Acute exercise rewires the proteomic landscape of human immune cells

Acute exercise rewires the proteomic landscape of human immune cells | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Resource Acute exercise rewires the proteomic landscape of human immune cells Philipp Zimmer, David Walzik, Niklas Joisten, Alan Metcalfe, Sebastian Proschinger, and 9 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6864249/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 02 Jan, 2026 Read the published version in Nature Communications → Version 1 posted You are reading this latest preprint version Abstract Exercise-driven alterations of the immune system are a key mechanism in the prevention and treatment of various diseases. Here, we performed mass spectrometry-based proteomics analysis on peripheral blood mononuclear cells (PBMCs) at a depth of > 6000 proteins. Comparing time- and workload-matched high-intensity interval exercise (HIIE) and moderate-intensity continuous exercise (MICE) we discover versatile changes in the proteomic makeup of PBMCs and reveal profound alterations related to effector function and immune cell activation pathways within one hour after exercise. These changes were more pronounced after HIIE compared to MICE and occurred despite identical immune cell mobilization patterns between the two exercise conditions. We further identify an immunoproteomic signature that effectively predicts cardiorespiratory fitness. This study provides a reliable data resource that expands our knowledge on how exercise modulates the immune system, and delivers biological evidence supporting the WHO 2020 guidelines, which highlight exercise intensity as a relevant factor to maintain health. Biological sciences/Immunology/Translational immunology Health sciences/Medical research/Translational research Health sciences/Biomarkers/Predictive markers Biological sciences/Physiology exercise immune cells PBMCs proteostasis effector function high-intensity interval training exercise immunology proteomics flow cytometry Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction Physical exercise is one of the most powerful strategies to prevent and counteract acute and chronic diseases across the entire human lifespan. The health benefits of exercise are demonstrated by numerous clinical and observational trials, 1 , 2 but the underlying biological mechanisms are poorly understood. Given the broad implications of the immune system in protecting from disease, there is a clear rationale for comprehensive investigations of the impact of exercise on immune cells. Both, the exercise-induced mobilization of specific immune cells into circulation as well as an increased cytokine release are well-described phenomena in exercise immunology. 3 However, few investigations have evaluated the molecular alterations in immune cells triggered by exercise. Bulk RNA sequencing revealed a complex interplay of up- and downregulated transcripts that peaked two minutes after exercise and returned to baseline within 30–60 minutes in peripheral blood mononuclear cells (PBMCs). Enriched pathways were predominantly related to inflammatory signaling and immune activation, but also included other processes such as cell growth and mobility. 4 While the number and kinetics of transcripts suggest that many expected and some novel signaling pathways are initiated by acute exercise, the proteomic response of PBMCs remains unclear. Since the proteomic makeup of cells defines their phenotype and function, it is crucial to evaluate the impact of exercise on the proteome of immune cells. Here, we provide a comprehensive characterization of the impact of two different endurance exercise paradigms on the proteomic makeup of PBMCs. The exercise sessions were designed based on the World Health Organization 2020 guidelines on physical activity, which highlight exercise intensity as a crucial variable for health promotion. 5 Using state-of-the-art mass spectrometry-based proteomics in a standardized experimental setting, we demonstrate that the proteomic makeup of PBMCs is reshaped by acute exercise. Interestingly, proteomic alterations are more potently induced by high-intensity interval exercise (HIIE) compared to moderate-intensity continuous exercise (MICE). The observed changes indicate altered immune cell activation and effector functions in the recovery phase following HIIE, which underlines the immunomodulatory impact of exercise on a proteomic level. Of note, flow cytometry-based immune cell phenotyping suggests that these adaptions occur independent of immune cell mobilization. Using prediction models, we additionally identify an immunoproteomic signature associated with cardiorespiratory fitness. Our data indicates that exercising at higher intensity is necessary to induce proteomic changes associated with immune function, providing biological support for exercise intensity as a crucial variable in the WHO 2020 physical activity guidelines. Results Study design and participant characteristics We designed a randomized crossover study comparing time- and workload-matched HIIE and MICE to investigate the impact of exercise intensity on the proteome of immune cells. PBMCs were isolated at baseline, immediately after, and 1h after each exercise condition (Fig. 1 A). In total, data from 23 overnight-fasted recreationally active runners (12 female, 11 male) was collected. Participants exhibited a mean (± SD) age of 30 ± 4 years, a body mass index (BMI) of 22.2 ± 2.38 kg m − 2 , and a cardiorespiratory fitness (measured as peak oxygen uptake, V̇O 2peak ) of 56.64 ± 6.43 mL min −1 kg −1 (Table S1). The collected samples were analyzed via state-of-the-art liquid-chromatography mass-spectrometry (LC-MS/MS) and spectral flow-cytometry. Comparable to other studies in the field of immunology, our proteomics analysis yielded an excellent coverage with a total of 7,385 identified and 6,759 quantified proteins. So far, large-scale proteomic analyses on immune cells have mostly been applied in animal models 6 or on resting samples from healthy donors, 7 making our study the first to apply these methods in a randomized interventional setting with repeated baselines. After data preprocessing, our dataset comprised 6,039 proteins across 23 participants in 2 exercise conditions with 3 measurement timepoints, respectively. This makes our dataset the largest immune cell proteomics data source available in exercise context to date (Fig. 1 B). Immune cell phenotyping by spectral flow cytometry was performed on a total of 3,537,855 vital lymphocytes (Table S2) to assess exercise-induced shifts within the PBMC compartment (Figure S1A). So far, this has been disregarded in exercise studies applying omics approaches on immune cells. 4 Immune cell mobilization and redistribution is independent of exercise intensity The mobilization and redistribution of immune cells in response to acute exercise is one of the core phenomena of exercise immunology and it is nowadays agreed upon that the recovery phase following exercise is characterized by a transmigration of lymphocytes from the bloodstream into peripheral tissues, with crucial implications in many disease settings, including anticancer immunity, 8 , 9 and immunological defense. 10 , 11 A remaining topic of debate, however, is whether exercise intensity influences the magnitude of immune cell mobilization since previous studies on this topic were matched for exercise duration, but not workload. 12 , 13 Thus, before dissecting proteomic alterations of PBMCs in response to exercise, we aimed to clarify whether immune cell kinetics differ in dependence on exercise intensity, since this would lead to a different composition of our PBMC samples in response to HIIE and MICE. By applying unsupervised immune cell clustering using self-organizing maps (SOM) we identified 6 main clusters in our PBMC samples, which were mapped to the corresponding immune cell populations based on their marker expression. Visual inspection of the identified clusters (Fig. 1 C) and quantification of exercise-induced cluster shifts resulted in a similar distribution pattern of immune cell clusters between HIIE and MICE with a mean delta of 0.004 ± 0.9% (Fig. 1 D; Table S2). Confirming these findings, absolute numbers of immune cell populations did not reveal time × condition interaction effects (Figure S1B; Table S3) and the proportional contribution of each immune cell population to the PBMC compartment was similar between HIIE and MICE (Fig. 1 E; Table S4). This suggest that exercise triggers similar mobilization and redistribution patterns independent of exercise intensity and indicates that exercise-induced changes in PBMC composition do not differ between HIIE and MICE. See also Figure S1 and Tables S2, S3, and S4. Measures of variability indicate high reliability of the generated proteomics dataset Inter-individual variability of all quantified proteins resulted in median coefficients of variation (CV) of < 5% for all measurement timepoints and conditions (Fig. 2 A). This is considerably lower than in other proteomics studies in exercise context 4 , 14 and underlines the homogeneity of our study population and the analytic quality of our proteomics pipeline. The applied crossover design additionally enabled us to calculate intra-individual protein variability. The overall mean difference between the two baselines amounted to 0.13 ± 0.75% for females and 0.06 ± 0.59% for males (Fig. 2 B). To assess variability on a per-protein level, we combined multiple measures of variability (i.e., mean CV at baseline, mean CV in response to exercise, and mean difference between the two baselines) into an integrated variability score. 99.34% of all quantified proteins revealed a proteomic variability of < 10% and 83.34% achieved a score of < 5% (Fig. 2 C, S2 A, and S2 B). In summary, the low variability emphasizes the high quality of our study setup, making our generated proteomics dataset highly reliable. Acute exercise alters the immune cell proteome To obtain first insights into the proteomic alterations induced by exercise, we performed principal component analysis (PCA). Visual inspection of the PCA suggested that the variation within our samples was mainly accounted for by measurement timepoints but not exercise condition per se (Fig. 2 D). PCA also suggested that sex and intervention day had little impact on the variation of our data (Figure S2C). Performing PCA separated by condition and measurement timepoint indicated that HIIE accounted for more variation 1h after exercise than MICE (Figure S2D and S2E). Next, we compared the impact of HIIE and MICE on proteomic alterations in PBMCs using linear mixed models. We identified 1,408 time effects, 119 group effects, and 27 time × group interaction effects (Fig. 2 E). Including sex as a fixed effect in our analysis did not yield significant results. Dissection of the obtained results revealed more time, group, and time × group interaction effects 1h after exercise compared to immediately after exercise and more alterations in HIIE compared to MICE (Fig. 2 E). In detail, HIIE was marked by 1,377 significantly altered proteins, while MICE only caused significant alterations in 64. The fact that immune cell counts and proportions did not differ between HIIE and MICE 1h after exercise (Fig. 1 E and S1 B), rules out the possibility that the proteomic differences are caused by a distinct PBMC composition. In line with our results obtained by PCA, this suggests that HIIE leads to a more profound reorganization of the PBMC proteome compared to MICE. Proteomic alterations differ between HIIE and MICE To evaluate proteins that were distinctly regulated by HIIE compared to MICE, we first dissected the interaction effects of our statistical analysis (Table S5). Immediately after exercise, we observed 5 proteins with distinct kinetics in HIIE compared to MICE (Fig. 2 F). Among these proteins, synaptotagmin-like protein 2 (SYTL2), a crucial contributor to cytotoxic granule exocytosis in lymphocytes, 15 displayed a strong increase in response to HIIE, while it remained unaltered in MICE. Similarly, bone marrow stromal antigen 2 (BST2) – known for its role in blocking virus release from infected cells 16 – increased in response to HIIE but decreased in MICE (Fig. 2 F and 2 G). This gives first insights into the immunomodulatory potential of HIIE and suggests immunological adaptions dependent on exercise intensity immediately after exercise. In the recovery phase after exercise, we observed 25 interaction effects. Hierarchical clustering yielded two major clusters of proteins marked by opposed kinetics in HIIE compared to MICE (Fig. 2 F and 2 G). For instance, toll-like receptor 1 (TLR1), BST2, and cluster of differentiation 302 (CD302) were increased 1h after HIIE but decreased in MICE. TLR1 is the most abundantly expressed TLR on NK cells 17 and serves as membrane-bound pattern recognition receptor for microbial lipopeptides that triggers cytokine production and NK cell cytotoxicity upon stimulation. 18 , 19 Several studies have demonstrated that TLR1 is crucial for antimicrobial defense, 20 , 21 suggesting that exercise-induced increases in TLR1 might reinforce NK cell-mediated immunity against invading pathogens. Of note, BST2 was the only protein that continued to increase from post exercise to 1h post exercise in HIIE, suggesting sustained intensity-dependent adaptions in immunological defense. In contrast, proteins such as SH2 domain-containing protein 1B (SH2D1B), which serves as a cytoplasmic adapter regulating NK cell effector functions, 22 or asparagine synthetase (ASNS), which was previously shown to regulate CD8 + T cell activation, differentiation, and effector function 23 , 24 were marked by a decrease in the recovery period following HIIE, while they remained unaltered or increased in MICE (Fig. 2 F and 2 G). Of note, our statistical analysis also yielded several group differences between HIIE and MICE (Fig. 2 H and 2 I, Table S5). In summary, our results suggest that the recovery phase following HIIE is marked by more profound alterations of the immune cell proteome compared to MICE. We provide evidence that several proteins related to immune effector function are differentially expressed over time between HIIE and MICE. Against the backdrop of our flow cytometry results, these effects occur despite identical immune cell mobilization patterns between the two exercise conditions. See also Figure S2 and Table S5. Exercise reshapes the immune cell proteome towards effector function To add a functional dimension to our results, we made use of the Gene Ontology (GO) Resource. 25 GO over-representation analysis yielded 27 enriched GO terms in HIIE and 9 enriched GO terms in MICE. Interestingly, enriched GO terms were centered around immune effector functions in both HIIE and MICE with biological processes like “disruption of cell in another organism” or “killing of cells of another organisms” yielding high enrichment (Fig. 3 A). For proteins altered by MICE, GO over-representation analysis additionally yielded several biological processes related to lymphocyte effector function, such as “lymphocyte mediated immunity” or “T cell mediated immunity”. Given that the over-representation analysis was conducted with much more proteins for HIIE, we additionally identified several cellular components and molecular functions in this analysis. Semantic evaluation of the identified GO terms underlined their association with immune effector function. For instance, “exogeneous protein binding”, “virus receptor activity”, and “endopeptidase activity” are known molecular functions in the context of immunological defense against viruses. 26 , 27 Similarly, “proteasome core complex” and “peroxisomal membrane” depict cellular components associated with such molecular function and biological processes (Fig. 3 A). Collectively, our GO over-representation analysis points towards enhanced regulation of immune effector functions in response to both HIIE and MICE (Table S6). Time-resolved protein changes differ between HIIE and MICE Within all proteins altered by exercise (n = 1,408), we found 1,344 proteins that were uniquely altered by HIIE, 31 proteins that were uniquely altered by MICE, and 33 proteins that were altered by both exercise conditions (Fig. 3 B). Analysis of proteins altered by HIIE suggested two major protein clusters that were characterized by increased or decreased protein abundance 1h after HIIE compared to baseline (Fig. 3 C). Considering the large number of proteins altered by HIIE compared to MICE, we took different approaches in analyzing the time effects of each exercise condition. For proteins altered by both exercise conditions, and proteins uniquely altered by MICE we performed hierarchical clustering to identify proteins displaying similar kinetics. Interestingly, when analyzing proteins that were altered by both exercise conditions, hierarchical clustering yielded 2 major protein clusters: one cluster containing proteins with similar kinetics between HIIE and MICE and one cluster containing proteins with different kinetics (Fig. 3 D). In absolute terms, most of the proteins responded similarly with only 4 proteins showing higher values in HIIE, including the antiviral protein BST2, which we previously identified in our analysis of time × group interaction effects (Fig. 2 F). Additionally, many of the proteins that were shared between HIIE and MICE were associated with immune effector functions, suggesting a shared regulation of several immunological processes by exercise. Examples of such proteins include granzymes (e.g., GZMB, GZMH, GZMM), perforin-1 (PRF1), or guanylate-binding proteins 5 (GBP5; Fig. 3 D). Similarly, hierarchical clustering of proteins uniquely altered by MICE identified 2 major clusters that separated proteins that decreased in response to MICE from proteins that increased, while showing no alterations in HIIE, respectively (Fig. 3 E). In line with the observed time × group interaction effects (Fig. 2 F) most proteins displayed lower abundance in response to MICE (Fig. 3 E). Of note, although the lower number of time effects and the decreased abundance of many proteins might suggest reduced immune effector functions in response to MICE, it is crucial to emphasize that several proteins with immunological functions, especially those jointly regulated between HIIE and MICE, revealed increased abundance in response to MICE as well. Thus, while our data suggests that the immunoproteomic impact of MICE seems to be less pronounced than that of HIIE, there is no conclusive evidence suggesting reduced immune effector function per se in response to MICE. See also Table S5 and S6. Identification of shared and unique immune effector functions regulated by HIIE To dissect the proteomic alterations in response to HIIE, we performed fuzzy c-means clustering and mapped the altered proteins (n = 1,377) to four distinct clusters by means of their relative membership (Fig. 4 A; Table S7). The four identified clusters confirmed what hierarchical clustering had previously suggested, i.e., two major clusters marked by increased or decreased protein abundance 1h after HIIE (Fig. 3 C and 4 A). We next leveraged biological theme comparisons 28 and identified 576 biological processes, 132 molecular functions, and 187 cellular components associated with the proteins altered by HIIE (Table S8). By generating gene-concept networks of the five most significant GO terms in each ontology, we observed both, shared and unique GO terms across our four protein clusters (Figure S3A – C). To quantify functional differences and similarities between the identified protein clusters, we performed gene set enrichment analyses (GSEA) 29 and observed a total of 169 enriched GO terms (Fig. 4 B and Table S9). Interestingly, cluster 4 did not yield any enriched GO terms and re-evaluation of the underlying statistics demonstrated that the individual GO terms did not reach the significance threshold. These findings were validated using the STRING resource. 