherlands) to determine \(\dot}_ }\) and maximal aerobic power (MAP). During the maximal exercise capacity test, expired gas was collected every 30 s via open-circuit respirometry (TrueOne 2400; Parvo Medics) with continuous HR monitoring (Polar Heart Rate Monitor; Polar Electro, Kempele, Finland). Before each test, gas analyzers were calibrated with commercially available gases (16% O2, 4% CO2), and volume flow was calibrated using a 3 L syringe. Following a 5 min warm-up at 1 W/kg, resistance was increased by 25 W every 150 s until volitional fatigue, determined as the inability to maintain a cadence > 60 rpm. Individual \(}} \)O2peak and MAP were determined, with MAP calculated as Wfinal + (t/150 × 25) if the final stage was not completed, to calculate the work rate for subsequent work (67.9 ± 10.2 kJ) and duration (10 min) matched HIIT and MICT exercise trials. At least 72 h prior to the first randomized exercise trial, participants returned to the laboratory for exercise trial familiarization. Following a 5 min warm-up at 1 W/kg, participants completed two cycling-based exercise sessions consisting of a single bout of HIIT and MICT (10 min total each) to confirm their ability to successfully complete exercise trials at the prescribed intensities.

2.3 Dietary Control and Standardized Meals

Participants recorded dietary information using the Easy Diet Diary mobile phone application [18]. Food and fluid intake for all meals and snacks was recorded over a 3-day period and analyzed by an accredited research dietician. The habitual diet record and baseline DXA/RMR data were used to prescribe a standardized meal, which was provided with cooking instructions and consumed at the participant’s home for dinner between 18:00–20:00 h before each trial day. The macronutrient composition of the standardized dinner was 50% carbohydrate, 30% fat, and 20% protein. Participants refrained from consuming any other food or fluids other than water from 20:00 h the evening prior to each trial. Participants also refrained from caffeine consumption after 12:00 h and alcohol consumption and ibuprofen 24 h prior to each trial.

2.4 Exercise Trials and Skeletal Muscle Biopsy Collection

In a randomized crossover design, participants were assigned their first exercise trial (i.e., HIIT or MICT, with half of participants randomly assigned to perform the HIIT session first) prior to commencing trial days and did not perform any exercise in the 72 h prior to each trial day. The HIIT session consisted of 10 min total cycling with 1-min intervals at 85 ± 0.1% of individual MAP (176 ± 34 W) interspersed with 1-min active recovery intervals at 50 W. The MICT protocol was work- and duration-matched and consisted of an acute bout of continuous cycling at 55 ± 2% of individual MAP (113 ± 17 W). A schematic of the overall study design is shown in Fig. 1A. The two HIIT and MICT exercise trials were separated by ≥ 10 days, and it was not possible to blind participants nor the principal researchers to the order of these trials. All trials were completed between June 2019 and November 2019.

On trial days participants arrived at the laboratory at 07:00–08:00 h, having fasted overnight since consuming the standardized dinner the evening prior, and only consumed water during the trials. Each participant’s preferred arm was cannulated for blood collections (detailed below) and local anesthetic (1% lignocaine hydrochloride in saline; McFarlane; Surrey Hills, Victoria Australia; 11037-AS) was administered to the vastus lateralis by a highly experienced medical doctor. A percutaneous skeletal muscle biopsy was collected at rest prior to commencing exercise (0 min) using a Bergstrom needle modified with suction and immediately snap-frozen, placed in liquid nitrogen and stored at − 80 °C until analysis. Additional skeletal muscle biopsy samples were collected from each participant mid-exercise (5 min) and immediately post-exercise (10 min), with the total 10 min cycling duration consistent for each exercise intensity. All three biopsies for each exercise trial were taken from the same leg, with each subsequent biopsy collected 3–5 cm distal to the previous biopsy(ies). For the HIIT trial, the mid-exercise biopsy was collected during an active recovery interval on a bed placed directly behind the cycle ergometer (~ 30 s), and participants re-commenced cycling immediately after the biopsy was collected. For the MICT trial, participants stopped cycling (~ 30 s) for the mid-exercise biopsy collection and re-commenced cycling immediately after the biopsy was collected. HR (Polar Heart Rate Monitor; Polar Electro) and rating of perceived exhaustion (RPE; Borg RPE Scale out of 20) were recorded at 1 min (i.e., following completion of the first HIIT “on” interval), 5 min, and 10 min during each participant’s HIIT and MICT trials.

