Lactobacillus plantarum ZJUFB2 (stored at China center for type culture collection, CCTCC M2020126) was screened from Chinese traditional fermented food (sourdough). The B2 tested sample was manufactured according to the reported guidelines (24). The concentration of the lyophilized B2 sample was ~10¹¹ colony forming units per gram (CFU/g).

Male C57BL/6 mice (n = 36) of 6-weeks were acquired from the Shanghai SLAC Laboratory Animal Co., Ltd. (Shanghai, China), housed at 60% relative humidity and 25°C, and provided with free access to water and diet. After a week of acclimatization, the mice were divided into three groups (n = 12, 04/cage). The mice of different groups were given a normal chow diet (NCD, 10% calories from fat, MD17111, Jiangsu Medicine Ltd., Supplementary Table 1), HFD (45% calories from fat, MD12032, Jiangsu Medicine Ltd., Supplementary Table 1), and an oral intervention of 0.2 ml of saline containing ~10⁹ CFU bacteria daily for a period of 16 weeks: (i) NCD group (NCD, saline), (ii) HFD group (HFD, saline), and (iii) B2 group (HFD, B2), respectively. Afterward, mice were fasted up to 12 h and euthanized using pentobarbital sodium (100 mg/kg, i.v.) to collect the samples. Serum samples were obtained by centrifugation of blood (3,000 × g, 15 min). The samples including mice livers, inguinal white adipose tissue (iWAT), epididymal white adipose tissue (eWAT), and brown adipose tissue (BAT) were obtained, weighed, and stored at −80°C.

After a 15-weeks feeding, male C57BL/6 mice were fasted (12 h) followed by an intraperitoneal glucose tolerance test (GTT) was conducted after glucose injection (2 g/kg of the bodyweight). Blood glucose was determined with an interval of 0, 30, 60, 90, and 120 min using a glucometer (Accu-Check glucometer, Roche, Mannheim, Germany). Fasting serum insulin and serum glucose levels were measured by ELISA and glucose oxidase assay kits (Jiyinmei, Wuhan, China, and Jiancheng Bioengineering Institute, Nanjing, China). Furthermore, the insulin resistance in terms of homeostasis model assessment was determined using this equation:

Homeostasis model assessment of insulin resistance (HOMA-IR) = Fasting serum glucose (mM) × Fasting serum insulin (mU/L)/22.5.

The serum biochemical parameters including low-density lipoproteins cholesterol (LDL-C), high-density lipoproteins cholesterol (HDL-C), total cholesterol (T-CHO), triacylglycerols (TG), total bile acids (TBAs), and hepatic glutathione peroxidase enzyme (GSH-Px), total superoxide dismutase enzyme (SOD), and catalase enzyme (CAT) were measured using corresponding kits (mention names) (Jiancheng Bioengineering Institute, Nanjing, China). Lipopolysaccharide (LPS) content was detected by a ToxinSensor Chromogenic Limulus Amebocyte Lysate (LAL) (Endotoxin Assay Kit, GenScript, Piscataway, New Jersey, USA).

Serum lipopolysaccharide-binding protein (LBP), leptin (LEP), adiponectin (ADP), glucagon-like peptide-1 (GLP-1), peptide YY (PYY), hemoglobin A1c (HbA1c), tumor necrosis factor-alpha (TNF-α), interleukin 10 (IL-10), and interleukin 6 (IL-6) levels were determined using commercial ELISA kits (Jiyinmei, Wuhan, China).

The hematoxylin and eosin (HandE) staining procedures and analysis were performed following the procedures of Zhong et al. (25). To determine the size of the adipocytes, 10 different random visual fields were computed for each HandE staining image.

All the reagents for extracting total RNA, synthesizing cDNA, and Quantitative PCR (qPCR) procedures were carried out using commercial kits (Vazyme Corp., Nanjing, China). The PCR program and calculation process were conducted following our previous study (25). The primers of sterol regulatory element-binding protein 1c (SREBP-1c); fatty acid synthase (FAS); peroxisome proliferator-activated receptors (PPARγ and PPARα), cholesterol 7α-hydroxylase (CYP7A1), and SHP used for performing RT-qPCR were listed in Supplementary Table 2.

