Forty 6-week-old SPF-grade healthy male SD rats (180 ± 20 g) were selected for this study, provided by Liaoning Changsheng Biotechnology Co. Ltd., with Laboratory Animal Production License No. [SCXK (Liao) 2020–0001] and Use License No. [SYXK (Hei) 2018–007]. The experimental procedures were housed in standard Special Pathogen Free (SPF) conditions in separate cages at the Animal Breeding Center of Formulary of Heilongjiang University of Traditional Chinese Medicine. All experimental animals were placed in the same environment with room temperature maintained at 24 ± 2°C, relative humidity of about 45–65%, 12 h/12 h circulating lighting day and night, and free access to food and water. All animal experiments in this study were approved by the Academic Ethics Committee of Harbin Institute of Physical Education and Sports, with the ethical approval number IACUC (2024038).

The following chemicals were obtained from Servicebio (Wuhan, China): phenylmethanesulfonyl fluoride (PMSF), radioimmunoprecipitation assay (RIPA) lysis buffer, enhanced chemiluminescence (ECL) reagent, polyvinylidene fluoride (PVDF) membranes, and an SDS-PAGE gel preparation kit. Rabbit monoclonal primary antibodies against: phosphorylated AMP-activated protein kinase (pAMPK); liver X receptor α (LXRα); sterol regulatory element binding protein 1c (SREBP-1c); carnitine palmitoyltransferase 1 (CPT-1); peroxisome proliferator-activated receptor α (PPARα); cytochrome P450 2E1 (CYP2E1) were purchased from Servicebio. Monoclonal antibodies targeting β-actin and Lamin B were also sourced from Servicebio. Fatty acid synthase (FAS) was obtained from Abcam (Cambridge, MA, USA), and acetyl-CoA carboxylase (ACC) was acquired from Proteintech Group, Inc. (Wuhan, China). Horseradish peroxidase (HRP)-Servicebio provided conjugated goat anti-rabbit and mouse immunoglobulin G (IgG) secondary antibodies. Antibodies targeting intestinal barrier-related proteins were purchased from Wuhan Three Eagles Biotechnology Co., Ltd. (Hubei, China). Enzyme-linked immunosorbent assay (ELISA) kits for LPS, interleukin-1β (IL-1β), interleukin-6 (IL-6), tumor necrosis factor-α (TNF-α), and leptin were obtained from Mlbio (Shanghai, China).

After 1 week of acclimatization, 40 SD rats were randomly divided into 5 groups of 8 rats each, and fed different experimental feeds: the control group (n = 8) was fed a normal fat feed with a fat content to energy ratio of 10 kcal% purchased from Liaoning Changsheng Biotech Co., Ltd. and the experimental feed production license No.: SCXK (Liao) 2015–003, which is the national standard rodent dry. The feed was a national standard dry rodent feed; high-fat control rats and rats in the exercise group (n = 32) were fed a high-fat feed with a fat content of 45 kcal% (China, Beijing, Huafukang Bioscience Co., Ltd., H10045). The steatosis disorder was stimulated for 8 consecutive weeks. The success of modeling was determined according to the blood biochemical indexes and liver histology of the rats. After successful modeling, the rats in the high-fat diet group were randomly divided into the HFD group, low-load aerobic exercise group (LEH), medium-load aerobic exercise group (MEH), and high-load aerobic exercise group (HEH), and the rats in the exercise group were subjected to the exercise interventions respectively: using the ZS-PT-III (Zongshi Dichuang Science and Technology Development Limited Liability Company, Beijing, China) running table, the rats were subjected to three kinds of running table exercise for 8 weeks, respectively, with different loads of exercise. During the 8-week experimental period, all rats were weighed weekly, and feed intake was recorded, and the training program model (23, 24) was slightly adjusted with reference to the method of Ortega-Santos (25) and others, as shown in Table 1. Finally, rats were fasted overnight and executed by injection of ether, blood was taken from the abdominal aorta, serum was extracted by centrifugation, and tissues such as liver, ileum, subcutaneous fat, and epididymal fat were taken and weighed. Liver tissues were immediately transferred to a formaldehyde solution to assess morphological changes, then frozen in liquid nitrogen and stored at - 80°C for the following assays.

