Hepatic metabolic adaptation to endurance exercise: temporal and sex differences by multiomics integration and validation

Yuan Gao; Zhangyi Tu; Qi Zhao; Jie Peng; Yuan Shu; Junlin Ren; Xueqiang Deng; Yongjun Chen; Yun Zhou; Haimin Zeng; Bofan Chen; Liang Hao

doi:10.1186/s12964-026-03154-x

. 2026 Aug 25;24:475. doi: 10.1186/s12964-026-03154-x

Hepatic metabolic adaptation to endurance exercise: temporal and sex differences by multiomics integration and validation

Yuan Gao ^1,^2,^#, Zhangyi Tu ^2,^3,^8,^#, Qi Zhao ^2,^3,^#, Jie Peng ^2,³, Yuan Shu ^2,³, Junlin Ren ^2,³, Xueqiang Deng ^2,⁴, Yongjun Chen ^2,³, Yun Zhou ^2,³, Haimin Zeng ^2,³, Bofan Chen ^2,³, Liang Hao ^2,^5,^6,^7,^✉

PMCID: PMC13545798 PMID: 42701184

Abstract

Background

Although endurance exercise benefits liver health, sex-specific adaptive trajectories remain unclear. This study mapped dynamic liver adaptation in males and females during prolonged training and identified underlying molecular programs.

Methods

Using publicly available time-resolved liver multi-omics data generated by the Molecular Transducers of Physical Activity Consortium (MoTrPAC), we established a computational pipeline for differential analysis of transcriptomic, proteomic, phosphoproteomic, and metabolomic data with FDR correction, followed by FGSEA pathway enrichment. Kinase activities were inferred through ortholog mapping and PhosphoSitePlus. Cross-omics co-expression networks were constructed using WGCNA and topological overlap to link omics features with physiological phenotypes. For experimental validation, liver tissues were collected from endurance-trained Sprague-Dawley rats, and key nodes were confirmed by Western blotting, qRT-PCR, and immunofluorescence/immunohistochemical staining. Public scRNA-seq data were further integrated to map multi-omics signals to single-cell resolution and assess functional changes in specific cell types.

Results

The hepatic response to exercise stress was stage-specific, shifting from early transcriptional activation to later proteomic and metabolic remodeling. Multi-omics integration revealed distinct sex-associated adaptive trajectories: males were more strongly associated with energy metabolism, redox-related programs, and amino acid/organic acid catabolism, whereas females showed prominent membrane lipid remodeling, proteostasis -related programs, and mitochondrial/ribosomal translational features. Single-cell analysis showed that tissue remodeling occurred without major lineage turnover, instead involving altered communication among pre-existing cell communities. Validation of PPP1R3G identified a protein-dominant exercise-responsive marker, supporting the contribution of post-transcriptional or protein-level regulation.

Conclusions

Hepatic adaptation to endurance stress follows a cross-omics evolutionary pattern with sex-specific reprogramming of energy supply and homeostatic maintenance. This time-resolved framework clarifies how exercise improves liver function and supports sex-oriented metabolic interventions and therapeutic target discovery.

Graphical Abstract

graphic file with name 12964_2026_3154_Figa_HTML.webp

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12964-026-03154-x.

Keywords: Endurance exercise, Liver, Metabolic adaptations, Multi-omics, Experimental validation

Highlights

Temporal multi-omics maps stage-specific liver adaptation to endurance training.

Exercise shifts hepatic responses from transcription to metabolic remodeling.

Males favor energy/redox adaptation, females favor lipids and proteostasis.

Single-cell analysis links remodeling to altered cell communication, not turnover.

PPP1R3G validation supports post-transcriptional control of liver adaptation.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12964-026-03154-x.

Background

Endurance training is a key non-pharmacological intervention for maintaining systemic metabolic homeostasis and extending health span, and it can significantly reduce the risk of cardiometabolic diseases. Studies have shown that regular exercise effectively prevents obesity, type 2 diabetes, and their related complications by enhancing cardiovascular function and metabolic flexibility [1–6]. However, the systemic metabolic benefits elicited by exercise are not driven by any single tissue alone, but rely substantially on dynamic crosstalk among organs. Within this complex cooperative network, the liver acts as a central metabolic hub that orchestrates whole-body fuel allocation, substrate selection, and integration of metabolic signals. In response to the energetic challenge imposed by exercise, the liver coordinates the balance between energy supply and demand across tissues through the regulation of lipid and glucose metabolism and related adaptive pathways [6–9]. Therefore, systematically elucidating the hepatic adaptive mechanisms to endurance training represents a logical cornerstone for uncovering the molecular basis of exercise-mediated improvement in whole-body metabolic homeostasis. In addition to exercise per se, the liver metabolic landscape is remodeled by diverse biological factors, with sexual dimorphism exerting an especially pronounced influence on hepatic homeostasis. The liver displays substantial sex-specific differences in gene regulation, xenobiotic metabolism, immune function, and disease susceptibility, which are closely linked to the onset, progression, and treatment responsiveness of metabolic disorders [10–12]. Meanwhile, exercise-induced metabolic benefits exhibit pronounced sex-specific differences, indicating that sex-related factors may substantially shape the liver’s molecular responses to exercise stimulation [13]. Therefore, investigating exercise adaptation mechanisms in the liver, a highly sex-sensitive organ, requires the systematic incorporation of the sex dimension and carries substantial biological value and clinical significance for developing precise metabolic intervention strategies.

However, existing studies have largely focused on a single omics layer or a single time point, making it difficult to systematically explain how exercise signals evolve from early transient events into later stable metabolic adaptations, or to resolve the coordinated relationships across different molecular layers. In recent years, revolutionary progress in high-throughput sequencing technologies has enabled access to molecular datasets from diverse omics layers, including genomics, epigenomics, transcriptomics, and proteomics, thereby allowing exercise-response networks to be reconstructed across multiple dimensions from signal transduction and gene regulation to metabolic reprogramming [14]. Large collaborative projects represented by Molecular Transducers of Physical Activity Consortium (MoTrPAC) have established standardized, high-temporal-resolution multi-omics resources under a strictly unified experimental design, offering unprecedented analytical depth and conceptual scope to integrate the molecular relay of signaling cascades with the temporal dynamics of long-term adaptation, and to reshape the understanding of exercise physiology by placing sex at the center of analysis [15, 16]. These publicly available consortium resources provide an opportunity for secondary integrative analyses aimed at addressing new biological questions beyond the original data release. In the present study, the primary time-resolved liver multi-omics data were obtained from MoTrPAC and reanalyzed to investigate sex-dependent hepatic adaptation to endurance training.

In addition, the application of single-cell transcriptomics (scRNA-seq) has enabled the dissection of changes in cell composition and the remodeling of transcriptional states within complex tissue microenvironments [17, 18]. This analytical framework helps connect systems-level physiology with cell-specific biology by projecting systemic metabolic features onto the local responses of defined hepatocyte populations and non-parenchymal cells [19]. The strength of cross-modal interrogation is reflected in its ability not only to uncover the adaptive logic of the liver within a multi-organ interactive framework, but also to provide a translational basis for the precise targeting of specific cell subsets in the development of exercise mimetics [20]. At the molecular level, molecules including fetuin A and FGF21 have been confirmed to contribute to exercise-induced metabolic remodeling in the liver [21, 22]. As a regulatory subunit associated with the PP1 holoenzyme, PPP1R3G is linked to glycogen metabolic regulation through interactions between PP1 and glycogen-associated substrates, including glycogen synthase, glycogen phosphorylase, and glycogen phosphorylase kinase [23–25]. Considering the pronounced tissue- and cell-specificity of PP1, its regulatory functions are likely to depend on distinct cellular environments [26]. Previous transcriptomic analysis in a mouse model of exhaustive exercise-induced liver injury identified hepatic Ppp1r3g as one of the most prominently altered genes; its expression increased after exhaustive exercise and was attenuated by SFN intervention. Together with the known role of PPP1R3G in glycogen regulation, these findings suggest that Ppp1r3g-related glycogen metabolic processes may be associated with the hepatoprotective effects of SFN [27]. However, the specific function of PPP1R3G in long-term hepatic adaptation to endurance training, particularly with respect to sex-dependent regulation and cell-type-specific responses, remains to be elucidated.

Based on the above evidence, we hypothesized that endurance training-induced cross-omics remodeling of the liver exhibits pronounced temporal stratification, and that its sexual dimorphism is rooted not only in baseline differences, but also in asymmetries in the magnitude, timing, and functional direction of exercise responses. To test this hypothesis, we reanalyzed publicly available time-resolved liver transcriptomic, proteomic, phosphoproteomic, and metabolomic data from MoTrPAC, quantified sex-biased training effects on the basis of baseline correction, and employed algorithms such as FGSEA and WGCNA to elucidate the organizational principles of the molecular networks. To support the findings from this public multi-omics analysis, we established sex-matched rat exercise-training models and collected corresponding liver tissues for experimental validation. By combining Western blotting, qRT-PCR, immunofluorescence/immunohistochemistry, and publicly available single-cell transcriptomic data, we identified cross-omics hubs represented by Ppp1r3g, deepened our understanding of hepatic adaptation in the context of multi-organ interactions, and provided a basis for the future development of sex-specific exercise mimetics and intervention strategies.

Methods

All animal experiments, omics analyses, and single-cell analyses were performed following standardized procedures, and detailed methods are described in the Supplementary Methods.

Description of MoTrPAC-derived data

The Molecular Transducers of Physical Activity Consortium (MoTrPAC) established a standardized, publicly accessible multi-omics resource for studying molecular responses to exercise across tissues and biological layers [15]. For the present study, we used the publicly available rat endurance-training dataset generated under a progressive 6-month training protocol, which included liver multi-omics profiles from Fischer 344 rats of both sexes and was described in detail in the corresponding dataset-specific publication [16]. Detailed information on sample preparation, data generation, preprocessing, and quality control was obtained from the published MoTrPAC reports [15, 16].

Randomization and blinding

Across all multi-omics platforms, samples were randomly assigned by unblinded batch managers and divided into appropriately sized batches according to the requirements of each analytical platform. Following randomization, all staff engaged in sample preparation, data acquisition, and initial data processing were blinded to sample identities. However, at the later stages of quality control and data analysis, experimental group information was no longer masked from the personnel involved.

Multi-omics data generation and processing

The complete experimental and analytical workflows for sample preparation, multi-omics data acquisition (transcriptomics, proteomics, phosphoproteomics, and metabolomics), normalization procedures, and batch effect correction have been described in detail in previous work [16] and are briefly summarized in the Supplementary Methods. The Supplementary Methods further provide extended technical details of the subsequent statistical analyses conducted in this study.

Differential expression analysis

We processed and statistically analyzed the multi-omics data using a workflow adapted from that reported by Many GM et al. [28] Because liver samples were collected from different animals at terminal training time points, the data were not analyzed as repeated measures. Instead, training duration was treated as a categorical time-point factor (SED, 1 W, 2 W, 4 W, 8 W) and modeled together with sex, forming a combined sex-by-time factor with 10 levels (5 time points × 2 sexes) that represented each sex-specific training group. Transcriptomic data were analyzed using the Bioconductor/R packages edgeR [29] and limma [30]. First, low-abundance transcripts were removed from the raw count data using edgeR::filterByExpr. Subsequently, multidimensional scaling (MDS) plots were generated based on log2-transformed TMM-normalized counts per million to explore the average log2-fold changes between samples. Differential expression analysis was performed using limma::voomWithQualityWeights, which combines observation-level precision weights from voom mean-variance modeling with sample-specific quality weights to account for heteroscedasticity and sample variability [31, 32]. RNA integrity number, median 5′-3′ bias, percent of reads mapping to globin and percent of PCR duplicates as quantified with unique molecular identifiers were included as covariates in the RNA-seq model after they had been mean-imputed and standardized. Specifically, for each transcript, a no-intercept cell-means linear model was fitted as follows: Y \sim 0 + exp_group + rin + pct_globin + pct_umi_dup + median_5_3_bias.

limma was also used to analyze the proteomic, phosphoproteomic, and metabolomic datasets. MDS plots for these datasets revealed differences in variance among samples; therefore, sample-specific quality weights were estimated using limma::arrayWeights and incorporated into the linear models to enhance robustness [33]. For all omics datasets, a no-intercept linear model was fit with the sex-by-time factor and relevant covariates. Following linear modeling with limma::lmFit, contrasts were constructed with limma::contrasts.fit to test: (1) sex-specific training responses (e.g., 1-week-trained versus SED males), (2) baseline sexual dimorphism (SED males versus SED females) and (3) sexually dimorphic training responses (the sex by training interaction effect). Subsequently, robust empirical Bayes moderation was performed using limma::eBayes, shrinking residual variances toward a common value (RNA-seq) or a global trend (proteomics and phosphoproteomics) [34], thereby improving the statistical power of differential analysis. For metabolomics, empirical Bayes moderation was performed separately for each analytical platform. This platform-specific analysis allowed the mean-variance relationship and residual variance structure to be estimated separately for each metabolomics platform before downstream integration of differential statistics. Finally, P values from the relevant comparisons were adjusted for multiple testing using the Benjamini-Hochberg method to control the false discovery rate. The numbers of differential features identified in comparisons at each time point (FDR < 0.05) were displayed using UpSet plots.

Fast Gene Set Enrichment Analysis (FGSEA)

Fast Gene Set Enrichment Analysis (FGSEA) was performed as previously described [35]. This method first ranks features and gene sets and uses signed -log10-transformed P values as the ranking metric, with the sign indicating the direction of the log2 fold change. In each comparison, the ranking metric was calculated at the level of individual molecular features, including proteins, transcripts, and metabolites. For proteomic and transcriptomic data, the ranking metric was further aggregated to the Entrez gene ID level using the arithmetic mean. Features that could not be mapped to Entrez genes before analysis were excluded. For proteomic and transcriptomic enrichment analyses, gene sets were obtained from the three subcollections of the C5:GO category in the Molecular Signatures Database (MSigDB, v7.5.1) [36], namely biological process (BP), molecular function (MF), and cellular component (CC) [37, 38]. For metabolomics data, FGSEA was conducted on the basis of metabolite groupings, which were defined using RefMet chemical subclass annotations from the Metabolomics Workbench RefMet database (https://www.metabolomicsworkbench.org) [39].

Kinase-substrate enrichment analysis

Kinase activity states were inferred from phosphoproteomic data using kinase–substrate annotation information provided by PhosphoSitePlus (v.6.6.0.4) [40]. First, 30,304 quantified phosphorylation sites in rat proteins were remapped to their corresponding sites in human orthologous proteins. Subsequently, FGSEA was performed for each phosphoproteomic comparison using site-level signed -log10-transformed P values as the ranking metric, with the sign indicating the direction of phosphorylation change. The kinase–substrate dataset was constructed based on the PSP kinase–substrate database by grouping all substrate sites phosphorylated by the same kinase into a single set, thereby generating the corresponding kinase sets. On this basis, changes in the activity of the corresponding kinases were inferred by evaluating the enrichment of their substrate sites in the ranked phosphoproteomic lists.

WGCNA module analysis and over-representation analysis (ORA)

WGCNA was used to analyze the data in order to characterize non-overlapping groups of correlated transcripts, proteins, and metabolites, referred to as modules [41]. Modules were ranked in descending order of size and annotated with the initials of the corresponding omics type (P, T, or M). Module eigengenes (MEs) were extracted, and Spearman correlation coefficients were calculated between metabolomic/lipidomic and proteomic module eigengenes, as well as between metabolomic/lipidomic and transcriptomic module eigengenes. To define the biological characteristics of the WGCNA-derived modules, ORA was performed in R using fgsea::fora with the same feature sets used for the FGSEA described above. For the hypergeometric tests, all Entrez gene IDs or RefMet IDs present in the corresponding WGCNA results, excluding features assigned to gray modules, were used as the background set.