30 We then focused our attention on GO terms that were shared across protein clusters 1–3 and obtained 11 shared biological processes. Semantic evaluation confirmed their close connection to immune function, as exemplified by GO terms like “cell killing”, “leukocyte activation”, or “defense response” (Fig. 4 B). Analysis of the underlying proteins resulted in a core proteome consisting of 369 proteins, most of which changed in abundance in the recovery phase following HIIE (Fig. 4 C). This suggests that the biological processes regulated by HIIE are driven by proteomic alterations in the recovery phase. We observed similar results for the 27 GO terms that were shared between clusters 1 and 2, and the 15 GO terms that were shared between clusters 2 and 3 (Fig. 4 B and S4 A-C). Concerning GO terms that were uniquely enriched in a specific protein cluster, we identified 29 GO terms uniquely enriched in cluster 1, 62 GO terms uniquely enriched in cluster 2, and 25 GO terms uniquely enriched in cluster 3 (Fig. 4 B and D). Among the GO terms enriched in cluster 1 we found enriched regulation of “endopeptidase activity” and “cellular response to organic substance” (Fig. 4 D). Similarly, cluster 2 demonstrated enriched regulation of “glycosaminoglycan binding”, and “cell migration”, which are crucial processes in the context of exercise-induced transmigration of immune cells from the bloodstream into peripheral tissues. In contrast, cluster 3 was characterized by a decreased regulation of several cellular components and biological processes 1h after exercise, which can be attributed to the underlying protein kinetic (Fig. 4 A). In summary, GSEA suggested a profound regulation of immune effector processes in the recovery phase following HIIE, which were in part shared and in part unique for specific protein kinetics. See also Figure S3 and S4, and Table S9. Identification of an immunoproteomic signature associated with cardiorespiratory fitness We ultimately leveraged our generated dataset to enable deeper insights into long-term adaptions to exercise training. Taking a data-driven approach, we started by pooling the baseline data of all our analyses, including participant characteristics as well as flow cytometry and LC-MS/MS results. This comprehensive dataset was then used to investigate potential pairwise associations with V̇O 2peak , a gold standard marker of cardiorespiratory fitness that is highly responsive to exercise training. In a first step we calculated Spearman’s rank correlation coefficients (r S ) and selected features that displayed moderate to high correlation (r S > 0.4 or < -0.4) with V̇O 2peak . This resulted in a reduction of our dataset from 6,063 to 260 features (Table S10). To establish an elaborate connection between these features and cardiorespiratory fitness, we next performed prediction analyses. Ridge regression yielded an R-squared of 0.61 and a mean squared error of 14.1 and visual inspection of the ranked coefficients revealed a homogeneous distribution of features with positive or negative impact on V̇O 2peak prediction, respectively (Fig. 5 A). By evaluating the 20 features with the highest predictive power (Fig. 5 B), we observed that nicotinamide phosphoribosyltransferase (NAMPT), a key enzyme of nicotinamide adenine dinucleotide (NAD + ) metabolism, demonstrated the highest positive impact on V̇O 2peak prediction. NAMPT plays a crucial role in salvaging intracellular NAD + and was previously shown to be exercise-responsive in skeletal muscle, 31– 33 but also other target tissues like immune cells. 34– 36 Our results support this notion and suggest that repeated exposure to exercise, which results in greater cardiorespiratory fitness, equips immune cells with a higher metabolic capacity, thereby linking to the immune effector functions previously identified in this work. Additionally, several studies have suggested a direct antiviral function of NAMPT in host defense. 37, 3 8 Similarly, we observed a positive impact on V̇O 2peak prediction for succinate receptor 1 (SUCNR1), a G-protein coupled receptor that was previously shown to control exercise capacity and systemic glucose homeostasis in mice. 39, 40 There are several reports that the effects of exercise-secreted succinate on skeletal muscle tissue adaptions are dependent on paracrine signaling to non-myofibrillar cells such as macrophages, that express SUCNR1. 40, 41 Besides features with positive impact on V̇O 2peak prediction, our analysis also yielded several proteins with a negative impact (Fig. 5 B). Among these features phosphatidylserine decarboxylase (PISD), an enzyme involved in lipid droplet biogenesis, 4 2 and caspase recruitment domain family, member 8 (CARD8), a pattern recognition receptor that regulates inflammasome activation and production of pro-inflammatory cytokines, 4 3 stood out due to their involvement in metabolism and inflammation. The negative impact of PISD suggests reduced cellular fat deposition with increasing cardiorespiratory fitness. Regarding CARD8, the negative impact on V̇O 2peak prediction might be explained by anti-inflammatory adaptions with higher cardiorespiratory fitness. 3, 4 4 Interestingly, besides correlations with V̇O 2peak , we also observed various inter-feature correlations (Fig. 5 C). Collectively our results suggest that the identified proteins associated with cardiorespiratory fitness reshape the phenotype of immune cells in response to exercise training. In a broader context these findings might serve as a molecular foundation for immunological health in the context of long-term training adaptions. Discussion A better understanding of the molecular underpinnings of physical exercise is needed to individualize exercise training recommendations and maximize their efficacy in mediating health benefits. While some human studies have addressed exercise responses in skeletal muscle and blood plasma using state-of-the-art systems biology approaches, 45 – 49 the impact of exercise on the immune system is less well understood. Here, we provide a comprehensive resource on how two different aerobic exercise stimuli rewire the proteomic makeup of PBMCs. We applied a robust randomized crossover design, including two standardized baseline measurements in a large sample size for proteomic analyses in humans. Our findings expand the literature by > 1000 proteomic changes in response to acute exercise in PBMCs. Particularly, immune effector function and cell activation pathways are regulated, and higher intensity is needed to stimulate these changes. Finally, we demonstrate that baseline proteomics data can predict V̇O 2peak and identify potential exercise-responsive targets in PBMCs that warrant further investigation. The acute exercise-induced mobilization of effector cells like NK cells and cytotoxic (CD8+) T cells is well-investigated 3 , 50 but less is known on changes in their proteomic makeup and the resulting cell functions. We identified comprehensive alterations in the immune cell proteome associated with cell function and activation pathways that match previous studies evaluating functional outcomes in different effector populations. 51 – 53 The regulation of immune effector functions suggests a transient state of immunomodulation following acute bouts of exercise, thereby elucidating the mechanisms of action underlying the benefits of exercise training for disease prevention. Interestingly, our results indicate that the observed changes in the proteomic makeup of PBMCs occur independent of exercise-induced mobilization and redistribution of immune cells. This is demonstrated by the fact that we observed far more proteomic alterations in response to HIIE compared to MICE, although the underlying PBMC composition did not differ between the two exercise conditions. Although previous investigations have neglected this crucial component of exercise immunology, our proteomics results are temporally in line with transcriptomic alterations identified before. 4 In this context, our open source web application, which can be found at https://sportsmedicine-dortmund.shinyapps.io/beat , offers the opportunity to mine the underlying dataset for specific proteins of interest, thus informing new hypothesis-driven research in the field of exercise immunology. Our results support the WHO recommendations on physical activity, which highlight the superior role of high exercise intensity for health promotion. 5 From an immunological perspective, we found distinct responses of HIIE and MICE when matching the interventions for duration and workload and thus conclude that exercising at higher intensity is crucial to induce more profound changes in the PBMC proteome. This might serve as a potential biological foundation for a recent comprehensive analysis revealing a superior effect of exercise intensity versus volume on longevity at a population-based level. 54 Finally, while previous work has elucidated the molecular underpinnings of cardiorespiratory fitness, 4 , 47 , 49 a possible link to immune cells has not yet been explored. We observed strong associations with V̇O 2peak for several proteins including NAMPT, which is crucial for cellular energy metabolism. Confirming these findings, we have recently demonstrated that NAMPT expression of human PBMCs increases in response to acute exercise. 3 5 Overall, this suggests an interrelation between acute exercise stimuli, immunometabolic competence, and cardiorespiratory fitness and suggests a putative role of PBMCs as peripheral mirror for systemic health. In conclusion, we identified > 1000 exercise-induced alterations in the PBMC proteome and provide a valuable data resource for future research. The identified changes were particularly related to immune effector function, serving as a mechanistic link for the preventive and therapeutic impact of regular exercise. In line with the WHO 2020 guidelines on physical activity, acute exercise at higher intensity elicited greater changes in the regulation of cell function and activation pathways, providing supportive biological evidence for the relevance of exercise intensity as an important factor when planning and structuring exercise training programs for health promotion. Finally, the associations between the PBMC proteome and V̇O 2peak shed light on potential molecular mediators of immunological health. Methods Participant recruitment and informed consent Prior to enrollment of the first participants the study received ethical approval by the local ethics committee of the German Sport University Cologne, which works according to the World Medical Association’s Declaration of Helsinki. The study meets the National Institutes of Health definition of a clinical trial and was prospectively registered in the German Clinical Trials Register (DRKS00017686). Study eligibility was assessed for 28 healthy adults aged between 18 and 35. To ensure complication-free execution of the high-intensity interval exercise on the treadmill, participants required a weekly running volume of 2–5 hours and a body mass index < 30. Any previous medical history of muscle disorders, cardiac or kidney diseases as well as regular intake of medication or nutritional supplements were treated as exclusion criteria. For female participants, breast-feeding or an ongoing pregnancy were also treated as exclusion criteria. Of the 28 subjects assessed for eligibility, two were considered ineligible due to acute infections. The remaining 26 participants provided written informed consent and were enrolled in the study. After baseline testing two further participants dropped out due to orthopedic problems while running (Achilles injuries). For one participant, biomaterial did not suffice to run analyses, which resulted in a total of 23 participants. An overview of all participant characteristics is displayed in Table S1. Study design Participants enrolled in this randomized crossover study were scheduled for three visits to an exercise physiology laboratory of the German Sport University Cologne: Baseline testing, a HIIE session, and a MICE session. For each visit, participants were asked to arrive overnight-fasted and refrain from alcohol and caffein intake in the 24h prior. Water intake was permitted ad libitum. All visits were scheduled between 07:00 and 10:00 am to account for a potential circadian impact on performance and biological outcomes. The minimum timeframe between each of the three visits was 72 hours, to prevent potential carryover effects. Baseline testing During baseline testing, written informed consent was obtained from participants and demographic, and anthropometric characteristics were recorded. Afterwards, participants underwent cardiopulmonary exercise testing. Cardiopulmonary exercise test (CPET) To standardize the exercise intensity between participants for the HIIE and MICE session, respectively, cardiorespiratory fitness was assessed as peak oxygen consumption (V̇O 2peak ) in an incremental CPET during baseline testing. The CPET was performed on a motorized treadmill (Woodway ELG 90, Weil am Rhein, Germany) that was set to 1% incline for all sessions. The warm-up consisted of 5 min at 6–8 km h −1 . Afterwards, participants began running at 8 km h −1 . The speed of the treadmill was then increased by 1 km h −1 every 60 seconds until participants reached volitional exhaustion. During the test, heart rate was recorded continuously (Polar FS1C, Kempele, Finland), and rate of perceived exertion was recorded prior to each increase in intensity. Participants were verbally encouraged to continue running by the supervising researcher. After reaching volitional exertion, participants were given a 5 min break before taking up exercise again for a V̇O 2peak verification test. For this test, the speed of the treadmill was set 1 km h −1 higher than what the participants had finished with. Just before the verification test, participants ran for 3 min at 8 km h −1 . The speed was then increased to the target speed within 20 seconds and participants were instructed to run as long as possible. During the entire CPET participants wore a face mask that was connected to a spirometer (Cortex Metalyzer 3B, CORTEX Biophysik GmbH, Leipzig, Germany) to collect breathing gases breath-by-breath. The highest 15-second interval during the CPET was used to calculate V̇O 2peak . Randomization To prevent sequence effects arising from the order in which HIIE and MICE were conducted, participants were randomized into one of two exercise intervention sequences after baseline testing: HIIE-MICE or MICE-HIIE. Following the minimization procedure by Pocock and Simon, 55 randomization was performed via concealed allocation (1:1) using the software Randomization in Treatment Arms (RITA; Evidat, Lübeck, Germany). Age, BMI, and V̇O 2peak were used as stratification factors. The intervention sequences did not differ in terms of participant characteristics, indicating that our randomization was unbiased (Table S1). Exercise interventions Exercise intensities for the HIIE and MICE session were calculated as percent of V̇O 2peak for each participant to ensure that all participants exercised at the same intensity. The exercise protocols for HIIE and MICE were designed in an time- and workload-matched manner as previously described 5 6 , 5 7 to isolate exercise intensity as the only differing variable between the two exercise conditions. This time- and workload-matched design is crucial to draw unbiased conclusions on the impact of exercise intensity. Both exercise sessions were performed on the same treadmill that was also used for the CPET at baseline (Woodway ELG 90, Weil am Rhein, Germany). During MICE participants performed a warm-up for 10 min at a self-selected intensity, followed by a 5 min break. Participants then ran for 50 min at 70% of their V̇O 2peak . During HIIE, participants performed 7 min of warm-up and cool-down at 70% V̇O 2peak with six bouts of high-intensity running at 90% V̇O 2peak in between. Each high-intensity bout lasted 3 min, followed by 3 min of active recovery at 50% V̇O 2peak . Blood collection and sample preparation Blood was drawn from a median antecubital vein in supine position at baseline, immediately after exercise, and 1h after exercise for the HIIE and MICE session, respectively. Each blood draw consisted of 24 mL of whole blood collected in EDTA tubes (Vacutainer, BD). After the last blood draw, peripheral blood mononuclear cells (PBMCs) were isolated via density gradient centrifugation. To achieve this, whole blood was first diluted with phosphate buffered saline (PBS) and then carefully layered on top of a lymphocyte separation medium (Cytiva Ficoll-Paque™ PLUS, Fisher Scientific). After centrifugation for 30 min at room temperature and 800 g − 1 , the PBMC-containing interphase was collected, washed with PBS, and centrifuged again for 10 min at room temperature and 800 g − 1 . PBMCs were then resuspended in freezing medium (Recovery™ cell culture freezing medium, Thermo Fisher Scientific) and stored at -80°C before being transferred to a -150°C freezer on the next day until further analysis. Flow cytometry Sample preparation and data acquisition Flow cytometry analysis was performed using a Cytek® Aurora full spectrum flow cytometer (Cytek Biosciences, California, USA). Cryopreserved PBMCs were gently thawed in a water bath at 37°C with a mean recovery of 81.28% viable cells assessed with the Zombie NIR™ Fixable Viability Kit (BioLegend, San Diego, CA, USA). After incubating 1 × 10 6 PBMCs in 2.5 µg Fc block for 10 min at room temperature, cells were stained with anti-CD3 (BUV395, clone SK7), anti-CD4 (PerCP, clone SK3), anti-CD8 (BV750, clone SK1), anti-CD16 (PE-Cy7, clone 3G8), anti-CD25 (BUV805, clone M-A251), anti-CD56 (BUV563, clone NCAM16.2), anti-CD20 (APC, clone L27), and anti-CD19 (APC, clone SJ25C1) antibodies (all from BD Biosciences, NJ, USA). In brief, cells were incubated in the dark with a master mix containing Brilliant Stain buffer (BD Biosciences) and antibodies against surface antigens for 30 min at 4°C. After washing with FACS buffer, the BD Pharmingen™ฏ Transcription Factor Buffer Set was used, and cells were fixed for 40 min at 4°C in the dark. Thereafter, intracellular staining was done by incubating cells with an anti-Foxp3 antibody (PE, clone 259D/C7) for 45 min at 4°C in the dark. After washing, cells were resuspended in FACS buffer and acquired on the flow cytometer within 2 hours after finishing the staining protocol. Data processing Gating was performed using FlowJo™ 10.10.0 (Fig. 1 C). B cells were phenotyped as CD3 − CD56 − CD19 + CD20 + , Natural Killer T (NKT) cells as CD3 + CD56 + , Natural Killer (NK) cells either as CD56 bright CD16 − (NK bright ) or CD56 dim CD16 + (NK dim ), and regulatory T cells (T regs ) as CD4 + CD25 + Foxp3 + . The person analyzing the samples was blinded to the participants’ group allocation. Analysis of total blood cell counts was performed from EDTA blood using a hematology analyzer (SYSMEX XP-300, Norderstedt, Germany). The lymphocyte count was then used to calculate the absolute number of peripherally circulating lymphocyte subsets according to the cell proportions derived by flow cytometry. LC-MS/MS-based untargeted proteomics Sample were processed and measured in a block randomized order 58 to prevent any technical bias that might occur during sample preparation or LC-MS/MS measurement. Sample preparation Isolated PBMCs were lysed in a RIPA buffer containing 10 mM sodium fluoride, 1 mM sodium orthovanadate, cOmplete™ Protease Inhibitor Cocktail (Merck KGaA), PhosSTOP™ (Merck KGaA), 250 U mL − 1 benzonase, and 10 U mL − 1 DNase I. Samples were incubated on ice for 1h and then centrifuged at 4°C and 13,000 g − 1 for 15 min. Protein concentration was determined in the supernatant with a BCA assay. An amount of 10 µg of protein per sample was digested (Trypsin) using an AssayMAP Bravo liquid handling system (Agilent technologies) running the autoSP3 protocol. 59 After sample preparation the remaining peptides were vacuum dried and stored at -20°C until LC-MS/MS analysis. MS method Orbitrap Exploris 480 The dried peptide sample was reconstituted (97.4% Water, 2.5% Hexafluoro-2-propanol and 0.1% trifluoroacetic acid (TFA)) and 10% of the sample were used. The LC-MS/MS analysis was carried out on an Ultimate 3000 UPLC system (Thermo Fisher Scientific) directly connected to an Orbitrap Exploris 480 mass spectrometer for a total of 120 min. Peptides were online desalted on a trapping cartridge (Acclaim PepMap300 C18, 5 µm, 300 Å wide pore; Thermo Fisher Scientific) for 3 min using 30 µl/min flow of 0.1% TFA in water. The analytical multistep gradient (300 nL/min) was performed using a nanoEase MZ Peptide analytical column (300Å, 1.7 µm, 75 µm x 200 mm, Waters) using solvent A (0.1% formic acid in water) and solvent B (0.1% formic acid in acetonitrile). For 102 min the concentration of B was linearly ramped from 4–30%, followed by a quick ramp to 78%, after two min the concentration of B was lowered to 2% and a 10 min equilibration step appended. Eluting peptides were analyzed in the mass spectrometer using data independent acquisition (DIA) mode. A full scan at 120 k resolution (380–1400 m/z, 300% AGC target, 45 ms maxIT) was followed 47 DIA windows. The DIA acquisition covered a mass range of 400–1000 m/z using windows of a variable width with 1 m/z overlap, an AGC target of 1000% with a maxIT set to 54 ms and recorded at a resolution of 30 k. Each sample was followed by a wash run (40 min) to avoid carry-over between samples. Instrument performance and suitability was monitored by regular (approx. one per 48 hours) injections of a standard sample and an in-house shiny application over the whole timeline of the experiment. Data analysis Analysis of DIA RAW files was performed with Spectronaut (Biognosys, version 19.1.240724.62635) 60 in directDIA+ (deep) library-free mode. Default settings were applied with the following adaptions. Within DIA Analysis under Identification the Precursor PEP Cutoff was set to 0.01, the Protein Qvalue Cutoff (Run) set to 0.01 and the Protein PEP Cutoff set to 0.01. In Quantification the Proteotypicity Filter was set to Only Protein Group Specific, the Protein LFQ Method was set to MaxLFQ and the quantification window was set to Not Synchronized (SN 17). The data was searched against the human proteome from Uniprot (human reference database with one protein sequence per gene, containing 20,597 unique entries from ninth of February 2024) and the contaminants FASTA from MaxQuant (246 unique entries from twenty-second of December 2022). Data processing Before further analysis, the obtained dataset was checked for proteins that were identified more than once. For these duplicate results, the event with the highest number of identified precursors across all samples was kept and all other events were deleted from the dataset. The data was then filtered for proteins that were quantified in ≥ 70% of the samples in at least one exercise condition and measurement timepoint (i.e., HIIE/MICE baseline, post exercise, 1h post exercise). Subsequently, we imputed the data separated by exercise condition and measurement timepoint using the missForest package. 