2.5 Blood Sampling and Analyses

Upon arrival to the laboratory, a cannula (22 G; Terumo, Tokyo, Japan) was inserted into an antecubital vein of each participant. Two vacutainers of venous blood (6 mL each) were collected via cannula at the same timepoints as skeletal muscle biopsies above, including pre-exercise (0 min), mid-exercise (5 min), and immediately post-exercise (10 min). Lipid panels (Roche Diagnostics, Basel, Switzerland; 6380115190) including triglycerides, total cholesterol, high-density lipoproteins (HDL) and low-density lipoproteins (LDL) were immediately measured from an aliquot of whole blood (~ 19 μL) using the COBAS b 101 system (Roche Diagnostics). Following inversion ten times, one EDTA-coated vacutainer collected for plasma (Interpath, Somerton, Victoria, Australia; 454036) was immediately placed on ice, centrifuged at 1500g for 10 min at 4 °C, aliquoted, and stored at − 80 °C until analysis. Simultaneous measurement of glucose and lactate from plasma aliquots was performed in duplicate using a calibrated YSI 2900 Biochemistry Analyzer (YSI Incorporated, Yellow Springs, OH, USA).

2.6 Immunoblotting

Snap-frozen human skeletal muscle biopsy samples were lysed in homogenization buffer containing 50 mM Tris–HCl (pH 7.5), 1 mM ethylenediaminetetraacetic acid (EDTA), 1 mM egtazic acid (EGTA), 10% glycerol, 1% Triton-X, 50 mM sodium fluoride, 5 mM sodium pyrophosphate with cOmplete Protease Inhibitor Cocktail and PhosSTOP phosphatase inhibitor (Sigma-Aldrich, St. Louis, MO, USA). Samples were centrifuged at 16,000g for 30 min at 4 °C and the protein content of the supernatant was determined using bicinchoninic acid (BCA) assay (Pierce, Rockford, IL, USA). Lysates (10 µg protein per well) suspended in Laemmli sample buffer were run on 4–15% pre-cast stain-free gels (Bio-Rad, Hercules, CA, USA) and transferred to polyvinylidene fluoride (PVDF) membranes (Merck Millipore, Burlington, MA, USA). Membranes were blocked with 7.5% bovine serum albumin (BSA) in Tris-buffered saline containing 0.1% Tween 20 (TBS-T) for 1 h at room temperature then incubated with primary antibodies overnight with rocking at 4 °C. After washing with TBS-T, membranes were incubated with secondary antibody for 1 h at room temperature. Antibodies against phospho-AMPK T172 (2535S), phospho-ACC S79 (11818S), and horseradish peroxidase-conjugated anti-rabbit (7074) IgG secondary antibodies were purchased from Cell Signaling Technology (Danvers, MA, USA). Proteins were detected via chemiluminescence using Bio-Rad Clarity Western ECL Substrate and imaged using the ChemiDoc Imaging System (Bio-Rad). The volume density of each target band was quantified using Bio-Rad Image Lab and normalized to total protein in each lane using stain-free imaging technology and Image Lab software (version 6.1, Bio-Rad).

