Muscle Fiber- and Cell Type-Specificity of Training Adaptation in Male Mice

Muscle Fiber- and Cell Type-Specificity of Training Adaptation in Male Mice | bioRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (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];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-M677548'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search New Results Muscle Fiber- and Cell Type-Specificity of Training Adaptation in Male Mice Sedat Dilbaz , Peter Škrinjar , Regula Furrer , Volkan Adak , Alina Stein , Stefan A. Steurer , Gesa Santos , View ORCID Profile Christoph Handschin doi: https://doi.org/10.1101/2025.11.04.686534 Sedat Dilbaz 1 Biozentrum, University of Basel , Basel, Switzerland Find this author on Google Scholar Find this author on PubMed Search for this author on this site Peter Škrinjar 1 Biozentrum, University of Basel , Basel, Switzerland 2 SIB, Swiss Institute of Bioinformatics , Basel, Switzerland Find this author on Google Scholar Find this author on PubMed Search for this author on this site Regula Furrer 1 Biozentrum, University of Basel , Basel, Switzerland Find this author on Google Scholar Find this author on PubMed Search for this author on this site Volkan Adak 1 Biozentrum, University of Basel , Basel, Switzerland Find this author on Google Scholar Find this author on PubMed Search for this author on this site Alina Stein 1 Biozentrum, University of Basel , Basel, Switzerland 3 Novartis AG , Basel, Switzerland Find this author on Google Scholar Find this author on PubMed Search for this author on this site Stefan A. Steurer 1 Biozentrum, University of Basel , Basel, Switzerland Find this author on Google Scholar Find this author on PubMed Search for this author on this site Gesa Santos 1 Biozentrum, University of Basel , Basel, Switzerland Find this author on Google Scholar Find this author on PubMed Search for this author on this site Christoph Handschin 1 Biozentrum, University of Basel , Basel, Switzerland Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Christoph Handschin For correspondence: christoph.handschin{at}unibas.ch Abstract Full Text Info/History Metrics Preview PDF ABSTRACT Skeletal muscle possesses extraordinary plasticity of structure, metabolism, and function in response to repeated contractile activity. As a syncytium embedded within a complex microenvironment, muscle relies on the coordination of distinct myonuclear populations and diverse mononucleated cell types. Here, we present a high-resolution single-nucleus RNA-sequencing atlas of 550’000 skeletal muscle nuclei, capturing the longitudinal transcriptional responses of 17 distinct myonuclear and 21 mononuclear cell populations at multiple time points after one bout of exhaustive exercise in trained and sedentary mice. The transcriptional programs of these populations are further shaped by training status into divergent adaptive trajectories. A subset of oxidative myonuclei enters a delayed regenerative state post-exercise, reflecting a bifurcated response to a disproportionate metabolic load on fibers during endurance exercise. Prior training accelerates homeostatic recovery and shields oxidative nuclei from exacerbated damage signatures. In parallel, mononucleated cells emerge as the primary mediators of intercellular communication during recovery. Together, this dataset establishes that training adaptation emerges through a coordinated interplay of intrinsic adaptive programs of multicellular remodeling, and provides a foundational resource for mechanistic insights into muscle plasticity. INTRODUCTION Skeletal muscle is among the most adaptable tissues in the body, capable of remodeling its structure, metabolism, and function in response to repeated contractile activity. This plasticity underlies the broad physiological benefits of exercise, which protects against many major non-communicable diseases, including cardiovascular disease, type 2 diabetes, obesity, neurodegenerative disorders, and certain cancers 1 . Despite undisputed evidence for these health-promoting benefits, the cellular and molecular mechanisms that orchestrate exercise adaptation remain incompletely understood. Major efforts have begun to chart the systemic and tissue-level transcriptional landscapes of exercise adaptation 2 – 4 , including multi-omics approaches from the Molecular Transducers of Physical Activity Consortium (MoTrPAC) 5 – 7 . These approaches however lack the cellular resolution to reveal how specific fiber types and diverse muscle-resident mononucleated cells (MMCs) contribute to tissue plasticity. This is particularly important in skeletal muscle, a structurally complex tissue composed of multinucleated fibers embedded in a heterogeneous microenvironment that includes vascular, immune, stromal, stem, and nerve-associated cells 8 . These mononuclear cell populations are spatially organized in specialized functional niches such as the neuromuscular and myotendinous junctions 8 – 10 —sites that likely serve as hubs for localized signaling and adaptation. Furthermore, diversity also exists among muscle fibers, which differ markedly in contractile and metabolic properties ranging from slow-twitch, low peak force generating, fatigue resistant and oxidative to fast-twitch, high peak force generating, fatigable and glycolytic, thereby enabling the tissue to meet the varying physiological demands of endurance- and strength-based movements 11 . In a simplified manner, the different fibers can be classified as type I, IIA, IIX, and, in rodents, type IIB according to the predominant expression of the corresponding myosin heavy chain gene ( Myh7 , Myh2 , Myh1 , and Myh4 , respectively). The morphological, functional and metabolic differences are reflected in distinct transcriptional programs, which likely shape how individual fibers and their associated cell types adapt to training. In principle, insights into this cellular heterogeneity and complexity could be gained with single-cell and single-nucleus transcriptomic approaches, but their application to skeletal muscle has been hindered by several technical challenges. Standard dissociation protocols for MMC isolation from muscle tissue 12 introduce strong stress-related artifacts that can mask genuine biological signals 13 , 14 . Moreover, because muscle fibers are multinucleated syncytia, muscle cells are severely underrepresented in single-cell RNA sequencing (scRNA-seq), and single-nucleus RNA sequencing (snRNA-seq) is therefore typically required for adequate coverage. Inversely however, a similar underrepresentation is often observed for rarer, but markedly more heterogenous MMC populations in snRNA-seq, in part due to limitations in enriching MMC nuclei during isolation. As a result, a holistic, unbiased, and temporally resolved framework capturing how distinct myonuclear and MMC populations respond to an acute bout of exercise and how they adapt over a period of endurance training has remained elusive. Here, we present an approach that overcomes these limitations, resulting in a snRNA-seq atlas of mouse skeletal muscle in trained and sedentary states, spanning multiple time points after one bout of exhaustive endurance exercise. Using the probe-based FLEX assay (10x Genomics) on fixed and sorted nuclei, we circumvent dissociation-induced artifacts and achieve high-resolution profiling across all major myonuclear and MMC types, including rare populations. Our approach uncovers longitudinal, fiber type-specific transcriptional trajectories, training-modulated regenerative programs, and the dynamics of multicellular orchestration with unprecedented granularity. Our results reveal that muscle adaptation is not governed by a uniform response but emerges through a finely tuned interplay of fiber- and cell type-intrinsic programs and multicellular coordination, offering new mechanistic insight into the spatial and temporal logic of muscle plasticity. To support future exploration and hypothesis generation, we provide this atlas as an easily accessible resource through the interactive web tool “XXX” ( https://XXX.scicore.unibas.ch/ ). RESULTS Fixed single-nucleus RNA profiling enables an atlas of muscle adaptation to endurance exercise To study fiber- and cell type-specific transcriptional responses to exercise before and after training adaptation, we provided mice with running wheels for 6 weeks ( Fig. 1a ). After this 6-week period, mice exhibited a significant increase in running performance on the treadmill ( Fig. 1b , Extended Data Fig. 1a ), typical increase of M. soleus muscle mass compared to control sedentary mice 15 , as well as a slightly higher lean mass, confirming that physiological adaptations to endurance exercise training have occurred ( Extended Data Fig. 1b–d ). Although single-nucleus transcriptomics are a powerful tool to study fiber- and cell type-specific perturbations in skeletal muscle, these studies can be hampered drastically by artificially induced stress signatures during sample preparation where enzymatic dissociations are performed 13 , 14 . In order to circumvent such potential biases, we applied the probe-based 10X Genomics FLEX assay to isolated and fixed nuclei of either muscle fiber or MMCs at 0h, 6h, and 12h post-exercise exhaustion ( Fig. 1a ). For this, myofiber and MMC nuclei were separated based on pericentriolar material 1 (PCM1) signal at the nuclear envelope, highly enriched on myonuclei ( Fig. 1c,d ) 16 , 17 . Compared to the conventional 3’ single-nucleus assays from 10X Genomics, the probe-based FLEX assay achieved higher gene detection per UMI, and a similar number of overall detected genes ( Extended Data Fig. 1e–g ), hence comparable sensitivity in transcript capture. Moreover, dissociation-induced transcriptional changes of genes like Hspa1a and Junb , markers of stress signatures in cells dissociated from their physiological niche, are prevented in the absence of exercise perturbation ( Extended Data Fig. 1h ). After extensive quality control, we retained high-quality RNA profiles of over 550’000 nuclei, including all known myonuclear and MMC nuclear populations ( Fig. 1e–g ). Additionally, unbiased clustering of our data shows exercise-induced subpopulations of myonuclei (named Ex0h_1–5 or Ex6h_1–2 ) with, at least in part, mixed myosin heavy chain ( Myh ) isoform expression ( Fig. 1h ), alluding to a partially shared, yet also bifurcated exercise response across fiber types. To shed further light on the fiber type-specific transcriptional responses, we categorized all body myonuclei based on their relative Myh expression (see Methods) into well-characterized fiber type groups. This digital myonuclear typing resulted in typical distributions with only small numbers of hybrid myonuclei ( Fig. 1i ), as previously reported 9 . In conclusion, our fixed single nuclear RNA profiling approach resulted in an unprecedented resolution of fiber and cell type-specific transcriptional dynamics post-exercise in trained and sedentary male mice. Download figure Open in new tab Extended Data Fig. 1: Supplementary cohort data and snRNA-seq technical comparison. a , Kaplan-Meier survival curves of mice lasting on the treadmill during the final maximal endurance capacity test. Significance calculated by two-sided log-rank test. b , c , Absolute muscle mass ( b ) and relative muscle mass normalized to body mass (BM) ( c ) of M. quadriceps (QUAD), M. soleus (SOL), M. tibialis anterior (TA), M. extensor digitorum longus (EDL), and M. gastrocnemius (GAS). Whenever available, average of contralateral muscles was taken. n = 16; data are mean ±sem; unpaired student’s t-test; parametric, when normality confirmed via Shapiro-Wilk, otherwise nonparametric Mann-Whitney test. d , From left to right: total body mass, EchoMRI-derived total lean mass, lean mass normalized to total body mass, EchoMRI-derived total fat mass, and fat mass normalized to total body mass. Data are mean ±sem; n = 16 per group, Two-way ANOVA with Tukey’s post hoc test with multiple comparisons. e , f , Number of genes detected per UMI, showing overall higher numbers of detected genes per sequencing reads in the probe-based FLEX assay in myonuclei ( e ) and MMCs ( f ). g , Number of detected genes (left) and UMIs (right) per nucleus or cell, comparing data of multiple snRNA-seq and scRNA-seq datasets. The FLEX assay reaches similar number of detected genes in myonuclei, and MMC nuclei, whereas cells as input in scRNA-seq experiments reach higher numbers of detected genes. h , Violin plots with expression levels of stress-related genes ( Hspa1a and Junb ) in distinct techniques and sample input types. Single-nucleus samples show absence of Hspa1a and Junb expression in non-exercised, basal samples. Single-cell samples display high levels of expression in these genes even in absence of an intervention. Download