Construction of the exercise animal model

All animal procedures were conducted in accordance with the guidelines approved by the Animal Platform of the Biomedical Testing Center of Nanchang University and were approved by the Animal Ethics Committee of Nanchang University (No. NCULAE-20250115001). Twelve 16-week-old male and female Sprague-Dawley (SD) rats were obtained from Jiangsu Huachuang Xinnuo Pharmaceutical Technology Co., Ltd. Rats were housed in same-sex cages under controlled conditions of 21 ± 1 °C and 50% humidity, with a 12-h light/12-h dark cycle. A 2 × 2 factorial design was used in this study to assess the effects of exercise and sex. Following adaptive training, the rats were randomly divided into four groups: male sedentary controls, male exercise-trained, female sedentary controls, and female exercise-trained. The experimental protocol followed the detailed description provided by the MoTrPAC research team [42].

Animal sample collection

At the end of the 8-week training period, all rats were anesthetized with isoflurane (1–2%) 48 h after the final exercise session. Food was removed 3 h before dissection. Tissue collection began at 8:30 a.m. During sample collection, blood was obtained by cardiac puncture and liver tissues were harvested. After collection, all tissues were immediately snap-frozen in liquid nitrogen and stored at -80 °C until further analysis.

Western blotting

Rat liver tissues were homogenized and lysed in RIPA lysis buffer containing PMSF (Solarbio, China) on ice. The lysates were centrifuged at 13,000 rpm for 10 min at 4 °C, and the supernatants were collected. The total protein concentration was measured using a BCA kit (Beyotime, China), and an appropriate amount of protein loading buffer (TransGen, China) was added before heating in boiling water for 10 min. Equal amounts of proteins were separated by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto nitrocellulose membranes (NC, Cytiva, USA). Membranes were blocked using 5% skim milk. The membranes were incubated overnight at 4 °C with primary antibodies against PPP1R3G (ABmart, China) and GAPDH (Proteintech, China), followed by incubation for 2 h with the corresponding HRP-conjugated anti-mouse or anti-rabbit secondary antibodies from ImmunoWay Biotechnology Company (USA). Membranes were washed with 1× TBST. Protein expression was visualized using enhanced chemiluminescence reagent (Solarbio, China) and a chemiluminescence imaging system (Bio-Rad, USA), and protein levels were quantified using ImageJ software and normalized to GAPDH. Relevant antibody information is provided in Table S1.

Quantitative real-time polymerase chain reaction (qRT-PCR)

Total RNA was extracted from liver tissues using TRIzol reagent (Invitrogen, USA) according to the manufacturer’s protocol. The purity and concentration of the extracted RNA were determined and used in subsequent experiments. Reverse transcription was performed using the PrimeScript RT Reagent Kit (TaKaRa, Japan). Quantitative real-time polymerase chain reaction was performed using a real-time PCR system (Applied Biosystems, USA) with SYBR Premix Ex Taq II (TaKaRa, Japan) according to the manufacturer’s instructions. Relative gene expression levels were calculated using the 2 − ΔΔCt method with GAPDH as the internal control. The primer sequences were as follows: PPP1R3G forward primer: ACAAGGTCACCAGGACGAA; reverse primer: AGCTGATCGTCAGGGATCGG. GAPDH forward primer: ACGGCAAGTTCAACGGCACAG; reverse primer: GAAGACGCCAGTAGACTCCACGAC.

Immunofluorescence staining

Paraffin-embedded liver sections were deparaffinized in xylene, rehydrated through graded ethanol, and subjected to antigen retrieval. The sections were then permeabilized with 0.1% Triton X-100 and blocked with 4% goat serum for 30 min at room temperature to reduce nonspecific binding. Subsequently, the sections were incubated overnight at 4 °C with a primary antibody against PPP1R3G (ABmart, China). After washing with PBS, the sections were incubated with an Alexa Fluor 488-conjugated IgG secondary antibody for 1 h at room temperature in the dark. Nuclei were counterstained with DAPI for 5 min. The sections were observed under a laser confocal microscope. Images were acquired using the DAPI and FITC channels with a 60× oil-immersion objective and a z-step interval of 0.5 μm.

Immunohistochemical staining

Liver tissues were fixed with paraformaldehyde, dehydrated through graded ethanol, embedded in paraffin, sectioned at a thickness of 5 μm, and deparaffinized in xylene. After rehydration and antigen retrieval, the sections were blocked with serum-free protein blocking buffer for 30 min at room temperature to reduce nonspecific binding. The sections were then incubated overnight at 4 °C with a primary antibody against PPP1R3G (ABmart, China). After washing, the sections were incubated with the corresponding HRP-conjugated secondary antibody at room temperature. Immunoreactivity was visualized using DAB substrate, and nuclei were counterstained with hematoxylin. Finally, the sections were dehydrated, cleared, mounted, and observed under a light microscope. Staining intensity was quantified using ImageJ software.

Single-cell RNA sequencing analysis

Single-cell RNA-seq data from livers of young mouse were downloaded from the Aging Atlas database (https://ngdc.cncb.ac.cn/aging/single-cell_marker?project=mouse_exercise_tissues), which provides an exercise-related single-cell atlas across multiple mouse tissues [43]. The liver dataset was derived from the published mouse exercise atlas and was processed according to the original publicly available workflow. This dataset was used as a cross-species reference to provide cell-type context for exercise-associated hepatic remodeling. Downstream analyses were performed in R using the Seurat package (v4.0.0) [44]. Raw count matrices were imported into Seurat. During initial object creation, genes expressed in fewer than 10 cells were excluded, and cells with fewer than 200 detected genes were removed. The percentage of mitochondrial transcripts was calculated for each cell using PercentageFeatureSet with the mitochondrial gene pattern ^MT-. Cells were retained for downstream analysis only if they met all of the following criteria: more than 200 and fewer than 2000 detected genes per cell (200 < nFeature_RNA < 2000), more than 500 and fewer than 5000 total transcript/UMI counts per cell (500 < nCount_RNA < 5000), and less than 15% mitochondrial RNA content (percent.mt < 15). After applying these QC criteria, the dataset was reduced from 9,098 cells to 4,269 high-quality cells. Following normalization (LogNormalize) and dimensionality reduction by PCA and t-SNE, cell clusters were identified with FindClusters and annotated using established marker genes. Differentially expressed genes (DEGs) in each major cell type after endurance training were identified using FindMarkers() with the Wilcoxon test. To account for multiple testing, Bonferroni-adjusted p-values provided by Seurat were used, and genes with adjusted p-values < 0.05 were considered statistically significant. GO and KEGG [45] enrichment analyses of the DEGs were conducted using the clusterProfiler package [46] based on the Bonferroni-adjusted significant DEG lists to characterize functional remodeling in representative cell types. Cell-cell communication networks were inferred using CellChat (v1.6.1) [47], and ligand–receptor interactions and signaling pathways were visualized among specified cell populations.

Statistical analyses

All statistical analyses and figure generation were performed in R, and the analytical workflow integrated multiple key R/Bioconductor packages. Specifically, the following packages were used: ComplexHeatmap (v2.12.0), edgeR (v3.40.1), emmeans (v1.8.5), fgsea (v1.26.0), limma (v3.54.0), msigdbr (v7.5.1), tidyverse (v2.0.0), and WGCNA (v1.71). In addition, the MoTrPAC rat endurance-training data and associated analysis tools used in this study were accessed through the MotrpacRatTraining6moData [48] and MotrpacRatTraining6mo [49] R packages, respectively. The packages are available from the MoTrPAC GitHub repositories:

https://github.com/MoTrPAC/MotrpacRatTraining6moData and

https://github.com/MoTrPAC/MotrpacRatTraining6mo. GraphPad Prism 9.3 was used for statistical analysis and plotting, and quantitative data are presented as mean ± standard error of the mean. For experimental designs involving two independent factors, two-way fixed-effects ANOVA was performed to evaluate the main effects of each factor and their interaction. For PPP1R3G validation experiments, Sex and Exercise were included as the two factors, and the Sex × Exercise interaction term was used to assess whether the exercise response differed between males and females. When appropriate, post-hoc pairwise comparisons were performed using Šidák’ s multiple-comparison test, with adjusted P values reported. Statistical significance was defined as P < 0.05. Post-hoc power and sensitivity analyses for the Sex × Exercise interaction in the PPP1R3G validation experiments were performed using G*Power 3.1.9.7 to further evaluate the robustness and detectable effect size of the interaction analysis.

Results

To delineate sex-specific hepatic adaptations to endurance training, we integrated longitudinal multi-omics analyses with targeted experimental validation.

Sexual dimorphism establishes a multi‑layered molecular baseline in the sedentary liver

To systematically characterize the sex-specific molecular baseline of the liver under sedentary conditions, we integratively analyzed the transcriptomic, proteomic, phosphoproteomic, and metabolomic datasets provided by MoTrPAC, and a schematic overview of the analytical workflow is shown in Fig. 1A. At the transcriptome level, sedentary male rats exhibited significant upregulation of several genes related to fatty acid and organic acid metabolism, including Cyp2c13 and Cyp4a2, whereas female livers showed preferential enrichment of transcripts involved in cell signaling and glycosylation control, such as Nrp2 and Man1a1 (Fig. 1B and Table S2 A). Consistently, the proteomic results partially mirrored the transcriptomic patterns, with males showing overall higher abundance of proteins related to lipid and steroid metabolism, whereas females displayed relatively increased expression of proteins involved in translational regulation and proteostasis (Fig. 1B and Table S2 B). Sex-biased differences were also evident at the phosphoproteomic and metabolomic levels, indicating that the sedentary liver exhibits multilayered molecular dimorphism (Fig. 1B and Table S2 C-D). To evaluate the functional concordance between transcriptional and protein expression differences, we further performed a joint pathway enrichment analysis of significantly altered molecules identified in both omics layers. Multiple GO terms were significantly enriched in both omics layers, including male-biased amino acid and organic acid catabolic processes, as well as sex-biased ribosomal, mitochondrial, and protein complex-related categories (Fig. 1C, D). Analysis of normalized enrichment scores indicated that males were preferentially enriched for catabolic and metabolic processes, whereas females showed relatively greater enrichment in several cellular maintenance, protein complex, and information-processing related terms. This divergence was also apparent in the protein-level GO-BP enrichment network. Male-biased pathways included amino acid catabolism, AMP metabolism, and chaperone-dependent protein refolding, whereas female-biased pathways were mainly associated with mRNA export, RNA polyadenylation, and rRNA maturation (Fig. 1E). FGSEA further supported these sex-specific patterns, with male-biased enrichment in sterol biosynthetic and organic acid catabolic processes and female-biased or layer-dependent enrichment in ribosomal, proteasomal, endopeptidase-complex, and mitochondrial component terms (Fig. 1F, G).

Fig. 1 — Sex‑specific molecular organization of the sedentary rat liver revealed by integrated multi‑omics analyses. (A) Analysis flowchart. (B) Volcano plots showing differential molecules between male and female rats in the transcriptome, proteome, phosphoproteome and metabolome under sedentary conditions. For sex-stratified analyses, the sedentary groups included n = 5 per sex for the transcriptome and metabolome, and n = 6 per sex for the proteome and phosphoproteome. (C) Cross-omics comparison of GO enrichment results between transcriptomic and proteomic datasets in sedentary male and female livers. Each point represents a GO term tested in both omics layers. Black points denote terms significantly enriched in both layers, BH-adjusted P < 0.05. (D) Normalized enrichment scores of GO terms identified in (C) as significantly enriched in both transcriptomic and proteomic datasets. Colors indicate male-biased (blue) or female-biased (pink) enrichment, and symbols denote proteomic (circles) or transcriptomic (triangles) results. (E) Protein-level GO-BP enrichment networks from the sedentary male versus female liver comparison. Each node represents an individual GO-BP term and is colored according to its normalized enrichment score, NES. Warmer colors indicate positive NES values, corresponding to male-enriched terms, whereas cooler colors indicate negative NES values, corresponding to female-enriched terms. Node size reflects pathway size. Edges connect GO-BP terms with overlapping protein members, and clustered networks represent functionally related terms sharing common components. (F) Cross-omics comparison of ranked-list FGSEA results between transcriptomic and proteomic datasets. Each point represents a GO term analyzed by FGSEA in both omics layers. Black points denote terms significantly enriched in both layers, BH-adjusted P < 0.05. (G) Normalized enrichment scores of GO terms identified in (F) as significantly enriched in both transcriptomic and proteomic FGSEA results. Colors indicate male-biased (blue) or female-biased (pink) enrichment, and symbols denote proteomic (circles) or transcriptomic (triangles) results

Temporal progression and sex bias of hepatic multi-omic remodeling during endurance training

To delineate the temporal molecular remodeling of the liver induced by endurance training, we analyzed transcriptomic, proteomic, phosphoproteomic, and metabolomic profiles obtained from the livers of male and female rats after 1, 2, 4, and 8 weeks of training, using sex-matched sedentary controls as references. We further implemented a baseline-adjusted approach, defined as [M_XW − M_SED] − [F_XW − F_SED], to eliminate inherent sex differences present in the sedentary state and markedly improve the resolution of sex-dependent differences in training responsiveness. At the pathway level, cross-omics analysis revealed sex-specific regulatory trajectories of endurance training across different time points. Furthermore, we integrated multi-omics data to compare exercise-induced changes in females and males relative to their respective sedentary controls in parallel across training stages, thereby systematically depicting the distinct molecular strategies underlying training adaptation in the two sexes.

Transcriptome

At the transcriptional level, endurance training induced relatively few significant gene-expression changes during the early phase, and transcriptomic remodeling became more evident mainly after prolonged training, particularly at week 8 (Fig. 2A and Table S3). Baseline-corrected sex-bias analysis revealed that sex-dependent training responses were already detectable at week 1 and showed modest temporal changes across training duration rather than a strictly monotonic increase (Fig. 2B and Table S4). At week 1, representative female-biased genes included Mast4 and Bdkrb2, whereas Atrip and Rn7sl1 showed male-biased responses. By week 2, Fam43a was positioned on the negative side of the sex-difference axis, indicating a relatively stronger training response in females, while Kcnip2 represented a male-biased response. At week 4, extracellular matrix- and tissue-remodeling-related transcripts, including Mmp11 and Ecm2, showed female-biased responses, whereas Atrip and Ralgds were male-biased. By week 8, detoxification- and signaling-related transcripts such as Gatc showed female-biased responses, while Faim and Clcn2 were male-biased, indicating that prolonged endurance training induced gene-specific sex-divergent transcriptional responses rather than a uniformly male-dominant pattern.

To characterize the temporal dynamics described above at the functional level, we performed FGSEA on the transcriptomic data (Fig. 2C-E). In the GO BP category (Fig. 2C), females showed early enrichment of metabolic pathways, and cellular ketone metabolic process remained positively enriched at 1 W, 2 W, and 8 W, while males displayed no significant enrichment in this pathway. In contrast, ribosome biogenesis and rRNA metabolic process were transiently enriched during the early phase in both sexes, particularly at weeks 1–2, and then declined, reflecting a shared but non-sustained protein synthesis-related transcriptional response to early training stimulation. Males exhibited a stronger bias in lipid-related pathways, with sterol metabolic/biosynthetic process significantly enriched from weeks 1 to 4 and attenuated by week 8, suggesting that males prominently initiate sterol-related transcriptional remodeling during the early-to-mid stages of training before it gradually subsides. In GO CC analysis (Fig. 2D), males showed significant enrichment of mitochondrial structures such as oxidoreductase complex and respirasome at week 1, consistent with the early training demand for oxidative metabolic capacity; females, by contrast, displayed transient enrichment of blood microparticle at weeks 2 and 8, which was attenuated at week 4, indicating a nonlinear temporal pattern. Structural terms such as cortical actin cytoskeleton and lamellipodium showed negative enrichment in females at selected time points, suggesting sex-dependent remodeling of cellular architecture. By the late stage, proteasome-related components became more prominent, and the sex-bias summary further indicated relatively stronger male-biased negative enrichment of proteasome-associated complexes at week 8. In the GO MF category (Fig. 2E), females showed enrichment of snoRNA binding and catalytic activity, acting on a tRNA during weeks 1–2, indicating enhanced post-transcriptional regulation, whereas males peaked in electron transfer activity at week 1 and maintained enrichment of acid-thiol ligase activity through weeks 1–4; by week 8, unfolded protein binding was significantly reduced in males, implying a lower requirement for UPR-associated factors in the late phase.

