L. rhamnosus strain B6 (CGMCC 13310) was provided by the State Key Laboratory of Dairy Biotechnology (Shanghai, China). The strain was routinely maintained on deMan–Rogosa–Sharpe (MRS) agar plates (Thermo Scientific™) and preserved long-term in MRS broth supplemented with 20% glycerol at −80 °C. Prior to the gavage experiments, the strain was anaerobically activated on MRS agar plates at 37 °C for 48 h in a Bugbox anaerobic workstation (Ruskinn, England) under an atmosphere of N₂:H₂:CO₂ = 80:10:10. Freshly activated B6 cultures were inoculated into sterile MRS broth and incubated at 37 °C for 18 h. L. rhamnosus B6 cells were harvested by centrifugation at 7,500×g and 4 °C for 15 min, washed, and resuspended in sterile 0.9% saline to a final concentration of 1.0 × 10⁸ colony-forming units (CFU) per milliliter. Viable cell counts were determined by spread plating serially diluted aliquots onto MRS agar plates, followed by anaerobic incubation at 37 °C for 48 h.

Escherichia coli BNCC307544 (ATCC 43895) was purchased from BeNa Culture Collection (Henan, China). E. coli SJTUF40005, Salmonella enterica SJTUF10458, and S. enterica SJTUF10464 were provided by Shanghai Jiao Tong University (Shanghai, China). All strains were cultured individually on trypticase phytone yeast extract (TPY) agar plates (Thermo Fisher Scientific™) at 37 °C under aerobic conditions for 24 h.

Indicator bacterial suspensions were diluted with sterile water to 10⁷ CFU/mL, and 30 μL of each suspension was spread on TPY agar plates. The cell-free culture supernatant of L. rhamnosus B6 was collected by centrifugation at 12,000 rpm for 10 min, followed by boiling for 3 min to inactivate enzymes and eliminate H₂O₂. The antibacterial activity against LPS-producing bacteria was determined using the Oxford cup diffusion method. Briefly, the Oxford cups were placed on indicator agar plates, 100 μL of the L. rhamnosus B6 cell-free supernatant was added to each cup, and the plates were incubated aerobically at 37 °C for 48 h. The diameters of the inhibition zones were measured using a Vernier caliper.

For the colonization experiment, following a 1-week acclimation period, 20 6-week-old male C57BL/6J mice (purchased from Chengqin Biotechnology Co., Ltd., Shanghai, China) were randomly divided into two groups. All mice received a 3-day antibiotic regimen (daily gavage of 0.2 mL of 25 mg/mL ampicillin solution, equivalent to 5 mg/mouse). Subsequent to ampicillin treatment, the mice were orally administered 0.2 mL of L. rhamnosus B6 suspension or 0.9% sterile saline via gavage once daily for 2 weeks. Fecal samples were collected at baseline and at the 1st and 2nd week post-interventions. At the end of the gavage period, five mice per group were euthanized via carbon dioxide (CO₂) asphyxiation in a 40 L euthanasia chamber (gas flow rate: 12 L/min), and the intestinal segments were harvested for further analysis. The microbial community composition was determined using 16S rRNA gene sequencing.

For the HFD-induced MAFLD animal experiment, 18 eight-week-old male C57BL/6J mice (Chengqin Biotechnology Co., Ltd.) were acclimated for 1 week and then randomly allocated to three groups (n = 6 per group): (1) negative control (SD group, standard diet; Research Diets, D12492i); (2) model control (HFD group, high-fat diet; Research Diets, D12450J); (3) HFD + L. rhamnosus B6 (HFD + B6 group, HFD plus daily gavage of 10⁹ CFU L. rhamnosus B6). Body weights and food intakes were recorded weekly. At the end of the 14-week study, the mice were euthanized via CO₂ asphyxiation. Blood samples were collected by cardiac puncture into plasma separation tube. Serum was harvested by centrifugation at 4,000 rpm for 15 min at 4 °C and stored at −80 °C. Subcutaneous adipose tissue, livers, and colons were collected, weighed, and snap-frozen at −80 °C.