61 Ultimately, proteins were annotated to match the gene names provided in the org.Hs.eg.db package for subsequent Gene Ontology (GO) analysis. Translation between gene names and Entrez gene identifiers was accomplished with the bitr function from the ClusterProfiler package. 28 , 62 Quantification and statistical analysis Samples from a total of 23 participants were available for statistical analyses. For one participant there was no sample from 1h after MICE due to difficulties during PBMC isolation. Statistical analysis and visualization were performed in R. If not otherwise noted, data wrangling was achieved using the dplyr 63 and tidyr 64 package and subsequently visualized with ggplot2 65 and ggpubr 66 . Unsupervised immune cell clustering using self-organizing maps Flow cytometry data of each sample was cleaned using the FlowAI plugin (v3.2.3) in FlowJo™ 10.10.0. The remaining events were gated as described above and live cells were downsampled to 3,000 events per sample using the DownSample plugin (v3.3.1). Subsequently, downsampled events were concatenated to obtain an overall dataset containing all exercise conditions (HIIE, MICE) and measurement timepoints (baseline, post exercise, 1h post exercise). This dataset was then used to perform unsupervised immune cell clustering using self-organizing maps (SOM) with the FlowSOM plugin (v4.1.0). The resulting 6 clusters were identified as CD4 + T cells, CD8 + T cells, NKT cells, CD56 dim cells, CD56 bright cells and B cells in the build-in Cluster Explorer in FlowJo™ 10.10.0. The overall dataset was visualized using Uniform Manifold Approximation and Projection (UMAP) via the UMAP plugin (v4.1.1) and FlowSOM clusters were superimposed via color-coding. This overall, color-coded immune cell clusters were used as a template map for subsequent clustering per exercise condition and measurement timepoint. To achieve this, the downsampled events were concatenated for baseline, post exercise and 1h post exercise in HIIE and MICE, respectively. Ultimately, FlowSOM clustering and UMAP were performed on each of these concatenated dataframes by applying them on the previously generated map. Exercise-induced mobilization of immune cells Exercise-induced alterations in immune cell counts were analyzed by applying linear mixed models to the flow cytometry results. Measurement timepoint and exercise condition were implemented as fixed effects and participant ID as random effect using the lmer function from the lme4 package. 67 Results of the linear mixed models were then analyzed for time and time × condition interaction effects via analyses of variance (ANOVAs) with the built-in ANOVA function from R stats. In case of significant results, pairwise comparisons of measurement timepoints and/or exercise conditions were performed by applying the emmeans function from the emmeans package. P-values were Bonferroni-corrected for multiple testing. Measures of variability All measures of variability were calculated with unimputed data to avoid potential bias arising from imputation. Inter-individual variability was calculated as coefficient of variation (CVs) for each protein across all participants separated by exercise condition (HIIE, MICE), and measurement timepoint (baseline, post exercise, 1h post exercise). CVs were calculated as the ratio of the standard deviation σ to the mean µ . Intra-individual variability was assessed by comparing the baseline values of the two intervention days. Relative differences between day 1 and day 2 were calculated (in percent) for each protein separated by study participant. Proteomic variability was quantified for each protein by calculating (i) the mean CV across all participants in HIIE and MICE at baseline, (ii) the mean CV across all participants in HIIE and MICE post exercise and 1h post exercise, and (iii) the mean difference between the two baselines across all participants. Proteomic variability was also calculated separated by exercise conditions and measurement timepoints (see Figure S2A and S2B). Principal component analysis Principal component analysis (PCA) was carried out using the built-in prcomp function from R stats. All samples were plotted with the fviz_pca_ind function from the factoextra package. Exercise condition, measurement timepoint, intervention day, and sex were used as metadata to color-code PCA results. PCAs were also computed on datasets separated by exercise condition or measurement timepoint to visually assess the impact of these variables on each other (see Figure S2C-E). Linear mixed models to identify proteins altered by HIIE and/or MICE To identify proteins altered by HIIE and/or MICE, a linear mixed model was fitted on the log 2 -transformed, normalized, and imputed protein intensities via the limma R package. 68 Intra-individual correlation was estimated via the duplicateCorrelation function. The model included the exercise condition (HIIE, MICE), the measurement timepoint (baseline, post exercise, 1h post exercise), and the interaction between both as fixed factors. A moderated t statistic 69 was obtained for each contrast of interest via the eBayes function with estimated variance trend and robustification. The resulting p-values for each contrast were adjusted with the Benjamini-Hochberg procedure 70 to control the false discovery rate and significance was declared at the adjusted 5% two-sided level. The model was subsequently extended to include sex and all two-way interactions. Gene ontology (GO) over-representation analysis Time effects of the statistical analysis with limma 68 were used to map proteins that were significantly altered by HIIE and MICE to GO terms, respectively. GO over-representation analysis was performed with the ClusterProfiler package. 28 , 62 For HIIE and MICE, significantly altered proteins were compared with the entire dataset of quantified proteins applying Benjamini-Hochberg correction of p-values with a p-value cutoff of 0.05 and a q-value cutoff of 0.2. Fuzzy c-means clustering Fuzzy c-means clustering was performed with the Mfuzz package. 71 , 72 Data was standardized using the standardise function and the optimal number of clusters was determined by calculating the minimum centroid distance for a range of cluster numbers using the Dmin function. The optimal fuzzifier was identified with the mestamiate function. Biological theme comparison Biological theme comparison was carried out using the compareCluster function from the ClusterProfiler package. 28 , 62 Entrez gene identifiers of the proteins contained in the identified clusters were used as input with the function command set to “enrichGO”. Benjamini-Hochberg correction was applied to p-values with a cutoff of 0.05 and minimum gene set size was set to 10. The results were simplified via the simplify function using a cutoff of 0.7 and visualized separated by ontology with the cnetplot function from the enrichplot package. 73 Gene ontology (GO) gene set enrichment analysis Gene set enrichment analysis was performed using the gseGO function from the ClusterProfiler package. 28 , 62 Entrez gene identifiers and fold changes from baseline of the proteins contained in the identified clusters were used as input with the minimum gene set size set to 10. In case fold changes were only positive or negative, the “scoreType” command was set to “pos” or “neg”, respectively. P-values were corrected using the Benjamini-Hochberg procedure with a p-value cutoff of 0.05. The underlying proteins mapping to each significant GO term were identified using the select function from the AnnotationDbi package. Shared and unique GO terms across the identified clusters were visualized with the UpSetR package. 74 V̇O 2peak prediction and correlation network analysis Preselection of features To identify features with high association to V̇O 2peak , we conducted a preselection in Python (v.3.9). 7 5 The features were standardized using z transformation and included the average of mass spectrometry-based proteomics data and flow cytometry-based immunophenotyping data at baseline of intervention day 1 and 2 as well as sex, height, weight and BMI. V̇O 2peak was scaled to body weight. Data from 2 participants were excluded from the analysis due to incomplete feature sets. Pairwise Spearman's rank correlations between all features and V̇O 2peak were calculated (Table S10) and features with a correlation coefficient of > 0.4 or < -0.4 were included in the subsequent analysis. From a total of 6,063 initial features, 260 remained after this selection. V̇O 2peak prediction modeling We ran LASSO 7 6 , 7 7 and ridge regression 7 8 as well as a random forest 7 9 as a non-linear, tree-based approach. A leave-one-out (LOO) cross-validation was performed in Python (v.3.9) to assess the predictive performance of these methods based on the 260 features. To optimize the hyperparameters for each model by grid search, a second inner cross validation was performed. For each training set, we selected the model that had the lowest test error. The predicted output value resulted from the cross validation iteration, where the corresponding output data point and its associated features were not included in the training set. These predicted values were used to calculate the mean squared error (MSE) and the r squared R 2 . Ridge regression outperformed the other models. All features with coefficients from the ridge regression are listed in the supplements (Table S10). Correlation network analysis The 20 features with the largest absolute mean value from the ridge regression were selected to create a weighted, undirected network using Spearman’s rank correlations. The network was visualized in R (v.4.4.1) with the packages Hmisc (v.5.2.1) and igraph (v.2.1.1.). Declarations Acknowledgements We want to thank Lars Donath and Ludwig Rappelt for their help in conducting the trial. We thank the team of the Proteomics Core Facility of the DKFZ particularly Adrian Stoegbauer und Alina Ertl for sample preparation and LC-MS/MS measurement. We also want to thank the German Sport University Cologne for supplying internal funds to A.J.M.. Figures were created with https://BioRender.com. Author contributions Conceptualization, N.J., A.J.M., and P.Z.; Methodology, N.J., A.J.M., A.S., and P.Z.; Software, D.W., C.We., M.S., and S.C.; Formal Analysis, D.W., S.P., C.We., M.S., and S.C.; Investigation, A.J.M., S.P., A.S., M.S., and D.H.; Resources, C.Wa., C.A.O., D.H., and P.Z.; Writing – Original Draft, D.W., N.J., S.P., C.We., M.S., and S.C.; Writing – Review & Editing, A.J.M., S.P., A.S., C.We., A.L.H., M.S., S.C., A.G, C.Wa., C.A.O., D.H., and P.Z.; Visualization, D.W. and C.We.; Supervision, P.Z., A.G., and D.H., Project Administration, P.Z.; Funding Acquisition, A.J.M Competing interests The authors declare no competing interests. RESSOURCE AVAILABILITY Lead contact Any requests for further pieces of information or resources should be directed to the Lead Contact Philipp Zimmer ( [email protected] ) Data and code availability All data associated with this article can be explored via our interactive web application at https://sportsmedicine-dortmund.shinyapps.io/beat. Raw data files of all samples processed in the proteomics analysis are hosted on the PRoteomics IDEntifications Database (PRIDE; https://www.ebi.ac.uk/pride) under the following PRIDE-ID: PXD058573. References Penedo FJ, Dahn JR (2005) Exercise and well-being: a review of mental and physical health benefits associated with physical activity. Curr Opin Psychiatry 18:189–193 Ruegsegger GN, Booth FW (2018) Health Benefits of Exercise. Cold Spring Harb Perspect Med 8:a029694 Walsh NP et al (2011) Position statement. Part one: Immune function and exercise. Exerc Immunol Rev 17:6–63 Contrepois K et al (2020) Molecular Choreography of Acute Exercise. Cell 181:1112–1130e16 Bull FC et al (2020) World Health Organization 2020 guidelines on physical activity and sedentary behaviour. Br J Sports Med 54:1451–1462 Howden AJM et al (2019) Quantitative analysis of T cell proteomes and environmental sensors during T cell differentiation. Nat Immunol 20:1542–1554 Rieckmann JC et al (2017) Social network architecture of human immune cells unveiled by quantitative proteomics. Nat Immunol 18:583–593 Campbell JP, Walzik D, Zimmer P (2024) What dictates tumour cell sensitivity to exercise? Nat Rev Immunol. 10.1038/s41577-024-01001-z Fiuza-Luces C et al (2024) The effect of physical exercise on anticancer immunity. Nat Rev Immunol 24:282–293 Campbell JP, Turner JE (2018) Debunking the Myth of Exercise-Induced Immune Suppression: Redefining the Impact of Exercise on Immunological Health Across the Lifespan. Front Immunol 9:648 Simpson RJ, Kunz H, Agha N, Graff R Exercise and the Regulation of Immune Functions. in Progress in Molecular Biology and Translational Science vol. 135 355–380 (Elsevier, 2015) Campbell JP et al (2009) Acute exercise mobilises CD8 + T lymphocytes exhibiting an effector-memory phenotype. Brain Behav Immun 23:767–775 Simpson RJ et al (2007) Apoptosis Does not Contribute to the Blood Lymphocytopenia Observed After Intensive and Downhill Treadmill Running in Humans. Res Sports Med 15:157–174 Savikj M et al (2022) Exercise timing influences multi-tissue metabolome and skeletal muscle proteome profiles in type 2 diabetic patients – A randomized crossover trial. Metabolism 135:155268 Ménasché G et al (2008) A newly identified isoform of Slp2a associates with Rab27a in cytotoxic T cells and participates to cytotoxic granule secretion. Blood 112:5052–5062 Jouvenet N et al (2009) Broad-Spectrum Inhibition of Retroviral and Filoviral Particle Release by Tetherin. J Virol 83:1837–1844 Chalifour A et al (2004) Direct bacterial protein PAMP recognition by human NK cells involves TLRs and triggers α-defensin production. Blood 104:1778–1783 Adib-Conquy M, Scott‐Algara D, Cavaillon J, Souza‐Fonseca‐Guimaraes F (2014) TLR‐mediated activation of NK cells and their role in bacterial/viral immune responses in mammals. Immunol Cell Biol 92:256–262 Noh J-Y, Yoon SR, Kim T-D, Choi I, Jung H (2020) Toll-Like Receptors in Natural Killer Cells and Their Application for Immunotherapy. Journal of Immunology Research 1–9 (2020) Lancioni CL et al (2011) Mycobacterium tuberculosis Lipoproteins Directly Regulate Human Memory CD4 + T Cell Activation via Toll-Like Receptors 1 and 2. Infect Immun 79:663–673 Triantafilou M et al (2006) Membrane Sorting of Toll-like Receptor (TLR)-2/6 and TLR2/1 Heterodimers at the Cell Surface Determines Heterotypic Associations with CD36 and Intracellular Targeting. J Biol Chem 281:31002–31011 Roncagalli R et al (2005) Negative regulation of natural killer cell function by EAT-2, a SAP-related adaptor. Nat Immunol 6:1002–1010 Gnanaprakasam JNR et al (2023) Asparagine restriction enhances CD8 + T cell metabolic fitness and antitumoral functionality through an NRF2-dependent stress response. Nat Metab 5:1423–1439 Wu J et al (2021) Asparagine enhances LCK signalling to potentiate CD8 + T-cell activation and anti-tumour responses. Nat Cell Biol 23:75–86 The Gene Ontology Consortium et al. The Gene Ontology knowledgebase in 2023. GENETICS 224, iyad031 et al (2023) Rouse BT, Mueller SN Host Defenses to Viruses. in Clinical Immunology 365–374.e1 (Elsevier, 2019). 10.1016/B978-0-7020-6896-6.00025-9 Rouse BT, Sehrawat S (2010) Immunity and immunopathology to viruses: what decides the outcome? Nat Rev Immunol 10:514–526 Yu G, Wang L-G, Han Y, He Q-Y (2012) clusterProfiler: an R Package for Comparing Biological Themes Among Gene Clusters. OMICS 16:284–287 Subramanian A et al (2005) Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl. Acad. Sci. U.S.A. 102, 15545–15550 Szklarczyk D et al (2019) STRING v11: protein–protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets. Nucleic Acids Res 47:D607–D613 Basse AL et al (2021) Nampt controls skeletal muscle development by maintaining Ca2 + homeostasis and mitochondrial integrity. Mol Metabolism 53:101271 Costford SR et al (2010) Skeletal muscle NAMPT is induced by exercise in humans. Am J Physiology-Endocrinology Metabolism 298:E117–E126 de Guia RM et al (2019) Aerobic and resistance exercise training reverses age-dependent decline in NAD + salvage capacity in human skeletal muscle. Physiol Rep 7 Basse AL et al (2023) NAMPT-dependent NAD + biosynthesis controls circadian metabolism in a tissue-specific manner. 120 Walzik D et al (2025) Acute exercise boosts NAD + metabolism of human peripheral blood mononuclear cells. Brain Behav Immun 123:1011–1023 Walzik D et al (2023) Tissue-specific effects of exercise as NAD+‐boosting strategy: Current knowledge and future perspectives. Acta Physiol. 10.1111/apha.13921 Brenner C (2022) Viral infection as an NAD + battlefield. Nat Metab 4:2–3 Feng S et al (2024) Cryptic phosphoribosylase activity of NAMPT restricts the virion incorporation of viral proteins. Nat Metab 6:2300–2318 Long JZ (2022) Molecular transducers and the cardiometabolic benefits of exercise. Nat Rev Endocrinol 18:77–78 Reddy A et al (2020) pH-Gated Succinate Secretion Regulates Muscle Remodeling in Response to Exercise. Cell 183:62–75e17 Abdelmoez AM et al (2023) Cell selectivity in succinate receptor SUCNR1 /GPR91 signaling in skeletal muscle. Am J Physiology-Endocrinology Metabolism 324:E289–E298 Kumar S, Chitraju C, Farese RV, Walther TC, Burd CG (2021) Conditional targeting of phosphatidylserine decarboxylase to lipid droplets. Biology Open 10:bio058516 Razmara M et al (2002) CARD-8 Protein, a New CARD Family Member That Regulates Caspase-1 Activation and Apoptosis. J Biol Chem 277:13952–13958 Gleeson M et al (2011) The anti-inflammatory effects of exercise: mechanisms and implications for the prevention and treatment of disease. Nat Rev Immunol 11:607–615 Blazev R et al (2022) Phosphoproteomics of three exercise modalities identifies canonical signaling and C18ORF25 as an AMPK substrate regulating skeletal muscle function. Cell Metabol 34:1561–1577e9 Deshmukh AS et al (2021) Deep muscle-proteomic analysis of freeze-dried human muscle biopsies reveals fiber type-specific adaptations to exercise training. Nat Commun 12:304 Perry AS et al (2024) Proteomic analysis of cardiorespiratory fitness for prediction of mortality and multisystem disease risks. Nat Med. 10.1038/s41591-024-03039-x Pillon NJ et al (2020) Transcriptomic profiling of skeletal muscle adaptations to exercise and inactivity. Nat Commun 11:470 Robbins JM et al (2021) Human plasma proteomic profiles indicative of cardiorespiratory fitness. Nat Metab 3:786–797 Schlagheck ML et al (2020) Cellular immune response to acute exercise: Comparison of endurance and resistance exercise. Eur J Haematol 105:75–84 Liu R et al (2017) Exercise Affects T-Cell Function by Modifying Intracellular Calcium Homeostasis. Med Sci Sports Exerc 49:29–39 Rumpf C et al (2021) The Effect of Acute Physical Exercise on NK-Cell Cytolytic Activity: A Systematic Review and Meta-Analysis. Sports Med 51:519–530 Rundqvist H et al (2020) Cytotoxic T-cells mediate exercise-induced reductions in tumor growth. eLife 9:e59996 Schwendinger F et al (2024) Intensity or volume: the role of physical activity in longevity. European Journal of Preventive Cardiology zwae295 10.1093/eurjpc/zwae295 Scott NW, McPherson GC, Ramsay CR, Campbell MK (2002) The method of minimization for allocation to clinical trials. Control Clin Trials 23:662–674 Bartlett JD et al (2011) High-intensity interval running is perceived to be more enjoyable than moderate-intensity continuous exercise: Implications for exercise adherence. J Sports Sci 29:547–553 Bartlett JD et al (2012) Matched work high-intensity interval and continuous running induce similar increases in PGC-1α mRNA, AMPK, p38, and p53 phosphorylation in human skeletal muscle. J Appl Physiol 112:1135–1143 Burger B, Vaudel M, Barsnes H (2021) Importance of Block Randomization When Designing Proteomics Experiments. J Proteome Res 20:122–128 Müller T et al (2020) Automated sample preparation with SP 3 for low-input clinical proteomics. Mol Syst Biol 16:e9111 Bruderer R et al (2015) Extending the Limits of Quantitative Proteome Profiling with Data-Independent Acquisition and Application to Acetaminophen-Treated Three-Dimensional Liver Microtissues. Mol Cell Proteom 14:1400–1410 Stekhoven DJ, Bühlmann P (2012) MissForest—non-parametric missing value imputation for mixed-type data. Bioinformatics 28:112–118 Wu T et al (2021) clusterProfiler 4.0: A universal enrichment tool for interpreting omics data. Innov 2:100141 Hadley Wickham R, François L, Henry (2023) Kirill Müller, & Davis Vaughan. dplyr: A Grammar of Data Manipulation. R package version 1.1.4 https://github.com/tidyverse/dplyr Hadley, Wickham (2024) Davis Vaughan, & Maximilian Girlich. tidyr: Tidy Messy Data. R package version 1.3.1 https://github.com/tidyverse/tidyr Wickham H (2011) ggplot2. WIREs Computational Stats 3, 180–185 Alboukadel Kassambara. ggpubr (2023) ‘ggplot2’ Based Publication Ready Plots. R package version 0.6.0 Bates D, Mächler M, Bolker B, Walker S (2015) Fitting Linear Mixed-Effects Models Using lme4. J Stat Soft 67 Ritchie ME et al (2015) limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res 43:e47–e47 Smyth GK (2004) Linear Models and Empirical Bayes Methods for Assessing Differential Expression in Microarray Experiments. Stat Appl Genet Mol Biol 3:1–25 Benjamini Y, Hochberg Y (1995) Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. J Royal Stat Soc Ser B: Stat Methodol 57:289–300 Futschik ME, Carlisle B (2005) Noise-Robust Soft Clustering of Gene Expression Time-Course Data. J Bioinform Comput Biol 03:965–988 Kumar L, Futschik ME, Mfuzz (2007) A software package for soft clustering of microarray data. Bioinformation 2:5–7 Guangchuang Yu enrichplot: Visualization of Functional Enrichment Result. R package version 1.26.2 Conway JR, Lex A, Gehlenborg N (2017) UpSetR: an R package for the visualization of intersecting sets and their properties. Bioinformatics 33:2938–2940 Van Rossum G, Drake FL (2009) Python 3 Reference Manual. CreateSpace, Scotts Valley, CA Friedman JH, Hastie T, Tibshirani R (2010) Regularization Paths for Generalized Linear Models via Coordinate Descent. J Stat Soft 33:1–22 Tibshirani R (1996) Regression Shrinkage and Selection Via the Lasso. J Royal Stat Soc Ser B: Stat Methodol 58:267–288 Hoerl AE, Kennard RW (1976) Ridge regression iterative estimation of the biasing parameter. Commun Stat - Theory Methods 5:77–88 Breiman L (2001) Random Forests. Mach Learn 45:5–32 Additional Declarations There is NO Competing Interest. Supplementary Files TableS1.xlsx Table S1 TableS3.xlsx Table S3 TableS7.xlsx Table S7 TableS4.xlsx Table S4 TableS8.xlsx Table S8 TableS10.xlsx Table S10 TableS5.xlsx Table S5 TableS2.xlsx Table S2 TableS6.xlsx Table S6 TableS9.xlsx Table S9 AdditionalInformation.docx Cite Share Download PDF Status: Published Journal Publication published 02 Jan, 2026 Read the published version in Nature Communications → Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6864249","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Resource","associatedPublications":[],"authors":[{"id":470332216,"identity":"4aa9efb8-97d7-42e2-8623-c1de7ac39e1e","order_by":0,"name":"Philipp Zimmer","email":"data:image/png;base64,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","orcid":"https://orcid.org/0000-0001-9631-4503","institution":"TU Dortmund University","correspondingAuthor":true,"prefix":"","firstName":"Philipp","middleName":"","lastName":"Zimmer","suffix":""},{"id":470332217,"identity":"23b0d86f-7412-4fc9-9ab2-185387fcce1c","order_by":1,"name":"David Walzik","email":"","orcid":"","institution":"TU Dortmund University","correspondingAuthor":false,"prefix":"","firstName":"David","middleName":"","lastName":"Walzik","suffix":""},{"id":470332218,"identity":"e9d6e37b-c3ad-454c-a6dc-dbaaa3b4a8e3","order_by":2,"name":"Niklas Joisten","email":"","orcid":"https://orcid.org/0000-0002-9947-8746","institution":"TU Dortmund University","correspondingAuthor":false,"prefix":"","firstName":"Niklas","middleName":"","lastName":"Joisten","suffix":""},{"id":470332219,"identity":"df67d361-dea1-445f-bcac-745fba8922ba","order_by":3,"name":"Alan Metcalfe","email":"","orcid":"","institution":"King's College London","correspondingAuthor":false,"prefix":"","firstName":"Alan","middleName":"","lastName":"Metcalfe","suffix":""},{"id":470332220,"identity":"6d48e2c8-2248-42a8-a105-bf62e163583d","order_by":4,"name":"Sebastian Proschinger","email":"","orcid":"","institution":"TU Dortmund University","correspondingAuthor":false,"prefix":"","firstName":"Sebastian","middleName":"","lastName":"Proschinger","suffix":""},{"id":470332221,"identity":"4c1cc053-141c-4de5-abc6-809b417a6086","order_by":5,"name":"Alexander Schenk","email":"","orcid":"","institution":"TU Dortmund University","correspondingAuthor":false,"prefix":"","firstName":"Alexander","middleName":"","lastName":"Schenk","suffix":""},{"id":470332222,"identity":"b957db13-69d8-4873-a860-5a3617b27cd7","order_by":6,"name":"Charlotte Wenzel","email":"","orcid":"","institution":"TU Dortmund University","correspondingAuthor":false,"prefix":"","firstName":"Charlotte","middleName":"","lastName":"Wenzel","suffix":""},{"id":470332223,"identity":"d43c68b6-98db-4173-b4e6-5578b9651536","order_by":7,"name":"Alessa Henneberg","email":"","orcid":"","institution":"German Cancer Research Center","correspondingAuthor":false,"prefix":"","firstName":"Alessa","middleName":"","lastName":"Henneberg","suffix":""},{"id":470332224,"identity":"a7ea2e0a-c78b-4f57-8908-8edcc217daf5","order_by":8,"name":"Martin Schneider","email":"","orcid":"","institution":"Proteomics Core Facility, German Cancer Research Center (DKFZ)","correspondingAuthor":false,"prefix":"","firstName":"Martin","middleName":"","lastName":"Schneider","suffix":""},{"id":470332225,"identity":"efb5743b-a196-4bbf-aed1-39502def18fd","order_by":9,"name":"Silvia Calderazzo","email":"","orcid":"","institution":"German Cancer Research Center","correspondingAuthor":false,"prefix":"","firstName":"Silvia","middleName":"","lastName":"Calderazzo","suffix":""},{"id":470332226,"identity":"8b7469bb-965d-44bf-b6da-e8ff22956577","order_by":10,"name":"Andreas Groll","email":"","orcid":"https://orcid.org/0000-0001-6787-9118","institution":"University of Alabama at Birmingham","correspondingAuthor":false,"prefix":"","firstName":"Andreas","middleName":"","lastName":"Groll","suffix":""},{"id":470332227,"identity":"50b86ece-7ccf-4180-be5b-bd93a818f50b","order_by":11,"name":"Carsten Watzl","email":"","orcid":"https://orcid.org/0000-0001-5195-0995","institution":"Department of Immunology, Leibniz Research Centre for Working Environment and Human Factors at the Technical University of Dortmund (IfADo), 44139 Dortmund, Germany","correspondingAuthor":false,"prefix":"","firstName":"Carsten","middleName":"","lastName":"Watzl","suffix":""},{"id":470332228,"identity":"57a3d2ae-d9e9-40b5-813b-eeb2fb19c6a9","order_by":12,"name":"Christiane Opitz","email":"","orcid":"","institution":"German Cancer Research Center","correspondingAuthor":false,"prefix":"","firstName":"Christiane","middleName":"","lastName":"Opitz","suffix":""},{"id":470332229,"identity":"ae2841c7-9a30-49eb-8f7e-090e2b751a80","order_by":13,"name":"Dominic Helm","email":"","orcid":"https://orcid.org/0000-0001-9321-2069","institution":"Technische Universitaet Muenchen","correspondingAuthor":false,"prefix":"","firstName":"Dominic","middleName":"","lastName":"Helm","suffix":""}],"badges":[],"createdAt":"2025-06-10 14:42:52","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6864249/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6864249/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41467-025-68101-9","type":"published","date":"2026-01-02T05:00:00+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":85664577,"identity":"f4a360b0-0070-4b77-95d9-fd358f03bcb3","added_by":"auto","created_at":"2025-06-30 12:26:26","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":1068233,"visible":true,"origin":"","legend":"\u003cp\u003eStudy design, analysis plan, and exercise-induced immune cell mobilization.\u003c/p\u003e\n\u003cp\u003e(A) Overview of the study design including time- and workload-matched high-intensity interval exercise (HIIE) and moderate-intensity continuous exercise (MICE).\u003c/p\u003e\n\u003cp\u003e(B) Overview of the bioanalytical and bioinformatic methods used to analyze peripheral blood mononuclear cells (PBMCs).\u003c/p\u003e\n\u003cp\u003e(C) Uniform Manifold Approximation and Projection (UMAP) of immune cell clusters identified by unsupervised clustering using self-organizing maps (SOM). Immune cell clusters are displayed color-coded and separated by exercise condition, and measurement timepoint. Each UMAP corresponds to 3,000 vital lymphocytes from 22 samples, resulting in a total of 66,000 events. For the UMAP representing 1h post MICE only 21 samples were available, resulting in 63,000 events.\u003c/p\u003e\n\u003cp\u003e(D) Comparison of exercise-induced shifts in the identified clusters.\u003c/p\u003e\n\u003cp\u003e(E) Proportions of lymphocyte subsets in response to HIIE and MICE.\u003c/p\u003e\n\u003cp\u003eSee also Figure S1 and Tables S2, S3, and S4.\u003c/p\u003e","description":"","filename":"image1.png","url":"https://assets-eu.researchsquare.com/files/rs-6864249/v1/e05f7b1259c9576622234cc8.png"},{"id":85663296,"identity":"261807da-0c40-4b59-8588-9722b99d04f6","added_by":"auto","created_at":"2025-06-30 12:10:26","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":808433,"visible":true,"origin":"","legend":"\u003cp\u003eMeasures of variability and proteomic changes in response to exercise.\u003c/p\u003e\n\u003cp\u003e(A) Inter-individual variability of all quantified proteins expressed as coefficients of variation (CVs) separated by exercise condition and timepoint.\u003c/p\u003e\n\u003cp\u003e(B) Intra-individual variability of all quantified proteins expressed as delta between the two baselines separated by sex and study participant.\u003c/p\u003e\n\u003cp\u003e(C) Proteomic variability of all quantified proteins.\u003c/p\u003e\n\u003cp\u003e(D) Principal component analysis of all samples using exercise condition and measurement timepoint as metadata. Small symbols indicate individual samples. Big symbols and circles indicate mean and 95 % confidence interval of the corresponding data subset.\u003c/p\u003e\n\u003cp\u003e(E) Quantification of proteomic changes in response to exercise. Linear mixed models containing exercise condition, measurement timepoint, and the interaction between both as fixed factors, were applied (n = 23).\u003c/p\u003e\n\u003cp\u003e(F) Interaction effects between time and exercise condition for HIIE and MICE. Dendrograms depict clusters identified by full-linkage hierarchical clustering.\u003c/p\u003e\n\u003cp\u003e(G) Kinetics of proteins displaying interaction effects separated by the identified clusters (1-3).\u003c/p\u003e\n\u003cp\u003e(H) Group differences between HIIE and MICE. Significant proteins are colored by exercise condition.\u003c/p\u003e\n\u003cp\u003e(I) Delta of protein quantities between HIIE and MICE. Significant proteins are colored by exercise condition.\u003c/p\u003e","description":"","filename":"image2.png","url":"https://assets-eu.researchsquare.com/files/rs-6864249/v1/da5fbaf7231c5f786f30eaf9.png"},{"id":85663772,"identity":"18736233-ac1f-40b3-a73f-0f5041dee4bb","added_by":"auto","created_at":"2025-06-30 12:18:26","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":661409,"visible":true,"origin":"","legend":"\u003cp\u003eExercise reshapes the immune cell proteome towards effector function and causes distinct alterations in protein abundance in response to HIIE and MICE\u003c/p\u003e\n\u003cp\u003e(A) Gene ontology (GO) over-representation analysis comparing significantly altered proteins in HIIE (n = 1,377) and MICE (n = 64) with all proteins quantified in this study (n = 6,039).\u003c/p\u003e\n\u003cp\u003e(B) Overview of proteins altered by HIIE, MICE, or both exercise conditions.\u003c/p\u003e\n\u003cp\u003e(C) Overview of proteins uniquely altered by HIIE.\u003c/p\u003e\n\u003cp\u003e(D) Overview of proteins altered by both exercise conditions. The first two clusters identified by hierarchical clustering separate proteins displaying different kinetics in HIIE and MICE (1) from proteins displaying similar kinetics (2).\u003c/p\u003e\n\u003cp\u003e(E) Overview of proteins uniquely altered by MICE. The first two clusters identified by hierarchical clustering separate proteins increasing in response to MICE (1) from proteins decreasing in response to MICE (2).\u003c/p\u003e","description":"","filename":"image3.png","url":"https://assets-eu.researchsquare.com/files/rs-6864249/v1/2001d86a18258e92cde12bb2.png"},{"id":85663297,"identity":"359943d5-8512-4b4c-9c3c-2ad882a7b1ee","added_by":"auto","created_at":"2025-06-30 12:10:26","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":767481,"visible":true,"origin":"","legend":"\u003cp\u003eIdentification of shared and cluster-specific immune effector functions in the recovery phase following HIIE\u003c/p\u003e\n\u003cp\u003e(A) Protein clusters identified by fuzzy c-means clustering in response to HIIE.\u003c/p\u003e\n\u003cp\u003e(B) Shared and unique GO terms across protein clusters 1 – 3.\u003c/p\u003e\n\u003cp\u003e(C) Identification of core biological processes shared across clusters 1 – 3 and temporal regulation of the underlying proteins.\u003c/p\u003e\n\u003cp\u003e(D) Cluster-specific GO terms regulated by HIIE.\u003c/p\u003e","description":"","filename":"image4.png","url":"https://assets-eu.researchsquare.com/files/rs-6864249/v1/8130c41d3c157697bb518e4c.png"},{"id":85663302,"identity":"2fc66197-c8d6-45e0-9986-b26c15c800fc","added_by":"auto","created_at":"2025-06-30 12:10:27","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":323738,"visible":true,"origin":"","legend":"\u003cp\u003eIdentification of an immunoproteomic signature associated with cardiorespiratory fitness\u003c/p\u003e\n\u003cp\u003e(A) Ridge regression coefficients of features showing moderate to high correlation with V̇O\u003csub\u003e2peak\u003c/sub\u003e (r\u003csub\u003eS\u003c/sub\u003e \u0026gt; 0.4 or \u0026lt; -0.4).\u003c/p\u003e\n\u003cp\u003e(B) Ridge regression coefficients of the 20 features with highest predictive power for V̇O\u003csub\u003e2peak\u003c/sub\u003e.\u003c/p\u003e\n\u003cp\u003e(C) Correlation network of the 20 features with highest predictive power for V̇O\u003csub\u003e2peak\u003c/sub\u003e. Dot sizes and colors represent the strength and direction of correlation with V̇O\u003csub\u003e2peak\u003c/sub\u003e. Connection width and colors represent the strength and direction of correlation between features. Connections are displayed for r\u003csub\u003eS\u003c/sub\u003e \u0026gt; 0.3 or \u0026lt; -0.3.\u003c/p\u003e","description":"","filename":"image5.png","url":"https://assets-eu.researchsquare.com/files/rs-6864249/v1/f60d42e5afe4807b5ef2e62f.png"},{"id":99676960,"identity":"66677666-c60e-4bc6-ba1e-2b5440a50345","added_by":"auto","created_at":"2026-01-07 08:10:12","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":4232580,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6864249/v1/1704f52c-d52c-42df-a641-e239f552fc9f.pdf"},{"id":85663293,"identity":"55f03709-a254-4f81-993f-bebeb7387c24","added_by":"auto","created_at":"2025-06-30 12:10:26","extension":"xlsx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":36825,"visible":true,"origin":"","legend":"Table S1","description":"","filename":"TableS1.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6864249/v1/d58fee249cb77c42b52ccaa1.xlsx"},{"id":85663294,"identity":"327b2ae4-21c6-4567-a3d4-8e493d4d71ba","added_by":"auto","created_at":"2025-06-30 12:10:26","extension":"xlsx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":27677,"visible":true,"origin":"","legend":"Table S3","description":"","filename":"TableS3.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6864249/v1/37cfe3d0d8c83e991cccde78.xlsx"},{"id":85664759,"identity":"b56a06ae-a2c3-4b04-83f6-86b65ea8742e","added_by":"auto","created_at":"2025-06-30 12:34:27","extension":"xlsx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":88981,"visible":true,"origin":"","legend":"Table S7","description":"","filename":"TableS7.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6864249/v1/11c3043511959fa5a2dccd06.xlsx"},{"id":85663770,"identity":"f09fb752-ca98-4566-be00-0eab1bc00d80","added_by":"auto","created_at":"2025-06-30 12:18:26","extension":"xlsx","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":37403,"visible":true,"origin":"","legend":"Table S4","description":"","filename":"TableS4.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6864249/v1/dbfc932016c4312d110c69d2.xlsx"},{"id":85663775,"identity":"dd021e48-92a1-4a48-9350-ee643e7369fd","added_by":"auto","created_at":"2025-06-30 12:18:27","extension":"xlsx","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":170092,"visible":true,"origin":"","legend":"Table S8","description":"","filename":"TableS8.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6864249/v1/90ed914b536850d1512e373d.xlsx"},{"id":85663304,"identity":"540ac913-2bdd-420e-ac19-ab8cfde075da","added_by":"auto","created_at":"2025-06-30 12:10:27","extension":"xlsx","order_by":6,"title":"","display":"","copyAsset":false,"role":"supplement","size":225139,"visible":true,"origin":"","legend":"Table S10","description":"","filename":"TableS10.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6864249/v1/f3b9bd73f478fb6649eec5cc.xlsx"},{"id":85663313,"identity":"b0268d15-234e-42f4-bd35-e6e42fb7542e","added_by":"auto","created_at":"2025-06-30 12:10:27","extension":"xlsx","order_by":7,"title":"","display":"","copyAsset":false,"role":"supplement","size":5202723,"visible":true,"origin":"","legend":"Table S5","description":"","filename":"TableS5.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6864249/v1/9e6d7521d8c0bebbbce8bc64.xlsx"},{"id":85664578,"identity":"cc7d535c-754d-487e-b48e-d2263d7b4f57","added_by":"auto","created_at":"2025-06-30 12:26:26","extension":"xlsx","order_by":8,"title":"","display":"","copyAsset":false,"role":"supplement","size":29645,"visible":true,"origin":"","legend":"Table S2","description":"","filename":"TableS2.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6864249/v1/417f0f59618c4c2225b85880.xlsx"},{"id":85663306,"identity":"40485074-c808-440c-9c3b-b841d000e482","added_by":"auto","created_at":"2025-06-30 12:10:27","extension":"xlsx","order_by":9,"title":"","display":"","copyAsset":false,"role":"supplement","size":18411,"visible":true,"origin":"","legend":"Table S6","description":"","filename":"TableS6.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6864249/v1/b69f49c9842a09906e897216.xlsx"},{"id":85664580,"identity":"2ff459c3-35d6-4809-bd49-71650ffb8cb3","added_by":"auto","created_at":"2025-06-30 12:26:27","extension":"xlsx","order_by":10,"title":"","display":"","copyAsset":false,"role":"supplement","size":75064,"visible":true,"origin":"","legend":"Table S9","description":"","filename":"TableS9.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6864249/v1/adb6836167490b1c7e5b46bc.xlsx"},{"id":85663321,"identity":"2a235a6d-eb39-4447-b978-08e50349bb46","added_by":"auto","created_at":"2025-06-30 12:10:27","extension":"docx","order_by":11,"title":"","display":"","copyAsset":false,"role":"supplement","size":2773562,"visible":true,"origin":"","legend":"","description":"","filename":"AdditionalInformation.docx","url":"https://assets-eu.researchsquare.com/files/rs-6864249/v1/87edb430438d40b9b7acb543.docx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"Acute exercise rewires the proteomic landscape of human immune cells","fulltext":[{"header":"Introduction","content":"\u003cp\u003ePhysical exercise is one of the most powerful strategies to prevent and counteract acute and chronic diseases across the entire human lifespan. The health benefits of exercise are demonstrated by numerous clinical and observational trials,\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e,\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e but the underlying biological mechanisms are poorly understood. Given the broad implications of the immune system in protecting from disease, there is a clear rationale for comprehensive investigations of the impact of exercise on immune cells.\u003c/p\u003e \u003cp\u003eBoth, the exercise-induced mobilization of specific immune cells into circulation as well as an increased cytokine release are well-described phenomena in exercise immunology.\u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e However, few investigations have evaluated the molecular alterations in immune cells triggered by exercise. Bulk RNA sequencing revealed a complex interplay of up- and downregulated transcripts that peaked two minutes after exercise and returned to baseline within 30\u0026ndash;60 minutes in peripheral blood mononuclear cells (PBMCs). Enriched pathways were predominantly related to inflammatory signaling and immune activation, but also included other processes such as cell growth and mobility.\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e While the number and kinetics of transcripts suggest that many expected and some novel signaling pathways are initiated by acute exercise, the proteomic response of PBMCs remains unclear. Since the proteomic makeup of cells defines their phenotype and function, it is crucial to evaluate the impact of exercise on the proteome of immune cells.\u003c/p\u003e \u003cp\u003eHere, we provide a comprehensive characterization of the impact of two different endurance exercise paradigms on the proteomic makeup of PBMCs. The exercise sessions were designed based on the World Health Organization 2020 guidelines on physical activity, which highlight exercise intensity as a crucial variable for health promotion.\u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e Using state-of-the-art mass spectrometry-based proteomics in a standardized experimental setting, we demonstrate that the proteomic makeup of PBMCs is reshaped by acute exercise. Interestingly, proteomic alterations are more potently induced by high-intensity interval exercise (HIIE) compared to moderate-intensity continuous exercise (MICE). The observed changes indicate altered immune cell activation and effector functions in the recovery phase following HIIE, which underlines the immunomodulatory impact of exercise on a proteomic level. Of note, flow cytometry-based immune cell phenotyping suggests that these adaptions occur independent of immune cell mobilization. Using prediction models, we additionally identify an immunoproteomic signature associated with cardiorespiratory fitness. Our data indicates that exercising at higher intensity is necessary to induce proteomic changes associated with immune function, providing biological support for exercise intensity as a crucial variable in the WHO 2020 physical activity guidelines.