A global summary of sex-biased pathway enrichment revealed temporally dynamic and functionally distinct training responses between males and females (Fig. 2F-H). In the BP category, male-biased enrichment was most evident during the early-to-middle stages, with representative terms including Sterol biosynthetic process at week 1 and week 2, and Dicarboxylic acid metabolic process, Lipid homeostasis, and amino acid catabolic processes at week 4. In contrast, female-biased enrichment became prominent at the later stage, particularly at week 8, with representative pathways including Fatty acid beta-oxidation, Fatty acid catabolic process, Lipid oxidation, Cellular lipid catabolic process, and Monocarboxylic acid catabolic process (Fig. 2F). In the CC category, male-biased terms included Cortical actin cytoskeleton at week 1 and mitochondrial-related components such as Inner mitochondrial membrane protein complex and Mitochondrial protein-containing complex at week 1, whereas female-biased enrichment was represented by Blood microparticle and Platelet alpha granule-related terms at week 2, and by Ribosome- and Proteasome-related complexes, including Ribosome, Proteasome complex, Proteasome core complex, and Proteasome regulatory particle, mainly at week 8 (Fig. 2G). In the MF category, the significant sex-biased terms were predominantly female-biased, with representative examples including Nuclease activity and Ribonuclease activity at week 1, Single-stranded DNA helicase activity and Unfolded protein binding at week 4, and Structural constituent of ribosome at week 8 (Fig. 2H).

Proteome

Proteomic alterations showed a dynamic, non-monotonic temporal pattern and became most pronounced at week 8, particularly in males (Fig. 3A and Table S5). Although the number of detected sex-biased proteins was relatively limited, the baseline-corrected proteomic analysis showed that male-biased responses predominated across training durations (Fig. 3B and Table S6). Importantly, the proteostasis and stress-response-related proteins Cryab, Hspb1, Hspa1b, and Hspa1l showed positive fold changes at week 8, indicating male-biased rather than female-biased responses during prolonged endurance training. Overall, pathway-level proteomic remodeling became more evident during the middle-to-late stages of endurance training, particularly in females, although males also showed early enrichment of selected mitochondrial, ribosomal, and redox-related terms (Fig. 3C-E). In the GO BP category, females showed positive enrichment of organic acid catabolic process and dicarboxylic acid metabolic process mainly during the middle-to-late stages, especially at weeks 4–8. By week 8, females also showed marked enrichment of mitochondrial respiratory chain complex assembly, NADH dehydrogenase complex assembly, inner mitochondrial membrane organization, and cellular amino acid catabolic process, indicating enhanced mitochondrial energy-generating and amino acid metabolic programs during prolonged training. In parallel, several RNA-processing-related terms, including RNA export from nucleus, mRNA export from nucleus, mRNA transport, RNA splicing via transesterification reactions, and RNA 3′-end processing, were negatively enriched in females at the late stage (Fig. 3C). In the GO CC category, females displayed middle-to-late positive enrichment of mitochondrial and proteasome-related components, including mitochondrial protein-containing complex, mitochondrial large ribosomal subunit, organelle ribosome, proteasome complex, and endopeptidase complex, suggesting coordinated remodeling of mitochondrial structures and protein turnover machinery. Males showed early enrichment of ribosomal and mitochondrial components, together with enrichment of microbody lumen during the early-to-middle stages. By week 8, outer mitochondrial membrane protein complex was positively enriched in males, whereas spliceosome-related complexes, including spliceosomal snRNP complex, spliceosomal complex, U2-type spliceosomal complex, and catalytic step 2 spliceosome, were negatively enriched (Fig. 3D). In the GO MF category, females showed positive enrichment of flavin adenine dinucleotide binding, oxidoreductase activity, metal cluster binding, electron transfer activity, and NADH dehydrogenase activity mainly during the middle-to-late stages, consistent with the enrichment of respiratory chain complex assembly in the BP category. Disulfide oxidoreductase activity was also positively enriched in females during the middle-to-late stages, indicating enhanced redox-associated protein regulation under prolonged training. In males, selected oxidoreductase-related activities and protein carrier chaperone activity were enriched mainly during the early-to-middle stages, but these responses generally weakened by week 8 (Fig. 3E). In the GO BP category, male-biased enrichment was mainly observed in ribosome/RNA-processing-related pathways, exemplified by Ribosomal small subunit biogenesis and Ribosome biogenesis, particularly at week 8. In contrast, female-biased enrichment became more evident from week 4 onward and was represented by Tricarboxylic acid cycle, Mitochondrial respiratory chain complex assembly, NADH dehydrogenase complex assembly, and Branched-chain amino acid catabolic process, suggesting stronger female-biased regulation of mitochondrial respiration and amino acid catabolism during the middle-to-late stages of training (Fig. 3F). In the GO CC category, males showed representative enrichment of structural and RNA-processing-associated components, including Actomyosin, Actin filament bundle, Preribosome, and Outer mitochondrial membrane protein complex, mainly at week 8. Females, by contrast, showed middle-to-late enrichment of mitochondrial and proteostasis-related components, such as Mitochondrial protein-containing complex, NADH dehydrogenase complex, Oxidoreductase complex, and Proteasome complex (Fig. 3G). In the GO MF category, male-biased enrichment was represented by CoA hydrolase activity and Fatty acid derivative binding during the early stages, with snoRNA binding becoming evident around week 8. Female-biased enrichment was more prominent at the late stage, particularly at week 8, and included Flavin adenine dinucleotide binding, NADH dehydrogenase activity, Oxidoreductase activity, and Electron transfer activity (Fig. 3H).

Phosphoproteome

The phosphoproteome was the fastest-responding omics layer to training, as significant alterations at multiple phosphosites were detected in both sexes by weeks 1–2, indicating that signaling transduction and rapid functional control represent the earliest molecular events activated by exercise (Fig. 4A and Table S7). Compared with the transcriptome, sex bias in the phosphoproteome was more pronounced and emerged earlier (Fig. 4B and Table S8). Sex-biased phosphoproteomic responses displayed distinct time-dependent patterns during training. At weeks 1 and 2, male-biased phosphosites predominated and included proteins associated with cytoskeletal organization, post-transcriptional regulation, energy metabolism, and redox-related processes. At week 4, the sex-biased response became more balanced, with both male- and female-biased phosphosites detected, including regulatory nodes related to growth and metabolic signaling. By week 8, male-biased sites became particularly prominent, including multiple proteins involved in stress tolerance and proteostasis maintenance, such as chaperone-associated molecules.

Fig. 4 — Dynamic Phosphoproteomic Landscape and Kinase Activity Regulation in the Liver During Endurance Training. (A)Volcano plots showing differential phosphorylation in the liver phosphoproteome of female and male rats after 1, 2, 4, and 8 weeks of endurance training relative to their respective sedentary controls (SED). (B) Volcano plots showing sex-dimorphic phosphoproteomic responses to endurance training at 1, 2, 4, and 8 weeks. Differences were calculated as (M_XW − M_SED) − (F_XW − F_SED). (C) Inferred kinase activity dynamics across 1, 2, 4, and 8 weeks of endurance training in female and male rats relative to sex-matched sedentary controls, based on KSEA of phosphoproteomic differential analysis results. (D) Top kinases most significantly enriched (FDR < 0.05) in any of the four comparisons of male and female training responses, based on phosphoproteomics KSEA results. Biological replicates were n = 12, 11, 12, 12, and 12 for the phosphoproteomic SED, 1 W, 2 W, 4 W, and 8 W groups, respectively; sex-stratified phosphoproteomic analyses used F = 6, 5, 6, 6, and 6 and M = 6, 6, 6, 6, and 6 for SED, 1 W, 2 W, 4 W, and 8 W, respectively

At the kinase activity level, exercise-induced phosphoproteomic remodeling was most evident at weeks 4 and 8, primarily involving signaling kinases associated with MAPK and mTOR pathways (Fig. 4C). KSEA-based kinase activity inference revealed no clustered positive kinase enrichment in females, whereas males showed significant activation of MAPK and mTOR signaling at week 4, involving kinases such as RPS6KA1, RPS6KB1, MAPK3/1, and MTOR. Although this enrichment was less prominent at week 8, partial activity appeared to be sustained by alternative kinases including MAPKAPK2, RPS6KB2, and PRKG1 (Fig. 4C). Within the sex-biased kinase activity framework, males displayed a more prominent mid-training peak in kinase activity at week 4, exemplified by CDK2 and CSNK2A1, whereas PRKACA remained negatively enriched in the late phase at week 8. In contrast, females showed no clustered positively enriched kinase pathways (Fig. 4D).

Metabolome

The liver metabolome displayed time-dependent remodeling during endurance training, with clearer separation and a larger number of differential metabolites emerging at later stages, particularly at week 8 in males (Fig. 5A and Table S9). Sex-biased training responses were detectable across training stages and were most pronounced at week 8, with additional stage-specific differences observed at weeks 2 and 4 (Fig. 5B and Table S10). Several acylcarnitine and lipid-related metabolites showed male-biased responses at specific stages, whereas female-biased responses involved selected amino acid-derived and lipid-associated metabolites, indicating sex-specific adaptation of substrate utilization.

At the metabolite subclass level, endurance training induced sex and stage-dependent remodeling of hepatic metabolites (Fig. 5C). In females, the enrichment pattern was heterogeneous across time. Early responses included positive enrichment of C24 bile acids and dipeptides at week 1, as well as sphingomyelins (SM) and ether phosphatidylcholines (O-PC) at week 2, whereas several lipid-related subclasses, including acyl CoAs, triglycerides (TG), and DG, showed negative enrichment at specific stages. During weeks 4–8, females exhibited selective enrichment of phospholipid-related subclasses such as ether phosphatidylethanolamines (O-PE), phosphatidylinositols (PI), PS, and phosphatidylcholines (PC), indicating stage-specific remodeling rather than a uniform lipid metabolic shift. In males, the response was more strongly associated with lipid-related subclasses. Acyl carnitines, SM, and acyl CoAs were positively enriched mainly during the early phase, particularly at weeks 1–2, whereas PC and phosphatidylethanolamines (PE) showed prominent positive enrichment at the late stage, especially at week 8. By contrast, DG, TG, PI, PS, O-PC, LPC, and unsaturated fatty acids showed negative enrichment at one or more time points. Sex-comparison enrichment analysis revealed a temporally dynamic divergence in hepatic metabolite subclass responses to endurance training (Fig. 5D). At week 1, males showed positive enrichment of acyl CoAs, while females showed stronger enrichment of PS, dipeptides, and DG. At week 2, male-biased enrichment became more evident for acyl carnitines and PC, indicating enhanced fatty acid transport and membrane lipid remodeling in males, whereas females showed enrichment of unsaturated fatty acids, dipeptides, TG, and DG. At week 4, several phospholipid classes, including PI, PS, O-PE, and O-PC, showed significant female-biased enrichment. By week 8, male-biased enrichment re-emerged prominently in lipid-related subclasses, including acyl CoAs, acyl carnitines, PE, SM, LPC, and O-PE, whereas TG and DG remained female-biased.

To systematically characterize the trajectories of key metabolite classes across time and sex, we performed class-based modular analysis (sFig. 1 A-D). The acylcarnitine profile exhibited clear sex-based divergence. Metabolites relatively elevated in females encompassed short-, medium-, and long-chain acylcarnitines, with the increase in long-chain acylcarnitines being most prominent. This pattern was evident in multiple modules, represented by CAR 2:0, 3-dehydroxycarnitine, CAR 5:0; OH, and CAR 8:1 in modules M0 and M1, and by CAR 16:3, CAR 16:1, CAR 18:2, CAR 14:1, and CAR 16:0; OH in modules M4 and M7. In contrast, metabolites relatively elevated in males were mainly concentrated among short- and medium-chain acylcarnitines and maintained a relatively consistent pattern across multiple training time points. Representative metabolites included CAR 4:0 and CAR 10:0; OH in the M0 module, CAR 5:0, CAR 6:0, and CAR 8:0 in the M1 module, and CAR 8:0; OH in the M2 module, with more pronounced increases at later stages of training. Compared with females, males showed overall lower levels of most long-chain acylcarnitines (sFig. 1 A). Sex differences in the acyl-CoA metabolomic profile displayed an opposing pattern. Male rats exhibited more extensive accumulation of acyl-CoAs, particularly short-chain, medium-chain, and multiple long-chain fatty acyl-CoAs, whereas female rats showed relative elevations only in a subset of unsaturated or modified acyl-CoAs (sFig. 1B).

Sex differences in amino acids and their derivatives were also evident. Metabolites with higher relative abundance in female rats were mainly various amino acid modification products and related metabolites; in addition, some sulfur-containing amino acids or amino acid derivatives were also relatively elevated. By contrast, male-enriched metabolites were dominated by multiple canonical free amino acids and related metabolites, a pattern that was more evident during the later training phase, while several downstream amino acid metabolites or modified amino acids were also relatively increased in males (sFig. 1 C). Nucleotides and their related metabolites displayed persistent and consistent sex differences throughout the training period. Most nucleotides, nucleoside diphosphates/triphosphates, and coenzyme nucleotide metabolites were relatively elevated in male samples, whereas these metabolites were generally present at lower levels in female samples (sFig. 1D).

By integrating transcriptomic, proteomic, phosphoproteomic, and metabolomic datasets, we visualized the temporal distribution of training-responsive pathway-associated differential features across omics layers. Figure 5E shows the female responses relative to female sedentary controls, whereas Fig. 5F shows the corresponding male responses relative to male sedentary controls. The UpSet plots display the intersection patterns of differential features identified at each time point across the four omics layers, revealing the dynamic changes in pathway-associated molecular responses during the training period. In females, differential pathway-associated features were detected across multiple time-point combinations, including both time-point-specific and shared features, with a relatively large number of differential features observed at 4 weeks. In males, differential features were more abundant at the early training stages, particularly at 1–2 weeks, and time-point-specific and overlapping features were also observed at later stages. Together, these results show that endurance training-induced pathway-associated features differed across training stages and between sexes, with contributions from multiple omics layers.

Cross-omics network modules reveal sex-divergent coordinated adaptation to endurance training

During training, the metabolomic/lipidomic modules exhibited sex-biased differentiation patterns. The first pattern was a female-enriched lipid- and amino acid–related module pattern, mainly composed of M1 and M5 (Fig. 6A). Specifically, M1 represented the largest metabolomic/lipidomic module and exhibited higher module eigengene values in females. The molecular classes overrepresented in this module included SM, O-PE, O-PC, amino acids, and ceramides (Cer) (Fig. 6A-B). M5 likewise showed higher ME values in females, with PC being the major overrepresented molecular class (Fig. 6A-B). The second pattern was a male-enriched lipid metabolism-related module pattern, mainly composed of M2, M3, and M4 (Fig. 6A). Specifically, M2 exhibited higher ME values in males and was primarily characterized by an overrepresentation of TG. M3 was enriched in unsaturated FA and PI, whereas M4 was enriched in PC and PE (Fig. 6B).

Fig. 6 — Characterization and cross-omics correlation of functional molecular modules across transcriptomic, proteomic, and metabolomic layers. (A) Boxplots display eigengene expression of 8 metabolomic modules (M1-M8). (B) Overrepresentation analysis of metabolite categories for metabolic/lipidomic modules. Dot size represents -log10-transformed adjusted or scaled p-values, and color represents overlap ratio. (C) Spearman correlation heatmap showing relationships between metabolic/lipidomic modules M1-M8 and transcriptomic and proteomic modules. Dot shape indicates BH-adjusted significance. (D) Boxplots show eigengene expression of 7 transcriptomic modules, T1-T7. (E-G) GO enrichment analysis of representative transcriptomic modules, including (E) biological process, (F) cellular component, and (G) molecular function categories. (H) Boxplots show eigengene expression of 9 proteomic modules, P1-P9. (I-K) GO enrichment analysis of representative proteomic modules, including (I) biological process, (J) cellular component, and (K) molecular function categories

We further examined the coordinated coupling between transcriptomic modules and the two metabolic states. The female-enriched T1 and T3 modules were both closely associated with female-biased metabolic features related to membrane lipids and sphingolipids. Specifically, T1 represented the largest transcriptomic module and exhibited the highest ME values in females (Fig. 6D). GO enrichment analysis showed that T1 was mainly enriched for Biological Process terms including Ribosome biogenesis, RNA splicing via transesterification reactions, rRNA metabolic process, and Regulation of mRNA metabolic process; Cellular Component terms including Spliceosomal complex, Preribosome, Methyltransferase complex, and ATPase complex; and Molecular Function terms including Helicase activity, GTPase binding, Transcription coactivator activity, and Ubiquitin-like protein ligase binding (Fig. 6E-G). T3 likewise displayed higher ME values in females and was primarily enriched for Biological Process terms including Sterol metabolic process, Secondary alcohol metabolic process, Thioester metabolic process, and Nucleoside bisphosphate metabolic process, as well as the Molecular Function term Guanyl-nucleotide exchange factor activity (Fig. 6E, G). Correlation analysis showed that both T1 and T3 were significantly positively correlated with the female-enriched metabolic modules M1 and M5 (Fig. 6C). Meanwhile, both modules were significantly negatively correlated with the male-enriched M2, M3, and M4 modules (Fig. 6C). In addition, T3 was also negatively correlated with M6 (Fig. 6C).