figure Open in new tab Fig. 1: A high-resolution single-nucleus transcriptomic atlas of muscle adaptation to endurance exercise. a , Scheme of experimental setup. Trained and Sedentary mice (n = 4 per condition and time point) underwent an exhaustive exercise bout on a treadmill and were sacrificed either 0h, 6h or 12h post-exhaustion or without acute exercise as control (PRE), followed by tissue dissection, nuclei fixation, PCM1-based FACS, and single-nucleus RNA sequencing (10X Genomics FLEX) on both myonuclear and mononucleated cell (MMC) nuclear fractions. Figures are in part generated with BioRender.com. b , Distance run during maximal endurance capacity test on the treadmill, showing increased distance run by trained mice. Data are mean ±sem, n = 12 (4 per time point excluding control groups to avoid acute exercise effects), Two-way ANOVA with Tukey’s post hoc test with multiple comparisons. c , Representative scatter plots with gating strategy for PCM1-positive and negative nuclei. Nuclei were sorted based on Hoechst 33342 and APC-conjugated anti-PCM1 antibody. d , Immunofluorescent images of FACS-sorted PCM1-positive myonucleus (left) and PCM1-negative MMC nucleus (right) with Höechst 33342-stained DNA (blue) and anti-PCM1 (red). e , Unifold Manifold Approximation and Projection (UMAP) plot representing the PCM1-positive myonuclear fraction with unsupervised Seurat clustering shows all myonuclear populations together with exercise-specific populations termed “ Ex0h_1–5 ” and “ Ex6h_1–2 ”. f , UMAP of myonuclear fraction with supervised clustering based on myonucleus typing, guided by the type-specific expression of myosin heavy chains ( Myh7 = I, Myh2 = IIA, Myh1 = IIX, Myh4 = IIB). Specialized myonuclei (NMJ, MTJ and spindle) were not assigned to fiber types. g , UMAP representation of PCM1-negative MMC fraction with unsupervised Seurat clustering. h , Dot plot showing average expression and percentage of myonuclei expressing Myh types across unbiased Seurat clustering. i , Dotplot showing average expression and percentage of myonuclei expressing Myh isoforms across clustering from myonucleus typing. IIX myonuclei expand in trained muscle via IIA/IIX hybrid fiber formation Repetitive bouts of endurance exercise induce well-characterized fiber type transitions (IIB→IIX→IIA) in skeletal muscle 2 , but how these adaptations manifest at the myonuclear level remains unclear. To address this gap, we quantified myonuclear populations in M. quadriceps of trained and sedentary mice under baseline conditions. Remarkably, Myh1 -expressing (IIX) myonuclei increased ∼1.8-fold in trained muscle (from 10.17% to 18.35%; q < 0.001), while Myh2 -expressing (IIA) nuclei decreased to ∼0.6 fold from baseline (from 6.62% to 4.16%; q = 0.0016) and Myh4 -expressing (IIB) nuclei declined to ∼0.9-fold from baseline (from 78.78% to 72.16%; q 0.05) for all isoforms, confirming that these proportional shifts reflect changes in myonuclear composition rather than altered Myh expression within a subset of nuclei ( Fig. 2b ). Using classical immunohistological fiber typing (staining for I, IIA and IIB, assign unstained as IIX) of M. quadriceps sections, we confirmed the expected shift toward oxidative types: IIA fibers increased ∼1.8-fold (from 11.14% to 20.82%, q < 0.001), IIX fibers increased ∼1.3-fold (from 13.27% to 17.57%; q = 0.018), and IIB fibers declined to ∼0.8-fold from baseline (from 74.75% to 60.71%; q < 0.001) ( Extended Data Fig. 2a,b ). However, due to the absence of specific staining, this approach neglects the potential of IIX hybrid fiber formation, leading to false-negatives for IIX fiber quantification. Therefore, an alternative staining (staining for I, IIA and IIX, assigning unstained to IIB) revealed that this shift toward a more oxidative distribution was predominantly driven by IIX/IIA hybrid fibers, which increased ∼2.8-fold (4.09% to 11.3%; q = 0.003), while IIB fibers decreased comparably to ∼0.8-fold from baseline (from 73.21% to 61.24%; q < 0.001) ( Fig. 2c,d ). Minimal Feret diameter analysis excluded hypertrophy as a contributor to IIX myonuclear expansion since only pure IIA fibers showed significant size increases (interaction p < 0.001) ( Fig. 2e ). Download figure Open in new tab Extended Data Fig. 2: Transcriptional and compositional baseline difference between trained and sedentary muscle. a , Immunofluorescent fiber typing of M. quadriceps muscle sections. Blue = MYH7 (I), Yellow = MYH2 (IIA), Unstained = IIX, Green = MYH4 (IIB). Insets show magnified regions highlighting fiber type differences. Scale bars, 500 μm (main images) and 100 μm (insets). b , Quantification of immunohistological fiber typing using antibodies against MYH7 (I), MYH2 (IIA), and MYH4 (IIB) showing percentage of total fiber count for each fiber type. IIA and IIX fibers are significantly enriched in trained muscles, at the expense of IIB fibers. Data are mean ±sem; n = 4; Two-way ANOVA with Tukey’s post hoc test. c , Heatmap of relative gene expression (pseudobulk) for mitochondrial and metabolic genes across major fiber types. Expression values counts per million (CPM) for cluster-normalization and are scaled by row. d , UpSet plots showing overlap of differentially expressed genes across fiber types (I, IIA, IIX, IIB) at baseline (PRE) in trained versus sedentary muscle. Genes upregulated (left) and downregulated (right) in trained mice are shown with selected transcription factors (TFs), ligands, and receptors annotated. Genes were identified via DESeq2 (Wald-test) on pseudobulk counts aggregated by biological replicate with p -value adjustment using Benjamini-Hochberg method. Genes with |log 2 FC| > 1 and q < 0.01 were annotated as significant. Download figure Open in new tab Fig. 2: Trained muscles display elevated IIX myonuclei by IIA/IIX hybrid fiber formation. a , Percentage of total myonuclei of each cluster derived by myonucleus typing showing significant increase of IIX myonuclei at the expense of IIA and IIB myonuclei in trained muscle compared to sedentary (n = 4). b , Violin plots showing Myh isoform expression in sedentary and trained muscle by myonucleus type. c , Quantification of immunohistological fiber typing using antibodies against MYH7 (I), MYH2 (IIA), and MYH1 (IIX) showing percentage of total fiber count for each fiber type (n = 4). d , Immunofluorescent fiber typing of M. quadriceps sections. Blue = MYH7 (I), Yellow = MYH2 (IIA), Red = MYH1 (IIX), Unstained = IIB; arrows indicate hybrid IIA/IIX fibers. Insets show magnified regions highlighting fiber type differences. Scale bars, 500 μm (main) and 100 μm (insets). e , Histomorphometry of fiber minimal Feret across fiber types. f , Number of significantly up- or downregulated genes per cluster between trained and sedentary muscle. Genes with |log 2 FC| > 1 and q < 0.01 were counted as significant. Unless otherwise indicated, data are mean ±sem. Significance for non-expression data was calculated using Two-way ANOVA with Tukey’s post hoc test. Significance for gene expression data was measured from sample- and cluster-wise aggregated pseudobulk counts (n = 4) using DESeq2 (Wald-test) with p -value adjustment using Benjamini-Hochberg. To assess functional consequences of these compositional changes, we compared baseline transcriptional profiles between individual myonucleus subpopulations. IIX myonuclei exhibited oxidative metabolic signatures with mitochondrial complex, TCA cycle, and fatty acid oxidation genes matching IIA levels while retaining intermediate levels of glycolytic genes, as well as calcium-handling and lactate metabolism transcripts akin to IIB nuclei ( Extended Data Fig. 2c ). Differential gene expression analysis between trained and sedentary mice of each population at rest, i.e. not affected by an acute exercise bout, revealed in general only few DEGs, with most of the changes occurring in IIA or IIX myonuclei (IIA: 68 DEGs; IIX: 98 DEGs, |log 2 FC| > 1, q < 0.01) ( Fig. 2f and Extended Data Fig. 2d ), reminiscent of the low number of gene expression changes previously observed in bulk tissue analysis in sedentary vs. trained muscle at rest 2 . Together, our results uncover an unexpected layer of exercise training adaptation with a selective enrichment of transcriptionally oxidative IIX myonuclei in IIA/IIX hybrid muscle fibers. Training status dictates fiber type- and cell type-specific transcriptional responses to exercise While chronic exercise training massively remodels muscle morphology and function 18 , 19 , persistent transcriptional changes are relatively rare, as seen in our single nucleus data as well as prior bulk analyses 2 . Nevertheless, sedentary and trained muscle respond differently to an acute exercise bout 2 . To decode this adaptive, training state-dependent programming on a fiber- and cell type-level, we profiled the transcriptional response to an exhaustive acute exercise bout in trained and sedentary mice. Immediately post-exhaustion (at the 0h time point), nearly all myonuclei were transcriptionally recruited, forming exercise-specific activation clusters (termed “ Ex0h_1–5 ”, “ Ex6h_1–2 ”), whereas MMCs exhibited only partial activation ( Fig. 3a and Extended Data Fig. 3a ). In sedentary mice, glycolytic IIB myonuclei returned to baseline within 6h, while oxidative IIA and IIX myonuclei remained perturbed even at 12h ( Fig. 3b ). Strikingly, in trained mice, oxidative myonuclei displayed a significantly faster recovery (IIAX and IIA: interaction p < 0.001), similar to that of glycolytic IIB fibers, reflecting an accelerated transcriptional rebound. This seems to be partially due to fewer nuclei transitioning into Ex6h_2 at 6h and 12h post-exercise ( Ex6h_2 : interaction p = 0.029) ( Fig. 3b ). Download figure Open in new tab Extended Data Fig. 3: Further analysis of exercise-induced gene expression dynamics. a , UMAP projections of MMC fraction from sedentary and trained mice across four time points (PRE, 0h, 6h, 12h). b , c , Stacked bar plots showing the temporal overlap of DEGs across time points, separated by training state, with absolute DEG numbers ( b ) and relative proportions ( c ). DEGs from all clusters are merged per time point. d , Line plots showing Augur -inferred perturbation scoring (AUC) of MMC fraction across time points and training states. e-g , Gene Ontology enrichment of DEGs in myonuclear ( e ), MMC ( f ), and shared fraction ( g ). Dot size indicates –log10( p -value); color denotes directionality (red: upregulated, blue: downregulated). Top five significant terms of each training state and directionality are displayed. SED, Sedentary; TRA, Trained. Download figure Open in new tab Fig. 3: Training remodels the temporal exercise response across muscle cell populations. a , UMAP visualization of all myonuclei across four time points (PRE, 0h, 6h, 12h), colored by unsupervised Seurat clustering, showing myonuclear subtypes and exercise-induced clusters. b , Training state-dependent temporal dynamics of myonuclei subtypes across time points. Data are mean proportion ±sd at each time point. Two-way ANOVA with Tukey’s post hoc test; n = 4 mice per group. Dashed line indicates mean of sedentary baseline (PRE) proportions. c , Bar plot summarizing differentially expressed genes (DEGs) between trained and sedentary groups over time, stratified by myonucleus and cell type, and colored by training-state specificity (unique to trained, unique to sedentary, or shared). DEGs were identified via DESeq2 (Wald-test) on pseudobulk counts aggregated by biological replicate with p -value adjustment using Benjamini-Hochberg method. Genes with |log 2 FC| > 1 and q < 0.01 were counted as significant. d , Line plots showing Augur -inferred perturbation scoring (AUC) of myonuclei subtypes across time points and training states. e , Stacked bar plots indicating the overlap of DEGs with myonuclear (MYO), mononucleated (MMC), or shared signatures separated by direction of regulation and training-state. DEGs from all time points were merged per training state. S, Sedentary; T, Trained. f , Stacked bar plot showing the fraction of up- and downregulated DEGs shared across clusters within the myonuclear (left) or MMC fraction (right). Color intensity reflects the number of clusters in which a given gene is regulated and therefore its cluster-specificity. DEGs from time points were merged. S, Sedentary; T, Trained. Pseudobulk differential gene expression analysis using DESeq2 revealed a robust transcriptional response at 0h in both myonuclei and MMCs (|log 2 FC| > 1, q < 0.01) ( Fig. 3c ). While high numbers of DEGs persisted in myonuclei at 6h and 12h, MMCs showed a sharp and long-lasting decline in DEGs by 6h, with the exception of adipocytes in which the number of transcripts is again elevated at 12h. Training adaptation resulted in the induction of a high number of novel, i.e. not