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eStudy design and participant characteristics\u003c/p\u003e\n\u003cp\u003eWe designed a randomized crossover study comparing time- and workload-matched HIIE and MICE to investigate the impact of exercise intensity on the proteome of immune cells. PBMCs were isolated at baseline, immediately after, and 1h after each exercise condition (Fig. \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003eA). In total, data from 23 overnight-fasted recreationally active runners (12 female, 11 male) was collected. Participants exhibited a mean (\u0026plusmn;\u0026thinsp;SD) age of 30\u0026thinsp;\u0026plusmn;\u0026thinsp;4 years, a body mass index (BMI) of 22.2\u0026thinsp;\u0026plusmn;\u0026thinsp;2.38 kg m\u003csup\u003e\u0026minus;\u0026thinsp;2\u003c/sup\u003e, and a cardiorespiratory fitness (measured as peak oxygen uptake, V̇O\u003csub\u003e2peak\u003c/sub\u003e) of 56.64\u0026thinsp;\u0026plusmn;\u0026thinsp;6.43 mL min\u003csup\u003e\u0026minus;1\u003c/sup\u003e kg\u003csup\u003e\u0026minus;1\u003c/sup\u003e (Table S1).\u003c/p\u003e\n\u003cp\u003eThe collected samples were analyzed via state-of-the-art liquid-chromatography mass-spectrometry (LC-MS/MS) and spectral flow-cytometry. Comparable to other studies in the field of immunology, our proteomics analysis yielded an excellent coverage with a total of 7,385 identified and 6,759 quantified proteins. So far, large-scale proteomic analyses on immune cells have mostly been applied in animal models\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e or on resting samples from healthy donors,\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e making our study the first to apply these methods in a randomized interventional setting with repeated baselines. After data preprocessing, our dataset comprised 6,039 proteins across 23 participants in 2 exercise conditions with 3 measurement timepoints, respectively. This makes our dataset the largest immune cell proteomics data source available in exercise context to date (Fig. \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003eB). Immune cell phenotyping by spectral flow cytometry was performed on a total of 3,537,855 vital lymphocytes (Table S2) to assess exercise-induced shifts within the PBMC compartment (Figure S1A). So far, this has been disregarded in exercise studies applying omics approaches on immune cells.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\n\u003cp\u003eImmune cell mobilization and redistribution is independent of exercise intensity\u003c/p\u003e\n\u003cp\u003eThe mobilization and redistribution of immune cells in response to acute exercise is one of the core phenomena of exercise immunology and it is nowadays agreed upon that the recovery phase following exercise is characterized by a transmigration of lymphocytes from the bloodstream into peripheral tissues, with crucial implications in many disease settings, including anticancer immunity,\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e8\u003c/span\u003e,\u003cspan class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e and immunological defense.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e10\u003c/span\u003e,\u003cspan class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e A remaining topic of debate, however, is whether exercise intensity influences the magnitude of immune cell mobilization since previous studies on this topic were matched for exercise duration, but not workload.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e12\u003c/span\u003e,\u003cspan class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e Thus, before dissecting proteomic alterations of PBMCs in response to exercise, we aimed to clarify whether immune cell kinetics differ in dependence on exercise intensity, since this would lead to a different composition of our PBMC samples in response to HIIE and MICE.\u003c/p\u003e\n\u003cp\u003eBy applying unsupervised immune cell clustering using self-organizing maps (SOM) we identified 6 main clusters in our PBMC samples, which were mapped to the corresponding immune cell populations based on their marker expression. Visual inspection of the identified clusters (Fig. \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003eC) and quantification of exercise-induced cluster shifts resulted in a similar distribution pattern of immune cell clusters between HIIE and MICE with a mean delta of 0.004\u0026thinsp;\u0026plusmn;\u0026thinsp;0.9% (Fig. \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003eD; Table S2). Confirming these findings, absolute numbers of immune cell populations did not reveal time \u0026times; condition interaction effects (Figure S1B; Table S3) and the proportional contribution of each immune cell population to the PBMC compartment was similar between HIIE and MICE (Fig. \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003eE; Table S4). This suggest that exercise triggers similar mobilization and redistribution patterns independent of exercise intensity and indicates that exercise-induced changes in PBMC composition do not differ between HIIE and MICE.\u003c/p\u003e\n\u003cp\u003eSee also Figure S1 and Tables S2, S3, and S4.\u003c/p\u003e\n\u003cp\u003eMeasures of variability indicate high reliability of the generated proteomics dataset\u003c/p\u003e\n\u003cp\u003eInter-individual variability of all quantified proteins resulted in median coefficients of variation (CV) of \u0026lt;\u0026thinsp;5% for all measurement timepoints and conditions (Fig. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eA). This is considerably lower than in other proteomics studies in exercise context\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e4\u003c/span\u003e,\u003cspan class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e and underlines the homogeneity of our study population and the analytic quality of our proteomics pipeline. The applied crossover design additionally enabled us to calculate intra-individual protein variability. The overall mean difference between the two baselines amounted to 0.13\u0026thinsp;\u0026plusmn;\u0026thinsp;0.75% for females and 0.06\u0026thinsp;\u0026plusmn;\u0026thinsp;0.59% for males (Fig. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eB). To assess variability on a per-protein level, we combined multiple measures of variability (i.e., mean CV at baseline, mean CV in response to exercise, and mean difference between the two baselines) into an integrated variability score. 99.34% of all quantified proteins revealed a proteomic variability of \u0026lt;\u0026thinsp;10% and 83.34% achieved a score of \u0026lt;\u0026thinsp;5% (Fig. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eC, \u003cspan class=\"InternalRef\"\u003eS2\u003c/span\u003eA, and \u003cspan class=\"InternalRef\"\u003eS2\u003c/span\u003eB). In summary, the low variability emphasizes the high quality of our study setup, making our generated proteomics dataset highly reliable.\u003c/p\u003e\n\u003cp\u003eAcute exercise alters the immune cell proteome\u003c/p\u003e\n\u003cp\u003eTo obtain first insights into the proteomic alterations induced by exercise, we performed principal component analysis (PCA). Visual inspection of the PCA suggested that the variation within our samples was mainly accounted for by measurement timepoints but not exercise condition \u003cem\u003eper se\u003c/em\u003e (Fig. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eD). PCA also suggested that sex and intervention day had little impact on the variation of our data (Figure S2C). Performing PCA separated by condition and measurement timepoint indicated that HIIE accounted for more variation 1h after exercise than MICE (Figure S2D and S2E).\u003c/p\u003e\n\u003cp\u003eNext, we compared the impact of HIIE and MICE on proteomic alterations in PBMCs using linear mixed models. We identified 1,408 time effects, 119 group effects, and 27 time \u0026times; group interaction effects (Fig. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eE). Including sex as a fixed effect in our analysis did not yield significant results. Dissection of the obtained results revealed more time, group, and time \u0026times; group interaction effects 1h after exercise compared to immediately after exercise and more alterations in HIIE compared to MICE (Fig. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eE). In detail, HIIE was marked by 1,377 significantly altered proteins, while MICE only caused significant alterations in 64. The fact that immune cell counts and proportions did not differ between HIIE and MICE 1h after exercise (Fig. \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003eE and \u003cspan class=\"InternalRef\"\u003eS1\u003c/span\u003eB), rules out the possibility that the proteomic differences are caused by a distinct PBMC composition. In line with our results obtained by PCA, this suggests that HIIE leads to a more profound reorganization of the PBMC proteome compared to MICE.\u003c/p\u003e\n\u003cp\u003eProteomic alterations differ between HIIE and MICE\u003c/p\u003e\n\u003cp\u003eTo evaluate proteins that were distinctly regulated by HIIE compared to MICE, we first dissected the interaction effects of our statistical analysis (Table S5). Immediately after exercise, we observed 5 proteins with distinct kinetics in HIIE compared to MICE (Fig. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eF). Among these proteins, synaptotagmin-like protein 2 (SYTL2), a crucial contributor to cytotoxic granule exocytosis in lymphocytes,\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e displayed a strong increase in response to HIIE, while it remained unaltered in MICE. Similarly, bone marrow stromal antigen 2 (BST2) \u0026ndash; known for its role in blocking virus release from infected cells\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e \u0026ndash; increased in response to HIIE but decreased in MICE (Fig. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eF and \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eG). This gives first insights into the immunomodulatory potential of HIIE and suggests immunological adaptions dependent on exercise intensity immediately after exercise.\u003c/p\u003e\n\u003cp\u003eIn the recovery phase after exercise, we observed 25 interaction effects. Hierarchical clustering yielded two major clusters of proteins marked by opposed kinetics in HIIE compared to MICE (Fig. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eF and \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eG). For instance, toll-like receptor 1 (TLR1), BST2, and cluster of differentiation 302 (CD302) were increased 1h after HIIE but decreased in MICE. TLR1 is the most abundantly expressed TLR on NK cells\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e and serves as membrane-bound pattern recognition receptor for microbial lipopeptides that triggers cytokine production and NK cell cytotoxicity upon stimulation.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e18\u003c/span\u003e,\u003cspan class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e Several studies have demonstrated that TLR1 is crucial for antimicrobial defense,\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e20\u003c/span\u003e,\u003cspan class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e suggesting that exercise-induced increases in TLR1 might reinforce NK cell-mediated immunity against invading pathogens. Of note, BST2 was the only protein that continued to increase from post exercise to 1h post exercise in HIIE, suggesting sustained intensity-dependent adaptions in immunological defense.\u003c/p\u003e\n\u003cp\u003eIn contrast, proteins such as SH2 domain-containing protein 1B (SH2D1B), which serves as a cytoplasmic adapter regulating NK cell effector functions,\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e or asparagine synthetase (ASNS), which was previously shown to regulate CD8\u0026thinsp;+\u0026thinsp;T cell activation, differentiation, and effector function\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e23\u003c/span\u003e,\u003cspan class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e were marked by a decrease in the recovery period following HIIE, while they remained unaltered or increased in MICE (Fig. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eF and \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eG). Of note, our statistical analysis also yielded several group differences between HIIE and MICE (Fig. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eH and \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eI, Table S5).\u003c/p\u003e\n\u003cp\u003eIn summary, our results suggest that the recovery phase following HIIE is marked by more profound alterations of the immune cell proteome compared to MICE. We provide evidence that several proteins related to immune effector function are differentially expressed over time between HIIE and MICE. Against the backdrop of our flow cytometry results, these effects occur despite identical immune cell mobilization patterns between the two exercise conditions.\u003c/p\u003e\n\u003cp\u003eSee also Figure S2 and Table S5.\u003c/p\u003e\n\u003cp\u003eExercise reshapes the immune cell proteome towards effector function\u003c/p\u003e\n\u003cp\u003eTo add a functional dimension to our results, we made use of the Gene Ontology (GO) Resource.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e GO over-representation analysis yielded 27 enriched GO terms in HIIE and 9 enriched GO terms in MICE. Interestingly, enriched GO terms were centered around immune effector functions in both HIIE and MICE with biological processes like \u0026ldquo;disruption of cell in another organism\u0026rdquo; or \u0026ldquo;killing of cells of another organisms\u0026rdquo; yielding high enrichment (Fig. \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003eA).\u003c/p\u003e\n\u003cp\u003eFor proteins altered by MICE, GO over-representation analysis additionally yielded several biological processes related to lymphocyte effector function, such as \u0026ldquo;lymphocyte mediated immunity\u0026rdquo; or \u0026ldquo;T cell mediated immunity\u0026rdquo;. Given that the over-representation analysis was conducted with much more proteins for HIIE, we additionally identified several cellular components and molecular functions in this analysis. Semantic evaluation of the identified GO terms underlined their association with immune effector function. For instance, \u0026ldquo;exogeneous protein binding\u0026rdquo;, \u0026ldquo;virus receptor activity\u0026rdquo;, and \u0026ldquo;endopeptidase activity\u0026rdquo; are known molecular functions in the context of immunological defense against viruses.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e26\u003c/span\u003e,\u003cspan class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e Similarly, \u0026ldquo;proteasome core complex\u0026rdquo; and \u0026ldquo;peroxisomal membrane\u0026rdquo; depict cellular components associated with such molecular function and biological processes (Fig. \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003eA). Collectively, our GO over-representation analysis points towards enhanced regulation of immune effector functions in response to both HIIE and MICE (Table S6).\u003c/p\u003e\n\u003cp\u003eTime-resolved protein changes differ between HIIE and MICE\u003c/p\u003e\n\u003cp\u003eWithin all proteins altered by exercise (n\u0026thinsp;=\u0026thinsp;1,408), we found 1,344 proteins that were uniquely altered by HIIE, 31 proteins that were uniquely altered by MICE, and 33 proteins that were altered by both exercise conditions (Fig. \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003eB). Analysis of proteins altered by HIIE suggested two major protein clusters that were characterized by increased or decreased protein abundance 1h after HIIE compared to baseline (Fig. \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003eC). Considering the large number of proteins altered by HIIE compared to MICE, we took different approaches in analyzing the time effects of each exercise condition.\u003c/p\u003e\n\u003cp\u003eFor proteins altered by both exercise conditions, and proteins uniquely altered by MICE we performed hierarchical clustering to identify proteins displaying similar kinetics. Interestingly, when analyzing proteins that were altered by both exercise conditions, hierarchical clustering yielded 2 major protein clusters: one cluster containing proteins with similar kinetics between HIIE and MICE and one cluster containing proteins with different kinetics (Fig. \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003eD). In absolute terms, most of the proteins responded similarly with only 4 proteins showing higher values in HIIE, including the antiviral protein BST2, which we previously identified in our analysis of time \u0026times; group interaction effects (Fig. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eF). Additionally, many of the proteins that were shared between HIIE and MICE were associated with immune effector functions, suggesting a shared regulation of several immunological processes by exercise. Examples of such proteins include granzymes (e.g., GZMB, GZMH, GZMM), perforin-1 (PRF1), or guanylate-binding proteins 5 (GBP5; Fig. \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003eD).\u003c/p\u003e\n\u003cp\u003eSimilarly, hierarchical clustering of proteins uniquely altered by MICE identified 2 major clusters that separated proteins that decreased in response to MICE from proteins that increased, while showing no alterations in HIIE, respectively (Fig. \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003eE). In line with the observed time \u0026times; group interaction effects (Fig. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eF) most proteins displayed lower abundance in response to MICE (Fig. \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003eE). Of note, although the lower number of time effects and the decreased abundance of many proteins might suggest reduced immune effector functions in response to MICE, it is crucial to emphasize that several proteins with immunological functions, especially those jointly regulated between HIIE and MICE, revealed increased abundance in response to MICE as well. Thus, while our data suggests that the immunoproteomic impact of MICE seems to be less pronounced than that of HIIE, there is no conclusive evidence suggesting reduced immune effector function \u003cem\u003eper se\u003c/em\u003e in response to MICE.\u003c/p\u003e\n\u003cp\u003eSee also Table S5 and S6.\u003c/p\u003e\n\u003cp\u003eIdentification of shared and unique immune effector functions regulated by HIIE\u003c/p\u003e\n\u003cp\u003eTo dissect the proteomic alterations in response to HIIE, we performed fuzzy c-means clustering and mapped the altered proteins (n\u0026thinsp;=\u0026thinsp;1,377) to four distinct clusters by means of their relative membership (Fig. \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003eA; Table S7). The four identified clusters confirmed what hierarchical clustering had previously suggested, i.e., two major clusters marked by increased or decreased protein abundance 1h after HIIE (Fig. \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003eC and \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003eA). We next leveraged biological theme comparisons\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e and identified 576 biological processes, 132 molecular functions, and 187 cellular components associated with the proteins altered by HIIE (Table S8). By generating gene-concept networks of the five most significant GO terms in each ontology, we observed both, shared and unique GO terms across our four protein clusters (Figure S3A \u0026ndash; C).\u003c/p\u003e\n\u003cp\u003eTo quantify functional differences and similarities between the identified protein clusters, we performed gene set enrichment analyses (GSEA)\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e29\u003c/span\u003e\u003c/sup\u003e and observed a total of 169 enriched GO terms (Fig. \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003eB and Table S9). Interestingly, cluster 4 did not yield any enriched GO terms and re-evaluation of the underlying statistics demonstrated that the individual GO terms did not reach the significance threshold. These findings were validated using the STRING resource.\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e We then focused our attention on GO terms that were shared across protein clusters 1\u0026ndash;3 and obtained 11 shared biological processes. Semantic evaluation confirmed their close connection to immune function, as exemplified by GO terms like \u0026ldquo;cell killing\u0026rdquo;, \u0026ldquo;leukocyte activation\u0026rdquo;, or \u0026ldquo;defense response\u0026rdquo; (Fig. \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003eB). Analysis of the underlying proteins resulted in a core proteome consisting of 369 proteins, most of which changed in abundance in the recovery phase following HIIE (Fig. \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003eC). This suggests that the biological processes regulated by HIIE are driven by proteomic alterations in the recovery phase. We observed similar results for the 27 GO terms that were shared between clusters 1 and 2, and the 15 GO terms that were shared between clusters 2 and 3 (Fig. \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003eB and \u003cspan class=\"InternalRef\"\u003eS4\u003c/span\u003eA-C).