2.7 Phosphoproteomic Sample Preparation

As depicted in Fig. 2A and detailed below, proteins from each human skeletal muscle biopsy sample were extracted, digested to peptides with trypsin, and isobarically labeled prior to phosphopeptide enrichment, fractionation, and analysis by liquid chromatography with tandem mass spectrometry (LC–MS/MS). Briefly, ~ 30 mg of each snap-frozen human skeletal muscle was lysed as previously described [17] in 6 M guanidine HCL (Sigma, St. Louis, MO, USA; G4505) and 100 mM Tris pH 8.5 containing 10 mM tris(2-carboxyethyl)phosphine (Sigma; 75259) and 40 mM 2-chloroacetamide (Sigma; 22790) using tip-probe sonication. The resulting lysate was heated at 95 °C for 5 min and centrifuged at 20,000g for 10 min at 4 °C, and the resulting supernatant was diluted 1:1 with water and precipitated with 5 volumes of acetone at − 20 °C overnight. Lysate was then centrifuged at 4000g for 5 min at 4 °C and the protein pellet was resuspended in Digestion Buffer (10% 2,2,2-trifluoroethanol [Sigma; 96924] in 100 mM HEPES pH 8.5). Protein content was determined using BCA (Thermo Fisher Scientific, Waltham, MA, USA). Four hundred μg protein (normalized to 100 μL final volume in Digestion Buffer) was digested with sequencing grade trypsin (Sigma; T6567) and LysC (Wako Chemicals, Richmond, VA, USA; 129-02541) at a 1:50 enzyme:substrate ratio at 37 °C overnight with shaking at 2000 rpm.

Fig. 2

Human skeletal muscle phosphoproteomic analysis reveals effective pre-exercise standardization and distinct signaling profile clusters in response to HIIT versus MICT after 5 min and 10 min. Vastus lateralis skeletal muscle biopsies were collected pre-exercise (0 min), mid-exercise (5 min), and immediately post-exercise (10 min) from each participant during the HIIT and MICT exercise trials (A). The 10 min HIIT cycling session consisted of alternating 1 min intervals at 85 ± 0.1% of individual MAP (176 ± 34 W) and 1 min active recovery intervals at 50 W. The total duration- and work-matched MICT cycling session consisted of 10 min cycling at 55 ± 2% of individual MAP (113 ± 17 W). Each of the 60 total muscle biopsy samples were prepared and subjected to LC–MS/MS analysis to accurately identify and quantity skeletal muscle protein phosphorylation sites at 0 min (pre-exercise), 5 min (mid-exercise), and 10 min (post-exercise) for both the HIIT and MICT exercise trials (A). Principal component analysis (B) and hierarchical clustering (C) of the phosphoproteomic datasets resulting from LC–MS/MS analysis of the 60 muscle biopsy samples were performed using the PhosR phosphoproteomic data analysis package (Kim et al. 2021 Cell Reports). Each individual point (B) or line (C) represents a unique biological sample, and samples are color-coded by exercise intensity and timepoint. Overall, 19% of the total variance in the overall phosphoproteomic dataset was explained by principal component (PC)1, while PC2 explained 6% of the variance. The total number of phosphopeptides and phosphosites identified and quantified using MS are shown (D), in addition to the number of differentially regulated phosphosites (± 1.5-fold change and adjusted P < 0.05) from each timepoint and/or exercise intensity comparison (F). Volcano plot shows the median phosphopeptide log2 fold change (x-axis) plotted against the − log10P-value (y-axis) for each pre-exercise condition, with no differentially regulated phosphosites at rest between crossover trials (E)

Digested peptides were labeled with 800 μg of 10-plex tandem mass tags (TMT) in 50% acetonitrile to a final volume of 200 μL at room temperature for 1.5 h. The TMT reaction was deacylated with 0.3% (w/v) of hydroxylamine for 10 min at room temperature and quenched to final volume of 1% trifluoroacetic acid (TFA). Each experiment consisting of seven TMT labeled peptides (ten total experiments, each including all six timepoints from a single participant’s HIIT and MICT trials and a pooled internal reference mix of peptides consisting of peptides from all ten participants) was then pooled, resulting in a final amount of 4 mg peptide per TMT 10-plex experiment. The sample identity and labeling channels have been uploaded as a table with the raw proteomic data to the ProteomeXchange Consortium via the PRIDE partner repository (see Resource Availability for login details).