observed in the sedentary muscle, genes in all myonuclear types, whereas this effect is less pronounced for MMC, with further suppressive effects on fibro-adipogenic progenitor (FAP) populations ( Fig. 3c ). Comparison of DEG overlaps across time points revealed that the early 0h transcriptional response is largely distinct from the later time points (6h and 12h). In contrast, approximately half of the DEGs detected at 6h and 12h were already induced at 0h, whereas nearly a quarter of DEGs persisted between the 6h and 12h time points, suggesting a sustained or sequential regulatory component within the recovery trajectory ( Extended Data Fig. 3b,c ). Using the Augur package, we quantified cluster-specific perturbations and found that oxidative myonuclei, particularly IIX myonuclei, exhibited the strongest perturbation post-exercise. Strikingly, this perturbation is further amplified by training adaptation for all myonuclei types ( Fig. 3d ). Although initial activation was higher in trained muscle, the data also suggest faster recovery, highlighting the dual role of training in enhancing both transcriptional responsiveness and resilience, leading to a more rapid regression to baseline. On the contrary, MMCs displayed only minor training-induced changes regarding cell activation, with FAPs becoming less activated, and resident immune cells and adipocytes showing higher responsiveness in trained muscles ( Extended Data Fig. 3d ). Moreover, comparative analysis of DEGs across fiber and cell types revealed striking specificity: only ∼20% of DEGs were shared between myonuclei and MMCs, whereas the majority of DEGs are either specific to myonuclei or MMCs ( Fig. 3e ). Gene Ontology (GO) analysis further confirms this functional dichotomy, with myonuclei-derived DEGs relating to regulation of membrane potential and synapse, while MMCs displayed pathways linked to angiogenesis and inflammation ( Extended Data Fig. 3e,f ). Shared DEGs were associated with general stress responses such as protein folding and apoptotic signaling, hinting towards a broad stress induction with dissimilar downstream consequences ( Extended Data Fig. 3g ). Furthermore, even within myonuclei and MMCs, the transcriptional responses were highly fiber- and cell type-dependent, with >50% of myonuclear DEGs restricted to one or two myonuclear types, and even greater specificity observed in MMCs ( Fig. 3f ). In summary, these findings demonstrate that training status shapes a distinct, fiber- and cell type-specific transcriptional landscape in response to an acute exercise bout, marked by heightened responsiveness in trained muscle. Contextual transcriptional networks drive functional fiber type adaptation While the preceding analyses demonstrated how endurance training modifies the transcriptional responsiveness of distinct myonuclear populations, the underlying functional adaptations driving these differences remain to be elucidated. To uncover these changes, we analyzed the immediate transcriptional response (0h post-exercise) in myonuclei from trained and sedentary mice, clustering DEGs into modules based on directionality and training state-dependency ( Fig. 4a–f and Extended Data Fig. 4a–h ). Download figure Open in new tab Extended Data Fig. 4: Additional transcriptional modules reveal expanded functional rewiring by training. a, Line plot showing normalized and scaled expression patterns of training-state-independently downregulated genes at 0h post-exercise (left) with respective UpSet plot showing contribution of individual myonuclear subtypes (right). b, As in a but for genes with downregulation in trained muscle. c , d , Gene Ontology enrichment networks for training state-independently downregulated gene module ( c ) and training-suppressed gene module ( d ). Node size reflects gene set size; node color shows contributing myonuclear subtypes. e , f , As in a and b but for genes showing upregulation ( e ) or downregulation ( f ) only in sedentary muscle at the 0h time point. g , h , Enrichment networks for gene modules in ( e ) and ( f ). i , Principal component analysis (PCA) of pseudobulk myonuclear transcriptomes at each time point. Each point represents a biological replicate colored by time and shaped by myonuclear subtype. Fill indicates training state. In a , b , e , f curves represent mean expression across myonuclear subtypes ±sd. Download figure Open in new tab Fig. 4: Training reprograms transcriptional networks from general stress response to contextual functional adaptation. a , Line plot displaying normalized and scaled expression of a training-state independently activated gene module at the 0h time point. Gene modules were defined by k-means clustering of expression profiles within each myonuclear subtype (see Methods). Modules showing similar temporal dynamics across subtypes were grouped. Line plot data represent the mean of module-level averages across subtypes ±sd. b , UpSet plot showing overlap of training state-independently upregulated genes at 0h across individual myonuclear clusters. c , Enrichment analysis network for training state-independently upregulated genes at 0h. Node size reflects gene set size; circles are colored by contributing myonucleus subtypes. d , Line plots showing temporal expression patterns of selected stress-related genes across myonuclei subtypes. e , f , As in a , b but for genes with enhanced induction in trained compared to sedentary myonuclei. g , Enrichment analysis networks for training-enhanced genes at 0h. Node size reflects gene set size; circles are colored by contributing myonucleus subtypes. h , Line plots showing temporal expression patterns of selected training-enhanced transcriptional regulators across myonuclei subtypes. i , Heatmaps showing expression levels and SCENIC-inferred transcription factor regulon activity of PGC-1α interaction partners across all time points in major myonucleus subtypes. Dots represent significant changes in comparison to condition-intrinsic PRE controls. S, Sedentary; T, Trained. j , Expression trajectories of genes involved in nutrient-sensing via mTORC1 at the lysosome, including Tfeb and Mtor . Genes were grouped based on functional compartments as indicated. For gene sets, a module depicting the average expression ±sd is shown. For d , h , i : Data are mean ±sem from pseudobulk profiles (n = 4). Statistical significance assessed by DESeq2 (Wald-test) comparing time point expression profiles to training state-specific control; p- values adjusted with the Benjamini–Hochberg method; * q < 0.05, ** q < 0.01, *** q < 0.001. Overall interaction via likelihood ratio test. We first identified a training state-independent module of genes significantly upregulated immediately after exhaustive exercise (sedentary log 2 FC > 1, q < 0.01; interaction |log 2 FC| < 0.5) ( Fig. 4a ). This module displayed both relatively broad fiber type overlap, with many genes common across myonuclear subtypes ( Fig. 4b ). Gene Ontology enrichment highlighted general stress-response pathways, reflected in the regulation of key transcription factors (e.g., Fos , Jun , Atf3 , Egr1 ), protein folding ( Hsp90aa1 , Hspa1a , Dnaja1 , Dnajb1 ), inflammatory signaling ( Nfkb1 , Clu ), and glucose metabolism ( Irs2 , Foxo1 , Csrp3 ) ( Fig. 4c ). Despite this shared response, nuanced fiber type specificity emerged: some genes were broadly upregulated but showed preferential induction in specific fiber types (e.g., Jun , Atf3 in oxidative types), while others were strictly confined to certain types (e.g., Nr3c1 in IIB myonuclei) ( Fig. 4d ). Although trained muscles showed enhanced transcriptional perturbations at 0h, genes involved in general stress-response were dampened ( Fig. 4d ), suggesting that training shifts myonuclei towards a more targeted response, facilitated by greater resilience that blunts the need for a generic stress program. The requirement for repeated exercise bouts to initiate such a specification has been proposed 20 . To explore these training-enhanced adaptations, we isolated a distinct gene module with significantly enhanced activation in trained mice (interaction |log 2 FC| > 0.5, q < 0.01) ( Fig. 4e ). This module showed heightened fiber type-specificity, notably enriched within IIX myonuclei, where 565 of the 1484 identified genes were uniquely changed ( Fig. 4f ). In accordance, principal component analysis revealed a shift toward more oxidative transcriptional profile post-exercise in all type-II myonuclei, with a pronounced training effect in IIX myonuclei ( Extended Data Fig. 4i ). Functional enrichment analysis of the training-enhanced genes uncovered more specific functional programs including transcriptional regulation through the PPAR signaling axis and RNA splicing, structural remodeling and cell junction organization, as well as enhanced membrane trafficking involving vesicle-mediated transport and secretion capacity ( Fig. 4g ). The transcriptional coactivator Ppargc1a (PGC-1α), a versatile and well-established regulator of endurance training adaptation 2 , 21 , 22 , together with several of its interaction partners ( Ppard , Ppara , Foxo1 , Foxo3 , Rxra/b/g , Rora/c , Esrrg , Thrb ) is intricately linked to PPAR signaling. Interestingly, the expression patterns of these metabolic regulators were fiber type-dependent, exhibiting diverse profiles likely driven by distinct physiological demands ( Fig. 4h ). In accordance, analysis of expression levels, as well as SCENIC-inferred regulon activities of additional PGC-1α transcription factors binding partners, revealed contextualized expression patterns and transcription factor activities driven by fiber type identity, general training state, or fiber type-specific adaptive programs ( Fig. 4i ). Downstream expression of genes involved in mitochondrial oxidative metabolism, lipid droplet formation, and ROS handling peaked at 6h as a consequence of their transcriptional regulation at 0h, particularly in trained muscles ( Extended Data Fig. 5 ). Download figure Open in new tab Extended Data Fig. 5: Transcriptional profiles of oxidative metabolism linked to contextual TF networks. Line plot showing time and condition-dependent CPM expression profiles of genes related to oxidative metabolism across main fiber types. Transcription factor names on the right indicate regulons. Data are mean ±sem from pseudobulk profiles (n = 4). Statistical significance assessed by DESeq2 (Wald-test) comparing time point expression profiles to training state-specific control; p- values adjusted with the Benjamini–Hochberg method; * q < 0.05, ** q < 0.01, *** q < 0.001. Overall interaction via likelihood ratio test. TF: transcription factor. For another illustration of fiber type-specific functional outcomes, we focused on the training-enhanced terms “Amino acids regulate mTORC1” and “lysosome organization” which consist of transcripts including Tfeb (a core regulator of lysosomal gene expression 23 ), the recruitment of mTORC1 (encoded by Mtor ) to the lysosome via Ragulator ( Flcn , Fnip1 , Rragc / d , Lamtor1 , Lamtor3 , Rheb) 24 , and lysosomal proteins (v-ATPase subunits, Lamp1/2 , Hexa , etc.) 25 ( Fig. 4j ). Notably, these genes show-training-enhanced expression and display the heighest activation in IIX myonuclei, accompanied with an exclusive significant interaction of Tfeb expression at 0h (interaction log 2 FC = 0.72, q = 0.003) and subsequent Mtor expression (interaction log 2 FC = 0.39, q = 0.008) at the 6h time point ( Fig. 4j ). Together, this may indicate a heightened sensitivity to amino acids, enabling anabolic pathways via mTORC1 initiation to facilitate a quicker transition from the immediate post-exhaustion catabolic state. In conclusion, our data offer detailed insights into how endurance training orchestrates fiber type-specific functional transcriptional networks, facilitating tailored adaptive responses within distinct myonuclear populations while mitigating general stress-responses seen in sedentary counterparts at early time points. Training shields oxidative muscle fibers from exacerbated damage Given our observation that endurance training reshapes the immediate transcriptional activation of myonuclei evoked by an exhaustive exercise bout, we next sought to understand how these adaptations affect myonuclear behavior in the subsequent return to baseline homeostasis at 6h and 12h post-exercise. In our unbiased clustering analysis ( Fig. 3a,b ), we identified two distinct myonuclear populations, Ex6h_1 and Ex6h_2 , that emerged exclusively at later time points. After near-complete transcriptional activation at 0h, ∼10% of myonuclei remained within these activation states at 6h, suggesting a delayed, functionally and spatially bifurcated response, likely reflecting myonuclei responding to greater mechanical, metabolic or other types of burden. To assess fiber