For gut microbiota analysis, the procedure was carried out according to our study (25). Briefly, fresh fecal samples (ca. 30 mg, n = 8) on 16 weeks were collected and immediately frozen using dry ice and stored at −80°C. Total DNA was extracted from fecal samples using a QIAamp DNA Stool Mini Kit (Qiagen, Venlo, Netherlands) according to the instructions of the manufacturer. The V3-V4 hypervariable regions of bacteria 16S rRNA gene were amplified with primers 338F (5′- ACTCCTACGGGAGGCAGCAG-3′) and 806R (5′-GGACTACHVGGGTWTCTAAT-3′) by thermocycler PCR system (GeneAmp 9700, ABI, USA). Library quality was checked using a Thermo NanoDrop 2000 ultraviolet microspectrophotometer, and 2% agarose gel electrophoresis as well as Qubit for library quantification (Thermo Scientific, USA). Illumina MiSeq platform (Majorbio Bio-pharm Technology Co. Ltd., Shanghai, China) was used to sequence. The analysis protocol was performed following our previous work (25). Sequences with 97% similarity were clustered into operational taxonomic units (OTUs) using UPARSE (V7.0.1001) (http://drive5.com/uparse/). A representative sequence was picked for each OTU and the Greengenes reference database was used to annotate taxonomic information for each representative sequence. The OTU absolute abundance table was extracted from the pipeline and converted to relative abundances by normalizing to total OTU clustering for analyzing the composition of gut microbiota by Quantitative Insights into Microbial Ecology software (QIIME) (http://qiime.org). In order to compare diversity, the OTU table was rarified to calculate the metrics as Chao1, Observed species, and Shannon index. 3D principal coordinate analysis (PCoA) plot was constructed for the evaluation of the dissimilarity and the community composition between samples from weighted UniFrac distances. Linear discriminant analysis (LDA) effect size (LEfSe) was performed to identify the difference between groups. The correlation analysis between the intestinal bacteria and significant biomarkers in mice was calculated by the R software (R version 3.6.3) (Free Software Foundation, USA). Raw Illumina MiSeq reads were publicly accessible in the NCBI Sequence Read Archive (SRA) database with accession number PRJNA723728.

The short-chain fatty acids (SCFAs) analysis was performed following the protocol of our previous study (25). The results of SCFAs were shown in μmol/g of a fecal sample.

The obtained data were analyzed statistically to determine the level of significance using one-way ANOVA along with a post-hoc Tukey's test. The statistical analysis was conducted through SPSS 22 (SPSS Inc., Chicago, IL, USA). GraphPad Prism 6 (GraphPad Software Inc., San Diego, CA, USA) was used to shape the experimental data, and the data shown in the manuscript indicating mean ± SE (n = 8–12). Means with the same letters in a table or a figure differ non-significantly (P > 0.05).

After 8 weeks, the body weight and total body weight gain (TBWG) of HFD and B2 groups were increased significantly (Figures 1A,B) as compared with the NCD group. At the end of the experiment, mice fed on B2 showed a relatively higher percentage of body weight gain compared with the NCD group, but no significant difference was observed in comparison to the HFD group (Figure 1C). However, no significant difference was observed in daily energy intakes among the groups, indicating that an HFD did not change energy intakes (Figure 1D). Similarly, there was no significance found in the weight of liver and fat tissues, and the size of eWAT between HFD and B2 groups (Figures 1E–G). Intriguingly, Figure 1H showed that the number of multi-locular lipid droplets in the livers of the B2 group were lower than in the HFD group. Based on the findings of this study, B2 exhibited strong inhibitory effects on the lipid formation in the liver of HFD-fed mice without disturbing their body weight.

High-fat diet group did not exhibit any effects on the serum TG (Figure 2A), but TCHO contents increased (from 4.16 to 6.4 mmol/L) in HFD-fed mice compared with the NCD group (Figure 2B). However, mice who received B2 intervention exhibited a decline in LDL-C (Figure 2C). Likewise, the B2 group showed a slight increase in the HDL-C in comparison to the NCD group (Figure 2D). In the GTT result, Figure 2E demonstrated that less glucose tolerance was observed in HFD-fed mice. The B2 intervention did not show any improvement in the glucose tolerance in comparison with the HFD group (Figure 2F). An HFD significantly elevated fasting serum glucose in mice but fasting serum insulin and HbA1c of NCD- and HFD-fed mice showed no significant differences (Table 1; Figure 2H). However, the data of HOMA-IR (from 3.25 to 1.74) and serum fasting glucose (from 6.96 to 3.7 mmol/L) showed that B2 intervention significantly decreased insulin resistance and hyperglycemia (Figure 2G; Table 1).

To evaluate changes in the hormone levels with B2 treatment, the serum LEP, ADP, PYY, and GLP-1 levels were determined. Figures 3A,B, indicate that the mice group fed with B2 showed no significant change in the LEP levels compared with the HFD group, but a decrease was observed in the ADP levels. B2 supplementation in HFD-fed mice significantly increased the serum GLP-1 and PYY levels compared with the NCD and HFD groups (Figures 3C,D). The serum LPS levels in the HFD group were significantly increased compared with the NCD group (Figure 3E). Overall, B2 intervention significantly improved HFD-induced increased levels of LPS and LBP (Figures 4E,F). Although there were some positive modifications in TNF-α, IL-6, and IL-10 levels after B2 administration, but no significant differences were observed (P > 0.05) (Figures 3G–I). Mice fed with B2 exhibited a slightly higher content of IL-10 (Figure 3I) and a lower concentration of IL-6 (Figure 3H) than that of the HFD group. Figures 3J–L showed that there were no significant differences in liver antioxidant enzymes between HFD and B2 groups.