In total, 20 μg of TMT-labeled peptide was removed for total proteome analysis (data not shown, as only 5–10 min of exercise does not affect total muscle protein content) and phosphopeptides were enriched from the remaining digestion of pooled peptides from each experiment using a modified version of the EasyPhos protocol [19]. Briefly, samples were diluted to a final concentration of 50% isopropanol containing 5% TFA and 0.8 mM KH2PO4. Dilutions were then incubated with 15 mg of TiO2 beads (GL Sciences, Tokyo, Japan; 5010–21315) for 8 min at 40 °C with shaking at 2000 rpm. Beads were washed four times with 60% isopropanol containing 5% TFA and resuspended in 60% isopropanol containing 0.1% TFA. The bead slurry was transferred to in-house packed C8 microcolumns (3 M Empore; 11913614) and phosphopeptides were eluted with 40% acetonitrile containing 5% ammonium hydroxide. The enriched phosphopeptides and 20 μg aliquot for total proteome analysis were acidified to a final concentration of 1% TFA in 90% isopropanol and purified by in-house packed SDB-RPS (Sigma; 66886-U) microcolumns. The purified peptides and phosphopeptides were resuspended in 2% acetonitrile in 0.1% TFA and stored at − 80 °C prior to offline fractionation using neutral phase C18BEH HPLC as previously described [17].

2.8 LC–MS/MS Data Acquisition and Processing

Peptides were analyzed on a Dionex 3500 nanoHPLC, coupled to an Orbitrap Eclipse mass spectrometer (Thermo Fisher Scientific) via electrospray ionization in positive mode with 1.9 kV at 275 °C and RF set to 30%. Separation was achieved on a 50 cm × 75 μm column (PepSep, Marslev, Denmark) packed with C18-AQ (1.9 μm; Dr Maisch, Ammerbuch, Germany) over 120 min at a flow rate of 300 nL/min. The peptides were eluted over a linear gradient of 3–40% Buffer B (Buffer A: 0.1% formic acid; Buffer B: 80% acetonitrile, 0.1% v/v FA) and the column was maintained at 50 °C. The instrument was operated in data-dependent acquisition (DDA) mode with an MS1 spectrum acquired over the mass range 350–1550 m/z (120,000 resolution, 1 × 106 automatic gain control [AGC] and 50 ms maximum injection time) followed by MS/MS analysis with a fixed cycle time of 3 s via HCD fragmentation mode and detection in the orbitrap (50,000 resolution, 1 × 105 AGC, 150 ms maximum injection time, and 0.7 m/z isolation width). Only ions with charge state 2–7 triggered MS/MS with peptide monoisotopic precursor selection and dynamic exclusion enabled for 30 s at 10 ppm.

DDA data were searched against the UniProt human database (June 2020; UP000005640_9606 and UP000005640_9606_additional) with MaxQuant v1.6.7.0 using default parameters with peptide spectral matches, peptide, and protein false discovery rate (FDR) set to 1% [20]. All data were searched with oxidation of methionine set as a variable modification and cysteine carbamidomethylation set as a fixed modification. For analysis of phosphopeptides, phosphorylation of serine, threonine, and tyrosine was set as a variable modification, and for analysis of TMT-labeled peptides, TMT was added as a fixed modification to peptide N-termini and lysine. First search MS1 mass tolerance was set to 20 ppm followed by recalibration and main search MS1 tolerance set to 4.5 ppm, while MS/MS mass tolerance was set to 20 ppm. MaxQuant output data were initially processed with Perseus [21] to remove decoy data, potential contaminants, and proteins only identified with a single peptide containing oxidized methionine. The “expand site” function was additionally used for phosphoproteomic data to account for multi-phosphorylated peptides prior to statistical analysis.