type-specific engagement in the late response, we quantified the fraction of IIA, IIX, and IIB myonuclei across time and condition within Ex6h_1 and Ex6h_2 , and all other myonuclei ( Fig. 5a ). In sedentary mice, ∼53% of IIA and ∼46% of IIX myonuclei transitioned into Ex6h_1 , and ∼5.5% and ∼7.5%, respectively, into Ex6h_2 at 6h post-exercise ( Fig. 5a ). In stark contrast, only ∼2% of IIB myonuclei entered Ex6h_1 , and ∼0.4% Ex6h_2 ( Fig. 5a ). Training significantly attenuated these transitions of IIA and IIX myonuclei (IIA- Ex6h_1 , interaction p = 0.009; IIA- Ex6h_2 , interaction p < 0.001; IIX- Ex6h_1 , interaction p < 0.001; IIX- Ex6h_2 , interaction p 75% of Ex6h_1 nuclei originated from oxidative types (IIA, IIXA, IIX), whereas Ex6h_2 showed a more heterogeneous profile, with ∼40% oxidative nuclei and ∼30% unclassified myonuclei due to low Myh transcript expression ( Fig. 5b ). This represents the highest proportion of unclassified nuclei across all clusters ( Extended Data Fig. 6a ), suggesting that a subset of oxidative myonuclei lose fiber identity within Ex6h_2 , potentially due to intensified transcriptional perturbations. Download figure Open in new tab Extended Data Fig. 6: Extended characterization of late-responding myonuclei following exercise. a , Bar plot showing the proportion of unclassified myonuclei across all identified myonuclear clusters. Ex6h_2 exhibits the highest percentage of Myh -unclassifiable nuclei, consistent with transcriptional perturbations. b , Heatmap showing the relative expression of inflammatory mediators across myonuclear clusters. c , Temporal expression profiles of selected inflammatory genes Csf1r and Edar across myonuclear subtypes. d , Expression trajectories of canonical atrophy-related genes ( Klf15 , Trim63 , Fbxo32 ) across myonuclear subtypes. e , Heatmap showing relative expression of transcription factors inferred to be active in Ex6h_2 via SCENIC analysis (see Fig. 5I ). In b , e pseudobulk aggregated counts were normalized for cluster size (CPM) and scaled per row (z-score). In c , d data represent pseudobulk expression (CPM mean ±sem, n = 4 mice per group). Statistical significance was assessed using DESeq2 (Wald-test); Benjamini–Hochberg adjusted p -values are shown: * q < 0.05, ** q < 0.01, *** q < 0.001. Download figure Open in new tab Fig. 5: Training remodels bifurcated recovery by restricting entry into highly perturbed regenerative states. a , Line plots showing fraction of nuclei in clusters Ex6h_1 , Ex6h_2 and all other clusters merged (“ Other ”) across time points and conditions and separated by fiber type. Two-way ANOVA with Tukey’s post hoc test, n = 4 mice per group. b , Donut plots showing the myonuclear subtype composition within the Ex6h_1 and Ex6h_2 clusters, which emerge specifically during the 6h and 12h post-exercise time points. c , Bar plot summarizing the number of DEGs in Ex6h_1 and Ex6h_2 compared to PRE baseline, stratified by regulation direction and training-specific interaction effects. Genes with absolute |log 2 FC| > 1 and q < 0.01 were counted as significant. Wald-test with p- value adjustment based on Benjamini-Hochberg. d , Gene set enrichment analysis of DEGs from Ex6h_1 and Ex6h_2 , showing functional signatures enriched in trained, sedentary, or both groups. Dot size indicates significance (–log10( p- value)); color indicates directionality. e , Heatmap showing the expression of selected genes related to sarcomere assembly and repair, including cytoskeletal, scaffold, and membrane repair factors. f , Line plots showing temporal expression trajectories of selected regeneration-related genes from ( d ) across myonuclear subtypes. g , Heatmap of genes involved in proteasome and muscle atrophy. While proteasome genes are elevated at 6h, canonical atrophy regulators (e.g., Trim63 , Fbxo32 , Klf15 ) peak at 0h. h , Heatmap of neuromuscular junction (NMJ)-related genes including neurotrophic and structural components elevated in Ex6h_2 . i , Line plots showing temporal expression trajectories of representative NMJ-associated genes. Trained muscle displays a blunted expression profile compared to sedentary. j , SCENIC-inferred transcription factor activity (AUC scores) for the top regulons in Ex6h_2 . In e , g , h , j : Pseudobulk aggregated counts or SCENIC-inferred AUC values were normalized for cluster size (CPM) and scaled per row by z-scoring for relative representation. In f , i : Data represent pseudobulk expression (CPM mean ±sem, n = 4 mice per group). Statistical significance was assessed using DESeq2 (Wald-test); Benjamini–Hochberg adjusted p -values are shown: * q < 0.05, ** q < 0.01, *** q < 0.001. To explore the function of these clusters, we performed pseudobulk DEG analysis, comparing Ex6h_1 and Ex6h_2 to all baseline (PRE) myonuclei. We identified 522 and 1’409 training-independent DEGs upregulated in Ex6h_1 and Ex6h_2 , respectively (log 2 FC > 1, q 0.5, q < 0.01). Gene set enrichment analysis revealed regenerative and metabolic terms in both clusters (“plasma membrane repair”, “aerobic respiration” in Ex6h_1 ; “angiogenesis”, “wound healing”, “positive regulation of cell migration” in Ex6h_2 ) ( Fig. 5d ). In trained muscle, these were accompanied by the upregulation of “sensory organ morphogenesis” and “cilium organization,” and downregulation of inflammatory processes (“macrophage chemotaxis”). We identified a repertoire of genes particularly in Ex6h_2 , which were previously associated with sarcomere assembly and repair 10 , 26 , 27 , including cytoskeletal and scaffold proteins ( Enah , Flnc , and Xirp2 ), developmental myosin heavy chains ( Myh9 , Myh10) , and membrane repair factors ( Dysf , Ahnak) ( Fig. 5e ). In parallel, we observed inflammatory mediators, including immune receptors (e.g. Csf1r, Edar) ( Extended Data Fig. 6b ), suggesting that Ex6h_2 myonuclei may integrate both structural and inflammatory cues during regeneration. These genes were primarily activated in oxidative myonuclei, reaching highest levels in unclassified Myh -low nuclei ( Fig. 5f and Extended Data Fig. 6c ). This suggests that oxidative myonuclei, while being more engaged in endurance exercise than their glycolytic counterparts, may mount for an amplified transcriptional response that suppresses Myh isoform expression under excessive stress. Notably, regeneration signatures were significantly attenuated in trained muscle, suggesting a protective effect of training adaptation in preventing exercise-induced damage ( Fig. 5f and Extended Data Fig. 6c ). To assess whether these regenerative populations also exhibited catabolic signatures, we analyzed the expression of proteolytic and atrophy-associated genes. Ex6h_2 showed robust upregulation of proteasomal subunits (log 2 FC > 1, q < 0.01), suggesting ongoing structural remodeling ( Fig. 5g ). In contrast, key atrophy regulators ( Klf15 , Trim63 , Fbxo32 ) and canonical autophagy genes were not elevated in Ex6h_1 or Ex6h_2 , but peaked immediately post-exercise at 0h ( Fig. 5g ). Intriguingly, trained muscle even showed enhanced transient expression of atrophy-associated genes ( Foxo1 , Foxo3 , Klf15 , Trim63 , Fbxo32) at 0h ( Fig. 4i and Extended Data Fig. 6d ), implying that rapid clean-up mechanisms may reduce the need for prolonged regenerative programs later. Beyond structural remodeling, Ex6h_2 also displayed elevated expression of neuromuscular junction (NMJ) and innervation-related genes, including neurotrophic factors, axon guidance regulators and structural proteins ( Utrn , Nav3 , Gdnf ) ( Fig. 5h ). The induction of some of these NMJ-related genes was significantly dampened in trained muscle ( Fig. 5i ), suggesting that neuromuscular integrity may represent an additional layer of exercise-induced damage that training can protect from. Finally, to identify upstream regulators of the transcriptional programs observed in Ex6h_2 , we applied SCENIC-based inference of transcription factor activity. Based on the top 10 transcription factors ranked by regulon activity in Ex6h_2 , we identified elevated activity and matching transcript expression for known regulators of muscle regeneration ( Runx1 , Smad1 ) and inflammation ( Rel , Irf5 ), along with transcription factors with uncharacterized or emerging roles in skeletal muscle, such as Glis3 , Ets2 , and Ikzf5 ( Fig. 5j and Extended Data Fig. 6e ). These results underscore how single-nucleus resolution enables the discovery of context-specific regulatory programs within rare regenerative myonuclear populations. In summary, our data suggest that regeneration after exercise entails a bifurcation in myonuclear fate, with the majority of myonuclei reverting to a homeostatic state while a subset of oxidative myonuclei enter a specialized regenerative program. Furthermore, training adaptation appears to protect myonuclei from transitioning into highly perturb states of regeneration. Training rewires multicellular crosstalk during muscle recovery Increasing evidence highlights the pivotal role of MMCs in orchestrating exercise-induced adaptations 28 . Immune, FAPs, and vascular cells contribute to structural remodeling and metabolic reprogramming, emphasizing the importance of non-fiber populations 29 – 38 . To investigate whether this multicellular orchestration is transcriptionally encoded, we leveraged our snRNA-seq dataset to infer intercellular signaling and its modulation by training. We applied the CellChat 39 , 40 algorithm to model receptor-ligand-based cell-cell communication networks. Myonuclei were classified by type, while MMCs were grouped into functional categories: “Vasculature” (capillary, venous, arterial, and lymphatic ECs, pericytes, smooth muscle), “FAP” ( Dpp4 ⁺, Gdf10 ⁺, Mme ⁺), “Immune” (lymphocytes, mast cells, innate immune cells), and “Glial” (myelinating and terminal Schwann cells, perineurial epithelium, endoneurial cells, and NMJ-capping kranocytes). Kranocytes, a subpopulation of Mme + FAPs, localized to NMJs but not nerves, as confirmed by RNA FISH for Adamtsl2 and Shisa3 ( Extended Data Fig. 7a–c ), and previous studies 41 , 42 . Quantification of total ligand–receptor interactions revealed a drop of interactions at 0h post-exercise, followed by an increase at 6h and 12h, representing an inverse pattern to the peak transcriptional perturbations at 0h. In trained muscle, interaction counts remained consistently lower across all time points, yet followed the same temporal trend ( Fig. 6a ). To identify the main contributors to intercellular communication, we next separated interactions by myonuclear (MYO) and MMC origin. MMCs dominated the signaling landscape throughout the time course, with the increase observed at 6h and 12h being driven exclusively by MMC-MMC interactions ( Fig. 6b ). In accordance, incoming and outgoing signaling strength confirmed a cell type-intrinsic propensity for multicellular engagement ( Fig. 6c ). FAPs and tenocytes showed the highest outgoing signal strength, while vascular cells and muscle stem cells (MuSCs) were the primary recipients. In contrast, myonuclei showed limited signaling capacity throughout recovery ( Extended Data Fig. 7d ). Download figure Open in new tab Extended Data Fig. 7: Identification of kranocytes and extended analysis of exercise-induced intercellular signaling. a , Violin plots showing Adamtsl2 and Shisa3 expression in the sciatic nerve atlas (left) and in the current study (right). Kranocytes ( Mme ⁺ FAP subset) specifically express both markers and are distinct from canonical nerve-associated glia. SC, Satellite cell (neuronal); EpC, Epithelial cell; Per, Pericyte. b , Representative RNA FISH images showing colocalization of Adamtsl2 and Shisa3 transcripts in kranocytes localized at neuromuscular junctions (NMJs). Inset shows high-magnification views of signal specificity. Blue = DAPI; cyan = Neurofilament H (NFH) for nerve; yellow: Adamtsl2 transcript; red = Shisa3 transcript, green = Chrne transcript marking the NMJ region. c , RNA FISH images from sciatic nerve sections reveal the absence of Adamtsl2 and Shisa3 expression in peripheral nerves, confirming that kranocytes are spatially restricted to muscle. d , Scatter plots showing incoming and outgoing intercellular signaling strength across cell types as predicted by CellChat , for sedentary (top row) and trained (bottom row) mice at all time points (PRE, 0h, 6h, 12h). Point size reflects the number of inferred ligand–receptor events per cell type. MMC populations (e.g., FAPs, glial cells, vasculature) exhibit the highest levels of outgoing and incoming signaling. e , Heatmap of differential ligand–receptor interaction numbers between trained and sedentary PRE conditions, stratified by sender–receiver cell type combinations. f , Scaled pathway information flow across selected CellChat signaling categories, showing temporal and training-dependent