In contrast, the male-enriched T4 and T5 modules exhibited another type of transcription-metabolism coupling pattern. T4 was primarily enriched for Biological Process terms including Cellular amino acid metabolic process, Organic acid catabolic process, and Protein targeting; Cellular Component terms including Mitochondrial protein-containing complex and Endoplasmic reticulum protein-containing complex; and the Molecular Function term Flavin adenine dinucleotide binding (Fig. 6E-G). T5 likewise displayed higher ME values in males and was mainly annotated with Biological Process terms including Cellular amino acid metabolic process, Organic acid catabolic process, Alpha-amino acid metabolic process, and Monocarboxylic acid catabolic process (Fig. 6E). Both T4 and T5 were significantly positively correlated with the male-enriched modules M2, M3, and M4, and were generally significantly negatively correlated with the female-enriched modules M1 and M5 (Fig. 6C).

The proteomic modules further supported the two sex-biased metabolic states described above. P1 was the largest proteomic module and showed higher ME values in females (Fig. 6H). GO enrichment analysis revealed that P1 was primarily enriched for Biological Process terms including Mitochondrial translation, Maturation of 5.8 S rRNA, Maturation of 5.8 S rRNA from tricistronic rRNA transcript (GO:0000466), and acetyl-CoA metabolic process (Fig. 6I). For Cellular Component, P1 was mainly associated with Organellar ribosome, Preribosome, Vesicle tethering complex, Mitochondrial large ribosomal subunit, and Small-subunit processome (Fig. 6J). P3 likewise displayed higher ME values in females and was primarily enriched for Biological Process terms including RNA export from nucleus, mRNA export from nucleus, and RNA polyadenylation, together with Cellular Component terms including Spliceosomal complex, Catalytic step 2 spliceosome, U2-type spliceosomal complex, Spliceosomal snRNP complex, and Precatalytic spliceosome (Fig. 6I-J). P1 was significantly positively correlated with the female-enriched metabolic modules M1 and M5, but significantly negatively correlated with the male-enriched metabolic modules M2, M3, and M4 (Fig. 6C). P3 showed a similar female-biased correlation pattern, with positive correlations with M1/M5 and negative correlations with M2/M3/M4.

Conversely, P2 and P4 represented male-enriched proteomic modules. P2 was one of the major proteomic modules and exhibited higher ME values in males (Fig. 6H). P2 was primarily enriched for Biological Process terms including Cellular amino acid catabolic process, Alpha-amino acid catabolic process, Branched-chain amino acid metabolic process, Serine family amino acid metabolic process, and Branched-chain amino acid catabolic process (Fig. 6I). For Cellular Component, P2 was enriched in Proteasome complex, whereas for Molecular Function, it was enriched in Translation factor activity, RNA binding and Translation initiation factor activity (Fig. 6J-K). P4 likewise displayed higher ME values in males, with GO enrichment mainly involving Biological Process terms including Retrograde vesicle-mediated transport, Inner mitochondrial membrane organization, Peroxisomal transport, Vesicle budding from membrane, and Tail-anchored membrane protein (GO:0071816) (Fig. 6I). Both P2 and P4 were significantly positively correlated with the male-enriched metabolic modules M2, M3, and M4, and significantly negatively correlated with the female-enriched metabolic modules M1 and M5 (Fig. 6C).

By integrating ME correlation structures across metabolomics/lipidomics, transcriptomics, and proteomics, two sex-biased cross-omics coordinated adaptation axes were identified during endurance training. The female-biased axis mainly involved M1/M5, T1/T3, and P1/P3, linking lipid and amino acid-related metabolic modules, including O-PE, O-PC, SM, Cer, PC, and amino acids, with RNA processing, ribosome biogenesis, mitochondrial translation, and acetyl-CoA metabolism. In contrast, the male-biased axis mainly involved M2/M3/M4, T4/T5, and P2/P4, connecting TG, unsaturated FA, PI, PC, and PE-related metabolic modules with amino acid and organic acid catabolism, proteasome function, protein targeting, membrane transport, mitochondrial membrane organization, and peroxisomal transport. These findings suggest that endurance training-induced molecular changes across omics layers were not independent, but organized into directionally coordinated networks associated with female- and male-biased metabolic states.

Notably, these coordinated programs also showed layer-specific features. RNA metabolism was reflected in both transcriptomic and proteomic layers, but T1 was more related to RNA splicing and rRNA metabolism, whereas P3 was associated with RNA export, polyadenylation, and spliceosomal complexes. Amino acid-related signals also differed by layer: male-associated T4/T5 and P2 pointed to amino acid and organic acid catabolism, while metabolite-level amino acids were enriched in the female-associated M1 module. Similarly, mitochondrial adaptation appeared as multiple regulatory dimensions, including mitochondrial translation in P1, mitochondrial protein-containing complexes in T4, and inner mitochondrial membrane organization in P4. Together, these results indicate that sex-differential adaptations to endurance training involved both cross-omics coordinated metabolic programs and molecular layer-specific regulation (sFig. 2).

Sex-specific multi-omics networks and cell type-specific liver adaptation

To dissect sex differences in molecular responses to endurance training, we first performed global multi-omics integrative network and functional enrichment analyses using transcripts, proteins, and metabolites that showed significant sex differences under exercise conditions (Fig. 7A-C). OmicsNet further incorporated related interacting molecules to construct the integrated molecular interaction network. This network showed that sex-differential signals were centered on highly connected hub nodes, including Fos, Ptprr, Sqle, Cyp2c13, Bcl2, and Gria3, which were extensively connected with metabolites related to energy metabolism and redox balance, such as ATP, ADP, H2O, and oxygen (Fig. 7A). Further pathway network analysis revealed that the sex-differentially enriched functional terms were modularized (Fig. 7B), clustering mainly into categories related to development and morphogenesis, immune chemotaxis, protein folding homeostasis, and neural structure, with functional relatedness among terms. The distribution of enriched functional categories showed that the largest fraction of terms was assigned to digestive tract development (55.06%), followed by chemokine (C-X-C motif) ligand 2 production (16.85%), digestive system development (13.48%), and negative regulation of cell growth (4.49%) (Fig. 7C). Categories related to neural structure, immune organ development, and signal transduction were also significant, although they accounted for smaller proportions of enriched terms.

Fig. 7 — Molecular interaction network and pathway enrichment analysis of Liver during endurance training. (A) Integrated molecular interaction network generated by OmicsNet using transcriptomic, proteomic, and metabolomic data. Nodes represent identified molecules or related interacting molecules, and edges indicate predicted molecular interactions. (B) Pathway network of significantly enriched GO and KEGG terms derived from differentially altered transcriptomic genes and proteomic proteins. Nodes represent enriched GO or KEGG terms, and edges indicate functional relatedness based on shared associated genes/proteins. (C) Pie chart showing the proportional distribution of significantly enriched functional terms across biological categories, with percentages indicating the relative contribution of each term group

With respect to sex-specific adaptation, we independently analyzed the network characteristics underlying the 8-week exercise response in males and females. In males, the interaction network constructed from transcriptomic, proteomic, and metabolomic data (8w vs. SED) revealed a highly connected architecture centered on Cyp2c55 and Gsta3, extensively coupled with NAD+/NADH, oxygen, H2O, and multiple epoxide and aromatic hydrocarbon metabolites (sFig. 3 A), indicating systemic remodeling organized around detoxification and redox regulation. The pathway network generated by semantic similarity clustering was concentrated on cyclin-dependent kinase activity, solute transport and ionic homeostasis, protein folding and proteostatic maintenance, iron ion transport, as well as DNA conformational changes and apoptotic signaling (sFig. 3B). In the functional category contributions, cyclin-dependent protein serine/threonine kinase activity (20.59%) and solute: sodium symporter activity (17.65%) showed the highest contributions, followed by iron ion transport, DNA conformation change, and negative regulation of extrinsic apoptotic signaling pathway (8.82% each) (sFig. 3 C), together depicting a multi-axis progression across metabolism, stress, and homeostasis.

By contrast, the female 8-week exercise-response network was metabolite-dominated, with PEP, D-erythrose-4-phosphate, D-xylulose-5-phosphate, inosine, and xanthosine serving as hubs that were extensively connected to enzymes involved in glycolysis, the pentose phosphate pathway, and nucleotide metabolism (sFig. 4 A). The pathway network clustered around negative regulation of megakaryocyte differentiation and processes related to chromatin/centromere structure, with dense internal connectivity (sFig. 4B). The distribution of enriched functional terms was highly concentrated, with negative regulation of megakaryocyte differentiation accounting for 86.67% of terms, accompanied by negative regulation of lipid localization and hydrogen peroxide metabolic process (6.67% each) (sFig. 4 C).

Cross-omics comparison identified PPP1R3G as a protein-dominant exercise-responsive molecule during exercise-induced hepatic adaptation. In the proteomic dataset, PPP1R3G abundance increased markedly and dynamically across the training period in both female and male rats, whereas the corresponding transcript, Ppp1r3g, showed no comparable significant induction (sFigs. 5, 6, 7 and 8). This transcript–protein divergence suggested that PPP1R3G may be regulated through post-transcriptional mechanisms, rather than through transcriptional activation alone. On this basis, PPP1R3G was selected as a candidate molecule for independent validation. We established animal models of exercised and sedentary female and male rats, as schematically illustrated in Fig. 8A, and examined the expression level and tissue distribution of PPP1R3G under sedentary and exercise conditions in both sexes. Western blot analysis showed that PPP1R3G protein was clearly detectable in liver tissue and was higher in both the female and male exercise groups than in their respective sedentary control groups (Fig. 8B and sFig. 9). Quantitative analysis further showed that exercise significantly increased PPP1R3G protein expression in the livers of both females and males, with exercised females exhibiting significantly higher PPP1R3G protein abundance than exercised males (Fig. 8C and Table S11-12). At the tissue level, immunohistochemical staining showed that PPP1R3G was predominantly localized in the cytoplasmic region of hepatocytes (Fig. 8D). Relative to the sedentary condition, exercise significantly expanded the PPP1R3G-positive area detected by IHC in both female and male livers (Fig. 8E). Of note, while the sex difference was not significant in the sedentary state, the PPP1R3G-positive area was significantly larger in female than in male liver under exercise conditions. By contrast, qPCR analysis showed that Ppp1r3g mRNA levels remained largely unchanged after exercise and did not reach statistical significance in either sex, with only a slight nonsignificant increase observed in males (Fig. 8F and Table S13-14). These results indicate that the exercise-responsive regulation of PPP1R3G is primarily manifested at the protein level and in its spatial tissue distribution, rather than at the transcript level. Immunofluorescence staining further corroborated these protein-level and spatial distribution patterns (Fig. 8G). PPP1R3G fluorescence was predominantly localized to the hepatocyte cytoplasm and was clearly intensified after exercise in both female and male rats. Quantitative fluorescence analysis showed that exercise significantly increased the fluorescence-positive area of PPP1R3G, Under exercise conditions, the PPP1R3G-positive area was greater in females than in males, whereas no significant sex difference was observed under sedentary conditions (Fig. 8H). Together, these validation results support PPP1R3G as an exercise-responsive protein whose regulation is mainly reflected at the protein and tissue-distribution levels, raising the possibility of post-transcriptional regulation.

Fig. 8 — Endurance exercise induces PPP1R3G protein expression and alters its hepatic distribution in female and male rats. (A) Schematic illustration of the experimental design using female and male rats under sedentary and endurance exercise conditions. (B) Representative Western blot images showing PPP1R3G protein expression in liver tissues from female and male rats under sedentary and exercise conditions. GAPDH was used as a loading control. (C) Quantification of PPP1R3G protein levels from Western blot analysis. PPP1R3G levels were normalized to GAPDH. (D) Representative immunohistochemical staining of PPP1R3G in liver sections from female and male rats under sedentary and exercise conditions. Upper panels show low-magnification views, and lower panels show higher-magnification images of the boxed regions. Scale bar = 100 μm. (E) Quantitative analysis of PPP1R3G-positive areas from IHC staining. (F) Relative mRNA expression levels of *Ppp1r3g* in liver tissues measured by qPCR. (G) Representative immunofluorescence images of PPP1R3G in liver sections. PPP1R3G is shown in red, nuclei are stained with DAPI in blue, and merged images illustrate cellular localization and exercise-induced changes. Scale bar = 100 μm. (H) Quantification of PPP1R3G-positive fluorescence areas from immunofluorescence analysis in female and male rats under sedentary and exercise conditions. Significance is indicated as follows: ns, not significant; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001

To provide cellular context for the rat liver multi-omics findings, the single-cell dataset described above was used as a cross-species reference for evaluating exercise-associated changes in major hepatic cell populations and potential cellular contributors to bulk liver signals. This analysis was interpreted primarily at the level of cell-type composition and population-level patterns. Following quality control, dimensionality reduction, and unsupervised clustering, 18 robust cell clusters were identified across the integrated liver control group (LC) and liver exercise group (LE) datasets (Fig. 9A). Based on cluster-specific differentially expressed genes and established canonical marker genes, these clusters were assigned to seven major cell-type categories: hepatocytes, Kupffer cells, hepatic stellate cells, endothelial cells, cholangiocytes, plasma cells, and basophils (Fig. 9B and Table S15). The detailed cluster-to-cell-type mapping and marker gene evidence are provided in Table S15. Integrated t-SNE visualization showed that cells from LC and LE were broadly distributed across shared embedding regions, without strong condition-driven separation at the global cell population level (Fig. 9C). When visualized separately by condition, all major annotated cell types were detected in both LC and LE, with no emergence of new cell types or obvious loss of existing cell populations observed (Fig. 9D). Although certain differences were observed in the relative distribution and density of cell populations within the t-SNE space, the overall cell-type composition remained stable between LC and LE, indicating that endurance training did not markedly reshape liver cell lineage composition, but instead more likely modulated the functional states of existing cell types. Building on these observations, we next compared liver intercellular interaction patterns between LC and LE. Cell communication analysis revealed that under the LC condition, hepatocytes engaged in broad and dense interaction networks with Kupffer cells, hepatic stellate cells, and endothelial cells (Fig. 9E). Under the LE condition, the topology of the global intercellular communication network was substantially altered, and the interaction strengths among distinct cell types were redistributed. Differential analysis between LE and LC further indicated that exercise-induced changes in cell communication displayed pronounced cell type-specific features, mainly involving interactions between hepatocytes and endothelial cells, cholangiocytes, and hepatic stellate cells.