GTT and ITT were performed at the end of the exercise period (weeks 15–16) to evaluate sustained metabolic improvements, with tissue collection following the tests to correlate functional outcomes with histological/biochemical changes. At weeks 15 and 16 of the trial, the rats were intraperitoneally administered glucose (2 g/kg) and insulin (0.75 U/kg) following 12- and 6-h fasts, respectively (26). Blood glucose concentrations were measured at multiple time points (0, 30, 60, 90, and 120 min) using a blood glucose meter (GA-3, Sinocare, Changsha, China) to assess the pharmacodynamics of the administered substances. The area under the blood glucose concentration curve (AUC) at the specified time intervals was used to assess each cohort’s glycemic tolerance and insulin sensitivity.

The liver and ileum tissue specimens were fixed with formalin and then covered with paraffin. The application of hematoxylin and eosin (H&E) staining was performed immediately on the tissue samples. Subsequent examination of the slides was facilitated by means of a Nikon E100 photomicrography imaging device, which enabled the recording of images and the formulation of observations (27). The tissue extracted from the fixed liver was then subjected to desiccation, a process that was deemed essential for the subsequent oil-red O staining experiment. Following this, the specimen was cut on a cryostat after being immersed in an optimum cutting temperature (OCT) embedding agent. The stained specimens were then observed under a 200 × microscope. The histopathological analysis was conducted by investigators who had not been previously informed of the experimental design.

The expression levels of claudin 1, occludin, and occludens zone 1 (ZO-1) proteins in the ileum were evaluated using IHC (immunohistochemistry) as outlined in the extant literature. HRP-conjugated goat anti-rabbit IgG was added after tissue sections were incubated with primary antibodies at a dilution of 1:100. DAB (diaminobenzidine) staining revealed discernible signals, whilst haematoxylin staining delineated the nucleus. The integrated optical density of IHC was measured with ImagePro Plus.

Gut microbiota analysis was performed by collecting fresh faecal samples from the final three days of the intervention. The subsequent extraction of DNA was conducted utilising the NucleoSpin Soil DNA Kit (Macherey-Nagel, Düren, Germany). Primers 341F (5′- CCTACGGGGNGGCWGCAG-3′) and 805R (5′-GACTACHVGGGGTATCTAATCC-3′) were utilized on a Mastercycler Pro Thermal Cycler (Thermo Fisher Scientific, Waltham, MA, USA) for the purpose of PCR amplification of the V3-V4 hypervariable region of the bacterial 16S rRNA gene. The PCR conditions comprised an initial denaturation step for 5 min at 95°C, followed by 35 cycles of 30 s at 95°C, 30 s at 50°C, and 40 s at 72°C, and a final extension step for 7 min at 72°C. The PCR protocol comprised the initial denaturation step at 95°C for 5 min, followed by 35 cycles of 30 s at 95°C, 30 s at 50°C, and 40 s at 72°C. The protocol was concluded with a final extension step at 72°C for 7 min. Amplicons were then purified using the QIAquick PCR Purification Kit (Qiagen, Hilden, Germany) and quantified using a Qubit 4.0 fluorometer (Thermo Fisher Scientific). Equimolar pooled amplicons (2 nM) were then subjected to sequencing on the Illumina MiSeq platform (2 × 300 bp paired-end) from Beijing Biomarker Biotechnology Co. The resulting raw sequences were then pair-end merged using FLASH v1.2.11 and subjected to a quality filter using Trimmomatic v0.33 (sliding window: 4 bp; average quality > 20). Microbiota data were analyzed using Amplicon Sequence Variant (ASV). Classification assignments were made according to the SILVA 138.1database. Alpha diversity (Shannon and Simpson indices) and beta diversity (PCoA, NMDS based on unweighted UniFrac distances) were calculated using QIIME2 v2022.8. Linear discriminant analysis effect size (LEfSe) was utilized to identify differentially abundant taxa (LDA score > 3) from phylum to genus. Group comparisons employed the Kruskal-Wallis test (p < 0.05).

All data were analyzed using GraphPad Prism 10 (San Diego, CA, USA) and are expressed as mean ± standard deviation (SD) unless otherwise specified. For the gut microbiome 16S rRNA sequencing data, principal component analysis (PCA) was performed and visualised using the R language platform (vegan package v 2.6–4). The assessment of between-group differences was conducted via one-way analysis of variance (ANOVA), followed by Tukey’s post hoc test for multi-group comparisons. Statistical significance was defined at p < 0.05.