remodeling of signaling dynamics. Color intensity reflects inferred information flow magnitude (z-score values per row). SED, Sedentary; TRA, Trained. Download figure Open in new tab Fig. 6: Multicellular communication networks during muscle recovery are mainly driven by MMCs. a , Line plot showing the total number of predicted ligand–receptor events across training states and time points, based on CellChat inference. b , Bar plot quantifying ligand–receptor interactions originating from myonuclei (MYO) or mononucleated cells (MMC), across time points and training states. MMCs are the predominant mediators of multicellular signaling at 6h and 12h post-exercise. c , Heatmaps displaying differential ligand–receptor interaction frequencies across cell types in sedentary and trained muscle at 0h, 6h, and 12h post-exercise. Signaling is quantified as differential events compared to respective PRE control. RL: Receptor-Ligand. d , Alluvial plot summarizing upregulated secreted factors based on the overlap of DEG analysis (log 2 FC > 0.5, q < 0.01) and curated predictions of secreted factors (including Ensembl BioMart, Uniprotkb, GO:Extracellular Space, and the Human Protein Atlas). Genes are shown in relation of cell compartment of origin (MYO, MMC, or both), regulation (training-sensitive or -independent), and cluster specificity (single vs multiple populations). e , Gene Ontology enrichment analysis of predicted exercise-induced secreted factors with annotation in at least three of the four reference databases. f , Cell type-colored network graph of upregulated secreted genes and their functional associations. Nodes represent secreted factors; edges indicate functional or pathway co-annotation; inner pie chart denotes cell type contribution; outer ring indicates training-dependence. For each GO term, up to 30 genes were selected based on maximal log 2 FC induction (see Methods). In sedentary mice, the signaling increase at 6h and 12h post-exercise was broadly distributed across FAPs, vascular, immune, adipocytes, and glial populations, indicating widespread MMCs engagement in multicellular communication during regeneration ( Fig. 6c ). In contrast, trained muscles exhibited fewer differential interactions with a shift toward glial signaling. Notably, this represented a re-engagement, as baseline trained muscle showed reduced communication with glial cells, NMJs, and adipocytes ( Extended Data Fig. 7e ). Beyond quantitative changes, exercise and training also reshaped the qualitative nature of intercellular signaling. Global pathway information flow analysis revealed distinct remodeling of major signaling pathways depending on training status and time ( Extended Data Fig. 7f ). To exemplify this complexity, we mapped significantly altered receptor–ligand interactions from myonuclei to glial cells (log₂FC > 0.5, q < 0.01 for ligand or receptor), revealing a dynamic signaling network shaped by time and training ( Extended Data Fig. 8a ). Download figure Open in new tab Extended Data Fig. 8: Exercise-dependent ligand-receptor signaling from myonuclei to glia. a , Circular chord plots visualizing differentially expressed ligand–receptor pairs between myonuclei and glial subpopulations at 0h, 6h, and 12h following exhaustive exercise compared to the respective PRE control of sedentary and trained conditions. Genes were filtered for log₂FC > 0.5 and q < 0.01 for either ligand or receptor. Ligands are shown on the bottom, receptors on the upper side. Colors indicate sender/receiver identity. b , Overlap of all DEGs with predicted secreted proteins from external databases. Secretome prediction was based on union across UniProtkb, BioMart, GO:Extracellular space, the Human Protein Atlas (HPA), and CellChat ’s built-in receptor–ligand database. Red line shows all genes with annotation in at least three of the four databases (independent of CellChat ) for subsequent GO analysis. To extend beyond CellChat ’s curated receptor–ligand pairs, we incorporated external annotations for extracellular protein (see methods). Intersecting these predictions with our DEG dataset revealed a diverse set of exercise-regulated secreted factors ( Extended Data Fig. 8b ). While most of these secreted genes are robustly induced by exercise regardless of training status, a subset exhibits clear training sensitivity. Notably, most secreted factors were expressed from one or few cell types, underscoring the cell type-specificity of their regulation ( Fig. 6d ). GO enrichment analysis revealed hallmarks of structural and functional tissue adaptation, including ECM, immune responses, skeletal system development, and vascular remodeling ( Fig. 6e ). A network view highlights the extent and specificity of secreted factor regulation across cell types, offering a rich resource for discovering novel myo- and exerkines involved in muscle adaptation ( Fig. 6f ). These findings iterate that exercise triggers a delayed, training-dependent multicellular signaling response mainly driven by MMCs. Notably, these interactions likely mark the onset of broader microenvironmental remodeling, extending beyond the acute recovery phase. Our dataset provides a valuable resource to navigate and dissect the cellular communication networks underlying these adaptations. DISCUSSION Skeletal muscle exhibits remarkable plasticity, which underlies the profound health benefits of regular exercise 18 , 19 , 43 – 46 . As syncytia, muscle adaptation emerges from the coordinated transcriptional activity of spatially and functionally distinct myonuclei and diverse MMCs. While recent efforts have begun to address fiber type-specific adaptations 28 , 37 , 47 – 51 , most studies rely on bulk profiling or focus on myofibers in isolation, overlooking the involvement of specialized populations and MMCs, as well as their longitudinal response to acute exercise. To fill this gap, we present the first snRNA-seq atlas capturing trained and sedentary states across multiple time points post-exercise, enabling population-resolved dissection of both myonuclear and MMC compartments. This dataset not only serves as a basis for mechanistic dissection of muscle plasticity but also uncovers previously unrecognized features of training adaptation. Endurance exercise elicits population-specific transcriptional responses and adaptations across both myonuclear and MMC populations. We previously demonstrated that trained muscle exhibits only modest transcriptional changes at rest, yet responds to acute exercise with a markedly altered expression profile, shaped in part by epigenetic remodeling 2 . Building on this, our current single-nucleus dataset reveals that all myonuclear subtypes, including those of different fiber types, engage in distinct transcriptional trajectories post-exercise ( Fig. 3c–f and 4a–h ). For MMCs, cell type-specificity of exercise response was even more striking. ( Fig. 3f ). Prior training further modulated the myonuclear response, dampening broad stress programs while initiating fiber type-intrinsic adaptive signatures. Among all myonuclear subtypes, IIX myonuclei exhibited the strongest transcriptional activation and the most pronounced modulation by training ( Fig. 3d and 4e-g ). Consistent with their role in fiber type transitions along the IIB → IIX → IIA axis in endurance adaptation 15 , 52 , we observed an increased abundance of IIX myonuclei in trained muscle, largely through the emergence of hybrid IIA/IIX fibers ( Fig. 2a and 2c ). Together, these findings add key layers to the existing framework of muscle plasticity in response to exercise, demonstrating that both acute responses and adaptations are shaped by population-specific transcriptional programs that can now be systematically dissected using our dataset. At 6h post-exercise, a bifurcated response emerged among oxidative myonuclei (IIA and IIX). While the majority began reverting to baseline states, a subset of oxidative myonuclei transitioned into a regenerative trajectory ( Fig. 5a ). This state was marked by co-expression of genes associated with structural remodeling, inflammation, proteostasis, partial denervation, and loss of clear fiber type identity, suggesting a deprogramming process during extensive regeneration ( Fig. 5c,e,g,h ). These signatures resemble transcriptional states previously observed in development, aging, and muscular dystrophies, pointing to a conserved response to structural stress 10 , 26 , 27 . Notably, training dampened the expression of these genes and prevented the emergence of a highly perturbed states ( Fig. 5a ). While chronic exercise has been shown to promote damage resilience 2 , 53 and support regeneration 54 , our data now capture these transcriptional dynamics in their specialized myonuclear context. These findings suggest that the physiological demands of endurance exercise, i.e. metabolic, mechanical, thermic, redox, proteostatic or of other nature 19 , are unevenly distributed across fiber types, with oxidative myonuclei disproportionately burdened by, and engaged in structural recovery. Our dataset reveals that even myonuclei of the same fiber type can diverge in their transcriptional fate, enabling finely tuned regulation of recovery and plasticity. MMCs emerge as the primary drivers of multicellular coordination during post-exercise muscle remodeling. While their early transcriptional perturbation mirrored that of myonuclei ( Fig. 3c,d ), their intercellular activity intensified over time, even as their expression profiles largely returned to baseline ( Fig. 3c and Fig. 6b,c ). These findings suggest a dual role for MMCs: an acute, cell-intrinsic response, followed by coordinated participation in multicellular remodeling during recovery. Previous exercise studies have proposed MMCs as paracrine modulators of fiber function 30 , 38 and as structural regulators of the microenvironment, supporting vascular remodeling 29 , immune coordination 33 , and extracellular matrix dynamics 35 . However, technical limitations have hindered an unbiased and comprehensive understanding of their contributions. Single-cell approaches introduce strong dissociation-induced artifacts 13 , 14 , overshadowing genuine exercise responses. Conversely, single-nucleus methods often underrepresent MMC diversity due to their scarcity and high heterogeneity. By preserving native transcriptional states and enabling robust detection of diverse MMC populations, our dataset provides a framework to dissect the intricate intercellular signaling events that underpin multicellular tissue remodeling during exercise recovery. By generating the first comprehensive, single-nucleus transcriptomic atlas of skeletal muscle across training states and time points, our study reveals new layers of complexity in muscle adaption to endurance exercise. Rather than following a uniform program, adaptation emerges from the interplay of fiber- and cell type-specific trajectories, spatially coordinated myonuclear fate decisions, and shifts in myonuclear composition. By capturing how each compartment of muscle adapts to repeated contractile demands, this atlas lays the foundation for future discoveries in muscle plasticity and offers a blueprint for developing therapeutic strategies grounded in the physiological benefits of exercise. Our study has several limitations. While we generated a holistic dataset covering early and intermediate time points (0h, 6h, 12h), it does not capture the full arc of exercise response, which may last up to a week 33 . Transcriptional perturbation was still ongoing after 12h and multicellular interactions likely continue beyond this window. Additionally, the 10X Genomics FLEX assay targets a subset of protein-coding genes and excludes non-protein-coding RNAs, which may also contribute to training adaptation 55 – 57 . Finally, our data were obtained in male mice, M. quadriceps , and endurance exercise, thus reflecting the physiological demands of this context. Further studies using additional muscle beds, resistance exercise paradigms, and female animals will be essential to uncover muscle-, sex- and modality-dependent aspects of muscle plasticity 5 . By generating the first comprehensive, single-nucleus transcriptional atlas of skeletal muscle across training states and time points, our study reveals new layers of complexity in muscle adaption to endurance exercise. Rather than following a uniform program, adaptation emerges from the interplay of fiber- and cell-type-specific trajectories, spatially coordinated myonuclear fate decisions, and shifts in myonuclear composition. By capturing how each compartment of muscle adapts to repeated contractile demands, this atlas lays the foundation for future discoveries in muscle plasticity and offers a blueprint for developing therapeutic strategies grounded in the physiological benefits of exercise. MATERIAL AND METHODS