Fig. 9 — Single-cell transcriptomic analysis reveals cell type-specific transcriptional and intercellular communication remodeling in the liver after endurance exercise. (A) t-SNE visualization of all liver cells from liver samples of the liver control group (LC) and liver exercise group (LE) after quality control and unsupervised clustering. Cells are colored by unsupervised cluster identity, with clusters numbered 0–17. (B) Annotation of the identified clusters based on canonical marker genes. Hep, hepatocytes; Kup, Kupffer cells; Hep-SC, hepatic stellate cells; EC, endothelial cells; Cho, cholangiocytes; Pla, plasma cells; Bas, basophils. (C) t-SNE visualization of all liver cells colored by experimental condition, LC or LE. (D) t-SNE plots showing the distribution of annotated liver cell types in LC and LE. (E) Cell–cell communication network analysis comparing LC and LE. (F) Cell type-specific differential gene expression analysis between LC and LE. (G) Gene Ontology (GO) and KEGG pathway enrichment analysis of differentially expressed genes in each cell type. LC, liver control group; LE, liver exercise group

Given that changes in intercellular communication are often accompanied by remodeling of intracellular transcriptional states, we next performed a systematic analysis of differentially expressed genes in each major cell type under exercise conditions. The results showed that exercise-induced transcriptional responses in the liver exhibited marked cell type specificity (Fig. 9F). In particular, hepatocytes, Kupffer cells, and hepatic stellate cells exhibited relatively abundant differentially expressed genes, with clear differences in the magnitude and distribution of upregulated versus downregulated genes, indicating that these cell types are highly transcriptionally responsive to exercise. By contrast, endothelial cells, cholangiocytes, and plasma cells showed relatively fewer differentially expressed genes, with a more limited overall magnitude of change.To further elucidate the functional implications of transcriptional changes across different cell types, we performed GO and KEGG enrichment analyses of the differentially expressed genes in each cell type (Fig. 9G). The results showed that multiple cell types were commonly enriched in metabolism-related biological processes, including lipid metabolism, fatty acid metabolism, and protein processing, indicating that metabolic remodeling is a central feature of the exercise-induced hepatic transcriptional response. Meanwhile, pronounced differences in functional enrichment were observed among cell types: hepatocytes were primarily enriched in lipid metabolic and mitochondrial pathways, Kupffer cells in immune and phagocytosis-related pathways, and hepatic stellate cells in processes related to the extracellular matrix and protein processing.

Discussion

Using longitudinal multi-omics profiling, cross-omics network integration, scRNA-seq analysis, and targeted validation of PPP1R3G, we found that endurance training was associated with a temporally ordered and sex-dependent hepatic molecular remodeling pattern. This adaptation was superimposed on a sexually dimorphic sedentary baseline, with males showing stronger metabolic and bioenergetic signatures and females showing greater enrichment of RNA processing, ribosomal organization, and cellular maintenance pathways. During training, phosphoproteomic responses emerged earliest, whereas transcriptomic, proteomic, and metabolomic remodeling became more pronounced during the middle-to-late stages. Cross-omics module analysis revealed pronounced sex divergence in endurance training-induced molecular adaptations, with females preferentially showing coordinated changes between membrane lipid/sphingolipid- and amino acid-related metabolism and RNA processing, ribosome biogenesis, and mitochondrial translation, whereas males preferentially exhibited coordinated remodeling of TG, fatty acid, and phospholipid metabolism together with amino acid/organic acid catabolism, protein homeostasis, membrane trafficking, and mitochondrial/peroxisomal function.

The baseline molecular dimorphism observed in the sedentary liver may provide a pre-existing molecular context for interpreting sex differences in subsequent endurance-training responses. Our findings showed that, at baseline, males exhibited stronger enrichment of fatty acid and organic acid metabolic programs, whereas females showed transcriptomic enrichment of signal transduction and glycosylation-related features, together with proteomic and pathway-level enrichment of translational regulation, RNA processing, ribosomal organization, and proteostasis-related programs. The partial concordance of these patterns across the transcriptome and proteome suggests that they are unlikely to represent single-layer noise and may reflect broader sex-associated organizational features. Earlier studies have indicated that hepatic metabolic dimorphism is associated with, and may be shaped by, the concerted effects of sex hormones, circadian regulation, and transcriptional network regulation [50, 51]. Our multi-omics atlas further suggests that these baseline network differences are also reflected in functional modules involving energy metabolism, ribosome biogenesis, mitochondrial components, protein complex organization, and proteostasis, which may provide molecular contexts in which training-associated responses occur. In addition, our results suggest that, at the male baseline, enrichment of catabolic metabolism, AMP metabolism, and chaperone-dependent protein refolding is consistent with a molecular state associated with greater substrate catabolism and energy metabolism. In contrast, the enrichment of RNA-processing and translational-infrastructure features in females is consistent with a molecular state associated with post-transcriptional regulation and structural maintenance. Compared with earlier studies proposing that female may rely more heavily on fatty acid oxidation during exercise, our data add the perspective that baseline translational and proteostatic architecture may also be associated with subsequent remodeling patterns [52]. This shifts the interpretation from a purely stimulus-driven view toward a model in which baseline molecular states may contribute to sex-associated differences in training-related adaptive trajectories.

Through high-resolution time-series analysis, we observed distinct temporal trajectories in exercise-associated liver remodeling between the sexes. Rather than showing a uniform sex-dominant response, the data indicated that sex differences varied across omics layers, biological processes, and training stages. At the transcriptomic level, early global gene-expression changes were relatively limited, but males showed earlier enrichment of selected sterol, lipid, mitochondrial, and energy-related programs, whereas females showed more prominent late-stage enrichment of lipid oxidation, ribosome, proteasome, and related functional terms. This observation is broadly consistent with the results of Nicholas Mwebaze et al., who reported that biological sex is associated with differences in exercise responses in females and males [53]. However, it remains controversial whether females or males exhibit greater sensitivity or functional adaptation to exercise stimuli in certain measures, as Schultz et al., for example, suggested that females may be more sensitive to exercise stimuli or that males may show a greater magnitude of adaptation [54]. We speculate that some of the inconsistency across studies may stem from the frequent use of single-time-point sampling in previous research, which may capture only partial snapshots of temporally distinct molecular changes in the two sexes. Using a baseline-correction model, we found that the observed sex-associated differences were not limited to response amplitude, but also involved differences in the timing and coordination of molecular changes across training.

In addition, integrated cross-omics analysis revealed marked sex-associated differences in liver molecular pathways linked to training. In males, the molecular profile was characterized by early enrichment of features related to mitochondrial components, oxidoreductase/respirasome-associated processes, electron transfer activity, and sterol metabolism, patterns that are consistent with associations with lipid metabolism, redox regulation, and stress-response processes reported previously [55]. In contrast, females showed stronger enrichment of molecular features related to structural remodeling and substrate-selective metabolism, including early ketone body-related signatures and membrane lipid remodeling. This pattern refines previous studies that emphasized a female advantage in lipid utilization by suggesting that such differences may depend on training stage, substrate class, and omics layer [56]. Our proteomic and phosphoproteomic data identified a female-associated pattern involving proteostasis networks and phospholipid membrane remodeling, which is consistent with a role in cellular homeostasis. Furthermore, kinase-centered phosphorylation signatures in males were most prominent at week 4 of training, a temporal pattern that differs from the transient kinase responses commonly reported in skeletal muscle and suggests that liver signaling responses to exercise may unfold over a more extended time window than those described in skeletal muscle [57].

Multi-omics modular network analysis revealed that sex-biased molecular adaptations during endurance training were not represented by isolated pathways alone, but rather reflected directionally concordant coordinated associations among the metabolome/lipidome, transcriptome, and proteome. Prior studies have reported that females typically show a higher tendency toward lipid oxidation during endurance exercise [58]. Given that ether lipids and sphingolipids are important structural components of cellular membranes, particularly mitochondrial membranes [59, 60], and play essential roles in maintaining membrane fluidity and supporting mitochondrial function [61], the coordinated changes between the aforementioned female-associated lipid modules and ribosomal and mitochondrial translation programs suggest a coupling between alterations in hepatic membrane lipid composition and macromolecular biosynthetic processes. By contrast, male-biased hepatic networks were more prominently enriched for modules associated with triacylglycerols, unsaturated FA, PI, and PE, and were linked to amino acid and organic acid catabolism, proteasomal function, translation factor activity, and mitochondrial inner membrane organization, in line with prior findings that males show greater amino acid oxidation and energy mobilization during exercise stress [62, 63]. Amino acid–related molecules were enriched in female metabolic modules, whereas transcriptional and protein modules associated with amino acid catabolism were primarily observed in males, suggesting that hepatic amino acid abundance and catabolic utilization may reflect distinct metabolic states between sexes. Previous population metabolomics studies have shown that multiple amino acids and their derivatives differ significantly between males and females, and that sex-specific metabolic pathways include amino acid metabolism-related processes [64]. Therefore, the cross-omics differences observed in this study are consistent with previous reports of sex-dependent regulation of amino acid metabolism and hepatic metabolic gene expression. Similarly, hepatic mitochondrial-related changes in females were primarily manifested as mitochondrial translation and rRNA maturation programs, whereas in males they more often involved mitochondrial inner membrane organization and transport processes, suggesting that regulation of mitochondrial function during training adaptation in the liver involves distinct molecular layers between sexes.

To further characterize the cellular context of the bulk multi-omics divergence observed during endurance training, scRNA-seq showed that endurance training was associated with stable cellular composition but marked shifts in cell-type-specific transcriptional states and inferred intercellular communication. Following prolonged exercise intervention, the relative abundances of the annotated liver cell clusters and major cell-type categories remained largely stable, with no marked emergence of new cell populations or disappearance of pre-existing groups detected. This differs from the findings of Y. Tsutsui et al. [65], who reported that exercise suppresses the accumulation of bone marrow-derived macrophages and PD-1 + CD8+ T cells in a NASH model, as well as from those of Ikuru Miura et al. [66], who reported exercise-associated phenotypic changes in Kupffer cells. This discrepancy may reflect differences between physiological training models and disease or injury contexts, in which changes in immune-cell composition or phenotype may be more prominent. Although the proportional composition of liver cell types remained relatively constant, inferred cell-cell communication patterns showed marked changes, including altered predicted interactions involving hepatocytes and endothelial cells, cholangiocytes, and hepatic stellate cells; this finding is consistent with previous studies [67]. Within this multi-layer framework, PPP1R3G served as a protein-dominant cross-omics and histological validation candidate and provided an example of transcript-protein discordance potentially related to post-transcriptional regulation. Although molecular and histological validation showed a marked increase in PPP1R3G protein abundance and cytoplasmic distribution in the livers of both sexes after exercise, its corresponding mRNA levels remained largely unchanged; this lack of transcript-protein concordance differs from transcription-centered models of exercise adaptation, such as those exemplified by PGC-1α-related signaling [68, 69]. Further analysis suggests that this pattern may be related to the annotated role of PPP1R3G as a protein phosphatase 1 regulatory subunit involved in glycogen metabolism [70], and may be consistent with exercise-associated post-transcriptional or protein-level regulation of glycogen-related processes, a possibility that will require perturbation-based validation in future studies.

Novelty and limitations

This study established a liver adaptation atlas for endurance training that integrates multi-omics layers with single-cell resolution, uncovering sex-specific molecular strategies underlying shared metabolic benefits. The integrated analysis identified distinct yet potentially complementary molecular programs associated with hepatic metabolic and homeostatic adaptation in males and females, thereby providing a foundation for future sex-aware experimental and translational studies. In particular, PPP1R3G emerged as an exercise-responsive candidate showing a more pronounced change at the protein level than at the transcript level, highlighting the potential contribution of post-transcriptional and protein-level regulation. Although this integrative analysis provides convergent evidence for sex-specific hepatic adaptations to endurance training, several important limitations should be acknowledged. A significant limitation of this study is the small size of the experimental validation cohort. The validation experiments included only three animals per group, which limited the statistical power to detect modest Sex × Exercise interaction effects. Sensitivity analysis showed that this design had 80% power only for very large interaction effects. Thus, although the validation assays provided supportive evidence for the predicted PPP1R3G response, these findings should be viewed as preliminary evidence that requires further validation in larger, adequately powered cohorts to confirm the sex-dependent regulation of PPP1R3G and to obtain more reliable effect-size estimates. Second, while the single-cell analyses revealed notable remodeling of intercellular communication, its in vivo functional relevance remains to be clarified by ligand–receptor inhibition experiments. Third, our experimental validation used Sprague-Dawley rats while the MoTrPAC discovery dataset used Fischer 344 rats. The experimental confirmation of key predicted responses in SD rats suggests these signatures may reflect conserved mechanisms; however, strain differences in body composition, metabolic phenotype, and exercise capacity may influence response magnitude or pattern. Future matched-strain validation would help quantify strain-specific versus conserved effects. Furthermore, the single-cell transcriptomic analysis used public mouse liver exercise data. Although major hepatic cell types and many metabolic pathways are evolutionarily conserved between mouse and rat, the cross-species nature of this analysis limits direct gene-level inference. Therefore, the scRNA-seq results are interpreted as cell-type contextualization and hypothesis generation. From the perspective of study design, the post-exercise sampling strategy was intended to capture durable training adaptations rather than acute responses. However, because age-matched sedentary controls were included only at the terminal time point, time-dependent or aging-related effects cannot be fully excluded. Future studies with time-matched controls across sampling points will be needed to more precisely distinguish training-induced adaptations from temporal effects. In addition, because energy intake and energy balance were not fully controlled, these factors may have contributed to the observed sex-dependent network divergence. Overall, the present study provides a resource and hypothesis-generating framework for understanding sex-associated hepatic adaptations to endurance training. Future studies in larger cohorts with matched genetic backgrounds and controlled nutritional conditions should validate the prioritized candidates and test their functional roles. In particular, the protein-dominant exercise response of PPP1R3G warrants targeted gain- and loss-of-function studies to determine its effects on relevant molecular pathways and phenotypes, including hepatic glycogen metabolism, glucose homeostasis, mitochondrial function, and systemic metabolic adaptation, as well as potential differences between sexes. Validation in human populations will ultimately be required to establish the translational relevance of these findings for sex-aware exercise strategies.

Conclusions

In summary, our study systematically demonstrates that the multi-omics response of liver tissue to chronic endurance intervention exhibits sex-associated temporal dynamics and cross-omics coordination. Our findings delineate a cross-scale remodeling trajectory of liver endurance adaptation, in which males show prominent associations with energy metabolism, sterol/lipid metabolic programs, redox-related pathways, and amino acid/organic acid catabolism, whereas females display coordinated membrane lipid remodeling, substrate-selective metabolism, RNA/ribosome-related regulation, mitochondrial translation, and proteostasis-associated programs. At the microscopic level, this remodeling was accompanied by stable major cell-type composition but altered cell-type-specific transcriptional states and inferred intercellular communication networks, suggesting that endurance training modulates the functional states and interactions of pre-existing hepatic cell populations rather than inducing major lineage turnover. PPP1R3G further provided a protein-dominant validation example of transcript-protein non-concordance, supporting the contribution of post-transcriptional or protein-level regulation to exercise-induced hepatic adaptation. Together, this time-resolved multi-omics atlas provides a framework for future perturbation-based studies aimed at dissecting causal links between specific molecular hubs and exercise-induced metabolic benefits.

Supplementary Information

Supplementary Material 1.^{(818.9KB, pdf)}

Supplementary Material 2.^{(21.7MB, docx)}

Supplementary Material 3.^{(10.5KB, xlsx)}

Supplementary Material 4.^{(5MB, xlsx)}

Supplementary Material 5.^{(7.3MB, xlsx)}

Supplementary Material 6.^{(11.1KB, xlsx)}

Supplementary Material 7.^{(10.7KB, xlsx)}

Supplementary Material 8.^{(11.6KB, xlsx)}

Supplementary Material 9.^{(1.8MB, xlsx)}

Supplementary Material 10.^{(10.8KB, xlsx)}

Supplementary Material 11.^{(910.1KB, xlsx)}

Supplementary Material 12.^{(16.3KB, xlsx)}

Supplementary Material 13.^{(1.9MB, xlsx)}

Supplementary Material 14.^{(9.7MB, xlsx)}

Supplementary Material 15.^{(10.9MB, xlsx)}

Supplementary Material 16.^{(13.9MB, xlsx)}

Supplementary Material 17.^{(25.8MB, xlsx)}

Supplementary Material 18.^{(49.9MB, xlsx)}

Supplementary Material 19.^{(866.8KB, tif)}

Supplementary Material 20.^{(866.8KB, tif)}

Supplementary Material 21.^{(1.9MB, tif)}

Supplementary Material 22.^{(1.9MB, tif)}

Supplementary Material 23.^{(2.2MB, tif)}

Supplementary Material 24.^{(2MB, tif)}

Acknowledgements

Data used in the preparation of this article were obtained from the Molecular Transducers of Physical Activity Consortium (MoTrPAC) database, which is available for public access at motrpac-data.org. We also gratefully acknowledge the Aging Atlas database, developed and maintained by the Beijing Institute of Genomics, Chinese Academy of Sciences, for providing high-quality liver single-cell transcriptomic data. This valuable resource greatly facilitated our cell-type-specific analysis of exercise-induced adaptation. The graphical abstract and Figures 1 A and 8 A were created with BioRender.com.