Liver tissues were processed for hematoxylin and eosin (H&E) staining and scanning electron microscopy (SEM). For H&E staining, the tissues were fixed in 4% paraformaldehyde for 24 h, embedded in paraffin, sectioned into 5 μm slices, dewaxed, rehydrated, stained with H&E, and observed under a light microscope (Olympus). For SEM, liver tissues were perfused with a fixative at 4 °C for 4 h, post-fixed in 0.1% osmium tetroxide at room temperature for 2 h, rinsed with PBS, dehydrated, embedded, and sectioned into 60 nm–80 nm ultrathin slices. The slices were stained with 2% uranyl acetate and lead citrate, air-dried overnight, and examined under a scanning electron microscope (Hitachi, Tokyo, Japan).

Serum total cholesterol (TCH) and low-density lipoprotein cholesterol (LDL-C) were measured using a Beckman Chemistry Analyzer AU2700 System (Beckman Coulter, Tokyo, Japan). Hepatic triglyceride (TG) levels were assayed by homogenizing liver tissues, centrifuging to collect supernatants, and using a triglyceride quantification kit (Abcam, ab65336) according to the manufacturer’s instructions.

Serum levels of tumor necrosis factor (TNF)-α, interleukin (IL)-1β, and IL-10 were quantified using enzyme-linked immunosorbent assay (ELISA) (MEIMIAN, Jiangsu, China), following the manufacturer’s instructions.

Genomic DNA was extracted from murine colon contents and fresh fecal samples using the OMEGA Soil DNA kit (M5635-02; Omega Bio-Tek, Norcross, GA, USA), following the manufacturer’s instructions. DNA quantity and quality were assessed using a NanoDrop NC2000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA) and agarose gel electrophoresis. The V3–V4 hypervariable regions of the 16S rRNA gene were amplified by polymerase chain reaction using the primers 338F (5′-ACTCCTACGGGAGGCAGCAG-3′) and 806R (5′-GGACTACHVGGGTWTCTAAT-3′). The PCR reaction contained 5 μL 5×buffer, 0.25 μL FastPfu DNA polymerase (5 U/μL), 2 μL dNTP mix (2.5 mM each), 1 μL of each primers (10 μM), 1 μL DNA template, and 14.75 μL ddH₂O. The thermal cycling program was as follows: initial denaturation at 98 °C for 5 min; 30 cycles of denaturation at 98 °C for 30 s, annealing at 55 °C for 30 s, and extension at 72 °C for 45 s; and a final extension step at 72 °C for 5 min. PCR amplicons were purified using Vazyme VAHTS™ DNA Clean Beads (Vazyme, Nanjing, China) and quantified using the Quant-iT PicoGreen dsDNA Assay Kit (Invitrogen, Carlsbad, CA, USA). Equal amounts of purified PCR products were pooled and subjected to paired-end sequencing (2 × 250 bp) on the Illumina NovaSeq platform (NovaSeq 6000 SP Reagent Kit, 500 cycles) at Genekinder Medical Technology Co., Ltd. (Shanghai, China).

Sequence data were analyzed using QIIME2 and R software (v3.2.0). ASV-level alpha diversity (Chao1 index) was calculated based on the ASV tables in QIIME2. Beta diversity was assessed using Bray–Curtis dissimilarity and visualized using principal coordinate analysis (PCoA). Phylogenetic analysis was performed using MEGA-X, and trees were visualized using iTOL (https://itol.embl.de/). Microbial functional predictions were performed using PICRUSt2 against the MetaCyc (https://metacyc.org/) and KEGG (https://www.kegg.jp/) databases.