\u003c/p\u003e\n\u003cp\u003eConcerning GO terms that were uniquely enriched in a specific protein cluster, we identified 29 GO terms uniquely enriched in cluster 1, 62 GO terms uniquely enriched in cluster 2, and 25 GO terms uniquely enriched in cluster 3 (Fig. \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003eB and D). Among the GO terms enriched in cluster 1 we found enriched regulation of \u0026ldquo;endopeptidase activity\u0026rdquo; and \u0026ldquo;cellular response to organic substance\u0026rdquo; (Fig. \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003eD). Similarly, cluster 2 demonstrated enriched regulation of \u0026ldquo;glycosaminoglycan binding\u0026rdquo;, and \u0026ldquo;cell migration\u0026rdquo;, which are crucial processes in the context of exercise-induced transmigration of immune cells from the bloodstream into peripheral tissues. In contrast, cluster 3 was characterized by a decreased regulation of several cellular components and biological processes 1h after exercise, which can be attributed to the underlying protein kinetic (Fig. \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003eA). In summary, GSEA suggested a profound regulation of immune effector processes in the recovery phase following HIIE, which were in part shared and in part unique for specific protein kinetics.\u003c/p\u003e\n\u003cp\u003eSee also Figure S3 and S4, and Table S9.\u003c/p\u003e\n\u003cp\u003eIdentification of an immunoproteomic signature associated with cardiorespiratory fitness\u003c/p\u003e\n\u003cp\u003eWe ultimately leveraged our generated dataset to enable deeper insights into long-term adaptions to exercise training. Taking a data-driven approach, we started by pooling the baseline data of all our analyses, including participant characteristics as well as flow cytometry and LC-MS/MS results. This comprehensive dataset was then used to investigate potential pairwise associations with V̇O\u003csub\u003e2peak\u003c/sub\u003e, a gold standard marker of cardiorespiratory fitness that is highly responsive to exercise training. In a first step we calculated Spearman\u0026rsquo;s rank correlation coefficients (r\u003csub\u003eS\u003c/sub\u003e) and selected features that displayed moderate to high correlation (r\u003csub\u003eS\u003c/sub\u003e \u0026gt; 0.4 or \u0026lt; -0.4) with V̇O\u003csub\u003e2peak\u003c/sub\u003e. This resulted in a reduction of our dataset from 6,063 to 260 features (Table S10).\u003c/p\u003e\n\u003cp\u003eTo establish an elaborate connection between these features and cardiorespiratory fitness, we next performed prediction analyses. Ridge regression yielded an R-squared of 0.61 and a mean squared error of 14.1 and visual inspection of the ranked coefficients revealed a homogeneous distribution of features with positive or negative impact on V̇O\u003csub\u003e2peak\u003c/sub\u003e prediction, respectively (Fig. \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003eA). By evaluating the 20 features with the highest predictive power (Fig. \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003eB), we observed that nicotinamide phosphoribosyltransferase (NAMPT), a key enzyme of nicotinamide adenine dinucleotide (NAD\u003csup\u003e+\u003c/sup\u003e) metabolism, demonstrated the highest positive impact on V̇O\u003csub\u003e2peak\u003c/sub\u003e prediction. NAMPT plays a crucial role in salvaging intracellular NAD\u003csup\u003e+\u003c/sup\u003e and was previously shown to be exercise-responsive in skeletal muscle,\u003csup\u003e31\u0026ndash;\u003cspan class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e but also other target tissues like immune cells.\u003csup\u003e34\u0026ndash;\u003cspan class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e Our results support this notion and suggest that repeated exposure to exercise, which results in greater cardiorespiratory fitness, equips immune cells with a higher metabolic capacity, thereby linking to the immune effector functions previously identified in this work. Additionally, several studies have suggested a direct antiviral function of NAMPT in host defense.\u003csup\u003e37, \u003cspan class=\"CitationRef\"\u003e3\u003c/span\u003e8\u003c/sup\u003e Similarly, we observed a positive impact on V̇O\u003csub\u003e2peak\u003c/sub\u003e prediction for succinate receptor 1 (SUCNR1), a G-protein coupled receptor that was previously shown to control exercise capacity and systemic glucose homeostasis in mice.\u003csup\u003e39,\u003cspan class=\"CitationRef\"\u003e40\u003c/span\u003e\u003c/sup\u003e There are several reports that the effects of exercise-secreted succinate on skeletal muscle tissue adaptions are dependent on paracrine signaling to non-myofibrillar cells such as macrophages, that express SUCNR1.\u003csup\u003e40,\u003cspan class=\"CitationRef\"\u003e41\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\n\u003cp\u003eBesides features with positive impact on V̇O\u003csub\u003e2peak\u003c/sub\u003e prediction, our analysis also yielded several proteins with a negative impact (Fig. \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003eB). Among these features phosphatidylserine decarboxylase (PISD), an enzyme involved in lipid droplet biogenesis,\u003csup\u003e4\u003cspan class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e and caspase recruitment domain family, member 8 (CARD8), a pattern recognition receptor that regulates inflammasome activation and production of pro-inflammatory cytokines,\u003csup\u003e4\u003cspan class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e stood out due to their involvement in metabolism and inflammation. The negative impact of PISD suggests reduced cellular fat deposition with increasing cardiorespiratory fitness. Regarding CARD8, the negative impact on V̇O\u003csub\u003e2peak\u003c/sub\u003e prediction might be explained by anti-inflammatory adaptions with higher cardiorespiratory fitness.\u003csup\u003e3,\u003cspan class=\"CitationRef\"\u003e4\u003c/span\u003e4\u003c/sup\u003e Interestingly, besides correlations with V̇O\u003csub\u003e2peak\u003c/sub\u003e, we also observed various inter-feature correlations (Fig. \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003eC).\u003c/p\u003e\n\u003cp\u003eCollectively our results suggest that the identified proteins associated with cardiorespiratory fitness reshape the phenotype of immune cells in response to exercise training. In a broader context these findings might serve as a molecular foundation for immunological health in the context of long-term training adaptions.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eA better understanding of the molecular underpinnings of physical exercise is needed to individualize exercise training recommendations and maximize their efficacy in mediating health benefits. While some human studies have addressed exercise responses in skeletal muscle and blood plasma using state-of-the-art systems biology approaches,\u003csup\u003e\u003cspan additionalcitationids=\"CR46 CR47 CR48\" citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e\u003c/sup\u003e the impact of exercise on the immune system is less well understood. Here, we provide a comprehensive resource on how two different aerobic exercise stimuli rewire the proteomic makeup of PBMCs. We applied a robust randomized crossover design, including two standardized baseline measurements in a large sample size for proteomic analyses in humans. Our findings expand the literature by \u0026gt;\u0026thinsp;1000 proteomic changes in response to acute exercise in PBMCs. Particularly, immune effector function and cell activation pathways are regulated, and higher intensity is needed to stimulate these changes. Finally, we demonstrate that baseline proteomics data can predict V̇O\u003csub\u003e2peak\u003c/sub\u003e and identify potential exercise-responsive targets in PBMCs that warrant further investigation.\u003c/p\u003e \u003cp\u003eThe acute exercise-induced mobilization of effector cells like NK cells and cytotoxic (CD8+) T cells is well-investigated\u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e,\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e\u003c/sup\u003e but less is known on changes in their proteomic makeup and the resulting cell functions. We identified comprehensive alterations in the immune cell proteome associated with cell function and activation pathways that match previous studies evaluating functional outcomes in different effector populations.\u003csup\u003e\u003cspan additionalcitationids=\"CR52\" citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e\u003c/sup\u003e The regulation of immune effector functions suggests a transient state of immunomodulation following acute bouts of exercise, thereby elucidating the mechanisms of action underlying the benefits of exercise training for disease prevention.\u003c/p\u003e \u003cp\u003eInterestingly, our results indicate that the observed changes in the proteomic makeup of PBMCs occur independent of exercise-induced mobilization and redistribution of immune cells. This is demonstrated by the fact that we observed far more proteomic alterations in response to HIIE compared to MICE, although the underlying PBMC composition did not differ between the two exercise conditions. Although previous investigations have neglected this crucial component of exercise immunology, our proteomics results are temporally in line with transcriptomic alterations identified before.\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e In this context, our open source web application, which can be found at \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://sportsmedicine-dortmund.shinyapps.io/beat\u003c/span\u003e\u003cspan address=\"https://sportsmedicine-dortmund.shinyapps.io/beat\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e, offers the opportunity to mine the underlying dataset for specific proteins of interest, thus informing new hypothesis-driven research in the field of exercise immunology.\u003c/p\u003e \u003cp\u003eOur results support the WHO recommendations on physical activity, which highlight the superior role of high exercise intensity for health promotion.\u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e From an immunological perspective, we found distinct responses of HIIE and MICE when matching the interventions for duration and workload and thus conclude that exercising at higher intensity is crucial to induce more profound changes in the PBMC proteome. This might serve as a potential biological foundation for a recent comprehensive analysis revealing a superior effect of exercise intensity versus volume on longevity at a population-based level.\u003csup\u003e\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eFinally, while previous work has elucidated the molecular underpinnings of cardiorespiratory fitness,\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e,\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e,\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e\u003c/sup\u003e a possible link to immune cells has not yet been explored. We observed strong associations with V̇O\u003csub\u003e2peak\u003c/sub\u003e for several proteins including NAMPT, which is crucial for cellular energy metabolism. Confirming these findings, we have recently demonstrated that NAMPT expression of human PBMCs increases in response to acute exercise.\u003csup\u003e3\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e Overall, this suggests an interrelation between acute exercise stimuli, immunometabolic competence, and cardiorespiratory fitness and suggests a putative role of PBMCs as peripheral mirror for systemic health.\u003c/p\u003e \u003cp\u003eIn conclusion, we identified\u0026thinsp;\u0026gt;\u0026thinsp;1000 exercise-induced alterations in the PBMC proteome and provide a valuable data resource for future research. The identified changes were particularly related to immune effector function, serving as a mechanistic link for the preventive and therapeutic impact of regular exercise. In line with the WHO 2020 guidelines on physical activity, acute exercise at higher intensity elicited greater changes in the regulation of cell function and activation pathways, providing supportive biological evidence for the relevance of exercise intensity as an important factor when planning and structuring exercise training programs for health promotion. Finally, the associations between the PBMC proteome and V̇O\u003csub\u003e2peak\u003c/sub\u003e shed light on potential molecular mediators of immunological health.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003eParticipant recruitment and informed consent\u003c/p\u003e \u003cp\u003e Prior to enrollment of the first participants the study received ethical approval by the local ethics committee of the German Sport University Cologne, which works according to the World Medical Association\u0026rsquo;s Declaration of Helsinki. The study meets the National Institutes of Health definition of a clinical trial and was prospectively registered in the German Clinical Trials Register (DRKS00017686). Study eligibility was assessed for 28 healthy adults aged between 18 and 35. To ensure complication-free execution of the high-intensity interval exercise on the treadmill, participants required a weekly running volume of 2\u0026ndash;5 hours and a body mass index\u0026thinsp;\u0026lt;\u0026thinsp;30. Any previous medical history of muscle disorders, cardiac or kidney diseases as well as regular intake of medication or nutritional supplements were treated as exclusion criteria. For female participants, breast-feeding or an ongoing pregnancy were also treated as exclusion criteria. Of the 28 subjects assessed for eligibility, two were considered ineligible due to acute infections. The remaining 26 participants provided written informed consent and were enrolled in the study. After baseline testing two further participants dropped out due to orthopedic problems while running (Achilles injuries). For one participant, biomaterial did not suffice to run analyses, which resulted in a total of 23 participants. An overview of all participant characteristics is displayed in Table S1.\u003c/p\u003e \u003cp\u003eStudy design\u003c/p\u003e \u003cp\u003eParticipants enrolled in this randomized crossover study were scheduled for three visits to an exercise physiology laboratory of the German Sport University Cologne: Baseline testing, a HIIE session, and a MICE session. For each visit, participants were asked to arrive overnight-fasted and refrain from alcohol and caffein intake in the 24h prior. Water intake was permitted ad libitum. All visits were scheduled between 07:00 and 10:00 am to account for a potential circadian impact on performance and biological outcomes. The minimum timeframe between each of the three visits was 72 hours, to prevent potential carryover effects.\u003c/p\u003e \u003cp\u003eBaseline testing\u003c/p\u003e \u003cp\u003e During baseline testing, written informed consent was obtained from participants and demographic, and anthropometric characteristics were recorded. Afterwards, participants underwent cardiopulmonary exercise testing.\u003c/p\u003e\n\u003ch3\u003eCardiopulmonary exercise test (CPET)\u003c/h3\u003e\n\u003cp\u003eTo standardize the exercise intensity between participants for the HIIE and MICE session, respectively, cardiorespiratory fitness was assessed as peak oxygen consumption (V̇O\u003csub\u003e2peak\u003c/sub\u003e) in an incremental CPET during baseline testing. The CPET was performed on a motorized treadmill (Woodway ELG 90, Weil am Rhein, Germany) that was set to 1% incline for all sessions. The warm-up consisted of 5 min at 6\u0026ndash;8 km h\u003csup\u003e\u0026minus;1\u003c/sup\u003e. Afterwards, participants began running at 8 km h\u003csup\u003e\u0026minus;1\u003c/sup\u003e. The speed of the treadmill was then increased by 1 km h\u003csup\u003e\u0026minus;1\u003c/sup\u003e every 60 seconds until participants reached volitional exhaustion. During the test, heart rate was recorded continuously (Polar FS1C, Kempele, Finland), and rate of perceived exertion was recorded prior to each increase in intensity. Participants were verbally encouraged to continue running by the supervising researcher. After reaching volitional exertion, participants were given a 5 min break before taking up exercise again for a V̇O\u003csub\u003e2peak\u003c/sub\u003e verification test. For this test, the speed of the treadmill was set 1 km h\u003csup\u003e\u0026minus;1\u003c/sup\u003e higher than what the participants had finished with. Just before the verification test, participants ran for 3 min at 8 km h\u003csup\u003e\u0026minus;1\u003c/sup\u003e. The speed was then increased to the target speed within 20 seconds and participants were instructed to run as long as possible. During the entire CPET participants wore a face mask that was connected to a spirometer (Cortex Metalyzer 3B, CORTEX Biophysik GmbH, Leipzig, Germany) to collect breathing gases breath-by-breath. The highest 15-second interval during the CPET was used to calculate V̇O\u003csub\u003e2peak\u003c/sub\u003e.\u003c/p\u003e \u003cp\u003eRandomization\u003c/p\u003e \u003cp\u003eTo prevent sequence effects arising from the order in which HIIE and MICE were conducted, participants were randomized into one of two exercise intervention sequences after baseline testing: HIIE-MICE or MICE-HIIE. Following the minimization procedure by Pocock and Simon,\u003csup\u003e\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e\u003c/sup\u003e randomization was performed via concealed allocation (1:1) using the software Randomization in Treatment Arms (RITA; Evidat, L\u0026uuml;beck, Germany). Age, BMI, and V̇O\u003csub\u003e2peak\u003c/sub\u003e were used as stratification factors. The intervention sequences did not differ in terms of participant characteristics, indicating that our randomization was unbiased (Table S1).\u003c/p\u003e \u003cp\u003eExercise interventions\u003c/p\u003e \u003cp\u003eExercise intensities for the HIIE and MICE session were calculated as percent of V̇O\u003csub\u003e2peak\u003c/sub\u003e for each participant to ensure that all participants exercised at the same intensity. The exercise protocols for HIIE and MICE were designed in an time- and workload-matched manner as previously described\u003csup\u003e5\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, 5\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e to isolate exercise intensity as the only differing variable between the two exercise conditions. This time- and workload-matched design is crucial to draw unbiased conclusions on the impact of exercise intensity. Both exercise sessions were performed on the same treadmill that was also used for the CPET at baseline (Woodway ELG 90, Weil am Rhein, Germany). During MICE participants performed a warm-up for 10 min at a self-selected intensity, followed by a 5 min break. Participants then ran for 50 min at 70% of their V̇O\u003csub\u003e2peak\u003c/sub\u003e. During HIIE, participants performed 7 min of warm-up and cool-down at 70% V̇O\u003csub\u003e2peak\u003c/sub\u003e with six bouts of high-intensity running at 90% V̇O\u003csub\u003e2peak\u003c/sub\u003e in between. Each high-intensity bout lasted 3 min, followed by 3 min of active recovery at 50% V̇O\u003csub\u003e2peak\u003c/sub\u003e.\u003c/p\u003e \u003cp\u003eBlood collection and sample preparation\u003c/p\u003e \u003cp\u003eBlood was drawn from a median antecubital vein in supine position at baseline, immediately after exercise, and 1h after exercise for the HIIE and MICE session, respectively. Each blood draw consisted of 24 mL of whole blood collected in EDTA tubes (Vacutainer, BD). After the last blood draw, peripheral blood mononuclear cells (PBMCs) were isolated via density gradient centrifugation. To achieve this, whole blood was first diluted with phosphate buffered saline (PBS) and then carefully layered on top of a lymphocyte separation medium (Cytiva Ficoll-Paque\u0026trade; PLUS, Fisher Scientific). After centrifugation for 30 min at room temperature and 800 g\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e, the PBMC-containing interphase was collected, washed with PBS, and centrifuged again for 10 min at room temperature and 800 g\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e. PBMCs were then resuspended in freezing medium (Recovery\u0026trade; cell culture freezing medium, Thermo Fisher Scientific) and stored at -80\u0026deg;C before being transferred to a -150\u0026deg;C freezer on the next day until further analysis.\u003c/p\u003e \u003cp\u003eFlow cytometry\u003c/p\u003e\n\u003ch3\u003eSample preparation and data acquisition\u003c/h3\u003e\n\u003cp\u003eFlow cytometry analysis was performed using a Cytek\u0026reg; Aurora full spectrum flow cytometer (Cytek Biosciences, California, USA). Cryopreserved PBMCs were gently thawed in a water bath at 37\u0026deg;C with a mean recovery of 81.28% viable cells assessed with the Zombie NIR\u0026trade; Fixable Viability Kit (BioLegend, San Diego, CA, USA). After incubating 1 \u0026times; 10\u003csup\u003e6\u003c/sup\u003e PBMCs in 2.5 \u0026micro;g Fc block for 10 min at room temperature, cells were stained with anti-CD3 (BUV395, clone SK7), anti-CD4 (PerCP, clone SK3), anti-CD8 (BV750, clone SK1), anti-CD16 (PE-Cy7, clone 3G8), anti-CD25 (BUV805, clone M-A251), anti-CD56 (BUV563, clone NCAM16.2), anti-CD20 (APC, clone L27), and anti-CD19 (APC, clone SJ25C1) antibodies (all from BD Biosciences, NJ, USA). In brief, cells were incubated in the dark with a master mix containing Brilliant Stain buffer (BD Biosciences) and antibodies against surface antigens for 30 min at 4\u0026deg;C. After washing with FACS buffer, the BD Pharmingen\u0026trade;ฏ Transcription Factor Buffer Set was used, and cells were fixed for 40 min at 4\u0026deg;C in the dark. Thereafter, intracellular staining was done by incubating cells with an anti-Foxp3 antibody (PE, clone 259D/C7) for 45 min at 4\u0026deg;C in the dark. After washing, cells were resuspended in FACS buffer and acquired on the flow cytometer within 2 hours after finishing the staining protocol.