Authors’ contributions

L.H. conceived and supervised the project, acquired funding, and coordinated manuscript revision. Y.G. designed the study, performed data analysis and experiments, generated figures, and drafted the manuscript. ZY.T. conducted multi-omics data analysis. J.P. contributed to experiments, data analysis, and manuscript writing. Y.S. processed and analyzed single-cell RNA-seq data. Q.Z. participated in experimental procedures. JL.R., XQ.D., Y.Z., and YJ.C. assisted with animal modeling and sample processing. HM.Z. and BF.C. participated in data collection. All authors reviewed and approved the final manuscript.

Funding

This work was supported by the National Natural Science Foundation of China Regional Program (No. 82260497), the International Science and Technology Cooperation Project of the Jiangxi Provincial Department of Science and Technology (No. 20253BDH410004), and the Science and Technology Project of Jiangxi Administration of Traditional Chinese Medicine (No. 2023Z023).

Data availability

Publicly available datasets used in this study include the multi-omics exercise response reference from the Molecular Transducers of Physical Activity Consortium (MoTrPAC) database and liver single-cell RNA-seq data from the Aging Atlas. No new transcriptomic, proteomic, phosphoproteomic, metabolomic, or single-cell RNA-seq datasets were generated in this study. We have made the data processing and analysis workflows used in this study publicly available at GitHub: https://github.com/t2572116649-glitch/MotrpacRatTrainingLiver. The liver multi-omic datasets were obtained from the MoTrPAC database, which is publicly accessible at motrpac-data.org, and from its associated public repositories. The liver transcriptomic data are available from SRA under accession PRJNA908279, and processed transcriptomic outputs are available from GEO under accession GSE242358. The liver proteomics data were obtained from the MassIVE dataset MSV000092922, and the liver phosphoproteomics data were obtained from the MassIVE dataset MSV000092923. The metabolomics data are available from Metabolomics Workbench under project PR001020, DOI: 10.21228/M8V97D. The liver single-cell RNA-seq data used in this study were obtained from the Aging Atlas, which is publicly accessible at https://ngdc.cncb.ac.cn/aging. Experimental validation data generated in this study, including Western blotting, qPCR, immunofluorescence, and immunohistochemistry data, are presented in the manuscript and Supplementary Materials. Additional raw data supporting these validation experiments are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

All animal procedures were approved by the Animal Ethics Committee of Nanchang University. The experiments were conducted in accordance with institutional guidelines and national regulations for the care and use of laboratory animals (No. NCULAE-20250115001).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Yuan Gao, Zhangyi Tu and Qi Zhao contributed equally to this work.

References

1.Warburton DE, Nicol CW, Bredin SS. Health benefits of physical activity: the evidence. CMAJ. 2006;174:801–9. 10.1503/cmaj.051351. [DOI] [PMC free article] [PubMed] [Google Scholar]
2.Aune D, Norat T, Leitzmann M, Tonstad S, Vatten LJ. Physical activity and the risk of type 2 diabetes: a systematic review and dose-response meta-analysis. Eur J Epidemiol. 2015;30:529–42. 10.1007/s10654-015-0056-z. [DOI] [PubMed] [Google Scholar]
3.Lee IM, Shiroma EJ, Lobelo F, Puska P, Blair SN, Katzmarzyk PT. Effect of physical inactivity on major non-communicable diseases worldwide: an analysis of burden of disease and life expectancy. Lancet. 2012;380:219–29. 10.1016/s0140-6736(12)61031-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Stensel DJ. How can physical activity facilitate a sustainable future? Reducing obesity and chronic disease. Proc Nutr Soc. 2023;82:286–97. 10.1017/s0029665123002203. [DOI] [PubMed] [Google Scholar]
5.Dempsey PC, Rowlands AV, Strain T, Zaccardi F, Dawkins N, Razieh C, Davies MJ, Khunti KK, Edwardson CL, Wijndaele K, et al. Physical activity volume, intensity, and incident cardiovascular disease. Eur Heart J. 2022;43:4789–800. 10.1093/eurheartj/ehac613. [DOI] [PubMed] [Google Scholar]
6.Silva MG, Nunes P, Oliveira P, Ferreira R, Fardilha M, Moreira-Gonçalves D, Duarte JA, Oliveira MM, Peixoto F. Long-Term Aerobic Training Improves Mitochondrial and Antioxidant Function in the Liver of Wistar Rats Preventing Hepatic Age-Related Function Decline. Biology (Basel). 2022;11:1750. 10.3390/biology11121750. [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Thyfault JP, Rector RS. Exercise Combats Hepatic Steatosis: Potential Mechanisms and Clinical Implications. Diabetes. 2020;69:517–24. 10.2337/dbi18-0043. [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Smart NA, King N, McFarlane JR, Graham PL, Dieberg G. Effect of exercise training on liver function in adults who are overweight or exhibit fatty liver disease: a systematic review and meta-analysis. Br J Sports Med. 2018;52:834–43. 10.1136/bjsports-2016-096197. [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Brouwers B, Hesselink MK, Schrauwen P, Schrauwen-Hinderling VB. Effects of exercise training on intrahepatic lipid content in humans. Diabetologia. 2016;59:2068–79. 10.1007/s00125-016-4037-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Burra P, Zanetto A, Schnabl B, Reiberger T, Montano-Loza AJ, Asselta R, Karlsen TH, Tacke F. Hepatic immune regulation and sex disparities. Nat Rev Gastroenterol Hepatol. 2024;21:869–84. 10.1038/s41575-024-00974-5. [DOI] [PubMed] [Google Scholar]
11.Matz-Soja M, Berg T, Kietzmann T. Sex-related variations in liver homeostasis and disease: From zonation dynamics to clinical implications. J Hepatol. 2026;84:181–93. 10.1016/j.jhep.2025.08.007. [DOI] [PubMed] [Google Scholar]
12.Burra P, Bizzaro D, Gonta A, Shalaby S, Gambato M, Morelli MC, Trapani S, Floreani A, Marra F, Brunetto MR, et al. Clinical impact of sexual dimorphism in non-alcoholic fatty liver disease (NAFLD) and non-alcoholic steatohepatitis (NASH). Liver Int. 2021;41:1713–33. 10.1111/liv.14943. [DOI] [PubMed] [Google Scholar]
13.Antunes GC, Cunha CVA, Oharomari LK, Vieira RFL, Fanti M, Rios TDS, Conceição de Mattis LR, Azevêdo Macêdo AP, da Silva ASR, Ropelle ER, et al. Time-restricted feeding combined with exercise improves hepatic and glycaemic metabolism in obese mice: A sex-dependent study. J Physiol. 2025;603:5455–75. 10.1113/jp287681. [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Ji Q, Jiang X, Wang M, Xin Z, Zhang W, Qu J, Liu GH. Multimodal Omics Approaches to Aging and Age-Related Diseases. Phenomics. 2024;4:56–71. 10.1007/s43657-023-00125-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
15.MoTrPAC Study Group., Lead Analysts & MoTrPAC Study Group. Molecular Transducers of Physical Activity Consortium (MoTrPAC): Mapping the Dynamic Responses to Exercise. Cell. 2020;181:1464–74. 10.1016/j.cell.2020.06.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
16.MoTrPAC Study Group., Lead Analysts & MoTrPAC Study Group. Temporal dynamics of the multi-omic response to endurance exercise training. Nature. 2024;629:174–83. 10.1038/s41586-023-06877-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Nadukkandy AS, Kalaiselvan S, Lin L, Luo Y. Clinical application of single-cell RNA sequencing in disease and therapy. Clin Transl Med. 2025;15:e70512. 10.1002/ctm2.70512. [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Halpern KB, Shenhav R, Matcovitch-Natan O, Toth B, Lemze D, Golan M, Massasa EE, Baydatch S, Landen S, Moor AE, et al. Single-cell spatial reconstruction reveals global division of labour in the mammalian liver. Nature. 2017;542:352–6. 10.1038/nature21065. [DOI] [PMC free article] [PubMed] [Google Scholar]
19.D’Alessandro LA, Hoehme S, Henney AM, Drasdo D, Klingmüller U. Unraveling liver complexity from molecular to organ level: challenges and perspectives. Prog Biophys Mol Biol. 2015;117:78–86. 10.1016/j.pbiomolbio.2014.11.005. [DOI] [PubMed] [Google Scholar]
20.Alabdan R, Shili H, Elhessewi GMS, Ghaleb M, Alanazi EM, Alharbi NH, Alharbi RM, Alhashmi AA. High-throughput dissection of inter-organ genetic networks: A multi-omic systems biology approach. SLAS Technol. 2026;36:100376. 10.1016/j.slast.2025.100376. [DOI] [PubMed] [Google Scholar]
21.Gao Y, Zhang W, Zeng LQ, Bai H, Li J, Zhou J, Zhou GY, Fang CW, Wang F, Qin XJ. Exercise and dietary intervention ameliorate high-fat diet-induced NAFLD and liver aging by inducing lipophagy. Redox Biol. 2020;36:101635. 10.1016/j.redox.2020.101635. [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Gonzalez-Gil AM, Elizondo-Montemayor L. The Role of Exercise in the Interplay between Myokines, Hepatokines, Osteokines, Adipokines, and Modulation of Inflammation for Energy Substrate Redistribution and Fat Mass Loss: A Review. Nutrients. 2020;12:1899. 10.3390/nu12061899. [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Gasa R, Jensen PB, Berman HK, Brady MJ, DePaoli-Roach AA, Newgard CB. Distinctive regulatory and metabolic properties of glycogen-targeting subunits of protein phosphatase-1 (PTG, GL, GM/RGl) expressed in hepatocytes. J Biol Chem. 2000;275:26396–403. 10.1074/jbc.M002427200. [DOI] [PubMed] [Google Scholar]
24.Newgard CB, Brady MJ, O’Doherty RM, Saltiel AR. Organizing glucose disposal: emerging roles of the glycogen targeting subunits of protein phosphatase-1. Diabetes. 2000;49:1967–77. 10.2337/diabetes.49.12.1967. [DOI] [PubMed] [Google Scholar]
25.Liu QH, Wang ZY, Tang JW, Mou JY, Ma ZW, Deng B, Liu Z, Wang L. Comparative transcriptome analysis of diurnal alterations of liver glycogen structure: A pilot study. Carbohydr Polym. 2022;295:119710. 10.1016/j.carbpol.2022.119710. [DOI] [PubMed] [Google Scholar]
26.Aggen JB, Nairn AC, Chamberlin R. Regulation of protein phosphatase-1. Chem Biol. 2000;7:R13–23. 10.1016/s1074-5521(00)00069-7. [DOI] [PubMed] [Google Scholar]
27.Yang J, Guo X, Li T, Xie Y, Wang D, Yi L, Mi M. Sulforaphane Inhibits Exhaustive Exercise-Induced Liver Injury and Transcriptome-Based Mechanism Analysis. Nutrients. 2023;15:3220. 10.3390/nu15143220. [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Many GM, Sanford JA, Sagendorf TJ, Hou Z, Nigro P, Whytock KL, Amar D, Caputo T, Gay NR, Gaul DA, et al. Sexual dimorphism and the multi-omic response to exercise training in rat subcutaneous white adipose tissue. Nat Metab. 2024;6:963–79. 10.1038/s42255-023-00959-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Robinson MD, McCarthy DJ, Smyth GK. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics. 2010;26:139–40. 10.1093/bioinformatics/btp616. [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, Smyth GK. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res 2015, 43:e47. 10.1093/nar/gkv007 [DOI] [PMC free article] [PubMed]
31.Law CW, Chen Y, Shi W, Smyth GK. voom: Precision weights unlock linear model analysis tools for RNA-seq read counts. Genome Biol. 2014;15:R29. 10.1186/gb-2014-15-2-r29. [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Liu R, Holik AZ, Su S, Jansz N, Chen K, Leong HS, Blewitt ME, Asselin-Labat M-L, Smyth GK, Ritchie ME. Why weight? Modelling sample and observational level variability improves power in RNA-seq analyses. Nucleic Acids Res. 2015;43:e97–97. 10.1093/nar/gkv412. [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Ritchie ME, Diyagama D, Neilson J, van Laar R, Dobrovic A, Holloway A, Smyth GK. Empirical array quality weights in the analysis of microarray data. BMC Bioinformatics. 2006;7:261. 10.1186/1471-2105-7-261. [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Phipson B, Lee S, Majewski IJ, Alexander WS, Smyth GK. Robust hyperparameter estimation protects against hypervariable genes and improves power to detect differential expression. Ann Appl Stat. 2016;10:946–63. 10.1214/16-aoas920. [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Korotkevich G, Sukhov V, Sergushichev A. Fast gene set enrichment analysis. bioRxiv [Preprint]. 2019. 10.1101/060012. [DOI] [Google Scholar]
36.Liberzon A, Birger C, Thorvaldsdóttir H, Ghandi M, Mesirov JP, Tamayo P. The Molecular Signatures Database (MSigDB) hallmark gene set collection. Cell Syst. 2015;1:417–25. 10.1016/j.cels.2015.12.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
37.Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, Davis AP, Dolinski K, Dwight SS, Eppig JT, et al. Gene ontology: tool for the unification of biology. Gene Ontology Consortium Nat Genet. 2000;25:25–9. 10.1038/75556. [DOI] [PMC free article] [PubMed] [Google Scholar]
38.Gene Ontology Consortium. The Gene Ontology resource: enriching a GOld mine. Nucleic Acids Res. 2021;49:D325–34. 10.1093/nar/gkaa1113. [DOI] [PMC free article] [PubMed] [Google Scholar]
39.Fahy E, Subramaniam S. RefMet: a reference nomenclature for metabolomics. Nat Methods. 2020;17:1173–4. 10.1038/s41592-020-01009-y. [DOI] [PubMed] [Google Scholar]
40.Hornbeck PV, Zhang B, Murray B, Kornhauser JM, Latham V, Skrzypek E. PhosphoSitePlus, 2014: mutations, PTMs and recalibrations. Nucleic Acids Res. 2015;43:D512–520. 10.1093/nar/gku1267. [DOI] [PMC free article] [PubMed] [Google Scholar]
41.Zhang B, Horvath S. A general framework for weighted gene co-expression network analysis. Stat Appl Genet Mol Biol. 2005;4:17. 10.2202/1544-6115.1128. [DOI] [PubMed] [Google Scholar]
42.Schenk S, Sagendorf TJ, Many GM, Lira AK, de Sousa LGO, Bae D, Cicha M, Kramer KS, Muehlbauer M, Hevener AL, et al. Physiological Adaptations to Progressive Endurance Exercise Training in Adult and Aged Rats: Insights from the Molecular Transducers of Physical Activity Consortium (MoTrPAC). Function (Oxf). 2024;5:zqae014. 10.1093/function/zqae014. [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Sun S, Ma S, Cai Y, Wang S, Ren J, Yang Y, Ping J, Wang X, Zhang Y, Yan H, et al. A single-cell transcriptomic atlas of exercise-induced anti-inflammatory and geroprotective effects across the body. Innov (Camb). 2023;4:100380. 10.1016/j.xinn.2023.100380. [DOI] [PMC free article] [PubMed] [Google Scholar]
44.Stuart T, Butler A, Hoffman P, Hafemeister C, Papalexi E, Mauck WM III, Hao Y, Stoeckius M, Smibert P, Satija R. Comprehensive Integration of Single-Cell Data. Cell. 2019;177:1888–e19021821. 10.1016/j.cell.2019.05.031. [DOI] [PMC free article] [PubMed] [Google Scholar]
45.Kanehisa M, Furumichi M, Sato Y, Ishiguro-Watanabe M, Tanabe M. KEGG: integrating viruses and cellular organisms. Nucleic Acids Res. 2021;49:D545–51. 10.1093/nar/gkaa970. [DOI] [PMC free article] [PubMed] [Google Scholar]
46.Yu G, Wang LG, Han Y, He QY. clusterProfiler: an R package for comparing biological themes among gene clusters. Omics. 2012;16:284–7. 10.1089/omi.2011.0118. [DOI] [PMC free article] [PubMed] [Google Scholar]
47.Jin S, Guerrero-Juarez CF, Zhang L, Chang I, Ramos R, Kuan CH, Myung P, Plikus MV, Nie Q. Inference and analysis of cell-cell communication using CellChat. Nat Commun. 2021;12:1088. 10.1038/s41467-021-21246-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
48.Gay N, Jean Beltran P, Amar D, Jimenez-Morales D, MoTrPAC Study Group. MotrpacRatTraining6moData: Data for analysis of the MoTrPAC endurance exercise training study in 6-month-old rats (2025). R package version 2.0.0. https://motrpac.github.io/MotrpacRatTraining6moData/
49.Gay N, Amar D, Jean Beltran P, MoTrPAC Study Group. MotrpacRatTraining6mo: Analysis of the MoTrPAC endurance exercise training data in 6-month-old rats (2024). R package version 1.6.6, https://motrpac.github.io/MotrpacRatTraining6mo/, https://github.com/MoTrPAC/MotrpacRatTraining6mo
50.Li S, Lin JD. Transcriptional control of circadian metabolic rhythms in the liver. Diabetes Obes Metab. 2015;17(Suppl 1):33–8. 10.1111/dom.12520. [DOI] [PMC free article] [PubMed] [Google Scholar]
51.Goldfarb CN, Karri K, Pyatkov M, Waxman DJ. Interplay Between GH-regulated, Sex-biased Liver Transcriptome and Hepatic Zonation Revealed by Single-Nucleus RNA Sequencing. Endocrinology. 2022;163:bqac059. 10.1210/endocr/bqac059. [DOI] [PMC free article] [PubMed] [Google Scholar]
52.Chenevière X, Borrani F, Sangsue D, Gojanovic B, Malatesta D. Gender differences in whole-body fat oxidation kinetics during exercise. Physiologie Appliquée Nutr Et Métabolisme. 2011;36:88–95. 10.1139/H10-086. [DOI] [PubMed] [Google Scholar]
53.Mwebaze N, Makubuya T, Kamwebaze M, Mwase M, Ojara RR, Opio P, Lumbuye L, Nahwera L. Physiological sex differences in response to exercise. Turkish J Kinesiol. 2025;11:241. 10.31459/turkjkin.1692902. [DOI] [Google Scholar]
54.Schulze ATV, McCoin CS, Onyekere C, Allen JA, Geiger PC, Dorn GW, Morris EM, Thyfault JP. Hepatic mitochondrial adaptations to physical activity: impact of sexual dimorphism, PGC1α and BNIP3-mediated mitophagy. J Physiol. 2018;596:4361–76. 10.1113/JP276539. [DOI] [PMC free article] [PubMed] [Google Scholar]
55.Varlamov O, Bethea CL, Roberts CT Jr. Sex-specific differences in lipid and glucose metabolism. Front Endocrinol (Lausanne). 2014;5:241. 10.3389/fendo.2014.00241. [DOI] [PMC free article] [PubMed] [Google Scholar]
56.Abo SMC, Casella E, Layton AT. Sexual Dimorphism in Substrate Metabolism During Exercise. Bull Math Biol. 2024;86:1–38. 10.1007/s11538-023-01242-4. [DOI] [PubMed] [Google Scholar]
57.Heinonen IHA, Kalliokoski KK, Hannukainen JC, Duncker DJ, Nuutila P, Knuuti J. Organ-specific physiological responses to acute physical exercise and long-term training in humans. Physiology. 2014;29:421–36. 10.1152/physiol.00067.2013. [DOI] [PubMed] [Google Scholar]
58.Purdom T, Kravitz L, Dokladny K, Mermier C. Understanding the factors that effect maximal fat oxidation. J Int Soc Sports Nutr. 2018;15:3. 10.1186/s12970-018-0207-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
59.Dean JM, Lodhi IJ. Structural and functional roles of ether lipids. Protein Cell. 2018;9:196–206. 10.1007/s13238-017-0423-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
60.Hernández-Corbacho MJ, Salama MF, Canals D, Senkal CE, Obeid LM. Sphingolipids in mitochondria. Biochim Biophys Acta Mol Cell Biol Lipids. 2017;1862:56–68. 10.1016/j.bbalip.2016.09.019. [DOI] [PMC free article] [PubMed] [Google Scholar]
61.Casares D, Escribá PV, Rosselló CA. Membrane Lipid Composition: Effect on Membrane and Organelle Structure, Function and Compartmentalization and Therapeutic Avenues. Int J Mol Sci. 2019;20:2167. 10.3390/ijms20092167. [DOI] [PMC free article] [PubMed] [Google Scholar]
62.Phillips SM, Atkinson SA, Tarnopolsky MA, MacDougall JD. Gender differences in leucine kinetics and nitrogen balance in endurance athletes. J Appl Physiol (1985) 1993, 75:2134–41. 10.1152/jappl.1993.75.5.2134 [DOI] [PubMed]
63.Andsell P, Thomas K, Hicks KM, Hunter SK, Howatson G, Goodall S. Physiological sex differences affect the integrative response to exercise: acute and chronic implications. Exp Physiol. 2020;105:2007–21. 10.1113/ep088548. [DOI] [PubMed] [Google Scholar]
64.Krumsiek J, Mittelstrass K, Do KT, Stückler F, Ried J, Adamski J, Peters A, Illig T, Kronenberg F, Friedrich N, et al. Gender-specific pathway differences in the human serum metabolome. Metabolomics. 2015;11:1815–33. 10.1007/s11306-015-0829-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
65.Tsutsui Y, Mori T, Yoshio S, Sato M, Sakata T, Yoshida Y, Kawai H, Yoshikawa S, Yamazoe T, Matsuda M, et al. Exercise changes the intrahepatic immune cell profile and inhibits the progression of nonalcoholic steatohepatitis in a mouse model. Hepatol Commun. 2023;7:e0236. 10.1097/hc9.0000000000000236. [DOI] [PMC free article] [PubMed] [Google Scholar]
66.Miura I, Komine S, Okada K, Wada S, Warabi E, Uchida F, Oh S, Suzuki H, Mizokami Y, Shoda J. Prevention of non-alcoholic steatohepatitis by long-term exercise via the induction of phenotypic changes in Kupffer cells of hyperphagic obese mice. Physiol Rep. 2021;9:e14859. 10.14814/phy2.14859. [DOI] [PMC free article] [PubMed] [Google Scholar]
67.Kaffe E, Roulis M, Zhao J, Qu R, Sefik E, Mirza H, Zhou J, Zheng Y, Charkoftaki G, Vasiliou V, et al. Humanized mouse liver reveals endothelial control of essential hepatic metabolic functions. Cell. 2023;186:3793–e38093726. 10.1016/j.cell.2023.07.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
68.Landgraf A, Okada J, Horton M, Liu L, Solomon S, Qiu Y, Kurland IJ, Sidoli S, Pessin JE, Shinoda K. Widespread discordance between mRNA expression, protein abundance and de novo lipogenesis activity in hepatocytes during the fed-starvation transition. bioRxiv [Preprint]. 2025. 10.1101/2025.04.15.649020.40909747 [DOI] [Google Scholar]
69.Haase TN, Ringholm S, Leick L, Biensø RS, Kiilerich K, Johansen ST, Nielsen MM, Wojtaszewski JFP, Hidalgo J, Pedersen PA, Pilegaard H. Role of PGC-1α in exercise and fasting-induced adaptations in mouse liver. Am J Physiol Regul Integr Comp Physiol. 2011;301(5):R1501–1509. 10.1152/ajpregu.00775.2010. [DOI] [PubMed] [Google Scholar]
70.Kumar GS, Choy MS, Koveal DM, Lorinsky MK, Lyons SP, Kettenbach AN, Page R, Peti W. Identification of the substrate recruitment mechanism of the muscle glycogen protein phosphatase 1 holoenzyme. Sci Adv. 2018;4:eaau6044. 10.1126/sciadv.aau6044. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material 1.^{(818.9KB, pdf)}

