The animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of Guangxi University and were conducted in accordance with the animal welfare and ethical rules. Three-week-old male C57BL/6 mice were purchased from Beijing Vital River Laboratory Animal Technology Company (Beijing, China). Mice were maintained in individually ventilated cages (IVCs) under 12 h light/12 h dark cycles at an ambient temperature of 22 ± 2°. Mice were ad libitum accessed to food and water.

After acclimating for one week, the mice were randomly divided into four groups with eight mice each in group (control, HFD, HFD + MGE300, and HFD + MGE600). The control mice were fed with a chow diet (10% energy from fat, D12450, Research Diets, New Brunswick, NJ) and the other mice were challenged with an HFD (60% energy from fat, D12492, Research Diets). Mice in the control and HFD groups were given orally 100 μl water at 9:00 a.m., while the HFD + MGE300 and HFD + MGE600 mice were orally administered 300 mg/kg or 600 mg/kg body weight MGE, respectively. During treatment for 18 weeks, the body weight and food intake were recorded weekly during treatment.

TheMGE was provided by Layn Biotech, Inc. (Guilin, China). Its main constituents were described in our previous publication (14).

The body fat percentage of mice was assayed by using Small Animal Body Composition Analysis and Imaging System NMR Analyzer (MesoQMR23-060H, Niumag Corporation, Shanghai, China).

Mice were subjected to Glucose Tolerance Test (GTT) (at 19 weeks of age) and Insulin Tolerance Test (ITT) (at 20 weeks of age) as previously described (30). Briefly, mice were fasted for 12 h or 6 h and then intraperitoneally injected with 2 g/kg body weight D-glucose or 1 IU/kg body weight insulin for GTT and ITT, respectively. Blood was collected from the tail tip at 0, 15, 30, 60, and 120 min post-injection. Glucose concentration was measured using an Accu-Chek Performa (Roche Diagnostics, United States).

The triglyceride (TG) content in the liver was measured using a TG colorimetric assay kit (Applygen Technologies, Beijing, China) according to the manufacturer’s instructions. Fifty milligrams of liver tissue were grounded in liquid nitrogen, and then TG was extracted. TG concentration was measured using a BioTek Epoch full-wavelength microplate reader (Epoch, United States) at a 550 nm wavelength.

Fat and liver tissues were fixed in 4% paraformaldehyde (PFA) overnight at 4°. The samples were embedded in paraffin and cut into 5 μm thick sections according to routine procedures. Hematoxylin and Eosin (H&E) staining was performed according to a standard procedure. Images were captured under a light microscope (Nikon, Eclipse 50i).

The structures of the colonic microbiota were assessed by 16S rRNA amplicon gene sequencing, which was performed by Biomarker Technologies (Beijing, China). Samples were collected for the colons and total genomic DNA was extracted by using the TIANamp Stool DNA Kit (TIANGEN Biotech, Beijing, China), and the V3-V4 regions of the 16S rRNA gene were amplified by using the forward primer 338F (5′-ACTCCTACGGGAGGCAGCA-3′) and reverse primer 806R (5′-GGACTACHVGGGTWTCTAAT-3′). The operational taxonomic unit (OTU) table and taxonomic analysis were obtained from Qiime2 (2019.10) software (31).

Post hoc analyses were performed using Prism 7 software (GraphPad, San Diego, CA, United States) when significance was achieved in the main or simple effects model. Multiple comparisons were corrected using Tukey–Kramer or Dunnett–Hsu adjustment in mixed model procedures. Data are presented as means ± SEM. P < 0.05 was considered statistically significant for a null hypothesis of no difference between study test meals. Bioinformatics analyses were carried out at the Multifunction Computer Center of Guangxi University and the R platform (version 4.0.3).

After challenge with a HFD for eighteen weeks, the mice had vastly higher body weights (Con vs. HFD; 25.1 ± 0.15 vs. 45.9 ± 0.95 g; P < 0.0001) and body fat percentage (Con vs. HFD; 7.7 ± 0.25 vs. 44.4 ± 0.41%; P < 0.0001) than the mice fed the control diet. There were no significant differences in body weight or body fat content between mice in the HFD and HFD + MGE300 groups, indicating that 300 mg/kg MGE does not attenuate obesity. However, when the mice were treated with 600 mg/kg MGE, the body weight and body fat content were significantly reduced compared with those of the HFD group (Figures 1A,B), indicating that 600 mg/kg MGE can attenuate HFD-induced obesity. In addition, compared with the control group, the high-fat diet increased the average weekly calorie intake of mice. However, there was no difference in calorie intake between MGE treated group (MGE300, MGE600) and the HFD group (Figure 1C). For individual adipose tissue, when mice were fed an HFD, the inguinal white adipose tissue (iWAT) and epididymal white adipose tissue (eWAT) weights were notably increased compared with those of the mice fed a control diet. A 300 mg/kg MGE did not decrease the iWAT or eWAT weight in the mice fed an HFD. However, 600 mg/kg MGE significantly reduced iWAT weight but had no impact on eWAT weight in the mice fed a HFD (Figure 1D). Consistently, the HFD challenge obviously increased the adipocyte size of iWAT and eWAT in the mice. Interestingly, 600 mg/kg MGE treatment reduced the adipocyte size of iWAT but did not change the adipocyte size of eWAT in DIO mice (Figures 1E–G). Considering the effects of 600 mg/kg MGE on alleviating obesity, this dose was used in the subsequent analyses.