The weight loss effect could be observed with different loads of aerobic exercise (Figure 1A). After 8 weeks of exercise intervention, the mean body weight of rats fed high-fat chow was significantly increased by about 112.1% (353.8 g) compared to the first week, whereas the mean body weight of rats on normal diet was only increased by about 68.4% (195.5 g), and the mean body weight of rats that underwent LEH was increased by 96.2% (284.4 g), the mean body weight of rats in the HEH group was increased by 91.4% (264.1 g), and rats in the HEH group gained an average of 84.1% (246.3 g). Comparing the HFD groups, HEH was more able to attenuate HFD-induced weight gain relative to LEH and MEH (Figures 1A,B). In addition, all three exercise groups significantly suppressed HFD-induced epididymal fat volume (p < 0.01) (Figures 1C,D). HEH and MEH reduced epididymal fat weight more significantly than LEH (p < 0.05).

After exercise intervention, HFD-induced accumulation of lipid vesicles and lipid droplets was significantly reduced (Figures 2A,B). Compared with the NFD group, hepatic lipid content (TG, TC) and liver function indexes (ALT, AST) were significantly elevated in the HFD group. After performing three groups of exercise training with different loads, the hepatic lipid levels gradually returned to normal levels, and ALT and AST levels were significantly reduced (Figures 2C–F). In addition, MEH had a more significant improvement effect on TC and TG indexes compared with LEH (p < 0.001), and HEH had a more significant improvement effect on TC and TG indexes compared with MEH (p < 0.001).

The results showed that the liver wet weight and Leptin level of rats in HFD group were significantly higher than that of NFD (p < 0.0001). MEH was better at improving Leptin level than LEH, while HEH was more effective at improving Leptin level compared with MEH (Figure 1F). The present study further measured the effects of three sets of aerobic exercise with different loads on several representative lipid metabolism genes and proteins. All three sets of exercise suppressed Lxrα and its downstream lipid synthesis-related genes (Srebp-1c, Fas and Accα), while increasing the expression of lipid degradation-related genes (PPARα and CPT-1). In addition, exercise activated the phosphorylation of AMPK and inhibited the overexpression of lipid peroxidation regulatory protein (Cytochrome P450 2E1, CYP2E1) (Figure 3). In addition, HEH was better at regulating lipid metabolism compared to the three aerobic exercise groups.

HFD significantly increased serum TC, TG, and LDL levels, whereas serum HDL levels were significantly decreased, and exercise significantly improved dyslipidemia (Figures 4A–D). The areas under the GTT and ITT curves confirmed that HFD-induced abnormalities of glucose tolerance and insulin resistance were significantly improved by the three-group aerobic exercise intervention.

In liver biochemical analysis, lipid peroxides malondialdehyde content, superoxide dismutase content, and glutathione peroxidase activity decreased with exercise intervention, and HEH was the most effective in modulating hepatic oxidative stress among the three groups of exercise (Figures 5A–C).

By blood biochemical analysis, it was found that exercise significantly reduced the serum levels of LPS and pro-inflammatory factors (Figures 5D–G), indicating that exercise could effectively alleviate HFD-stimulated endotoxemia and systemic inflammation and that HEH had the best effect among the three exercise groups. In addition, exercise improved the intestinal epithelial structure of the ileum (Figures 6A, F–H), effectively attenuating HFD-induced loss of intestinal tight junction proteins (Figures 6B–E). Importantly, compared with the LEH group, the MEH group had a significant effect on the restoration of ileal villus length and crypt depth and a better effect on the improvement of intestinal epithelial structure.

HFD-induced rats deregulated hepatic BA uptake and significantly upregulated the expression of Na + −taurocholate cotransporting polypeptide (NTCP) and bile salt export pump (BSEP). These effects were associated with enhanced hepatic expression of the nuclear Farnesoid X Receptor (FXR), leading to a significant induction of the small heterodimer partner (SHP) in HFD-induced impairment of the enterohepatic BA cycle by a mechanism involving microbiota composition and dysfunction. Among them, FXR appeared to be upregulated, leading to increased SHP expression. The exercise intervention group restored HFD-induced hepatic and intestinal FXR expression levels to those observed in sedentary control rats, which eliminated hepatic and intestinal SHP gene expression and similarly counteracted HFD-mediated NTCP and BSEP hepatic overexpression (Figure 5I). Notably, MEH and HEH were superior to LEH for the regulation of genes related to bile acid metabolism.

A total of 545 common OTUs were found in the four groups, 676 specific OTUs in the HFD group, 1,470 specific OTUs in the NFD group, 1,092 specific OTUs in the LEH group, 1,534 specific OTUs in the MEH group, and 1,184 specific OTUs in the MEH group. The HFD diet significantly decreased the abundance of gut flora, and the results of the PCoA of the OTU levels showed a significant separation effect between the normal and the model groups, while the LEH, MEH, and HEH groups were closer to the NFD group. Exercise significantly induced a change in the gut microbiota composition in HFD rats. Groups were separated significantly. LEH, MEH, and HEH groups were closer to NFD. Exercise significantly induced the abundance and diversity of gut microbiota in HFD rats, with microbiota composition tending to favor the NFD group (Figures 7A–E). According to the histograms as well as heatmaps, at the phylum level, Fusobacteriota and Firmicutes abundance as well as Firmicutes/Bacteroidetes ratio were significantly higher in the HFD group compared to the NFD group (Figures 7E,J), and after the exercise intervention, the abundance of thick-walled Firmicutes and the Firmicutes/Bacteroidetes ratio were significantly lower and the MEH and HEH groups were more convergent to the NFD level (Figure 7K); the MEH and HEH groups were more convergent to the NFD level (Figure 7K). phylum (Bacteroidetes) ratio as well as Fusobacteriota abundance were significantly lower and more convergent to the NFD level in the MEH and HEH groups (Figure 7K); at the phylum level, the expression of Clostridia and Bacteroidia was significantly lower in the HFD group compared to the NFD group, and the same three groups of aerobic exercise reversed the above situation (Figure 6J); at the family level, Oscillospiraceae had lower expression in the HFD group compared to the NFD group; at the genus level, the HFD group showed that Ruminococcaceae_NK4A214_group, Lachnospiraceae_NK4A136_group, Colidextribacter, Oscillibacter, Prevotellaceae_UCG-003, UCG-005, Akkermansia decreased in abundance, Romboutsia, Clostridium_sensu_stricto_1, Turicibacter, Allobaculum increased in abundance, and the HFD-induced abnormal pattern of abundance of the above phyla was somewhat undone by the exercise intervention (Figures 7B,D), and it is noteworthy that the improvement effect of exercise on most of the phyla was best in the MEH, HEH group.

In the obesity and MASLD model, the relative abundance of Bacteroidia, Clostridia, Prevotellaceae_UCG-003 and Oscillibacter was negatively correlated with weight gain (p < 0.05), ALT (p < 0.05), AST (p < 0.05) and LDL (p < 0.05) plasma levels were negatively correlated with and positively correlated with HDL levels (p < 0.05). Oscillibacter and Colidextribacter were also negatively correlated with LDL levels (p < 0.05), body weight (p < 0.05), and ALT (p < 0.05) levels, and positively correlated with HDL; Firmicutes, Rumboutsia abundance was negatively correlated with HDL levels, where Firmicutes was positively correlated with body weight gain (p < 0.05), TC (p < 0.05), TG (p < 0.05), ALT (p < 0.05) and AST (p < 0.05) plasma levels, Allobaculum was positively correlated with body weight (p < 0.01), LDL (p < 0.05) correlation, and Romboutsia was positively correlated with ALT (p < 0.01) and LDL (p < 0.05) plasma levels (Figure 8A).

The abundance of Bacteroidia, Prevotellaceae_UCG-003 was positively correlated with lipolytic genes CPT-1, PPAR-α (p < 0.01), where UCG-003 was positively correlated with bile acid pathway related genes NTCP, BSEP (p < 0.01), and negatively correlated with adipogenic genes SREBP-1, ACC-α, LXR-α (p < 0.01). ACC-α, LXR-α (p < 0.01), both of which were negatively correlated with bile acid pathway-related genes FXR (p < 0.01); the abundance of firmicutes was positively correlated with the levels of LXR-α, SREBP-1, ACC-α and FXR (p < 0.05), and negatively correlated with the levels of CPT-1 (p < 0.05), PPAR-α (p < 0.01). Oscillibacter abundance was negatively correlated with the levels of LXR-α, ACC-α (p < 0.05), and positively correlated with the levels of CPT-1 and PPAR-α (p < 0.05) (Figure 8B).

This study has several strengths, not the least of which is its design based on the enterohepatic axis, which sets it apart from other studies in the field. By comparing different types of exercise in the context of MASLD, it provides a more comprehensive assessment of the bile acid pathway, lipid metabolism, and other relevant metrics. It also explains possible mechanisms. However, it is also important to understand its limitations. For example, it did not assess liver function markers and genes associated with lipid metabolism prior to training, and it lacked baseline data, which could affect the reliability of the results. In addition, it did not examine how lipid-related processes in the body are affected during training. Therefore, we were unable to get a full picture of the effects of the three exercises on these processes. This study also did not perform a lipidomics study to further clarify the effects of exercise on lipogenesis.