Animals Male C57BL/6J mice (Janvier Labs) were housed in the animal facility of the Biozentrum, University of Basel, under controlled conditions (22°C, 12-hour light/dark cycle) with ad libitum access to food (standard chow) and water. All experimental procedures were approved by the Cantonal Veterinary Office Basel-Stadt and conducted in accordance with Swiss regulations for animal care and experimentation. Mice were acclimatized to the facility for at least one week prior to handling. At 11 weeks of age, animals were randomly assigned to either a trained or sedentary group. The trained group was provided with voluntary running wheels equipped with hourly distance monitoring over a 6-week period. Sedentary control mice received comparable environmental enrichment in the form of plastic housing structures. Body composition was assessed using an EchoMRI TM -100 analyzer in conscious, restrained animals at the start of the running wheel period and after 4 weeks of training. Exhaustive exercise treadmill test A maximal endurance capacity test was performed on a motorized treadmill (5° incline) at the beginning and end of the 6-week training period. The final test served both as an assessment of endurance adaptation and as an acute exercise intervention, with exhaustion marking the 0 h time point for molecular profiling. Running wheels were removed 48h prior to treadmill acclimatization to minimize acute effects of voluntary activity. Mice were acclimatized to the treadmill over two consecutive days with 5 minutes each at 0, 5, and 8 m/min for the first day and 5 minutes each at 5, 8, and 12 m/min on the second day. On the third day, mice underwent the maximal exercise protocol, starting with 5 minutes at 5 m/min, followed by 5 minutes at 8 m/min, then an increase of 2 m/min every 3 minutes until reaching 32 m/min. Then, speed was increased by 2 m/min every 10 minutes until exhaustion. Mice were considered exhausted when they were unable to resume running despite mild electrical stimuli (0.5 mA, 200 ms pulses at 1 Hz) and gentle prodding. Animals were sacrificed either immediately (0h), 6h or 12h after exhaustion. Muscles were removed, weighted, snap frozen in liquid nitrogen and stored at −80°C until further analysis. Muscles for histology were embedded in OCT and frozen in isopentane at −160°C. Mice assigned to the control “PRE” time point (from both sedentary and trained groups) were not subjected to the treadmill test to avoid confounding acute transcriptional responses. Nuclei Isolation and FACS Matrix-free nuclei isolation from frozen skeletal muscle tissue was established for subsequent 10X Genomics Flex processing. In short, 50 mg of frozen and crushed M. quadriceps was lysed using a bead-based tissue homogenizer and lysis buffer containing 10 mM NaCl, 3 mM MgCl2, 10 mM TRIS, Protease Inhibitor (Roche - cOmplete), 0.1 % Triton X-100, 1 nM DTT, and 1 U/µl RNAse inhibitor (Promega - RNasin). Nuclei suspension was filtered through 70 µm and 30 µm cell strainers with a washing and centrifugation step (500g, 5 min, 4 °C) after each filtering using a Wash Buffer with 2 % BSA and RNAse inhibitor (0.2U/µl) in PBS. The final nuclei pellet was fixed for 1h at room temperature and quenched using the 10X Genomics Fixation Kit (1000497). Afterwards, fixed nuclei were stained with Hoechst 33342 and an APC-conjugated anti-PCM1 antibody (#HPA023370, conjugated with Lightning-Link Alexa Fluor-647 Kit #336-0010) for 30 min at 4°C. After final washing, the nuclei suspension was sorted for PCM1-positive and –negative fractions using a BD Aria Fusion Sorter. A small sample of sorted nuclei was taken for imaging. Sorted nuclei fractions where prepared for cryo-storage according to the 10X Genomics Flex protocol and stored at −80°C until the day of sequencing library preparation. Single-nucleus library preparation Single-nucleus RNA-seq libraries were generated using the Chromium Next GEM Fixed RNA kit (10X Genomics, User Guide Rev D CG000522). Nuclei were quantified using Acridine Orange/Propidium Iodide staining and counted with a LUNA-FL fluorescence cell counter. Following probe hybridization, samples were pooled for equal representation (pool and wash option). A total of 16 samples were multiplexed per run, each starting with at least 80’000 nuclei. One biological replicate from each group was included in every run. Myonuclei and muscle-resident mononucleated cell (MMC) nuclei libraries were prepared separately. Sequencing library quality was confirmed via Bioanalyzer. Sequencing was done with a NovaSeq6000 (Illumina) targeting 15’000 reads per nucleus. Fiber typing via immunofluorescence To determine muscle fiber type composition, two immunofluorescence staining protocols were performed in parallel on consecutive 10 μm cryosections of the M. quadriceps . One protocol targeted MYH7 (I), MYH2 (IIA), and MYH4 (IIB), with fibers negative for all three classified as IIX. The second protocol targeted MYH7 (I), MYH2 (IIA), and MYH1 (IIX), with triple-negative fibers classified as IIB. Sections were air-dried for 10 minutes at room temperature, followed by 5 minutes of rehydration in PBS. Blocking was performed for 30 minutes at room temperature in PBS containing 0.4% Triton X-100 and 10% goat serum. After three washes in PBS (5 minutes each), sections were incubated for 1h at room temperature with a primary antibody mix diluted in PBS + 10% goat serum. The mix for the MYH4-staining protocol contained BA-D5 (anti-MYH7, DSHB, 1:50), SC-71 (anti-MYH2, DSHB, 1:200), and BF-F3 (anti-MYH4, DSHB, 1:100), whereas the MYH1-staining protocol contained BA-D5 and SC-71 with 6H1 (anti-MYH1, DSHB, 1:25). After primary incubation, sections were washed in PBS (3 × 5 min), followed by a 1h incubation with the appropriate secondary antibody mix in PBS + 10% goat serum. The MYH4 protocol used Alexa Fluor 568 goat anti-mouse IgG1 (A21124, Invitrogen), Alexa Fluor 488 goat anti-mouse IgM (A21042, Invitrogen), and DyLight 405 goat anti-mouse IgG2b (115-475-207, Jackson ImmunoResearch). The MYH1 protocol was identical except that Alexa Fluor 647 goat anti-mouse IgM (A21238, Invitrogen) was used in place of Alexa Fluor 488. All secondary antibodies were used at 1:100 dilutions. After final PBS washes (3 × 5 min), sections were dehydrated in 70% ethanol (2 × 2 min) followed by 100% ethanol (2 × 2 min), mounted with mounting medium, and imaged using a Zeiss Axio Scan.Z1 slide scanner. For fiber type quantification, cryosections were selected at comparable anatomical depth across all samples. Automatic muscle fiber detection was performed using a customized version of the “1_identify_fibers.py” script from the Myosoft Fiji toolkit (Hyojung-Choo/Myosoft), adapted for in-house use. Following segmentation, individual fibers were manually classified based on their staining in all fluorescence channels. Morphometric parameters, including cross-sectional area and minimal Feret diameter, were extracted with Fiji. RNA-FISH RNA-FISH was performed on 10 μm cryosections of M. quadriceps using the RNAscope Multiplex Fluorescent v2 assay (ACD Bio) according to the manufacturer’s standard protocol. Briefly, sections were fixed with 4% paraformaldehyde for 10 minutes on ice, rinsed with PBS, and dehydrated through graded ethanol. Protease IV treatment was applied for 5 minutes, followed by probe hybridization, signal amplification, and fluorophore development as described in the user manual. Subsequently, immunofluorescence staining was performed to label either the basal lamina or neuronal structures using an anti-Laminin antibody (ab11575, Abcam, 1:200) or an anti-Neurofilament H antibody (AB1991, Merck, 1:200), respectively. Sections were rinsed twice with TBST (TBS containing 1:20,000 Tween-20, pH 7.6), blocked for 1h at room temperature in TBS supplemented with 10% goat serum and 0.1% BSA, and then incubated overnight at 4°C with the primary antibody diluted in TBS + 0.1% BSA. After three washes with TBST, sections were incubated for 30 minutes at room temperature with the secondary antibody (Alexa Fluor 750 goat anti-rabbit IgG, A21039, Invitrogen; 1:500 dilution). Following final TBST washes, sections were mounted with DAPI-containing mounting medium and imaged. Data Analysis Preprocessing, quality control, data integration, and clustering Feature-barcode matrices for each sample were generated using the multi pipeline from CellRanger (version 7.1.0) using the mouse GRCm39 (mm10) reference genome. Ambient RNA contamination was removed using the remove-background function of CellBender 58 (version 0.3.0). Processed count matrices were imported into R (version 4.3.2) for downstream analysis with Seurat 59 (version 5). Initial quality control included filtering out nuclei with fewer than 500 or more than 5’000 detected features, total UMI counts exceeding 15’000, or mitochondrial transcript content greater than 3%. Putative doublets were identified and scored using the DoubletFinder algorithm, and samples were subsequently merged, treating myonuclei and MMC nuclei separately. Integration across samples was performed using Seurat ’s IntegrateLayers function with batch correction based on Harmony . Clusters enriched for high doublet scores were excluded, followed by reclustering of the cleaned dataset. In the MMC dataset, additional filtering was applied to remove contaminating myonuclei by excluding nuclei with Titin ( Ttn ) expression above a threshold of 1. “Digital” myonucleus typing To enable fiber type–specific differential expression analysis, myonuclei were classified into subtypes based on the expression of the four major myosin heavy chain ( Myh ) isoforms: Myh1 (IIX), Myh2 (IIA), Myh4 (IIB), and Myh7 (I). For each nucleus, the total Myh expression was calculated as the sum of log-normalized expression values obtained from Seurat ’s NormalizeData function. A myonucleus was assigned to a specific fiber type if at least 45% of its total Myh signal originated from a single isoform and the expression level of that isoform exceeded 1.5. Hybrid fiber types were defined when two isoforms each contributed >45% of the total Myh expression and both had expression levels above 1.5. Nuclei annotated as specialized subtypes, including neuromuscular junction (NMJ), myotendinous junction (MTJ), and spindle fiber nuclei, were retained under their original annotation. Myonuclei that did not meet any of the above criteria were labeled as “Unclassified”. Pseudobulk differential gene expression analysis Differential gene expression analysis was performed using DESeq2 60 in a pseudobulk framework. For each cell type or myonuclear subtype, raw UMI counts were aggregated by biological replicate (n = 4), condition, time point, and cluster identity. The resulting pseudobulk count matrices were used to construct DESeq2 datasets, including the interaction term ( ∼ condition + time + condition:time ). Prior to model fitting, low-expressed genes were filtered out, keeping genes with cluster-size adjusted minimal counts in at least 4 pseudobulk samples. After filtering, DESeq2 was run using default settings, followed by log 2 FC shrinkage with the lfcShrink function using the “ ashr ” 61 method to improve effect size estimation. Result tables were extracted for each time point in comparison to the condition-specific “PRE” baseline, as well as the respective interaction terms. To identify genes enriched in the Ex6h_1 and Ex6h_2 clusters, we adjusted the existing unsupervised Seurat clustering by combining all myonuclei of the “PRE” time point as control cluster (clustering termed “PRE.combined”). We then applied pseudobulk aggregation based on replicates, condition, and the adjusted “PRE.combined” clustering and ran DESeq2 with the interaction term (∼ condition + PRE.combined + condition:PRE.combined ), allowing assessment of the main effect in the Ex6h_1 and Ex6h_2 clusters, as well as the training-state-dependent differences. Genes with at least 20 counts in 4 samples were kept and log 2 FC was shrunk as before. Genes were considered significantly regulated in comparisons if they had a baseMean ≥ 10, absolute log 2 FC ≥ 1, and q < 0.01. For interaction terms, a log 2 FC threshold of 0.5 was used, with the same criteria for baseMean and q -value. Gene expression heatmaps were generated based on aggregated expression values from Seurat ’s AggregateExpression function, grouped by time point and myonuclear subtype. Replicates were pooled for better heatmap visualization. Differences in sequencing depth and cluster size were normalized by conversion to counts per million (CPM). CPM values were then log-transformed and z-score normalized by gene (row-wise) to highlight relative expression pattern across groups. Heatmaps were generated using pheatmap . Line plots were generated by aggregating raw counts per replicate, cluster, and group, followed by conversion to