To investigate that B2 intervention would alter the gut microbiota and lipid metabolism, the contents of bile acids and lipometabolism-related genes expression levels were determined. The total bile acids data showed a significant increase in the liver tissues with B2 intervention compared with the NCD and HFD groups, while there was a slight increase in fecal and serum samples (Figures 4A–C). The results of RT-qPCR showed that the mRNA expression levels of SREBP-1c, CYP7A1, and PPARα have not been significantly altered after B2 intervention (Figures 4E,F,I). The expressions of the FAS were upregulated in HFD and B2 groups as compared with the NCD group (Figure 4D). The mRNA expression levels of PPARγ and SHP were found lower in the B2 group than HFD group (Figures 4G,H).

The α-diversity (diversity parameter of Sobs and Chao1) of the gut microbiota was significantly decreased in the HFD-fed group compared with the NCD group (Figures 5A,B). However, the results for Simpson and Shannon diversities showed a decrease in the B2 group (Figures 5C,D) compared with the HFD group, indicating that the B2 intervention slightly decreased the gut microbial α-diversity. To investigate the difference in the microbial structure among the different groups, a PCoA was performed to calculate the similarity of community composition (Figure 5E). HFD and B2 groups formed a peculiar composition of microflora independently from the NCD group. Moreover, the B2 group had a separated cluster as compared with the HFD group. Figure 5F showed that the HFD and B2 groups shared 448 and 415 common OTUs with the NCD group, respectively. The classification abundance analysis showed that the HFD group had a higher relative abundance of Firmicutes and a lower Bacteroidetes in comparison with NCD-fed mice. After B2 supplementation, the abundances of Firmicutes and Bacteroidetes showed a decline, and Actinobacteria and Proteobacteria increased compared with the HFD group (P > 0.05, Figure 5G). The results of community abundance on Family level showed that B2 had a lower abundance of Ruminococcaceae (P < 0.05), Lactobacillaceae, Desulfovibrionaceae, Lachnospiraceae, and Mogibacteriaceae (P < 0.05), and a higher abundance of Bifidobacteriaceae, Helicobacteraceae, and Deferribacteraceae in comparison with HFD group (Figure 5H).

The gut microbiota of experimental mice was mainly composed of 50 genera (Figure 6A). The consumption of an HFD significantly increased the relative abundance of Allobaculum, Oscillospira, Bifidobacterium, norank_f__Desulfovibrionaceae, unclassified_f__Helicobacteraceae, Ruminococcus, norank_f_Mogibacteriaceae, Mucispirillum, Streptococcus, unclassified_f__Lachnospiraceae, unclassified_f__Ruminococcaceae, Lactococcus, and Dehalobacterium, and significantly decreased the relative abundance of norank_f__S24-7, Prevotella, Turicibacter, Desulfovibrio, Clostridium, unclassified_o__Bacteroidales_c__Bacteroidia, Sutterella, Coprococcus, Adlercreutzia, and norank_o__YS2. To further examine the specific effects of B2 on the gut microbiota in HFD-fed mice, a LEfSe analysis was used to find the distinct bacteria in each group (Figure 6B). Figures 6C–K showed the modifications of the abundance of the characterized genera of bacteria base on the LEfSe report. The predominant genus in the NCD group was g_norank_f__S24-7 and in the HFD-fed mice was Allobaculum (Figures 6C,D). The administration of B2 dramatically decreased the obese-type gut microbiota (Oscillospira, norank_f__Desulfovibrionaceae, Ruminococcus, and norank_f_Mogibacteriaceae) were significantly decreased after B2 intervention (Figures 6E–H), and the alteration in their abundance was in line with the modification in HOMA-IR. The relative abundance of probiotics Bifidobacterium was also explored (Figure 6J). B2 treatment significantly elevated its relative abundance (P < 0.05) in comparison with the HFD group. Moreover, F16, a biomarker in the B2 group, was remarkably increased after probiotic administration (Figure 6I), while Akkermansia was decreased (Figure 6K). Finally, Spearman's correlation analysis was applied to investigate the relationship between the intestinal microflora and biochemical profiles (Figure 6L). Microbes such as F16 and g_norank_f__Lactobacillaceae, were found significantly enriched in the B2 group and were negatively correlated with HOMA-IR and LPS. In contrast, Oscillospira, Lactococcus, and Mucispirillum had positive correlations with LDL-C and LPS.

Figure 7 shows that B2 intervention significantly increased the total contents of fecal SCFAs as compared with the HFD group (Figure 7H). Specifically, B2 intervention significantly increased the levels of acetate, i-butyrate, and i-valerate in comparison with the HFD group (Figures 7A,C,E). Moreover, the B2 group showed a relatively higher amount of propionate and valerate than the HFD group (Figures 7B,F).