We next determined whether MGE treatment might ameliorate fatty liver in DIO mice. As shown in Figure 2A, the mice in the HFD group had higher liver weights than those in the control group (Con vs HFD; 1.3 ± 0.03 vs 1.7 ± 0.13 g; P < 0.05). H&E staining showed that there were many lipid droplets distributed in the liver of HFD-challenged mice (Figure 2B), suggesting that HFD feeding results in fatty liver. We further assayed TG content in the liver and found that the mice in the HFD group had significantly higher TG content than the mice in the control group (Figure 2C). This further confirmed that the HFD challenge leads to fatty liver in mice. As expected, MGE treatment markedly reduced the liver weight and TG content in the livers of mice challenged with an HFD, indicating that MGE can alleviate fatty liver in DIO mice (Figures 2A–C).

Compared with the control group, the HFD group had significantly higher AUCs of GTT and ITT (Figures 2D–G), indicating that the HFD challenge reduces glucose and insulin sensitivity in mice. However, MGE treatment notably reduced the AUCs of the GTT and ITT in the mice challenged with a HFD (Figures 2D–G). Taken together, MGE supplementation alleviated fatty liver and improved global metabolic capacities in DIO mice.

High-throughput sequencing generated 1,919,310 raw reads. After removing the low-quality sequences, there were 1,787,645 clean tags, and each sample produced an average of 74,485 clean tags. Based on a 97% similarity cutoff value, all the effective reads were clustered into operational taxonomic units (OTUs).

To identify the differences in intestinal microbiota structure among groups, the α-diversity and β-diversity were compared. As shown in Figures 3A–D, the α-diversity indexes (ACE, Shannon, and phylogenetic diversity) in the HFD group were significantly increased compared with those in the control group, but there were no differences between the HFD and MGE groups. On the other hand, non-metric multidimensional scaling (NMDS) and principal coordinates analysis (PCoA) of the unweighted UniFrac distance matrix were used to identify β-diversity. As shown in Figures 3E,F, NMDS and PCoA showed that the clusters of the HFD group were significantly different from those in the control group, but MGE treatment significantly ameliorated these alters. Taken together, the above results indicate that MGE can restore the changes of community structure in microbiota but not species types in DIO mice.

Combining the unweighted pair-group method with arithmetic mean (UPGMA), the relative abundance of the gut microbiota at the phylum level was identified by QIIME2 taxonomic analysis. As shown in Figure 4A, the HFD group had an apparent difference in taxonomic composition compared with the control group, but there was an apparent similarity between the MGE group and the control group. For individual phyla, the relative abundance of Firmicutes was significantly increased, but that of Bacteroidetes was significantly decreased due to HFD feeding, resulting in an increase in the ratio of Firmicutes to Bacteroidetes (F/B) in DIO mice (Figure 4B). However, MGE treatment restored the relative abundance of Firmicutes and Bacteroidetes, as well as the ratio of F/B to control levels, in DIO mice (Figure 4B). In addition, the linear discriminant analysis effect size (LEfSe) method was used to further determine differences in microbiota abundances (Figure 4C). There was 67 differentially abundant microbiota, and 49 out of 67 (73.1%) belonged to Bacteroidetes and Firmicutes, which further confirmed that MGE can restore the abundance of Firmicutes and Bacteroidetes in DIO mice.

To further explore the effect of MGE on bacterial taxonomic composition at the genus level, a clustering heatmap was constructed. Figure 5A displays the top 72 OTUs with significant differences (P < 0.01) selected for clustering analysis. Fifty-eight out of 72 OTUs were significantly less abundant in the control group and MGE group than in the HFD group. Interestingly, Lactococcus, Blautia, Lactobacillus, Ruminiclostridium_9, Oscillibacter, and Ruminiclostridium belong to Firmicutes and are obesogenic microbiota (Figure 5B). In contrast, 14 out of 72 OTUs, including Bacteroides, were significantly more abundant in the control group and MGE group than in the HFD group (Figure 5B). Taken together, MGE can restore the intestinal microbiota at the genus level in DIO mice.

We further performed a correlation analysis between the gut microbiota structures and obese phenotypes. Based on the Spearman index, we found that the abundances of Firmicutes, Lactobacillus, and Lactococcus were positively correlated with obese phenotypes, such as body weight, iWAT weight, body fat content, liver TG concentration, and AUC of GTT and ITT. However, the abundances of Bacteroidetes and Bacteroides were negatively correlated with the above obesity parameters (Figure 6A). The results of correlation analysis show that MGs alleviate obesity and maybe associated with the gut microbiota modulation (Figure 6B).