Untargeted metabolomics of murine fecal samples was performed using an UHPLC system (Vanquish, Thermo Fisher Scientific) coupled to an Orbitrap Exploris 120 mass spectrometer (Orbitrap MS, Thermo) with a Waters ACQUITY UPLC BEH Amide column (2.1 mm × 50 mm, 1.7 μm). Briefly, 25 mg of fecal samples were mixed with grinding beads and 0.5 mL of extraction solution (MeOH: ACN:H₂O, 2:2:1, v/v) containing deuterated internal standards. The mixtures were vortexed for 30 s, homogenized at 35 Hz for 4 min, sonicated in an ice-water bath at 4 °C for 5 min (repeated three times), incubated at −40 °C for 1 h to precipitate proteins, and centrifuged at 12,000 rpm for 15 min at 4 °C. The supernatants were transferred to new glass vials for further analysis.

Raw data were converted to mzXML format using ProteoWizard and processed using an in-house R program based on XCMS for peak detection, extraction, alignment, and integration (22, 23). Metabolite annotation was performed using the KEGG and Human Metabolome Database (HMDB). Principal component analysis (PCA), orthogonal partial least squares discriminant analysis (OPLS-DA), pathway analysis, volcano plots, and heatmaps (annotated with fold changes and adjusted P-values) were generated using MetaboAnalyst 6.0 (six biological replicates per group).

Total RNA was extracted from the liver samples using TRIzol reagent (T9108, Takara) following the manufacturer’s protocol. RNA concentrations were measured using a Nanodrop analyzer (Thermo Fisher Scientific), and the extracted RNA was reverse-transcribed into complementary DNA using the PrimeScript RT reagent kit (RR037A, Takara). Real-time PCR was performed on an Applied Biosystems™ 7500 (Thermo Fisher Scientific) with gene-specific primers (Supplementary Table S1). The results were normalized against the housekeeping gene GAPDH, and the data were analyzed using the 2^−ΔΔCt method.

Statistical analyses and graphical visualizations were performed using GraphPad Prism 10.4.1. After testing for normality and homogeneity of variances, an unpaired two-tailed Student’s t-test was employed for comparisons between the two groups. For multi-group comparisons, one-way analysis of variance (ANOVA) was conducted first, followed by Tukey’s multiple comparison test for post hoc analysis. Statistical significance was set at P <0.05.

Antibiotic-depleted mice were used to evaluate L. rhamnosus B6 colonization and the regulation of intestinal flora. Initially, fecal microbial diversity (Chao 1 index) was comparable between groups, but L. rhamnosus B6-supplemented mice exhibited a significantly higher Chao 1 index than controls after 1 week (P <0.05), though this difference diminished at 2 weeks (Figure 1A). These results demonstrate that L. rhamnosus B6 accelerates the restoration of intestinal flora after antibiotic treatment. After 2 weeks of L. rhamnosus B6 intervention, the gut microbiota structure became more diverse (Figure 1B), and the relative abundance of Lactobacillus increased (Figures 1C, D). Phylogenetic analysis of Lactobacillus-related OTUs confirmed that OTU6337 (16S rRNA V3-V4 sequence identical to L. rhamnosus) was present in the colonic and intestinal contents of mice after 2 weeks of L. rhamnosus B6 intervention, verifying successful colonization (Figures 1E, F).

To assess the alleviative effects of L. rhamnosus B6 on MAFLD, mice were fed a normal chow diet, a high-fat diet, or a high-fat diet supplemented with L. rhamnosus B6 for 14 weeks, following the experimental protocol illustrated in Figure 2A. Outcome indicators were measured at the end of the study. L. rhamnosus B6 supplementation reduced body weight gain in the HFD + B6 group to levels comparable to those observed in the ND group and significantly lower than those in the HFD group (Figure 2B). Regarding metabolic parameters, although serum TCH, LDL, and liver TG levels in the HFD + B6 group were higher than those in the ND group, they were markedly lower than those in the HFD group (Figures 2C–E). H&E staining revealed that hepatocytes in the HFD group exhibited swelling with prominent steatosis, accompanied by extensive accumulation and dispersion of lipid droplets in the cytoplasm. Mitochondrial deformation and rupture were observed, along with a disorganized rough endoplasmic reticulum. Notably, L. rhamnosus B6 intervention significantly ameliorated hepatocellular steatosis, reduced the number of lipid droplet vacuoles, tended to normalize mitochondrial structure, and alleviated degeneration and disorganization of the rough endoplasmic reticulum (Figures 2F–K). Collectively, these results demonstrate that L. rhamnosus B6 administration alleviates MAFLD, as corroborated by both metabolic indicators and histological observations.