CPM and visualization using ggplot2 . Mean expression values were displayed across time points for each condition, with error bars indicating the standard error of the mean. Definition of gene set modules Gene modules with coherent expression patterns across time and training were defined by selecting genes according to the following criteria across all myonuclear subtypes. We defined two categories: training-independent genes, which were significantly regulated in sedentary mice without a significant condition interaction, and training-dependent genes, which included sedentary-regulated genes with a significant interaction term as well as all other genes with significant interaction effects (|log 2 FC| ≥ 0.5, q < 0.01). For each cluster, aggregated expression matrices were generated and filtered to retain either the training-independent or training-dependent gene sets, resulting in two matrices per cluster. For each matrix, CPM values were log-transformed, mean-centered, and row-wise z-score normalized. Heatmaps were generated and clustered using k-means to identify distinct temporal expression profiles. Genes belonging to similar expression modules across clusters were then grouped into broader training-independent or training-dependent modules. For each module, average expression across clusters was calculated, and line plots were generated displaying the mean trajectory with ribbons representing the standard deviation across contributing clusters. The overlap of gene identities across myonuclear subtypes within each module was visualized using UpSet plots, highlighting the subtype-specific contributions to each expression program. Only nuclei assigned to clearly defined fiber types (I, IIA, IIX, IIB), NMJ, or MTJ subtypes were included; hybrid and unclassified myonuclei were excluded from this analysis. Enrichment analysis Ontology-based enrichment analysis was performed in a myonuclear subtype-specific manner using the Metascape 62 web platform ( metascape.org ) with default parameters. To account for differences in gene expression repertoires between myonuclear subtypes and MMC types, we provided custom background gene lists whenever applicable. Metascape clusters enriched terms by computing pairwise similarity based on shared gene membership and applies hierarchical clustering to group related terms. For visualization, only the representative term of each cluster was displayed. Enrichment network plots generated by Metascape were further processed and visually adapted using Cytoscape (version 3.10.2) . To functionally characterize the Ex6h_1 and Ex6h_2 clusters, we performed rank-based gene set enrichment analysis (GSEA) using the gseGO function from the clusterProfiler 63 R package (version 4.10.1) with the org.Mm.eg.db annotation database. For each cluster, GSEA was applied separately to both the main effect (Ex6h vs. PRE myonuclei) and the condition interaction results from the DESeq2 analysis. Gene lists were ranked by log 2 FC and enrichment was performed against the Gene Ontology (GO) Biological Process ontology with a gene set size between 10 and 500 and a p -value cutoff of 0.05. For visualization, we displayed only the top five enriched terms per analysis direction (up- or downregulated) and category (main effect or interaction term). Perturbation scoring Perturbation analysis was performed using the Augur 64 R package ( github/neurorestore/Augur , version 1.0.3) to quantify transcriptional activation between baseline (PRE) and post-exercise myonuclei and MMC at each time point (0h, 6h, 12h), separately for sedentary and trained mice. Augur’s calculate_auc() function was applied to compute area under the curve (AUC) scores per cell type, with permutation-based controls ( augur_mode = ‘permute’ ) used to establish null distributions. To assess training-dependent shifts in perturbation magnitude, we applied Augur ’s calculate_differential_prioritization() function to compare AUCs between sedentary and trained groups, using 100 subsamples and 1000 permutations. Unclassified and spindle nuclei were excluded from this analysis. AUC trajectories and subtype rankings were visualized across time points for each condition using line plots. Transcription factor regulon activity analysis Transcription factor (TF) activity was inferred using the pySCENIC 65 pipeline (version 0.12.1) with default settings. Raw UMI count matrices were exported from the Seurat object and used as input for gene regulatory network (GRN) inference with GRNBoost2 from the arboreto package. TF-target modules were identified using modules_from_adjacencies , followed by motif enrichment pruning using cisTarget databases (mm10 v10nr) to refine regulons. Regulon activity scores were then calculated using AUCell , which estimates the enrichment of each regulon’s targets within the expression profile of individual nuclei. The resulting AUC matrix was imported into Seurat as a new assay (“TF.AUC”) using CreateAssayObject , enabling visualization of regulon activity across UMAP embeddings and metadata variables via standard Seurat functions. Cell-cell communication and secreted factor analysis Intercellular communication analysis was performed using the CellChat 40 R package (version 2.1.2), following the standard comparative workflow. Prior to CellChat analysis, we merged the processed Seurat objects of myonuclei and MMC populations and re-ran the standard Seurat preprocessing pipeline, including normalization, integration, and dimensionality reduction. However, we retained the original clustering obtained from the individual datasets to preserve subtype resolution. For improved interpretability in the context of cell–cell communication, MMC clusters were grouped into broader functional categories as described in the main text. CellChat was run separately for each group defined by condition and time point to enable comparative analysis of signaling networks. The standard CellChat pipeline was applied, including inference of overexpressed ligand–receptor pairs, pathway-level communication probabilities, and visualization of dynamic signaling patterns. To extend beyond curated ligand–receptor pairs, we annotated secreted factors using four databases with secreted factor prediction: Ensembl BioMart, UniProtKB, GO:Extracellular Space (GO:0005615), and the Human Protein Atlas. Proteins with predicted transmembrane domains were excluded from these references whenever possible. We intersected this reference set with all DEGs showing a log 2 FC > 0.5 and q < 0.01 across all clusters and time points and displayed in an alluvial plot the cellular origin (myonuclei or MMCs), training-dependence, and cluster specificity of these predicted secreted DEGs. Among these, factors annotated by at least three databases were subjected GO enrichment using Metascape . From the top four GO terms, we extracted up to ten most strongly training-enhanced, training-suppressed, and training-independently upregulated genes and visualized them in a cluster-annotated network via Cytoscape . Statistics Differential gene expression analyses and time-point-specific interaction effects were calculated using DESeq2 with the Wald test and Benjamini–Hochberg correction for multiple testing. Overall interaction terms were evaluated using DESeq2 ’s built-in likelihood ratio test (LRT) framework to assess the effect of training condition on the temporal response. For pairwise comparisons of quantitative measurements, we used unpaired Student’s t-tests when normality was confirmed by the Shapiro–Wilk test; otherwise, the nonparametric Mann–Whitney test was applied. For multi-factorial comparisons (e.g., body composition, myonuclear cluster proportions, endurance performance, and regulon activity), two-way ANOVA followed by Tukey’s post hoc test was used. Unless otherwise stated, n = 4 mice per group. Author contributions C.H. and S.D. conceptualized and managed the study; S.D. performed snRNA-seq, including sample preparation, library preparation, and snRNA-seq data analysis. P.S. performed data analysis for transcription factor activity prediction; A.S., S.A.S., G.S. prepared histological samples; V.A., R.F., S.D. performed and analyzed histological experiments; S.D. performed RNA-FISH; S.D., A.S., S.A.S., G.S. performed animal experiments; S.D. assembled figures; C.H. and S.D. wrote the manuscript with the help of all other authors. Data availability Single-nucleus transcriptomics raw data and fully analyzed Seurat files have been deposited to GEO under the accession number XXX. The data can also be queried through the interactive web tool “XXX” ( https://XXX.scicore.unibas.ch/ ). All other data are available upon reasonable request. Code availability The authors declare no original code. Competing interests The authors declare no competing interests. Declaration of generative AI and AI-assisted technologies in the writing process During the preparation of the manuscript, the authors used ChatGPT (4o, 5) in order to improve readability. The final content of this manuscript was reviewed and edited as needed and the authors take full responsibility for the content of the publication. Acknowledgments This study was supported by the Swiss National Science Foundation (Grant no. CRSII5_209252). We thank the FACS Core Facility, the Imaging Core Facility and the Animal Facility of the Biozentrum/University of Basel and the Genomics Core Facility (D-BSSE, ETH Zürich) for their invaluable help with animal caretaking and sample analysis. Data processing and analysis were performed at sciCORE ( http://scicore.unibas.ch ) scientific computing center at the University of Basel, with support by the SIB (Swiss Institute of Bioinformatics). Some of the graphics were created with Biorender.com. Funder Information Declared Swiss National Science Foundation , CRSII5_209252 References ↵ Pedersen , B. K. & Saltin , B . Exercise as medicine - evidence for prescribing exercise as therapy in 26 different chronic diseases . Scand J Med Sci Sports 25 Suppl 3 , 1 – 72 ( 2015 ). doi: 10.1111/sms.12581 OpenUrl CrossRef ↵ Furrer , R. et al. Molecular control of endurance training adaptation in male mouse skeletal muscle . Nat Metab 5 , 2020 – 2035 ( 2023 ). doi: 10.1038/s42255-023-00891-y OpenUrl CrossRef Amar , D. et al. Time trajectories in the transcriptomic response to exercise - a meta-analysis . Nat Commun 12 , 3471 ( 2021 ). doi: 10.1038/s41467-021-23579-x OpenUrl CrossRef PubMed ↵ Pillon , N. J. et al. Transcriptomic profiling of skeletal muscle adaptations to exercise and inactivity . Nat Commun 11 , 470 ( 2020 ). doi: 10.1038/s41467-019-13869-w OpenUrl CrossRef PubMed ↵ MoTr , P. A. C. S. G. , Lead , A. & MoTr , P. A. C. S. G . Temporal dynamics of the multi-omic response to endurance exercise training . Nature 629 , 174 – 183 ( 2024 ). doi: 10.1038/s41586-023-06877-w OpenUrl CrossRef PubMed Sanford , J. A. et al. Molecular Transducers of Physical Activity Consortium (MoTrPAC): Mapping the Dynamic Responses to Exercise . Cell 181 , 1464 – 1474 ( 2020 ). doi: 10.1016/j.cell.2020.06.004 OpenUrl CrossRef PubMed ↵ Amar , D. et al. The mitochondrial multi-omic response to exercise training across rat tissues . Cell Metab 36 , 1411 – 1429 e1410 ( 2024 ). doi: 10.1016/j.cmet.2023.12.021 OpenUrl CrossRef ↵ Giordani , L. et al. High-Dimensional Single-Cell Cartography Reveals Novel Skeletal Muscle-Resident Cell Populations . Mol Cell 74 , 609 – 621 e606 ( 2019 ). doi: 10.1016/j.molcel.2019.02.026 OpenUrl CrossRef PubMed ↵ Dos Santos , M. et al. Single-nucleus RNA-seq and FISH identify coordinated transcriptional activity in mammalian myofibers . Nat Commun 11 , 5102 ( 2020 ). doi: 10.1038/s41467-020-18789-8 OpenUrl CrossRef PubMed ↵ Kim , M. et al. Single-nucleus transcriptomics reveals functional compartmentalization in syncytial skeletal muscle cells . Nat Commun 11 , 6375 ( 2020 ). doi: 10.1038/s41467-020-20064-9 OpenUrl CrossRef PubMed ↵ Schiaffino , S. & Reggiani , C . Fiber types in mammalian skeletal muscles . Physiol Rev 91 , 1447 – 1531 ( 2011 ). doi: 10.1152/physrev.00031.2010 OpenUrl CrossRef PubMed ↵ Liu , L. , Cheung , T. H. , Charville , G. W. & Rando , T. A . Isolation of skeletal muscle stem cells by fluorescence-activated cell sorting . Nat Protoc 10 , 1612 – 1624 ( 2015 ). doi: 10.1038/nprot.2015.110 OpenUrl CrossRef PubMed ↵ van den Brink , S. C. et al. Single-cell sequencing reveals dissociation-induced gene expression in tissue subpopulations . Nat Methods 14 , 935 – 936 ( 2017 ). doi: 10.1038/nmeth.4437 OpenUrl CrossRef PubMed ↵ Machado , L. et al. Tissue damage induces a conserved stress response that initiates quiescent muscle stem cell activation . Cell Stem Cell 28 , 1125 – 1135 e1127 ( 2021 ). doi: 10.1016/j.stem.2021.01.017 OpenUrl CrossRef PubMed ↵ Leuchtmann , A. B. , Afifi , Y. , Ritz , D. & Handschin , C . Effects of high-resistance wheel running