\u003c/p\u003e\n\u003ch3\u003eData processing\u003c/h3\u003e\n\u003cp\u003eGating was performed using FlowJo\u0026trade; 10.10.0 (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e1\u003c/span\u003eC). B cells were phenotyped as CD3\u003csup\u003e\u0026minus;\u003c/sup\u003eCD56\u003csup\u003e\u0026minus;\u003c/sup\u003eCD19\u003csup\u003e+\u003c/sup\u003eCD20\u003csup\u003e+\u003c/sup\u003e, Natural Killer T (NKT) cells as CD3\u003csup\u003e+\u003c/sup\u003eCD56\u003csup\u003e+\u003c/sup\u003e, Natural Killer (NK) cells either as CD56\u003csup\u003ebright\u003c/sup\u003eCD16\u003csup\u003e\u0026minus;\u003c/sup\u003e (NK\u003csup\u003ebright\u003c/sup\u003e) or CD56\u003csup\u003edim\u003c/sup\u003eCD16\u003csup\u003e+\u003c/sup\u003e (NK\u003csup\u003edim\u003c/sup\u003e), and regulatory T cells (T\u003csub\u003eregs\u003c/sub\u003e) as CD4\u003csup\u003e+\u003c/sup\u003eCD25\u003csup\u003e+\u003c/sup\u003eFoxp3\u003csup\u003e+\u003c/sup\u003e. The person analyzing the samples was blinded to the participants\u0026rsquo; group allocation. Analysis of total blood cell counts was performed from EDTA blood using a hematology analyzer (SYSMEX XP-300, Norderstedt, Germany). The lymphocyte count was then used to calculate the absolute number of peripherally circulating lymphocyte subsets according to the cell proportions derived by flow cytometry.\u003c/p\u003e \u003cp\u003eLC-MS/MS-based untargeted proteomics\u003c/p\u003e \u003cp\u003eSample were processed and measured in a block randomized order\u003csup\u003e\u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e58\u003c/span\u003e\u003c/sup\u003e to prevent any technical bias that might occur during sample preparation or LC-MS/MS measurement.\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eSample preparation\u003c/h2\u003e \u003cp\u003eIsolated PBMCs were lysed in a RIPA buffer containing 10 mM sodium fluoride, 1 mM sodium orthovanadate, cOmplete\u0026trade; Protease Inhibitor Cocktail (Merck KGaA), PhosSTOP\u0026trade; (Merck KGaA), 250 U mL\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e benzonase, and 10 U mL\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e DNase I. Samples were incubated on ice for 1h and then centrifuged at 4\u0026deg;C and 13,000 g\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e for 15 min. Protein concentration was determined in the supernatant with a BCA assay. An amount of 10 \u0026micro;g of protein per sample was digested (Trypsin) using an AssayMAP Bravo liquid handling system (Agilent technologies) running the autoSP3 protocol.\u003csup\u003e\u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e59\u003c/span\u003e\u003c/sup\u003e After sample preparation the remaining peptides were vacuum dried and stored at -20\u0026deg;C until LC-MS/MS analysis.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eMS method Orbitrap Exploris 480\u003c/h3\u003e\n\u003cp\u003eThe dried peptide sample was reconstituted (97.4% Water, 2.5% Hexafluoro-2-propanol and 0.1% trifluoroacetic acid (TFA)) and 10% of the sample were used. The LC-MS/MS analysis was carried out on an Ultimate 3000 UPLC system (Thermo Fisher Scientific) directly connected to an Orbitrap Exploris 480 mass spectrometer for a total of 120 min. Peptides were online desalted on a trapping cartridge (Acclaim PepMap300 C18, 5 \u0026micro;m, 300 \u0026Aring; wide pore; Thermo Fisher Scientific) for 3 min using 30 \u0026micro;l/min flow of 0.1% TFA in water. The analytical multistep gradient (300 nL/min) was performed using a nanoEase MZ Peptide analytical column (300\u0026Aring;, 1.7 \u0026micro;m, 75 \u0026micro;m x 200 mm, Waters) using solvent A (0.1% formic acid in water) and solvent B (0.1% formic acid in acetonitrile). For 102 min the concentration of B was linearly ramped from 4\u0026ndash;30%, followed by a quick ramp to 78%, after two min the concentration of B was lowered to 2% and a 10 min equilibration step appended. Eluting peptides were analyzed in the mass spectrometer using data independent acquisition (DIA) mode. A full scan at 120 k resolution (380\u0026ndash;1400 m/z, 300% AGC target, 45 ms maxIT) was followed 47 DIA windows. The DIA acquisition covered a mass range of 400\u0026ndash;1000 m/z using windows of a variable width with 1 m/z overlap, an AGC target of 1000% with a maxIT set to 54 ms and recorded at a resolution of 30 k. Each sample was followed by a wash run (40 min) to avoid carry-over between samples. Instrument performance and suitability was monitored by regular (approx. one per 48 hours) injections of a standard sample and an in-house shiny application over the whole timeline of the experiment.\u003c/p\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003eData analysis\u003c/h2\u003e \u003cp\u003eAnalysis of DIA RAW files was performed with Spectronaut (Biognosys, version 19.1.240724.62635)\u003csup\u003e\u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e60\u003c/span\u003e\u003c/sup\u003e in directDIA+ (deep) library-free mode. Default settings were applied with the following adaptions. Within DIA Analysis under Identification the Precursor PEP Cutoff was set to 0.01, the Protein Qvalue Cutoff (Run) set to 0.01 and the Protein PEP Cutoff set to 0.01. In Quantification the Proteotypicity Filter was set to Only Protein Group Specific, the Protein LFQ Method was set to MaxLFQ and the quantification window was set to Not Synchronized (SN 17). The data was searched against the human proteome from Uniprot (human reference database with one protein sequence per gene, containing 20,597 unique entries from ninth of February 2024) and the contaminants FASTA from MaxQuant (246 unique entries from twenty-second of December 2022).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eData processing\u003c/h2\u003e \u003cp\u003eBefore further analysis, the obtained dataset was checked for proteins that were identified more than once. For these duplicate results, the event with the highest number of identified precursors across all samples was kept and all other events were deleted from the dataset. The data was then filtered for proteins that were quantified in \u0026ge;\u0026thinsp;70% of the samples in at least one exercise condition and measurement timepoint (i.e., HIIE/MICE baseline, post exercise, 1h post exercise). Subsequently, we imputed the data separated by exercise condition and measurement timepoint using the missForest package.\u003csup\u003e\u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e61\u003c/span\u003e\u003c/sup\u003e Ultimately, proteins were annotated to match the gene names provided in the org.Hs.eg.db package for subsequent Gene Ontology (GO) analysis. Translation between gene names and Entrez gene identifiers was accomplished with the bitr function from the ClusterProfiler package.\u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e,\u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e62\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eQuantification and statistical analysis\u003c/p\u003e \u003cp\u003eSamples from a total of 23 participants were available for statistical analyses. For one participant there was no sample from 1h after MICE due to difficulties during PBMC isolation. Statistical analysis and visualization were performed in R. If not otherwise noted, data wrangling was achieved using the dplyr\u003csup\u003e\u003cspan citationid=\"CR63\" class=\"CitationRef\"\u003e63\u003c/span\u003e\u003c/sup\u003e and tidyr\u003csup\u003e\u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e64\u003c/span\u003e\u003c/sup\u003e package and subsequently visualized with ggplot2\u003csup\u003e65\u003c/sup\u003e and ggpubr\u003csup\u003e\u003cspan citationid=\"CR66\" class=\"CitationRef\"\u003e66\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eUnsupervised immune cell clustering using self-organizing maps\u003c/h2\u003e \u003cp\u003eFlow cytometry data of each sample was cleaned using the FlowAI plugin (v3.2.3) in FlowJo\u0026trade; 10.10.0. The remaining events were gated as described above and live cells were downsampled to 3,000 events per sample using the DownSample plugin (v3.3.1). Subsequently, downsampled events were concatenated to obtain an overall dataset containing all exercise conditions (HIIE, MICE) and measurement timepoints (baseline, post exercise, 1h post exercise). This dataset was then used to perform unsupervised immune cell clustering using self-organizing maps (SOM) with the FlowSOM plugin (v4.1.0). The resulting 6 clusters were identified as CD4\u0026thinsp;+\u0026thinsp;T cells, CD8\u0026thinsp;+\u0026thinsp;T cells, NKT cells, CD56\u003csup\u003edim\u003c/sup\u003e cells, CD56\u003csup\u003ebright\u003c/sup\u003e cells and B cells in the build-in Cluster Explorer in FlowJo\u0026trade; 10.10.0. The overall dataset was visualized using Uniform Manifold Approximation and Projection (UMAP) via the UMAP plugin (v4.1.1) and FlowSOM clusters were superimposed via color-coding. This overall, color-coded immune cell clusters were used as a template map for subsequent clustering per exercise condition and measurement timepoint. To achieve this, the downsampled events were concatenated for baseline, post exercise and 1h post exercise in HIIE and MICE, respectively. Ultimately, FlowSOM clustering and UMAP were performed on each of these concatenated dataframes by applying them on the previously generated map.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eExercise-induced mobilization of immune cells\u003c/h2\u003e \u003cp\u003eExercise-induced alterations in immune cell counts were analyzed by applying linear mixed models to the flow cytometry results. Measurement timepoint and exercise condition were implemented as fixed effects and participant ID as random effect using the lmer function from the lme4 package.\u003csup\u003e\u003cspan citationid=\"CR67\" class=\"CitationRef\"\u003e67\u003c/span\u003e\u003c/sup\u003e Results of the linear mixed models were then analyzed for time and time \u0026times; condition interaction effects via analyses of variance (ANOVAs) with the built-in ANOVA function from R stats. In case of significant results, pairwise comparisons of measurement timepoints and/or exercise conditions were performed by applying the emmeans function from the emmeans package. P-values were Bonferroni-corrected for multiple testing.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eMeasures of variability\u003c/h2\u003e \u003cp\u003eAll measures of variability were calculated with unimputed data to avoid potential bias arising from imputation. Inter-individual variability was calculated as coefficient of variation (CVs) for each protein across all participants separated by exercise condition (HIIE, MICE), and measurement timepoint (baseline, post exercise, 1h post exercise). CVs were calculated as the ratio of the standard deviation \u003cem\u003eσ\u003c/em\u003e to the mean \u003cem\u003e\u0026micro;\u003c/em\u003e. Intra-individual variability was assessed by comparing the baseline values of the two intervention days. Relative differences between day 1 and day 2 were calculated (in percent) for each protein separated by study participant. Proteomic variability was quantified for each protein by calculating (i) the mean CV across all participants in HIIE and MICE at baseline, (ii) the mean CV across all participants in HIIE and MICE post exercise and 1h post exercise, and (iii) the mean difference between the two baselines across all participants. Proteomic variability was also calculated separated by exercise conditions and measurement timepoints (see Figure S2A and S2B).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003ePrincipal component analysis\u003c/h2\u003e \u003cp\u003ePrincipal component analysis (PCA) was carried out using the built-in prcomp function from R stats. All samples were plotted with the fviz_pca_ind function from the factoextra package. Exercise condition, measurement timepoint, intervention day, and sex were used as metadata to color-code PCA results. PCAs were also computed on datasets separated by exercise condition or measurement timepoint to visually assess the impact of these variables on each other (see Figure S2C-E).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eLinear mixed models to identify proteins altered by HIIE and/or MICE\u003c/h2\u003e \u003cp\u003eTo identify proteins altered by HIIE and/or MICE, a linear mixed model was fitted on the log\u003csub\u003e2\u003c/sub\u003e-transformed, normalized, and imputed protein intensities via the limma R package.\u003csup\u003e\u003cspan citationid=\"CR68\" class=\"CitationRef\"\u003e68\u003c/span\u003e\u003c/sup\u003e Intra-individual correlation was estimated via the duplicateCorrelation function. The model included the exercise condition (HIIE, MICE), the measurement timepoint (baseline, post exercise, 1h post exercise), and the interaction between both as fixed factors. A moderated \u003cem\u003et\u003c/em\u003e statistic\u003csup\u003e\u003cspan citationid=\"CR69\" class=\"CitationRef\"\u003e69\u003c/span\u003e\u003c/sup\u003e was obtained for each contrast of interest via the eBayes function with estimated variance trend and robustification. The resulting p-values for each contrast were adjusted with the Benjamini-Hochberg procedure\u003csup\u003e\u003cspan citationid=\"CR70\" class=\"CitationRef\"\u003e70\u003c/span\u003e\u003c/sup\u003e to control the false discovery rate and significance was declared at the adjusted 5% two-sided level. The model was subsequently extended to include sex and all two-way interactions.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eGene ontology (GO) over-representation analysis\u003c/h2\u003e \u003cp\u003eTime effects of the statistical analysis with limma\u003csup\u003e\u003cspan citationid=\"CR68\" class=\"CitationRef\"\u003e68\u003c/span\u003e\u003c/sup\u003e were used to map proteins that were significantly altered by HIIE and MICE to GO terms, respectively. GO over-representation analysis was performed with the ClusterProfiler package.\u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e,\u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e62\u003c/span\u003e\u003c/sup\u003e For HIIE and MICE, significantly altered proteins were compared with the entire dataset of quantified proteins applying Benjamini-Hochberg correction of p-values with a p-value cutoff of 0.05 and a q-value cutoff of 0.2.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003eFuzzy c-means clustering\u003c/h2\u003e \u003cp\u003eFuzzy c-means clustering was performed with the Mfuzz package.\u003csup\u003e\u003cspan citationid=\"CR71\" class=\"CitationRef\"\u003e71\u003c/span\u003e,\u003cspan citationid=\"CR72\" class=\"CitationRef\"\u003e72\u003c/span\u003e\u003c/sup\u003e Data was standardized using the standardise function and the optimal number of clusters was determined by calculating the minimum centroid distance for a range of cluster numbers using the Dmin function. The optimal fuzzifier was identified with the mestamiate function.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003eBiological theme comparison\u003c/h2\u003e \u003cp\u003eBiological theme comparison was carried out using the compareCluster function from the ClusterProfiler package.\u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e,\u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e62\u003c/span\u003e\u003c/sup\u003e Entrez gene identifiers of the proteins contained in the identified clusters were used as input with the function command set to \u0026ldquo;enrichGO\u0026rdquo;. Benjamini-Hochberg correction was applied to p-values with a cutoff of 0.05 and minimum gene set size was set to 10. The results were simplified via the simplify function using a cutoff of 0.7 and visualized separated by ontology with the cnetplot function from the enrichplot package.\u003csup\u003e\u003cspan citationid=\"CR73\" class=\"CitationRef\"\u003e73\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003eGene ontology (GO) gene set enrichment analysis\u003c/h2\u003e \u003cp\u003eGene set enrichment analysis was performed using the gseGO function from the ClusterProfiler package.\u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e,\u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e62\u003c/span\u003e\u003c/sup\u003e Entrez gene identifiers and fold changes from baseline of the proteins contained in the identified clusters were used as input with the minimum gene set size set to 10. In case fold changes were only positive or negative, the \u0026ldquo;scoreType\u0026rdquo; command was set to \u0026ldquo;pos\u0026rdquo; or \u0026ldquo;neg\u0026rdquo;, respectively. P-values were corrected using the Benjamini-Hochberg procedure with a p-value cutoff of 0.05. The underlying proteins mapping to each significant GO term were identified using the select function from the AnnotationDbi package. Shared and unique GO terms across the identified clusters were visualized with the UpSetR package.\u003csup\u003e\u003cspan citationid=\"CR74\" class=\"CitationRef\"\u003e74\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003eV̇O\u003csub\u003e2peak\u003c/sub\u003e prediction and correlation network analysis\u003c/h2\u003e \u003cp\u003e \u003cb\u003ePreselection of features\u003c/b\u003e To identify features with high association to V̇O\u003csub\u003e2peak\u003c/sub\u003e, we conducted a preselection in Python (v.3.9).\u003csup\u003e7\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e The features were standardized using z transformation and included the average of mass spectrometry-based proteomics data and flow cytometry-based immunophenotyping data at baseline of intervention day 1 and 2 as well as sex, height, weight and BMI. V̇O\u003csub\u003e2peak\u003c/sub\u003e was scaled to body weight. Data from 2 participants were excluded from the analysis due to incomplete feature sets. Pairwise Spearman's rank correlations between all features and V̇O\u003csub\u003e2peak\u003c/sub\u003e were calculated (Table S10) and features with a correlation coefficient of \u0026gt;\u0026thinsp;0.4 or \u0026lt; -0.4 were included in the subsequent analysis. From a total of 6,063 initial features, 260 remained after this selection.\u003c/p\u003e \u003cp\u003e \u003cb\u003eV̇O\u003c/b\u003e \u003csub\u003e \u003cb\u003e2peak\u003c/b\u003e \u003c/sub\u003e \u003cb\u003eprediction modeling\u003c/b\u003e We ran LASSO\u003csup\u003e7\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, 7\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e and ridge regression\u003csup\u003e7\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e as well as a random forest\u003csup\u003e7\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e as a non-linear, tree-based approach. A leave-one-out (LOO) cross-validation was performed in Python (v.3.9) to assess the predictive performance of these methods based on the 260 features. To optimize the hyperparameters for each model by grid search, a second inner cross validation was performed. For each training set, we selected the model that had the lowest test error. The predicted output value resulted from the cross validation iteration, where the corresponding output data point and its associated features were not included in the training set. These predicted values were used to calculate the mean squared error (MSE) and the r squared R\u003csup\u003e2\u003c/sup\u003e. Ridge regression outperformed the other models. All features with coefficients from the ridge regression are listed in the supplements (Table S10).\u003c/p\u003e \u003cp\u003e \u003cb\u003eCorrelation network analysis\u003c/b\u003e The 20 features with the largest absolute mean value from the ridge regression were selected to create a weighted, undirected network using Spearman\u0026rsquo;s rank correlations. The network was visualized in R (v.4.4.1) with the packages Hmisc (v.5.2.1) and igraph (v.2.1.1.).\u003c/p\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003ch2\u003eAcknowledgements\u003c/h2\u003e\n\u003cp\u003eWe want to thank Lars Donath and Ludwig Rappelt for their help in conducting the trial. We thank the team of the Proteomics Core Facility of the DKFZ particularly Adrian Stoegbauer und Alina Ertl for sample preparation and LC-MS/MS measurement. We also want to thank the German Sport University Cologne for supplying internal funds to A.J.M.. Figures were created with https://BioRender.com.