2.9 Bioinformatic Analysis

For analysis of human muscle phosphopeptides with TMT-based quantification, data were first log2 transformed and each phosphosite abundance was corrected by subtracting the abundance of the pooled sample in the same TMT batch. The phosphoproteomic data were processed using the pipeline implemented in the PhosR package [22]. Filtering was performed to retain phosphosites present in at least three participants (out of ten total participants), in at least one timepoint (out of six total timepoints). Missing values in the retained phosphosites were imputed first by a site- and sample condition-specific imputation method, where for a phosphosite that contains missing values in a condition, if more than three samples were quantified in that condition, the missing values were imputed on the basis of these quantified values for that phosphosite in that condition, and then by a random-tail imputation method [23]. The imputed data were normalized using the “combat” function in the sva package [24] for removing batch effects and then the “RUVphospho” function in PhosR for the removal of additional unwanted variation with a set of stably phosphorylated sites as negative controls [25]. The batch-corrected data were further converted to ratios relative to the pre-exercise samples (i.e., “0 min” controls). Baseline reliability and variability of phosphosite quantification between trials were assessed using the intraclass correlation coefficient (ICC) and coefficient of variation (CV). CV was calculated as SD/|mean|× 100, and ICC was computed using a two-way random effects model, measuring absolute agreement between conditions (Supplementary Table 4). Reproducibility classification of ICC values followed Cicchetti’s thresholds (i.e., excellent ≥ 0.75, good 0.60–0.74, fair 0.40–0.59, and poor < 0.40) [26].

Differentially phosphorylated sites were identified using the limma R package [27]. Phosphosites with ± 1.5-fold change and FDR-adjusted P value < 0.05 from each timepoint and exercise intensity comparison were considered as differentially phosphorylated (Fig. 3A–H). Kinase activities at post- or mid-exercise for both HIIT and MICT were inferred on the basis of the changes in phosphorylation (relative to the corresponding pre-exercise control samples) of their known substrates using the KinasePA package [28] and the PhosphoSitePlus annotation database [29] (Fig. 4A). Pathway enrichment analysis was then performed, whereby phosphosites were first summarized into their host protein levels by taking the maximum log2 fold change for each comparison between conditions, and then the pathway enrichment was performed on the basis of the inferred host protein changes using the KinasePA package and Reactome database [30] (Fig. 4B).

Fig. 3

Human skeletal muscle phosphorylation sites differentially regulated by an acute bout of work- and duration-matched HIIT and/or MICT. Volcano plots showing the median phosphopeptide log2 fold change (x-axis) are plotted against the − log10P value (y-axis) for each individual exercise intensity versus the respective timepoint (A–F). From the ~ 15,000 total phosphopeptides detected, significantly up-regulated (red dots) and down-regulated (blue dots) phosphosites are shown (± 1.5-fold change and adjusted P < 0.05), with black dots representing phosphosites that were detected but not significantly regulated by exercise. Volcano plots comparing signaling responses with each exercise intensity (i.e., HIIT versus MICT) at each timepoint are shown in G, H

Fig. 4

Kinase and pathway enrichment uncovers common and unique kinases and pathways regulated by HIIT and/or MICT. Kinase activity (A) was inferred via direction analysis using kinase perturbation analysis (KinasePA; [28]) to annotate and visualize how kinases and their known substrates are perturbed by each exercise intensity and timepoint. Pathway enrichment analysis (B) was performed using the Reactome database [30] to determine biological pathways that are enriched within the lists of significantly regulated genes (corresponding to their respective phosphoproteins) for each exercise intensity and timepoint relative to its respective pre-exercise control. For kinase activity inference (A) and pathway enrichment (B), z-scores above and below the dotted lines (corresponding to |z-score|> 1.64) were considered as increased or decreased by exercise, respectively, as they correspond to a one-tailed P value of ~ 0.05 in normally distributed data

Putative substrates of kinases for HIIT and MICT were predicted using the “kinaseSubstrateScore” function in the PhosR package, and the results were represented as heatmaps (Fig. 5A, B). Pathway over-representation analysis was performed on protein sets identified from the kinase–substrate scoring analysis (i.e., kinase–substrate pairs with a score > 0.85 were selected) per kinase using the “enrichKEGG” function implemented in the clusterProfiler R package [31], and P values were adjusted for multiple testing using Benjamini–Hochberg FDR correction at α = 0.05 (Fig. 5C, D). The prediction scores were subsequently used for constructing signalome networks. Pearson’s correlation was performed on pairwise kinases, and then the correlation matrix was binarized on the basis of the correlation score threshold of 0.85. Undirected graphs were built from the binary adjacency matrix using the “graph_from_adjacency_matrix” function from the igraph package [32], and results from this analysis were presented as network diagrams (Fig. 6A, B).