on hallmarks of endurance and resistance training adaptations in mice . Physiol Rep 11 , e15701 ( 2023 ). doi: 10.14814/phy2.15701 OpenUrl CrossRef PubMed ↵ Bengtsen , M. et al. Comparing the epigenetic landscape in myonuclei purified with a PCM1 antibody from a fast/glycolytic and a slow/oxidative muscle . PLoS Genet 17 , e1009907 ( 2021 ). doi: 10.1371/journal.pgen.1009907 OpenUrl CrossRef PubMed ↵ Winje , I. M. et al. Specific labelling of myonuclei by an antibody against pericentriolar material 1 on skeletal muscle tissue sections . Acta Physiol (Oxf) 223 , e13034 ( 2018 ). doi: 10.1111/apha.13034 OpenUrl CrossRef PubMed ↵ Furrer , R. & Handschin , C . Molecular aspects of the exercise response and training adaptation in skeletal muscle . Free Radic Biol Med 223 , 53 – 68 ( 2024 ). doi: 10.1016/j.freeradbiomed.2024.07.026 OpenUrl CrossRef PubMed ↵ Furrer , R. , Hawley , J. A. & Handschin , C . The molecular athlete: exercise physiology from mechanisms to medals . Physiol Rev 103 , 1693 – 1787 ( 2023 ). doi: 10.1152/physrev.00017.2022 OpenUrl CrossRef ↵ Thomas , A. C. Q. , Stead , C. A. , Burniston , J. G. & Phillips , S. M . Exercise-specific adaptations in human skeletal muscle: Molecular mechanisms of making muscles fit and mighty . Free Radic Biol Med 223 , 341 – 356 ( 2024 ). doi: 10.1016/j.freeradbiomed.2024.08.010 OpenUrl CrossRef PubMed ↵ Handschin , C. et al. Skeletal muscle fiber-type switching, exercise intolerance, and myopathy in PGC-1alpha muscle-specific knock-out animals . J Biol Chem 282 , 30014 – 30021 ( 2007 ). doi: 10.1074/jbc.M704817200 OpenUrl Abstract / FREE Full Text ↵ Lin , J. et al. Transcriptional co-activator PGC-1 alpha drives the formation of slow-twitch muscle fibres . Nature 418 , 797 – 801 ( 2002 ). doi: 10.1038/nature00904 OpenUrl CrossRef PubMed Web of Science ↵ Sardiello , M. et al. A gene network regulating lysosomal biogenesis and function . Science 325 , 473 – 477 ( 2009 ). doi: 10.1126/science.1174447 OpenUrl Abstract / FREE Full Text ↵ Sancak , Y. et al. The Rag GTPases bind raptor and mediate amino acid signaling to mTORC1 . Science 320 , 1496 – 1501 ( 2008 ). doi: 10.1126/science.1157535 OpenUrl Abstract / FREE Full Text ↵ Zhang , C. S. et al. The lysosomal v-ATPase-Ragulator complex is a common activator for AMPK and mTORC1, acting as a switch between catabolism and anabolism . Cell Metab 20 , 526 – 540 ( 2014 ). doi: 10.1016/j.cmet.2014.06.014 OpenUrl CrossRef PubMed ↵ Petrany , M. J. et al. Single-nucleus RNA-seq identifies transcriptional heterogeneity in multinucleated skeletal myofibers . Nat Commun 11 , 6374 ( 2020 ). doi: 10.1038/s41467-020-20063-w OpenUrl CrossRef PubMed ↵ Sun , C. et al. Lineage tracing of nuclei in skeletal myofibers uncovers distinct transcripts and interplay between myonuclear populations . Nat Commun 15 , 9372 ( 2024 ). doi: 10.1038/s41467-024-53510-z OpenUrl CrossRef PubMed ↵ Murach , K. A. & Bagley , J. R . A primer on global molecular responses to exercise in skeletal muscle: Omics in focus . J Sport Health Sci , 101029 ( 2025 ). doi: 10.1016/j.jshs.2025.101029 OpenUrl CrossRef ↵ Fan , Z. et al. Exercise-induced angiogenesis is dependent on metabolically primed ATF3/4(+) endothelial cells . Cell Metab 33 , 1793 – 1807 e1799 ( 2021 ). doi: 10.1016/j.cmet.2021.07.015 OpenUrl CrossRef PubMed ↵ Van der Stede , T. et al. Cellular deconstruction of the human skeletal muscle microenvironment identifies an exercise-induced histaminergic crosstalk . Cell Metab 37 , 842 – 856 e847 ( 2025 ). doi: 10.1016/j.cmet.2024.12.011 OpenUrl CrossRef PubMed Reddy , A. et al. pH-Gated Succinate Secretion Regulates Muscle Remodeling in Response to Exercise . Cell 183 , 62 – 75 e17 ( 2020 ). doi: 10.1016/j.cell.2020.08.039 OpenUrl CrossRef PubMed Shi , Y. et al. Succinate Regulates Exercise-Induced Muscle Remodelling by Boosting Satellite Cell Differentiation Through Succinate Receptor 1 . J Cachexia Sarcopenia Muscle 16 , e13670 ( 2025 ). doi: 10.1002/jcsm.13670 OpenUrl CrossRef PubMed ↵ Langston , P. K. et al. Regulatory T cells shield muscle mitochondria from interferon-gamma-mediated damage to promote the beneficial effects of exercise . Sci Immunol 8 , eadi5377 ( 2023 ). doi: 10.1126/sciimmunol.adi5377 OpenUrl CrossRef PubMed Yang , J. et al. Single-cell dissection of the obesity-exercise axis in adipose-muscle tissues implies a critical role for mesenchymal stem cells . Cell Metab 34 , 1578 – 1593 e1576 ( 2022 ). doi: 10.1016/j.cmet.2022.09.004 OpenUrl CrossRef PubMed ↵ Peck , B. D. et al. A muscle cell-macrophage axis involving matrix metalloproteinase 14 facilitates extracellular matrix remodeling with mechanical loading . FASEB J 36 , e22155 ( 2022 ). doi: 10.1096/fj.202100182RR OpenUrl CrossRef PubMed Kaneshige , A. et al. Relayed signaling between mesenchymal progenitors and muscle stem cells ensures adaptive stem cell response to increased mechanical load . Cell Stem Cell 29 , 265 – 280 e266 ( 2022 ). doi: 10.1016/j.stem.2021.11.003 OpenUrl CrossRef PubMed ↵ Wen , Y. et al. Myonuclear transcriptional dynamics in response to exercise following satellite cell depletion . iScience 24 , 102838 ( 2021 ). doi: 10.1016/j.isci.2021.102838 OpenUrl CrossRef PubMed ↵ Knudsen , N. H. et al. Interleukin-13 drives metabolic conditioning of muscle to endurance exercise . Science 368 ( 2020 ). doi: 10.1126/science.aat3987 OpenUrl Abstract / FREE Full Text ↵ Jin , S. et al. Inference and analysis of cell-cell communication using CellChat . Nat Commun 12 , 1088 ( 2021 ). doi: 10.1038/s41467-021-21246-9 OpenUrl CrossRef PubMed ↵ Jin , S. , Plikus , M. V. & Nie , Q . CellChat for systematic analysis of cell-cell communication from single-cell transcriptomics . Nat Protoc 20 , 180 – 219 ( 2025 ). doi: 10.1038/s41596-024-01045-4 OpenUrl CrossRef PubMed ↵ Leinroth , A. P. et al. Identification of distinct non-myogenic skeletal-muscle-resident mesenchymal cell populations . Cell Rep 39 , 110785 ( 2022 ). doi: 10.1016/j.celrep.2022.110785 OpenUrl CrossRef PubMed ↵ Fuertes-Alvarez , S. et al. Unveiling the neuro-vascular interplay in the skeletal muscle in health, injury and disease . bioRxiv , 2025.2008.2012.669843 ( 2025 ). doi: 10.1101/2025.08.12.669843 OpenUrl Abstract / FREE Full Text ↵ Egan , B. & Sharples , A. P . Molecular responses to acute exercise and their relevance for adaptations in skeletal muscle to exercise training . Physiol Rev 103 , 2057 – 2170 ( 2023 ). doi: 10.1152/physrev.00054.2021 OpenUrl CrossRef Leuchtmann , A. B. , Adak , V. , Dilbaz , S. & Handschin , C . The Role of the Skeletal Muscle Secretome in Mediating Endurance and Resistance Training Adaptations . Front Physiol 12 , 709807 ( 2021 ). doi: 10.3389/fphys.2021.709807 OpenUrl CrossRef PubMed Qiu , Y. et al. Exercise sustains the hallmarks of health . J Sport Health Sci 12 , 8 – 35 ( 2023 ). doi: 10.1016/j.jshs.2022.10.003 OpenUrl CrossRef PubMed ↵ Pedersen , B. K . The Physiology of Optimizing Health with a Focus on Exercise as Medicine . Annu Rev Physiol 81 , 607 – 627 ( 2019 ). doi: 10.1146/annurev-physiol-020518-114339 OpenUrl CrossRef PubMed ↵ Stec , M. J. et al. Combined endurance and resistance exercise training alters the spatial transcriptome of skeletal muscle in young adults . iScience 28 , 113301 ( 2025 ). doi: 10.1016/j.isci.2025.113301 OpenUrl CrossRef PubMed Deshmukh , A. S. et al. Deep muscle-proteomic analysis of freeze-dried human muscle biopsies reveals fiber type-specific adaptations to exercise training . Nat Commun 12 , 304 ( 2021 ). doi: 10.1038/s41467-020-20556-8 OpenUrl CrossRef PubMed Raue , U. et al. Fast and slow muscle fiber transcriptome dynamics with lifelong endurance exercise . J Appl Physiol (1985) 136 , 244 – 261 ( 2024 ). doi: 10.1152/japplphysiol.00442.2023 OpenUrl CrossRef PubMed Van der Stede , T. et al. Transcriptomic signatures of human single skeletal muscle fibers in response to high-intensity interval exercise . Am J Physiol Cell Physiol 327 , C1249 – C1262 ( 2024 ). doi: 10.1152/ajpcell.00299.2024 OpenUrl CrossRef PubMed ↵ Rubenstein , A. B. et al. Integrated single-cell multiome analysis reveals muscle fiber-type gene regulatory circuitry modulated by endurance exercise . Genome Res 35 , 1664 – 1677 ( 2025 ). doi: 10.1101/gr.280051.124 OpenUrl Abstract / FREE Full Text ↵ Moreillon , M. et al. Hybrid fiber alterations in exercising seniors suggest contribution to fast- to-slow muscle fiber shift . J Cachexia Sarcopenia Muscle 10 , 687 – 695 ( 2019 ). doi: 10.1002/jcsm.12410 OpenUrl CrossRef PubMed ↵ Spanidis , Y. et al. Resistance-Trained Individuals Are Less Susceptible to Oxidative Damage after Eccentric Exercise . Oxid Med Cell Longev 2018 , 6857190 ( 2018 ). doi: 10.1155/2018/6857190 OpenUrl CrossRef ↵ Joanisse , S. et al. Exercise conditioning in old mice improves skeletal muscle regeneration . FASEB J 30 , 3256 – 3268 ( 2016 ). doi: 10.1096/fj.201600143RR OpenUrl CrossRef PubMed ↵ Wohlwend , M. et al. The exercise-induced long noncoding RNA CYTOR promotes fast-twitch myogenesis in aging . Sci Transl Med 13 , eabc7367 ( 2021 ). doi: 10.1126/scitranslmed.abc7367 OpenUrl CrossRef PubMed Trewin , A. J. et al. Long non-coding RNA Tug1 modulates mitochondrial and myogenic responses to exercise in skeletal muscle . BMC Biol 20 , 164 ( 2022 ). doi: 10.1186/s12915-022-01366-4 OpenUrl CrossRef PubMed ↵ Yue , Y. et al. The long noncoding RNA lnc-H19 is important for endurance exercise by maintaining slow muscle fiber types . J Biol Chem 299 , 105281 ( 2023 ). doi: 10.1016/j.jbc.2023.105281 OpenUrl CrossRef PubMed ↵ Fleming , S. J. et al. Unsupervised removal of systematic background noise from droplet-based single-cell experiments using CellBender . Nat Methods 20 , 1323 – 1335 ( 2023 ). doi: 10.1038/s41592-023-01943-7 OpenUrl CrossRef PubMed ↵ Hao , Y. et al. Dictionary learning for integrative, multimodal and scalable single-cell analysis . Nat Biotechnol 42 , 293 – 304 ( 2024 ). doi: 10.1038/s41587-023-01767-y OpenUrl CrossRef PubMed ↵ Love , M. I. , Huber , W. & Anders , S . Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2 . Genome Biol 15 , 550 ( 2014 ). doi: 10.1186/s13059-014-0550-8 OpenUrl CrossRef PubMed ↵ Stephens , M . False discovery rates: a new deal . Biostatistics 18 , 275 – 294 ( 2017 ). doi: 10.1093/biostatistics/kxw041 OpenUrl CrossRef PubMed ↵ Zhou , Y. et al. Metascape provides a biologist-oriented resource for the analysis of systems-level datasets . Nat Commun 10 , 1523 ( 2019 ). doi: 10.1038/s41467-019-09234-6 OpenUrl CrossRef PubMed ↵ Yu , G. , Wang , L. G. , Han , Y. & He , Q. Y . clusterProfiler: an R package for comparing biological themes among gene clusters . OMICS 16 , 284 – 287 ( 2012 ). doi: 10.1089/omi.2011.0118 OpenUrl CrossRef PubMed Web of Science ↵ Skinnider , M. A. et al. Cell type prioritization in single-cell data . Nat Biotechnol 39 , 30 – 34 ( 2021 ). doi: 10.1038/s41587-020-0605-1 OpenUrl CrossRef PubMed ↵ Van de Sande , B. et al. A scalable SCENIC workflow for single-cell gene regulatory network analysis . Nat Protoc 15 , 2247 – 2276 ( 2020 ). doi: 10.1038/s41596-020-0336-2 OpenUrl CrossRef PubMed View the discussion thread. Back to top Previous Next Posted November 05, 2025. Download PDF Email Thank you for your interest in spreading the word about bioRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Muscle Fiber- and Cell Type-Specificity of Training Adaptation in Male Mice Message Subject (Your Name) has forwarded a page to you from bioRxiv Message Body (Your Name) thought you would like to see this page from the bioRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Muscle Fiber- and Cell Type-Specificity of Training Adaptation in Male Mice Sedat Dilbaz , Peter Škrinjar , Regula Furrer , Volkan Adak , Alina Stein , Stefan A. Steurer , Gesa Santos , Christoph Handschin bioRxiv 2025.11.04.686534; doi: https://doi.org/10.1101/2025.11.04.686534 Share This Article: Copy Citation Tools Muscle Fiber- and Cell Type-Specificity of Training Adaptation in Male Mice Sedat Dilbaz , Peter Škrinjar , Regula Furrer , Volkan Adak , Alina Stein , Stefan A. Steurer , Gesa Santos , Christoph Handschin bioRxiv 2025.11.04.686534; doi: https://doi.org/10.1101/2025.11.04.686534 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Physiology Subject Areas All Articles Animal Behavior and Cognition (7626) Biochemistry (17654) Bioengineering (13878) Bioinformatics (41895) Biophysics (21430) Cancer Biology (18569) Cell Biology (25471) Clinical Trials (138) Developmental Biology (13366) Ecology (19876) Epidemiology (2067) Evolutionary Biology (24294) Genetics (15593) Genomics (22480) Immunology (17719) Microbiology (40331) Molecular Biology (17155) Neuroscience (88500) Paleontology (666) Pathology (2829) Pharmacology and Toxicology (4818) Physiology (7635) Plant Biology (15116) Scientific Communication and Education (2044) Synthetic Biology (4286) Systems Biology (9817) Zoology (2269)