Supplementary Material 2.^{(21.7MB, docx)}

Supplementary Material 3.^{(10.5KB, xlsx)}

Supplementary Material 4.^{(5MB, xlsx)}

Supplementary Material 5.^{(7.3MB, xlsx)}

Supplementary Material 6.^{(11.1KB, xlsx)}

Supplementary Material 7.^{(10.7KB, xlsx)}

Supplementary Material 8.^{(11.6KB, xlsx)}

Supplementary Material 9.^{(1.8MB, xlsx)}

Supplementary Material 10.^{(10.8KB, xlsx)}

Supplementary Material 11.^{(910.1KB, xlsx)}

Supplementary Material 12.^{(16.3KB, xlsx)}

Supplementary Material 13.^{(1.9MB, xlsx)}

Supplementary Material 14.^{(9.7MB, xlsx)}

Supplementary Material 15.^{(10.9MB, xlsx)}

Supplementary Material 16.^{(13.9MB, xlsx)}

Supplementary Material 17.^{(25.8MB, xlsx)}

Supplementary Material 18.^{(49.9MB, xlsx)}

Supplementary Material 19.^{(866.8KB, tif)}

Supplementary Material 20.^{(866.8KB, tif)}

Supplementary Material 21.^{(1.9MB, tif)}

Supplementary Material 22.^{(1.9MB, tif)}

Supplementary Material 23.^{(2.2MB, tif)}

Supplementary Material 24.^{(2MB, tif)}

Data Availability Statement

[CR1] 1.Warburton DE, Nicol CW, Bredin SS. Health benefits of physical activity: the evidence. CMAJ. 2006;174:801–9. 10.1503/cmaj.051351. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR2] 2.Aune D, Norat T, Leitzmann M, Tonstad S, Vatten LJ. Physical activity and the risk of type 2 diabetes: a systematic review and dose-response meta-analysis. Eur J Epidemiol. 2015;30:529–42. 10.1007/s10654-015-0056-z. [DOI] [PubMed] [Google Scholar]

[CR3] 3.Lee IM, Shiroma EJ, Lobelo F, Puska P, Blair SN, Katzmarzyk PT. Effect of physical inactivity on major non-communicable diseases worldwide: an analysis of burden of disease and life expectancy. Lancet. 2012;380:219–29. 10.1016/s0140-6736(12)61031-9. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR4] 4.Stensel DJ. How can physical activity facilitate a sustainable future? Reducing obesity and chronic disease. Proc Nutr Soc. 2023;82:286–97. 10.1017/s0029665123002203. [DOI] [PubMed] [Google Scholar]

[CR5] 5.Dempsey PC, Rowlands AV, Strain T, Zaccardi F, Dawkins N, Razieh C, Davies MJ, Khunti KK, Edwardson CL, Wijndaele K, et al. Physical activity volume, intensity, and incident cardiovascular disease. Eur Heart J. 2022;43:4789–800. 10.1093/eurheartj/ehac613. [DOI] [PubMed] [Google Scholar]

[CR6] 6.Silva MG, Nunes P, Oliveira P, Ferreira R, Fardilha M, Moreira-Gonçalves D, Duarte JA, Oliveira MM, Peixoto F. Long-Term Aerobic Training Improves Mitochondrial and Antioxidant Function in the Liver of Wistar Rats Preventing Hepatic Age-Related Function Decline. Biology (Basel). 2022;11:1750. 10.3390/biology11121750. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR7] 7.Thyfault JP, Rector RS. Exercise Combats Hepatic Steatosis: Potential Mechanisms and Clinical Implications. Diabetes. 2020;69:517–24. 10.2337/dbi18-0043. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR8] 8.Smart NA, King N, McFarlane JR, Graham PL, Dieberg G. Effect of exercise training on liver function in adults who are overweight or exhibit fatty liver disease: a systematic review and meta-analysis. Br J Sports Med. 2018;52:834–43. 10.1136/bjsports-2016-096197. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR9] 9.Brouwers B, Hesselink MK, Schrauwen P, Schrauwen-Hinderling VB. Effects of exercise training on intrahepatic lipid content in humans. Diabetologia. 2016;59:2068–79. 10.1007/s00125-016-4037-x. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR10] 10.Burra P, Zanetto A, Schnabl B, Reiberger T, Montano-Loza AJ, Asselta R, Karlsen TH, Tacke F. Hepatic immune regulation and sex disparities. Nat Rev Gastroenterol Hepatol. 2024;21:869–84. 10.1038/s41575-024-00974-5. [DOI] [PubMed] [Google Scholar]

[CR11] 11.Matz-Soja M, Berg T, Kietzmann T. Sex-related variations in liver homeostasis and disease: From zonation dynamics to clinical implications. J Hepatol. 2026;84:181–93. 10.1016/j.jhep.2025.08.007. [DOI] [PubMed] [Google Scholar]

[CR12] 12.Burra P, Bizzaro D, Gonta A, Shalaby S, Gambato M, Morelli MC, Trapani S, Floreani A, Marra F, Brunetto MR, et al. Clinical impact of sexual dimorphism in non-alcoholic fatty liver disease (NAFLD) and non-alcoholic steatohepatitis (NASH). Liver Int. 2021;41:1713–33. 10.1111/liv.14943. [DOI] [PubMed] [Google Scholar]

[CR13] 13.Antunes GC, Cunha CVA, Oharomari LK, Vieira RFL, Fanti M, Rios TDS, Conceição de Mattis LR, Azevêdo Macêdo AP, da Silva ASR, Ropelle ER, et al. Time-restricted feeding combined with exercise improves hepatic and glycaemic metabolism in obese mice: A sex-dependent study. J Physiol. 2025;603:5455–75. 10.1113/jp287681. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR14] 14.Ji Q, Jiang X, Wang M, Xin Z, Zhang W, Qu J, Liu GH. Multimodal Omics Approaches to Aging and Age-Related Diseases. Phenomics. 2024;4:56–71. 10.1007/s43657-023-00125-x. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR15] 15.MoTrPAC Study Group., Lead Analysts & MoTrPAC Study Group. Molecular Transducers of Physical Activity Consortium (MoTrPAC): Mapping the Dynamic Responses to Exercise. Cell. 2020;181:1464–74. 10.1016/j.cell.2020.06.004. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR16] 16.MoTrPAC Study Group., Lead Analysts & MoTrPAC Study Group. Temporal dynamics of the multi-omic response to endurance exercise training. Nature. 2024;629:174–83. 10.1038/s41586-023-06877-w. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR17] 17.Nadukkandy AS, Kalaiselvan S, Lin L, Luo Y. Clinical application of single-cell RNA sequencing in disease and therapy. Clin Transl Med. 2025;15:e70512. 10.1002/ctm2.70512. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR18] 18.Halpern KB, Shenhav R, Matcovitch-Natan O, Toth B, Lemze D, Golan M, Massasa EE, Baydatch S, Landen S, Moor AE, et al. Single-cell spatial reconstruction reveals global division of labour in the mammalian liver. Nature. 2017;542:352–6. 10.1038/nature21065. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR19] 19.D’Alessandro LA, Hoehme S, Henney AM, Drasdo D, Klingmüller U. Unraveling liver complexity from molecular to organ level: challenges and perspectives. Prog Biophys Mol Biol. 2015;117:78–86. 10.1016/j.pbiomolbio.2014.11.005. [DOI] [PubMed] [Google Scholar]

[CR20] 20.Alabdan R, Shili H, Elhessewi GMS, Ghaleb M, Alanazi EM, Alharbi NH, Alharbi RM, Alhashmi AA. High-throughput dissection of inter-organ genetic networks: A multi-omic systems biology approach. SLAS Technol. 2026;36:100376. 10.1016/j.slast.2025.100376. [DOI] [PubMed] [Google Scholar]