\u003c/p\u003e\n\u003ch2\u003eAuthor contributions\u003c/h2\u003e\n\u003cp\u003eConceptualization, N.J., A.J.M., and P.Z.; Methodology, N.J., A.J.M., A.S., and P.Z.; Software, D.W., C.We., M.S., and S.C.; Formal Analysis, D.W., S.P., C.We., M.S., and S.C.; Investigation, A.J.M., S.P., A.S., M.S., and D.H.; Resources, C.Wa., C.A.O., D.H., and P.Z.; Writing \u0026ndash; Original Draft, D.W., N.J., S.P., C.We., M.S., and S.C.; Writing \u0026ndash; Review \u0026amp; Editing, A.J.M., S.P., A.S., C.We., A.L.H., M.S., S.C., A.G, C.Wa., C.A.O., D.H., and P.Z.; Visualization, D.W. and C.We.; Supervision, P.Z., A.G., and D.H., Project Administration, P.Z.; Funding Acquisition, A.J.M\u003c/p\u003e\n\u003ch2\u003eCompeting interests\u003c/h2\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e\n\u003ch3\u003eRESSOURCE AVAILABILITY\u003c/h3\u003e\n\u003ch2\u003eLead contact\u003c/h2\u003e\n\u003cp\u003eAny requests for further pieces of information or resources should be directed to the Lead Contact Philipp Zimmer ([email protected])\u003c/p\u003e\n\u003ch2\u003eData and code availability\u003c/h2\u003e\n\u003cp\u003eAll data associated with this article can be explored via our interactive web application at https://sportsmedicine-dortmund.shinyapps.io/beat. Raw data files of all samples processed in the proteomics analysis are hosted on the PRoteomics IDEntifications Database (PRIDE; https://www.ebi.ac.uk/pride) under the following PRIDE-ID: PXD058573.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003ePenedo FJ, Dahn JR (2005) Exercise and well-being: a review of mental and physical health benefits associated with physical activity. Curr Opin Psychiatry 18:189\u0026ndash;193\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRuegsegger GN, Booth FW (2018) Health Benefits of Exercise. Cold Spring Harb Perspect Med 8:a029694\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWalsh NP et al (2011) Position statement. Part one: Immune function and exercise. Exerc Immunol Rev 17:6\u0026ndash;63\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eContrepois K et al (2020) Molecular Choreography of Acute Exercise. Cell 181:1112\u0026ndash;1130e16\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBull FC et al (2020) World Health Organization 2020 guidelines on physical activity and sedentary behaviour. Br J Sports Med 54:1451\u0026ndash;1462\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHowden AJM et al (2019) Quantitative analysis of T cell proteomes and environmental sensors during T cell differentiation. Nat Immunol 20:1542\u0026ndash;1554\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRieckmann JC et al (2017) Social network architecture of human immune cells unveiled by quantitative proteomics. Nat Immunol 18:583\u0026ndash;593\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCampbell JP, Walzik D, Zimmer P (2024) What dictates tumour cell sensitivity to exercise? Nat Rev Immunol. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41577-024-01001-z\u003c/span\u003e\u003cspan address=\"10.1038/s41577-024-01001-z\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFiuza-Luces C et al (2024) The effect of physical exercise on anticancer immunity. Nat Rev Immunol 24:282\u0026ndash;293\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCampbell JP, Turner JE (2018) Debunking the Myth of Exercise-Induced Immune Suppression: Redefining the Impact of Exercise on Immunological Health Across the Lifespan. Front Immunol 9:648\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSimpson RJ, Kunz H, Agha N, Graff R Exercise and the Regulation of Immune Functions. in \u003cem\u003eProgress in Molecular Biology and Translational Science\u003c/em\u003e vol. 135 355\u0026ndash;380 (Elsevier, 2015)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCampbell JP et al (2009) Acute exercise mobilises CD8\u0026thinsp;+\u0026thinsp;T lymphocytes exhibiting an effector-memory phenotype. Brain Behav Immun 23:767\u0026ndash;775\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSimpson RJ et al (2007) Apoptosis Does not Contribute to the Blood Lymphocytopenia Observed After Intensive and Downhill Treadmill Running in Humans. Res Sports Med 15:157\u0026ndash;174\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSavikj M et al (2022) Exercise timing influences multi-tissue metabolome and skeletal muscle proteome profiles in type 2 diabetic patients \u0026ndash; A randomized crossover trial. Metabolism 135:155268\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eM\u0026eacute;nasch\u0026eacute; G et al (2008) A newly identified isoform of Slp2a associates with Rab27a in cytotoxic T cells and participates to cytotoxic granule secretion. Blood 112:5052\u0026ndash;5062\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJouvenet N et al (2009) Broad-Spectrum Inhibition of Retroviral and Filoviral Particle Release by Tetherin. J Virol 83:1837\u0026ndash;1844\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChalifour A et al (2004) Direct bacterial protein PAMP recognition by human NK cells involves TLRs and triggers α-defensin production. Blood 104:1778\u0026ndash;1783\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAdib-Conquy M, Scott‐Algara D, Cavaillon J, Souza‐Fonseca‐Guimaraes F (2014) TLR‐mediated activation of NK cells and their role in bacterial/viral immune responses in mammals. Immunol Cell Biol 92:256\u0026ndash;262\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNoh J-Y, Yoon SR, Kim T-D, Choi I, Jung H (2020) Toll-Like Receptors in Natural Killer Cells and Their Application for Immunotherapy. \u003cem\u003eJournal of Immunology Research\u003c/em\u003e 1\u0026ndash;9 (2020)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLancioni CL et al (2011) \u003cem\u003eMycobacterium tuberculosis\u003c/em\u003e Lipoproteins Directly Regulate Human Memory CD4\u003csup\u003e+\u003c/sup\u003e T Cell Activation via Toll-Like Receptors 1 and 2. Infect Immun 79:663\u0026ndash;673\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTriantafilou M et al (2006) Membrane Sorting of Toll-like Receptor (TLR)-2/6 and TLR2/1 Heterodimers at the Cell Surface Determines Heterotypic Associations with CD36 and Intracellular Targeting. J Biol Chem 281:31002\u0026ndash;31011\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRoncagalli R et al (2005) Negative regulation of natural killer cell function by EAT-2, a SAP-related adaptor. Nat Immunol 6:1002\u0026ndash;1010\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGnanaprakasam JNR et al (2023) Asparagine restriction enhances CD8\u0026thinsp;+\u0026thinsp;T cell metabolic fitness and antitumoral functionality through an NRF2-dependent stress response. Nat Metab 5:1423\u0026ndash;1439\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWu J et al (2021) Asparagine enhances LCK signalling to potentiate CD8\u0026thinsp;+\u0026thinsp;T-cell activation and anti-tumour responses. Nat Cell Biol 23:75\u0026ndash;86\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eThe Gene Ontology Consortium et al. The Gene Ontology knowledgebase in 2023. GENETICS 224, iyad031 et al (2023)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRouse BT, Mueller SN Host Defenses to Viruses. in \u003cem\u003eClinical Immunology\u003c/em\u003e 365\u0026ndash;374.e1 (Elsevier, 2019). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/B978-0-7020-6896-6.00025-9\u003c/span\u003e\u003cspan address=\"10.1016/B978-0-7020-6896-6.00025-9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRouse BT, Sehrawat S (2010) Immunity and immunopathology to viruses: what decides the outcome? Nat Rev Immunol 10:514\u0026ndash;526\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYu G, Wang L-G, Han Y, He Q-Y (2012) clusterProfiler: an R Package for Comparing Biological Themes Among Gene Clusters. OMICS 16:284\u0026ndash;287\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSubramanian A et al (2005) Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles. \u003cem\u003eProc. Natl. Acad. Sci. U.S.A.\u003c/em\u003e 102, 15545\u0026ndash;15550\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSzklarczyk D et al (2019) STRING v11: protein\u0026ndash;protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets. Nucleic Acids Res 47:D607\u0026ndash;D613\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBasse AL et al (2021) Nampt controls skeletal muscle development by maintaining Ca2\u0026thinsp;+\u0026thinsp;homeostasis and mitochondrial integrity. Mol Metabolism 53:101271\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCostford SR et al (2010) Skeletal muscle NAMPT is induced by exercise in humans. Am J Physiology-Endocrinology Metabolism 298:E117\u0026ndash;E126\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ede Guia RM et al (2019) Aerobic and resistance exercise training reverses age-dependent decline in NAD \u003csup\u003e+\u003c/sup\u003e salvage capacity in human skeletal muscle. Physiol Rep 7\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBasse AL et al (2023) NAMPT-dependent NAD\u0026thinsp;+\u0026thinsp;biosynthesis controls circadian metabolism in a tissue-specific manner. 120\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWalzik D et al (2025) Acute exercise boosts NAD\u0026thinsp;+\u0026thinsp;metabolism of human peripheral blood mononuclear cells. Brain Behav Immun 123:1011\u0026ndash;1023\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWalzik D et al (2023) Tissue-specific effects of exercise as NAD+‐boosting strategy: Current knowledge and future perspectives. Acta Physiol. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1111/apha.13921\u003c/span\u003e\u003cspan address=\"10.1111/apha.13921\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBrenner C (2022) Viral infection as an NAD\u0026thinsp;+\u0026thinsp;battlefield. Nat Metab 4:2\u0026ndash;3\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFeng S et al (2024) Cryptic phosphoribosylase activity of NAMPT restricts the virion incorporation of viral proteins. Nat Metab 6:2300\u0026ndash;2318\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLong JZ (2022) Molecular transducers and the cardiometabolic benefits of exercise. Nat Rev Endocrinol 18:77\u0026ndash;78\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eReddy A et al (2020) pH-Gated Succinate Secretion Regulates Muscle Remodeling in Response to Exercise. Cell 183:62\u0026ndash;75e17\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAbdelmoez AM et al (2023) Cell selectivity in succinate receptor \u003cem\u003eSUCNR1\u003c/em\u003e /GPR91 signaling in skeletal muscle. Am J Physiology-Endocrinology Metabolism 324:E289\u0026ndash;E298\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKumar S, Chitraju C, Farese RV, Walther TC, Burd CG (2021) Conditional targeting of phosphatidylserine decarboxylase to lipid droplets. Biology Open 10:bio058516\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRazmara M et al (2002) CARD-8 Protein, a New CARD Family Member That Regulates Caspase-1 Activation and Apoptosis. J Biol Chem 277:13952\u0026ndash;13958\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGleeson M et al (2011) The anti-inflammatory effects of exercise: mechanisms and implications for the prevention and treatment of disease. Nat Rev Immunol 11:607\u0026ndash;615\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBlazev R et al (2022) Phosphoproteomics of three exercise modalities identifies canonical signaling and C18ORF25 as an AMPK substrate regulating skeletal muscle function. Cell Metabol 34:1561\u0026ndash;1577e9\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDeshmukh AS et al (2021) Deep muscle-proteomic analysis of freeze-dried human muscle biopsies reveals fiber type-specific adaptations to exercise training. Nat Commun 12:304\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePerry AS et al (2024) Proteomic analysis of cardiorespiratory fitness for prediction of mortality and multisystem disease risks. Nat Med. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41591-024-03039-x\u003c/span\u003e\u003cspan address=\"10.1038/s41591-024-03039-x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePillon NJ et al (2020) Transcriptomic profiling of skeletal muscle adaptations to exercise and inactivity. Nat Commun 11:470\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRobbins JM et al (2021) Human plasma proteomic profiles indicative of cardiorespiratory fitness. Nat Metab 3:786\u0026ndash;797\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchlagheck ML et al (2020) Cellular immune response to acute exercise: Comparison of endurance and resistance exercise. Eur J Haematol 105:75\u0026ndash;84\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu R et al (2017) Exercise Affects T-Cell Function by Modifying Intracellular Calcium Homeostasis. Med Sci Sports Exerc 49:29\u0026ndash;39\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRumpf C et al (2021) The Effect of Acute Physical Exercise on NK-Cell Cytolytic Activity: A Systematic Review and Meta-Analysis. Sports Med 51:519\u0026ndash;530\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRundqvist H et al (2020) Cytotoxic T-cells mediate exercise-induced reductions in tumor growth. eLife 9:e59996\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchwendinger F et al (2024) Intensity or volume: the role of physical activity in longevity. \u003cem\u003eEuropean Journal of Preventive Cardiology\u003c/em\u003e zwae295 \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/eurjpc/zwae295\u003c/span\u003e\u003cspan address=\"10.1093/eurjpc/zwae295\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eScott NW, McPherson GC, Ramsay CR, Campbell MK (2002) The method of minimization for allocation to clinical trials. Control Clin Trials 23:662\u0026ndash;674\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBartlett JD et al (2011) High-intensity interval running is perceived to be more enjoyable than moderate-intensity continuous exercise: Implications for exercise adherence. J Sports Sci 29:547\u0026ndash;553\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBartlett JD et al (2012) Matched work high-intensity interval and continuous running induce similar increases in PGC-1α mRNA, AMPK, p38, and p53 phosphorylation in human skeletal muscle. J Appl Physiol 112:1135\u0026ndash;1143\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBurger B, Vaudel M, Barsnes H (2021) Importance of Block Randomization When Designing Proteomics Experiments. J Proteome Res 20:122\u0026ndash;128\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eM\u0026uuml;ller T et al (2020) Automated sample preparation with SP 3 for low-input clinical proteomics. Mol Syst Biol 16:e9111\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBruderer R et al (2015) Extending the Limits of Quantitative Proteome Profiling with Data-Independent Acquisition and Application to Acetaminophen-Treated Three-Dimensional Liver Microtissues. Mol Cell Proteom 14:1400\u0026ndash;1410\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStekhoven DJ, B\u0026uuml;hlmann P (2012) MissForest\u0026mdash;non-parametric missing value imputation for mixed-type data. Bioinformatics 28:112\u0026ndash;118\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWu T et al (2021) clusterProfiler 4.0: A universal enrichment tool for interpreting omics data. Innov 2:100141\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHadley Wickham R, Fran\u0026ccedil;ois L, Henry (2023) Kirill M\u0026uuml;ller, \u0026amp; Davis Vaughan. dplyr: A Grammar of Data Manipulation. \u003cem\u003eR package version 1.1.4\u003c/em\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/tidyverse/dplyr\u003c/span\u003e\u003cspan address=\"https://github.com/tidyverse/dplyr\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHadley, Wickham (2024) Davis Vaughan, \u0026amp; Maximilian Girlich. tidyr: Tidy Messy Data. \u003cem\u003eR package version 1.3.1\u003c/em\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/tidyverse/tidyr\u003c/span\u003e\u003cspan address=\"https://github.com/tidyverse/tidyr\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWickham H (2011) ggplot2. \u003cem\u003eWIREs Computational Stats\u003c/em\u003e 3, 180\u0026ndash;185\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAlboukadel Kassambara. ggpubr (2023) \u0026lsquo;ggplot2\u0026rsquo; Based Publication Ready Plots. \u003cem\u003eR package version 0.6.0\u003c/em\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBates D, M\u0026auml;chler M, Bolker B, Walker S (2015) Fitting Linear Mixed-Effects Models Using lme4. J Stat Soft 67\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRitchie ME et al (2015) limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res 43:e47\u0026ndash;e47\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSmyth GK (2004) Linear Models and Empirical Bayes Methods for Assessing Differential Expression in Microarray Experiments. Stat Appl Genet Mol Biol 3:1\u0026ndash;25\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBenjamini Y, Hochberg Y (1995) Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. J Royal Stat Soc Ser B: Stat Methodol 57:289\u0026ndash;300\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFutschik ME, Carlisle B (2005) Noise-Robust Soft Clustering of Gene Expression Time-Course Data. J Bioinform Comput Biol 03:965\u0026ndash;988\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKumar L, Futschik ME, Mfuzz (2007) A software package for soft clustering of microarray data. Bioinformation 2:5\u0026ndash;7\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGuangchuang Yu enrichplot: Visualization of Functional Enrichment Result. \u003cem\u003eR package version 1.26.2\u003c/em\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eConway JR, Lex A, Gehlenborg N (2017) UpSetR: an R package for the visualization of intersecting sets and their properties. Bioinformatics 33:2938\u0026ndash;2940\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVan Rossum G, Drake FL (2009) Python 3 Reference Manual. CreateSpace, Scotts Valley, CA\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFriedman JH, Hastie T, Tibshirani R (2010) Regularization Paths for Generalized Linear Models via Coordinate Descent. J Stat Soft 33:1\u0026ndash;22\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTibshirani R (1996) Regression Shrinkage and Selection Via the Lasso. J Royal Stat Soc Ser B: Stat Methodol 58:267\u0026ndash;288\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHoerl AE, Kennard RW (1976) Ridge regression iterative estimation of the biasing parameter. Commun Stat - Theory Methods 5:77\u0026ndash;88\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBreiman L (2001) Random Forests. Mach Learn 45:5\u0026ndash;32\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"exercise, immune cells, PBMCs, proteostasis, effector function, high-intensity interval training, exercise immunology, proteomics, flow cytometry","lastPublishedDoi":"10.21203/rs.3.rs-6864249/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6864249/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eExercise-driven alterations of the immune system are a key mechanism in the prevention and treatment of various diseases. Here, we performed mass spectrometry-based proteomics analysis on peripheral blood mononuclear cells (PBMCs) at a depth of \u0026gt;\u0026thinsp;6000 proteins. Comparing time- and workload-matched high-intensity interval exercise (HIIE) and moderate-intensity continuous exercise (MICE) we discover versatile changes in the proteomic makeup of PBMCs and reveal profound alterations related to effector function and immune cell activation pathways within one hour after exercise. These changes were more pronounced after HIIE compared to MICE and occurred despite identical immune cell mobilization patterns between the two exercise conditions. We further identify an immunoproteomic signature that effectively predicts cardiorespiratory fitness. This study provides a reliable data resource that expands our knowledge on how exercise modulates the immune system, and delivers biological evidence supporting the WHO 2020 guidelines, which highlight exercise intensity as a relevant factor to maintain health.\u003c/p\u003e","manuscriptTitle":"Acute exercise rewires the proteomic landscape of human immune cells","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-06-30 12:10:22","doi":"10.21203/rs.3.rs-6864249/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"nature-communications","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"NCOMMS","sideBox":"Learn more about [Nature Communications](http://www.nature.com/ncomms/)","snPcode":"","submissionUrl":"https://mts-ncomms.nature.com/","title":"Nature Communications","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Communications","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"8dabb9ff-b04a-4ff6-a7a1-0171a2cabb61","owner":[],"postedDate":"June 30th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":49944142,"name":"Biological sciences/Immunology/Translational immunology"},{"id":49944143,"name":"Health sciences/Medical research/Translational research"},{"id":49944144,"name":"Health sciences/Biomarkers/Predictive markers"},{"id":49944145,"name":"Biological sciences/Physiology"}],"tags":[],"updatedAt":"2026-01-07T08:10:05+00:00","versionOfRecord":{"articleIdentity":"rs-6864249","link":"https://doi.org/10.1038/s41467-025-68101-9","journal":{"identity":"nature-communications","isVorOnly":false,"title":"Nature Communications"},"publishedOn":"2026-01-02 05:00:00","publishedOnDateReadable":"January 2nd, 2026"},"versionCreatedAt":"2025-06-30 12:10:22","video":"","vorDoi":"10.1038/s41467-025-68101-9","vorDoiUrl":"https://doi.org/10.1038/s41467-025-68101-9","workflowStages":[]},"version":"v1","identity":"rs-6864249","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6864249","identity":"rs-6864249","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}