Chronic inflammation is a key factor in the pathogenesis of MAFLD. The gut–liver axis, which mediates the translocation of bacterial products into the portal circulation, has been identified as an upstream trigger of inflammation in MAFLD (24, 25). Following L. rhamnosus B6 intervention, the expression of inflammatory factors in liver tissues and their concentrations in serum were determined. Compared with the HFD group, the L. rhamnosus B6 intervention group exhibited a marked reduction in the serum levels of the pro-inflammatory factors TNF-α and IL-1β, along with an increase in the anti-inflammatory factor IL-10, with a trend toward the levels observed in the ND group (Figures 3A–C). Furthermore, the expression of inflammatory factors in liver tissues was detected by RT-qPCR, which showed that L. rhamnosus B6 intervention significantly reduced the expression of pro-inflammatory factors TNF-α and IL-1β, as well as nuclear transcription factor NF-κB (Figures 3D–F). Collectively, these findings suggest that L. rhamnosus B6 intervention can reduce the serum concentrations of pro-inflammatory factors and subsequently suppress their expression in the liver, thereby alleviating hepatic inflammation.

Accumulating evidence has demonstrated that gut microbial dysbiosis is closely associated with MAFLD progression. Therefore, intestinal contents were collected from each mouse at the end of the experiment for 16S rRNA sequencing to elucidate the underlying mechanism through which L. rhamnosus B6 alleviates MAFLD. At the genus level, L. rhamnosus B6 intervention increased the relative abundance of Lactobacillus compared to the HFD group, although this was not statistically significant (Figure 4A). Notably, L. rhamnosus B6 intervention significantly decreased the abundance of Allobaculum and an unclassified Coriobacteriaceae, while significantly increasing the abundance of Akkermansia (Figure 4B). The top 10 bacterial ASVs in each group were identified using random forest analysis and ranked by importance scores. The results indicated that L. rhamnosus B6 intervention suppressed the abundance of several ASVs belonging to Allobaculum, Cupriavidus, Desulfovibrionaceae, and Enterobacteriacea (Figure 4C). PCoA based on the Bray–Curtis dissimilarity metric revealed that the gut microbial community structure in the L. rhamnosus B6-treated group differed from that in the HFD group but was closer to that in the ND group, highlighting the regulatory effect of L. rhamnosus B6 administration on gut microbial composition (Figure 4D). PICRUST2 was employed to predict the microbial functional profiles based on ASV abundance. The PWY-1422 and PWY-5948 pathways were upregulated, whereas pathways involved in myo-inositol degradation (PWY-7237 and PWY-562) and LPS synthesis (LPSSYN-PWY) were downregulated following L. rhamnosus B6 intervention, and these profiles differed significantly from those in the HFD group (Figure 4E). Spearman’s correlation analysis was performed to explore the associations between ASVs and predicted functional pathways, as illustrated in Figure 4F. The ASVs suppressed by L. rhamnosus B6 exhibited a significant positive correlation with the pathways inhibited by L. rhamnosus B6 administration. In addition, antibacterial assays revealed that the fermentation supernatant of L. rhamnosus B6 exerted a significant inhibitory effect against Enterobacteriaceae strains, including E. coli BNCC307544, E. coli SJTUF40005, S. enterica SJTUF10458, and S. enterica SJTUF10464, with inhibition zone diameters of 19.7 mm, 20.9 mm, 20.1 mm, and 19.1 mm, respectively (Supplementary Figure S1). Overall, these findings indicate that L. rhamnosus B6 intervention downregulates LPS synthesis and myo-inositol degradation functions of the gut microbiota by inhibiting the proliferation of ASVs belonging to Allobaculum, Cupriavidus, Desulfovibrionaceae, and Enterobacteriaceae.