Fig. 5

Kinase–substrate predictions and pathway enrichment analysis identify differential regulation of downstream substrates and pathways in response to HIIT versus MICT. Kinase–substrate associations were predicted in response to HIIT and MICT via the phosphoproteome signaling profiles and kinase recognition motif of known substrates using PhosR [22]. This analysis generated prediction matrices, with columns corresponding to kinases, rows corresponding to phosphosites, and values in the heatmaps denoting how likely a phosphosite is phosphorylated by a given kinase in response to HIIT (A) and MICT (B). Pathway enrichment analysis was performed using kinase–substrate predictions (i.e., phosphosites with a high prediction score for each kinase) to determine how kinases regulate common and/or distinct signaling pathways in response to HIIT (C) and MICT (D)

Fig. 6

Signalome network highlights distinct HIIT and MICT kinase clusters and differential signaling trajectories in response to each exercise intensity. Signalome networks for HIIT (A) and MICT (B) exercise were constructed using the PhosR phosphoproteomic data analysis package [22]. This “signalome” construction method utilized the phosphoproteome signaling profile and kinase recognition motif of known substrates to visualize the interaction of kinases and their collective actions on signal transduction. Kinases clustered together are highly correlated in terms of kinase–substrate predictions. Visualization of five phosphoprotein clusters from the phosphoproteomic dataset highlights distinct kinase–substrate regulation within the HIIT and MICT signaling networks, with shared and unique signaling trajectories shown for a panel of kinases in response to HIIT (C) and MICT (D)

Five protein modules were identified by clustering proteins with phosphosites sharing similar dynamic phosphorylation profiles and kinase regulation across both HIIT and MICT. The proportion of phosphosites that were phosphorylated by kinases for each protein module was calculated and presented as bubble plots (Fig. 6C, D). The activity of each protein module was then inferred. The regulated phosphosites were first obtained across all conditions (i.e., analysis of variance [ANOVA] test with adjusted P < 0.05) and the average log2 fold change of the regulated phosphosites for each of the five modules (relative to the corresponding pre-exercise control) were calculated (Supplementary Fig. 2A–E).

2.10 Statistical Analysis

Statistical analysis was performed using GraphPad Prism (version 9.4). A two-way ANOVA with repeated measurements was used to determine the effects of time and exercise intensity, with Tukey’s test applied for multiple comparisons (P < 0.05 considered as significant; sample size and statistical parameters are reported in the Fig. 1 legend). Spearman correlation of individual phosphorylation sites with plasma lactate concentrations at each timepoint and exercise intensity was performed to determine significantly correlated phosphosites, with Benjamini–Hochberg false discovery rate applied (q < 0.05 considered as significant; sample size and statistical parameters are reported in the Fig. 7 legend).

Fig. 7

Correlation of HIIT and MICT phosphosites and plasma lactate levels identifies > 3000 lactate-correlated sites including functional phosphosites that govern protein activity and metabolic regulation. Spearman correlation of individual phosphorylation sites (n = 8509 total) with plasma lactate concentrations at each timepoint and exercise intensity (n = 60 total plasma samples analyzed) are shown for four of the most significantly correlated sites (q < 0.05 with Benjamini–Hochberg FDR) with annotated functional roles in governing their respective phosphoprotein’s activation state and regulating a range of key metabolic processes (e.g., glycolysis, glucose transport, and mitochondrial biogenesis) including PDHA1 S201 (A), RPTOR S859 (B), TFEB S123/S128/S136 (C), and TBC1D4 S588 (D)