[CR21] 21.Gao Y, Zhang W, Zeng LQ, Bai H, Li J, Zhou J, Zhou GY, Fang CW, Wang F, Qin XJ. Exercise and dietary intervention ameliorate high-fat diet-induced NAFLD and liver aging by inducing lipophagy. Redox Biol. 2020;36:101635. 10.1016/j.redox.2020.101635. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR22] 22.Gonzalez-Gil AM, Elizondo-Montemayor L. The Role of Exercise in the Interplay between Myokines, Hepatokines, Osteokines, Adipokines, and Modulation of Inflammation for Energy Substrate Redistribution and Fat Mass Loss: A Review. Nutrients. 2020;12:1899. 10.3390/nu12061899. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR23] 23.Gasa R, Jensen PB, Berman HK, Brady MJ, DePaoli-Roach AA, Newgard CB. Distinctive regulatory and metabolic properties of glycogen-targeting subunits of protein phosphatase-1 (PTG, GL, GM/RGl) expressed in hepatocytes. J Biol Chem. 2000;275:26396–403. 10.1074/jbc.M002427200. [DOI] [PubMed] [Google Scholar]

[CR24] 24.Newgard CB, Brady MJ, O’Doherty RM, Saltiel AR. Organizing glucose disposal: emerging roles of the glycogen targeting subunits of protein phosphatase-1. Diabetes. 2000;49:1967–77. 10.2337/diabetes.49.12.1967. [DOI] [PubMed] [Google Scholar]

[CR25] 25.Liu QH, Wang ZY, Tang JW, Mou JY, Ma ZW, Deng B, Liu Z, Wang L. Comparative transcriptome analysis of diurnal alterations of liver glycogen structure: A pilot study. Carbohydr Polym. 2022;295:119710. 10.1016/j.carbpol.2022.119710. [DOI] [PubMed] [Google Scholar]

[CR26] 26.Aggen JB, Nairn AC, Chamberlin R. Regulation of protein phosphatase-1. Chem Biol. 2000;7:R13–23. 10.1016/s1074-5521(00)00069-7. [DOI] [PubMed] [Google Scholar]

[CR27] 27.Yang J, Guo X, Li T, Xie Y, Wang D, Yi L, Mi M. Sulforaphane Inhibits Exhaustive Exercise-Induced Liver Injury and Transcriptome-Based Mechanism Analysis. Nutrients. 2023;15:3220. 10.3390/nu15143220. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR28] 28.Many GM, Sanford JA, Sagendorf TJ, Hou Z, Nigro P, Whytock KL, Amar D, Caputo T, Gay NR, Gaul DA, et al. Sexual dimorphism and the multi-omic response to exercise training in rat subcutaneous white adipose tissue. Nat Metab. 2024;6:963–79. 10.1038/s42255-023-00959-9. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR29] 29.Robinson MD, McCarthy DJ, Smyth GK. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics. 2010;26:139–40. 10.1093/bioinformatics/btp616. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR30] 30.Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, Smyth GK. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res 2015, 43:e47. 10.1093/nar/gkv007 [DOI] [PMC free article] [PubMed]

[CR31] 31.Law CW, Chen Y, Shi W, Smyth GK. voom: Precision weights unlock linear model analysis tools for RNA-seq read counts. Genome Biol. 2014;15:R29. 10.1186/gb-2014-15-2-r29. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR32] 32.Liu R, Holik AZ, Su S, Jansz N, Chen K, Leong HS, Blewitt ME, Asselin-Labat M-L, Smyth GK, Ritchie ME. Why weight? Modelling sample and observational level variability improves power in RNA-seq analyses. Nucleic Acids Res. 2015;43:e97–97. 10.1093/nar/gkv412. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR33] 33.Ritchie ME, Diyagama D, Neilson J, van Laar R, Dobrovic A, Holloway A, Smyth GK. Empirical array quality weights in the analysis of microarray data. BMC Bioinformatics. 2006;7:261. 10.1186/1471-2105-7-261. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR34] 34.Phipson B, Lee S, Majewski IJ, Alexander WS, Smyth GK. Robust hyperparameter estimation protects against hypervariable genes and improves power to detect differential expression. Ann Appl Stat. 2016;10:946–63. 10.1214/16-aoas920. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR35] 35.Korotkevich G, Sukhov V, Sergushichev A. Fast gene set enrichment analysis. bioRxiv [Preprint]. 2019. 10.1101/060012. [DOI] [Google Scholar]

[CR36] 36.Liberzon A, Birger C, Thorvaldsdóttir H, Ghandi M, Mesirov JP, Tamayo P. The Molecular Signatures Database (MSigDB) hallmark gene set collection. Cell Syst. 2015;1:417–25. 10.1016/j.cels.2015.12.004. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR37] 37.Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, Davis AP, Dolinski K, Dwight SS, Eppig JT, et al. Gene ontology: tool for the unification of biology. Gene Ontology Consortium Nat Genet. 2000;25:25–9. 10.1038/75556. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR38] 38.Gene Ontology Consortium. The Gene Ontology resource: enriching a GOld mine. Nucleic Acids Res. 2021;49:D325–34. 10.1093/nar/gkaa1113. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR39] 39.Fahy E, Subramaniam S. RefMet: a reference nomenclature for metabolomics. Nat Methods. 2020;17:1173–4. 10.1038/s41592-020-01009-y. [DOI] [PubMed] [Google Scholar]

[CR40] 40.Hornbeck PV, Zhang B, Murray B, Kornhauser JM, Latham V, Skrzypek E. PhosphoSitePlus, 2014: mutations, PTMs and recalibrations. Nucleic Acids Res. 2015;43:D512–520. 10.1093/nar/gku1267. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR41] 41.Zhang B, Horvath S. A general framework for weighted gene co-expression network analysis. Stat Appl Genet Mol Biol. 2005;4:17. 10.2202/1544-6115.1128. [DOI] [PubMed] [Google Scholar]

[CR42] 42.Schenk S, Sagendorf TJ, Many GM, Lira AK, de Sousa LGO, Bae D, Cicha M, Kramer KS, Muehlbauer M, Hevener AL, et al. Physiological Adaptations to Progressive Endurance Exercise Training in Adult and Aged Rats: Insights from the Molecular Transducers of Physical Activity Consortium (MoTrPAC). Function (Oxf). 2024;5:zqae014. 10.1093/function/zqae014. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR43] 43.Sun S, Ma S, Cai Y, Wang S, Ren J, Yang Y, Ping J, Wang X, Zhang Y, Yan H, et al. A single-cell transcriptomic atlas of exercise-induced anti-inflammatory and geroprotective effects across the body. Innov (Camb). 2023;4:100380. 10.1016/j.xinn.2023.100380. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR44] 44.Stuart T, Butler A, Hoffman P, Hafemeister C, Papalexi E, Mauck WM III, Hao Y, Stoeckius M, Smibert P, Satija R. Comprehensive Integration of Single-Cell Data. Cell. 2019;177:1888–e19021821. 10.1016/j.cell.2019.05.031. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR45] 45.Kanehisa M, Furumichi M, Sato Y, Ishiguro-Watanabe M, Tanabe M. KEGG: integrating viruses and cellular organisms. Nucleic Acids Res. 2021;49:D545–51. 10.1093/nar/gkaa970. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR46] 46.Yu G, Wang LG, Han Y, He QY. clusterProfiler: an R package for comparing biological themes among gene clusters. Omics. 2012;16:284–7. 10.1089/omi.2011.0118. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR47] 47.Jin S, Guerrero-Juarez CF, Zhang L, Chang I, Ramos R, Kuan CH, Myung P, Plikus MV, Nie Q. Inference and analysis of cell-cell communication using CellChat. Nat Commun. 2021;12:1088. 10.1038/s41467-021-21246-9. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR48] 48.Gay N, Jean Beltran P, Amar D, Jimenez-Morales D, MoTrPAC Study Group. MotrpacRatTraining6moData: Data for analysis of the MoTrPAC endurance exercise training study in 6-month-old rats (2025). R package version 2.0.0. https://motrpac.github.io/MotrpacRatTraining6moData/

[CR49] 49.Gay N, Amar D, Jean Beltran P, MoTrPAC Study Group. MotrpacRatTraining6mo: Analysis of the MoTrPAC endurance exercise training data in 6-month-old rats (2024). R package version 1.6.6, https://motrpac.github.io/MotrpacRatTraining6mo/, https://github.com/MoTrPAC/MotrpacRatTraining6mo

[CR50] 50.Li S, Lin JD. Transcriptional control of circadian metabolic rhythms in the liver. Diabetes Obes Metab. 2015;17(Suppl 1):33–8. 10.1111/dom.12520. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR51] 51.Goldfarb CN, Karri K, Pyatkov M, Waxman DJ. Interplay Between GH-regulated, Sex-biased Liver Transcriptome and Hepatic Zonation Revealed by Single-Nucleus RNA Sequencing. Endocrinology. 2022;163:bqac059. 10.1210/endocr/bqac059. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR52] 52.Chenevière X, Borrani F, Sangsue D, Gojanovic B, Malatesta D. Gender differences in whole-body fat oxidation kinetics during exercise. Physiologie Appliquée Nutr Et Métabolisme. 2011;36:88–95. 10.1139/H10-086. [DOI] [PubMed] [Google Scholar]

[CR53] 53.Mwebaze N, Makubuya T, Kamwebaze M, Mwase M, Ojara RR, Opio P, Lumbuye L, Nahwera L. Physiological sex differences in response to exercise. Turkish J Kinesiol. 2025;11:241. 10.31459/turkjkin.1692902. [DOI] [Google Scholar]

[CR54] 54.Schulze ATV, McCoin CS, Onyekere C, Allen JA, Geiger PC, Dorn GW, Morris EM, Thyfault JP. Hepatic mitochondrial adaptations to physical activity: impact of sexual dimorphism, PGC1α and BNIP3-mediated mitophagy. J Physiol. 2018;596:4361–76. 10.1113/JP276539. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR55] 55.Varlamov O, Bethea CL, Roberts CT Jr. Sex-specific differences in lipid and glucose metabolism. Front Endocrinol (Lausanne). 2014;5:241. 10.3389/fendo.2014.00241. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR56] 56.Abo SMC, Casella E, Layton AT. Sexual Dimorphism in Substrate Metabolism During Exercise. Bull Math Biol. 2024;86:1–38. 10.1007/s11538-023-01242-4. [DOI] [PubMed] [Google Scholar]

[CR57] 57.Heinonen IHA, Kalliokoski KK, Hannukainen JC, Duncker DJ, Nuutila P, Knuuti J. Organ-specific physiological responses to acute physical exercise and long-term training in humans. Physiology. 2014;29:421–36. 10.1152/physiol.00067.2013. [DOI] [PubMed] [Google Scholar]

[CR58] 58.Purdom T, Kravitz L, Dokladny K, Mermier C. Understanding the factors that effect maximal fat oxidation. J Int Soc Sports Nutr. 2018;15:3. 10.1186/s12970-018-0207-1. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR59] 59.Dean JM, Lodhi IJ. Structural and functional roles of ether lipids. Protein Cell. 2018;9:196–206. 10.1007/s13238-017-0423-5. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR60] 60.Hernández-Corbacho MJ, Salama MF, Canals D, Senkal CE, Obeid LM. Sphingolipids in mitochondria. Biochim Biophys Acta Mol Cell Biol Lipids. 2017;1862:56–68. 10.1016/j.bbalip.2016.09.019. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR61] 61.Casares D, Escribá PV, Rosselló CA. Membrane Lipid Composition: Effect on Membrane and Organelle Structure, Function and Compartmentalization and Therapeutic Avenues. Int J Mol Sci. 2019;20:2167. 10.3390/ijms20092167. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR62] 62.Phillips SM, Atkinson SA, Tarnopolsky MA, MacDougall JD. Gender differences in leucine kinetics and nitrogen balance in endurance athletes. J Appl Physiol (1985) 1993, 75:2134–41. 10.1152/jappl.1993.75.5.2134 [DOI] [PubMed]

[CR63] 63.Andsell P, Thomas K, Hicks KM, Hunter SK, Howatson G, Goodall S. Physiological sex differences affect the integrative response to exercise: acute and chronic implications. Exp Physiol. 2020;105:2007–21. 10.1113/ep088548. [DOI] [PubMed] [Google Scholar]

[CR64] 64.Krumsiek J, Mittelstrass K, Do KT, Stückler F, Ried J, Adamski J, Peters A, Illig T, Kronenberg F, Friedrich N, et al. Gender-specific pathway differences in the human serum metabolome. Metabolomics. 2015;11:1815–33. 10.1007/s11306-015-0829-0. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR65] 65.Tsutsui Y, Mori T, Yoshio S, Sato M, Sakata T, Yoshida Y, Kawai H, Yoshikawa S, Yamazoe T, Matsuda M, et al. Exercise changes the intrahepatic immune cell profile and inhibits the progression of nonalcoholic steatohepatitis in a mouse model. Hepatol Commun. 2023;7:e0236. 10.1097/hc9.0000000000000236. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR66] 66.Miura I, Komine S, Okada K, Wada S, Warabi E, Uchida F, Oh S, Suzuki H, Mizokami Y, Shoda J. Prevention of non-alcoholic steatohepatitis by long-term exercise via the induction of phenotypic changes in Kupffer cells of hyperphagic obese mice. Physiol Rep. 2021;9:e14859. 10.14814/phy2.14859. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR67] 67.Kaffe E, Roulis M, Zhao J, Qu R, Sefik E, Mirza H, Zhou J, Zheng Y, Charkoftaki G, Vasiliou V, et al. Humanized mouse liver reveals endothelial control of essential hepatic metabolic functions. Cell. 2023;186:3793–e38093726. 10.1016/j.cell.2023.07.017. [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR68] 68.Landgraf A, Okada J, Horton M, Liu L, Solomon S, Qiu Y, Kurland IJ, Sidoli S, Pessin JE, Shinoda K. Widespread discordance between mRNA expression, protein abundance and de novo lipogenesis activity in hepatocytes during the fed-starvation transition. bioRxiv [Preprint]. 2025. 10.1101/2025.04.15.649020.40909747 [DOI] [Google Scholar]

[CR69] 69.Haase TN, Ringholm S, Leick L, Biensø RS, Kiilerich K, Johansen ST, Nielsen MM, Wojtaszewski JFP, Hidalgo J, Pedersen PA, Pilegaard H. Role of PGC-1α in exercise and fasting-induced adaptations in mouse liver. Am J Physiol Regul Integr Comp Physiol. 2011;301(5):R1501–1509. 10.1152/ajpregu.00775.2010. [DOI] [PubMed] [Google Scholar]

[CR70] 70.Kumar GS, Choy MS, Koveal DM, Lorinsky MK, Lyons SP, Kettenbach AN, Page R, Peti W. Identification of the substrate recruitment mechanism of the muscle glycogen protein phosphatase 1 holoenzyme. Sci Adv. 2018;4:eaau6044. 10.1126/sciadv.aau6044. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Hepatic metabolic adaptation to endurance exercise: temporal and sex differences by multiomics integration and validation

Yuan Gao

Zhangyi Tu

Qi Zhao

Jie Peng

Yuan Shu

Junlin Ren

Xueqiang Deng

Yongjun Chen

Yun Zhou

Haimin Zeng

Bofan Chen

Liang Hao

Abstract

Background

Methods

Results

Conclusions

Graphical Abstract

Supplementary Information

Highlights

Supplementary Information

Background

Methods

Description of MoTrPAC-derived data

Randomization and blinding

Multi-omics data generation and processing

Differential expression analysis

Fast Gene Set Enrichment Analysis (FGSEA)

Kinase-substrate enrichment analysis

WGCNA module analysis and over-representation analysis (ORA)

Construction of the exercise animal model

Animal sample collection

Western blotting

Quantitative real-time polymerase chain reaction (qRT-PCR)

Immunofluorescence staining

Immunohistochemical staining

Single-cell RNA sequencing analysis

Statistical analyses

Results

Sexual dimorphism establishes a multi‑layered molecular baseline in the sedentary liver

Fig. 1.

Temporal progression and sex bias of hepatic multi-omic remodeling during endurance training

Transcriptome

Fig. 2.

Proteome

Fig. 3.

Phosphoproteome

Fig. 4.

Metabolome

Fig. 5.

Cross-omics network modules reveal sex-divergent coordinated adaptation to endurance training

Fig. 6.

Sex-specific multi-omics networks and cell type-specific liver adaptation

Fig. 7.

Fig. 8.

Fig. 9.

Discussion

Novelty and limitations

Conclusions

Supplementary Information

Acknowledgements

Authors’ contributions

Funding

Data availability

Declarations

Ethics approval and consent to participate

Consent for publication

Competing interests

Footnotes

References

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles