Five-week-old male C57BL/6J mice (n = 50) were obtained from Beijing Huafukang Laboratory Animal Technology Co., Ltd. (Beijing, China). The mice were housed on a reverse 12:12-h dark-light cycle at 22°C–25°C with free access to food and water. All procedures were approved by Sport Science Experiment Ethics Committee of Beijing Sport University (approval number: 2022023H).

After 1-week acclimation, the mice were randomly divided into two groups: the normal control group (CON, n = 20) fed with regular chow diet (3.87 kcal/g, 12.79% energy as fat, 1,025, Huafukang, Beijing, China) and the high-fat diet (HFD, n = 30) group was fed with HFD (5.24 kcal/g, 60% energy as fat, H10060, Huafukang, Beijing, China). IR was diagnosed if the area under the curve (AUC) during glucose tolerance test (GTT) of the HFD group was 1.2 times higher than that of the CON group. After 12 weeks of diet, the mice were allocated into the following four groups: normal sedentary (NS, n = 8), normal exercise (NE, n = 8), HFD sedentary (HS, n = 8), and HFD exercise (HE, n = 8). The body weight was measured once a week using weighing scale. Food intake was measured twice a week by weighing the food provided and remaining in each cage. The consumed amount was calculated by multiplying by the caloric content to determine the weekly caloric intake. Fasting blood glucose levels were measured weekly from the tail tip after a 12-hour fasting period. (Figure 1A).

The exercise groups underwent acclimation for 3 days (10–15 m/min, 20–40 min/d) consecutively before beginning the exercise training sessions. Adhering to the established treadmill exercise model (Hoydal et al., 2007), the mice performed on a treadmill with a progressive increase in running speed: 15–20 m/min, 60 min/d, 5 days/week for 8 weeks. Mice in the sedentary groups were placed on the treadmill for the same duration as those in the exercise group.

The mice underwent a glucose tolerance test (GTT) 48 h after the completion of the exercise protocol, following an overnight fast. The insulin tolerance test (ITT) was then performed on the third day after the exercise protocol, following a 4-hour fasting period. Blood glucose concentrations were measured by a glucometer (Roche Ltd., Basel, Switzerland). GTT was measured after an intraperitoneal injection of glucose (2 g/kg). To perform the ITT, calculated the volume of insulin to be injected into each mouse based on a dose of 0.5 U/kg of body weight. Blood glucose levels were measured from the tail vein at 0, 30, 60, 90, and 120 min after the glucose challenge (Vinué and González-Navarro, 2015). Homeostasis model assessment of insulin resistance (HOMA-IR) was calculated using the following equation: HOMA-IR = [Fasting Insulin (mU/L) × Fasting Glucose (mmol/L)]/22.5. HOMA-IR was calculated to assess IR, with a normal range of 1–3 in healthy, non-obese mice (Lee et al., 2008).

The mice were euthanized by isoflurane inhalation 48 h after the final training session. Blood samples were procured after eyeball removal, and the entirety of the blood samples were allowed to clot undisturbed at room temperature for 30 min. Blood samples were centrifuged at 4°C for 15 min at 3,000 rpm (Eppendorf 5424, rotor F-45-24-11, 5424000014). Then, only the left gastrocnemius was dissected and rapidly frozen with liquid nitrogen and stored at −80°C for biochemical analyses (n = 6 per group). For immunofluorescence experiments, the freshly excised right gastrocnemius (n = 3 per group) was covered with optimum cutting temperature compound and submerged into precooled isopentane in liquid nitrogen for 30 s. The quadriceps muscles were dissected and stored at −80°C for Western blotting and RNA analysis (n = 3 per group).

Frozen muscle tissues were sectioned at 8 μm in thickness using a sliding cryostat microtome (Leica CM 1850; Germany). Three tissue sections were selected from each sample (n = 3/group). The cryosections were fixed with a 4% paraformaldehyde fixing solution for 10 min and then washed in PBS three times, permeabilized in 0.1% Triton X-100 for 15 min, and blocked with 5% normal goat serum for 1 h at room temperature. Then the sections were incubated overnight at 4°C with primary antibodies. The dilutions of primary antibodies were as follows: F4/80 (1:100, MCA497GA, Bio-Rad), CD11c (1:200, 173420-1-AP, Proteintech), CD206 (1:200, AF2535, R&D Systems). The sections were protected from light and incubated with secondary Alexa Fluor 488- or Alexa Fluor 555-labeled goat anti-rabbit or mouse antibodies (1:1,000, A-11034, A-21422, Molecular Probes) for 2 h. Nuclei were stained by DAPI. Results were observed with a fluorescence microscope (Leica, Germany). The fluorescence index was determined with Image Pro Plus 6.0 (Media Cybernetics, United States).

Four biochemical indicators of blood lipids were measured with kits (Njjcbio, Nanjing, China) using a clinical chemistry analyzer (AU5800, Beckman). IL-1β and IL-10 levels were measured using ELISA kits (AD3364Mo, AD2837Mo, Andygene). Procedure was followed according to the manufacturer’s instructions. Serum and tissue samples were homogenized in PBS with proteinase inhibitors, then centrifuged to collect the supernatant. Protein concentration was determined using the BCA Assay Kit. The samples were incubated in 96-well plates coated with capture antibodies, followed by detection with biotinylated antibodies and a substrate reaction. Absorbance at 450 nm was measured to determine cytokine concentrations using a standard curve.

Total proteins were isolated from skeletal muscle using RIPA protein buffer (89900, Thermo Scientific) and protease inhibitor cocktails (P8340, Sigma-Aldrich). Protein concentrations were detected using the bicinchoninic acid (BCA) protein assay kit (23227, Thermo Scientific). The samples were boiled with a loading buffer at 98°C for 10 min. The proteins were electrophoresed on 8%–12% SDS-polyacrylamide gels for separation, chosen based on target protein sizes, transferred to polyvinylidene difluoride (PVDF) membranes (88518, Cell Signaling), and blocked in 5% bovine serum albumin (BSA, A7906, Sigma-Aldrich) for 2 h. The proteins were transferred to PVDF membranes at 60 V for 1 h at 4°C. Primary antibodies (Table 1) were diluted in blocking buffer and incubated with the membranes overnight at 4°C. Then, the membranes were washed five times for 3 min in Tris-buffer saline (TBS, T1503, Sigma-Aldrich) containing 0.05% Tween 20 (TBST) and incubated with a secondary antibody conjugated to HRP for 1 h at ambient temperature. After the incubation, the membranes were washed five times for 3 min each using TBST. The density of each band was quantified using Image Lab densitometry software (Bio-Rad, Hercules, CA), and protein abundance was quantified by ImageJ. GAPDH served as a loading control to facilitate value normalization.

RNA was isolated from the quadriceps muscle using TRIzol. miRNAs sequence was sequenced by Sangon (Shanghai, China) using the Illumina sequencing platform (Illumina, San Diego, CA, United States). Briefly, the quality of raw sequencing reads was processed using FastQC and trimmed using CutAdapt and Trimmomatic to remove low-quality reads. Parameters for read quality filtering included sequence length >50 bp, average GC content between 30% and 70%, and a minimum Phred score of 30 for each read. Subsequently, the miRNA reads were aligned against established mature murine microRNA sequences using Bowtie (version 1.1.1). Before statistical analysis, all miRNA read counts were normalized and counted using the reads per million (RPM) method. Differential expression was performed using edgeR software and calculated using DESeq2. The threshold for DEGs was an |log2 (FC)| ≥ 1.0 and a P-value ≤0.05.

Total RNA was extracted from the quadriceps to the manufacturer’s instructions (B511321-0100, Dalian Baosheng). Conventional cDNA synthesis was performed using RevertAid Premium Reverse Transcriptase (EP0733, Thermo Fisher Scientific). qPCR reaction was carried out in 10 μL reactions using 2X SG Fast qPCR Master Mix (B639271, Dalian Baosheng). Real-time PCR of miRNA was performed using miDETECT A Track™ miRNA qRT-PCR Starter Kit (C10712, RIBOBIO). The data presented correspond to the mean of 2^−ΔΔCT from at least three independent experiments after being normalized to β-actin or U6. The primers were used in Table 2.

MiR-221-3p target gene prediction was conducted with TargetScan Mouse 7.2. To validate the interaction between miR-221-3p and Socs1, a luciferase reporter assay was performed. The 3′untranslated regions (UTRs) of Socs1, which contain potential miR-221-3p binding sites, were cloned into the downstream region of the Renilla luciferase gene. Mutations were introduced in the binding sites of the 3′UTR to disrupt miR-221-3p binding, and these mutated constructs were used to assess the specificity of miR-221-3p binding. The two reporter plasmids (wild-type (WT) and mutated (Mut)) were co-transfected into HEK293T cells with miR-221-3p overexpression plasmid or a negative control (NC) plasmid. Post 24 h of co-transfection, the Renilla luciferase activity was measured using the Dual-Glo luciferase reporter assay system (E2920, Promega, Madison, WI, United States) according to the manufacturer’s instructions. For controls, a luciferase assay kit negative control (NC plasmid) was used to ensure the specificity of the luciferase activity measurement. The luciferase activities of the reporter plasmids were normalized to those of the Firefly luciferase activity to account for transfection efficiency.

To assess statistical differences, the repeated measures analysis of variance (ANOVA) was analyzed by Bonferroni post hoc analysis for pairwise comparisons within the groups for body weight, food intake, and fasting blood glucose. The statistical significance of differences among groups was examined by two-way ANOVA with Bonferroni’s multiple-comparison test. For correlation analysis, the Pearson correlation test was employed. All data using Prism8 software (GraphPad software v8.0, Prism, La Jolla, CA). Values were expressed as mean ± SD and statistical significance was tested at P < 0.05.

To explore the contribution of a high-fat diet to the development of obesity and hyperglycemia, several parameters including body weight, fasting blood glucose, and food intake were meticulously measured over 12 weeks of intervention. Following a span of 3 weeks on the HFD, a significant rise in body weight and fasting blood glucose was evident, as compared to the normal control (CON) group (Figure 1B). Notably, the HFD group exhibited a notably larger area under the curve (AUC) in the glucose tolerance test (GTT) than the CON group (P < 0.01, Figure 1C). Upon IR model was confirmed, 8 weeks of exercise training was implemented in exercise groups. A substantial reduction in body weight was observed when contrasted with the CON group (P < 0.01, Figure 1D). This weight reduction was predominantly concentrated within visceral adipose tissue (VAT) and subcutaneous adipose tissue (SAT) (Figure 1G). Moreover, aerobic exercise reversed HFD diet-induced hyperglycemia (P < 0.01, Figure 1E), while the reduced caloric intake was not significant in the HE group compared with the HS group (Figure 1F).

To confirm that aerobic exercise can suppress dysglycemia and dyslipidemia, GTT, Insulin tolerance test (ITT), fasting insulin levels, and blood lipid levels were measured. The GTT and ITT demonstrated that exercise improved glucose tolerance and insulin tolerance in the HE group as compared with the HS group, which was indicated by a significantly decreased AUC (P < 0.01, Figures 2A,B). Serum insulin levels and HOMA-IR index were also significantly increased in the HS group, which is significantly higher than the normal range, indicating the occurrence of insulin resistance. Nonetheless, this perturbation was effectively reversed following an 8-week regimen of aerobic exercise (P < 0.01, Figures 2C,D). Next, we examined serum levels of lipids and observed that the contents of four measures were remarkedly increased. Most importantly, the content of total cholesterol (TC) and triglyceride (TG) decreased significantly with exercise treatment (P < 0.01, Figure 2E). These findings suggest that aerobic exercise intervention could ameliorate dyslipidemia and dysglycemia.

To investigate the effects of aerobic exercise on improving inflammatory status, the inflammatory cytokines IL-1β and IL-10 levels were assayed in both serum and skeletal muscle through employment of an ELISA kit. Levels of IL-1β in both serum and skeletal muscle exhibited a substantial increase as compared to the CON group. Remarkably, aerobic exercise precipitated a notable reduction in skeletal muscle IL-1β levels (P < 0.01, Figure 3A). Meanwhile, the M2 marker IL-10 was markedly reduced both in serum and skeletal muscle, and this effect was reversed by aerobic exercise training (P < 0.01, Figure 3B).

Macrophages are key cells in chronic inflammation, to test the phenotype of macrophages at the site of inflammation in skeletal muscle, we measured F4/80, CD11c, and CD206 protein using immunofluorescence, and counted the relative number of macrophages in each population. Skeletal muscle tissue was stained for macrophages (F4/80) and CD11c, marking M1-positive cells (Figure 4A), and for F4/80 with CD206, indicating M2-positive cells (Figure 4B). There was a significantly greater percentage of M1-type macrophages in skeletal muscle from mice fed with an HFD when compared to the CON group (P < 0.01, Figure 4C). Interestingly, aerobic exercise significantly increased the proportion of M2 macrophages when compared to the HS group (P < 0.01, Figure 4D). Further, the percentages of M1/M2 were remarkedly reduced after exercise training (P < 0.01, Figure 4E). Thus, HFD-induced IR is associated with increased macrophage infiltration and the chronic inflammation of skeletal muscle. Obesity-associated IR is related to the increased infiltration of phenotype of the macrophages.

To identify differentially expressed miRNAs by HFD diet and aerobic exercise treatment, we performed small RNA-sequence in skeletal muscle. 58 microRNAs were differentially expressed from HFD-exercise mice while 59 were differentially expressed from HFD mice in skeletal muscle, including 26 microRNAs that were concordantly regulated in both models (Figure 5A). miR-221-3p is a strong candidate that was increased 1.88-fold and decreased 1.25-fold in the HFD and exercise models, respectively (P < 0.001 and P = 0.016) (Supplementary Table S1).

To determine the possible mechanism underlying exercise-induced regulation of macrophage polarization, we used algorithms in TargetScan. Our investigation revealed that Socs1 emerges as a prospective target of miR-221-3p (Figure 5B). We then carried out a dual luciferase reporter assay for miR-221-3p. miR-221-3p overexpression significantly reduced the luciferase activity of the reporter genes containing wild-type 3′-UTR regions of Socs1, but no significant changes were observed in the mutated Socs1 region (Figure 5B).

Furthermore, miR-221-3p and Socs1 mRNA expression levels in skeletal muscle were detected using quantitative real-time PCR. Notably, an elevation in miR-221-3p levels was observed in the HFD group as compared to the CON group. Intriguingly, the implementation of aerobic exercise led to a significant reduction in miR-221-3p expression (P < 0.01, Figure 5C). Conversely, the levels of Socs1 were found to be attenuated in the skeletal muscle of HFD mice in contrast to the CON group. Remarkably, this reduction was counteracted through engagement in aerobic exercise (P < 0.01, Figure 5D). To further validate the negative regulatory role of miR-221 on Socs1, a comprehensive analysis of the correlation between miR-221-3p and Socs1 was undertaken. The outcomes unveiled a negative correlation between the expression levels of Socs1 and miR-221-3p, with statistical significance (P < 0.01, Figure 5E).

To determine whether SOCS1 is involved in regulating M1 and M2 macrophage polarization, we examined the protein levels of SOCS1 and macrophage polarization-related signaling factors (Figure 6A). The results showed that the expression of SOCS1 in HFD mice was significantly decreased compared with the CON group (P < 0.01). Aerobic exercise showed a greater increase of SOCS1 protein level compared with the HS group (P < 0.01, Figure 6B). The effects of exercise on improving M1 macrophage polarization were detected by measuring JAK1 and the levels of phosphorylated and total STAT1, STAT3. Along with reducing SOCS1 expression, HFD remarkedly elevated JAK1 and p-STAT1 and decreased p-STAT3 levels compared with the CON group (P < 0.01, Figures 6C–E). Meanwhile, the expression of JAK1(P < 0.01, Figure 6C) and p-STAT1 (P < 0.05, Figure 6D) were significantly reduced in the HE mice compared with the HS group.