Beyond inflammatory cytokines, the imbalance between fatty acid uptake/lipogenesis and export/metabolism leads to lipid accumulation, another key hallmark of MAFLD (26). To investigate the impact of L. rhamnosus B6 intervention on MAFLD-related metabolic alterations, untargeted metabolomics analysis was performed on fecal samples from mice in each group at the end of the experiment using LC-MS/MS, and the data were processed using the MetaboAnalyst platform (version 6.0). The PCA results revealed a clear separation between the three groups. Compared with the HFD group, the HFD + B6 group clustered closer to the ND group (Figure 5A), and the orthogonal partial least squares-discriminant analysis (OPLS-DA) score plot also displayed distinct grouping between the HFD and HFD + B6 groups (Figure 5B). A volcano plot was used to identify differential metabolites (DMs) between the HFD and HFD + B6 groups, and a total of 197 DMs were screened, including 116 upregulated and 81 downregulated metabolites (Figure 5C). These DMs were categorized into several classes, with lipids and lipid-like molecules being the most abundant, accounting for 49.2% of all significantly altered compounds modulated by L. rhamnosus B6. Among the differentially expressed lipids and lipid-like molecules, fatty acyls, glycophoropholipids, sterol lipids, and prenol lipids were the major components, representing 34.0%, 30.9%, 17.5%, and 12.4%, respectively (Figure 5D). A heatmap depicting the hierarchical clustering results of the top 30 DMs showed that nine and 21 metabolites were significantly enriched in the HFD and HFD + B6 groups, respectively (Figure 5E; Supplementary Table S2). Specifically, oleic acid, linoleic acid, and lipoxin A4 levels were elevated, whereas Cer[NS] 36:3 levels were reduced following L. rhamnosus B6 intervention. All DMs between the two groups were subjected to KEGG-based metabolic pathway enrichment analysis. The core metabolic pathways involving α-linolenic acid, linoleic acid, sphingolipids, phenylalanine, tyrosine metabolism, and mitochondrial β-oxidation of long-chain saturated fatty acids were identified as the core metabolic pathways associated with the ameliorative effects of L. rhamnosus B6 on MAFLD (Figure 5F).

To further investigate the correlations among gut microbiota, metabolic biomarkers, and key DMs in MAFLD mice, a Mantel test was performed to analyze the associations between the ASVs matrix (comprising ASVs with a relative abundance >0.1%) and major biomarkers, including body weight gain, serum biochemical indices, and hepatic inflammatory cytokine expression levels. As presented in Figure 6A, the ASV matrix of the HFD+B6 group was negatively correlated with body weight gain, whereas the ASV matrix of the HFD group showed a non-significant positive correlation with this parameter. In addition, marked differential correlations with ASV composition were observed between the HFD + B6 and HFD groups in terms of serum TCH, TNF-α, and IL-10 concentrations and hepatic IL-1β expression. Specifically, the HFD + B6 group exhibited negative correlations with serum TCH levels, serum TNF-α levels, and hepatic IL-1β expression and a positive correlation with serum IL-10 concentrations. In contrast, the HFD group showed opposite correlation patterns with the aforementioned biomarkers (Figure 6A). Spearman’s rank correlation analysis was conducted among the abundance of the 10 key discriminatory bacterial ASVs, serum biomarkers, and 12 significantly altered lipids and lipid-like metabolites. The results demonstrated that S24-7_ASV76453, Lactobacillus_ASV50156, and Akkermansia_ASV10889, all of which had higher relative abundances in the HFD + B6 group, were positively correlated with IL-10, oleic acid, linoleic acid, and lipoxin A4 levels. Conversely, the other discriminatory bacterial ASVs that predominated in the HFD group displayed opposite correlation trends (Figures 6B–D). Overall, these correlation analyses indicate that the ASVs significantly enriched by L. rhamnosus B6 intervention are closely associated with the core metabolic and inflammatory biomarkers, suggesting that the L. rhamnosus B6-modulated gut microbiota may play a crucial role in mediating the ameliorative effects on MAFLD.