Limosilactobacillus reuteri HM108 is derived from healthy breast milk and is presently conserved at the China General Microbiological Culture Collection Center (CGMCC No. 26166). Limosilactobacillus reuteri HM108 was grown in de Man–Rogosa–Sharpe broth (Hopebio Company, Qingdao, China) at 37°C for 12 h and passaged twice to ensure viability and promote optimal growth. After being rinsed twice with sterile phosphate-buffered saline (PBS), the bacteria were adjusted to the target concentration for gavage in rats.

All animal experiments were conducted in line with the Guidelines for the Care and Use of Laboratory Animals of Beijing Union University and received approval from the Animal Ethics Committee of Beijing Union University (JCZX11-2306-2, Beijing, China). Male SD rats aged 6 weeks of SPF grade were sourced from Vital River Laboratory Animal Technology Co., Ltd. (Beijing, China). A detailed outline of the experimental protocol is presented in Figure 1A.

During the first week of acclimatisation, the animals were maintained at a constant temperature of 22 ± 2°C and a humidity level of 55 ± 10%. The animals were maintained on a normal diet (ND) under a 12-h light/dark regimen. Following this initial phase, 50 rats weighing 200 ± 20 g were divided randomly into five separate groups (n = 10): (1) ND; (2) HFD; (3) high-fat + low-dose (HFD-L, 2.5 × 10⁸ CFU/rat); (4) high-fat + medium-dose (HFD-M, 5 × 10⁸ CFU/rat); (5) high-fat + high-dose (HFD-H, 1.5 × 10⁹ CFU/rat), and three intervention groups supplemented with different concentrations of L. reuteri HM108 in their diet simultaneously. The HFD consisted of 15% sucrose and 15% lard. Over the course of the 6-w experiment, the ND and HFD groups of rats received a daily dose of 2 mL of pure PBS solution via gavage. The remaining groups were administered 2 mL of freshly prepared PBS containing the specified concentrations of L. reuteri HM108.

Following gavage, faecal samples were collected following a 12-h fasting period. Blood was extracted via cardiac puncture under sodium pentobarbital-induced anaesthesia. After resting for 1 h at 25°C, the serum was separated by centrifugation. The rats were then dissected to collect liver tissue, which was promptly flash-frozen in liquid nitrogen and preserved at −80°C for future analysis.

Enzyme-linked immunosorbent assay (ELISA) kits (Nanjing Jiancheng Institute of Biological Engineering, Nanjing, China) were used to measure various biomarkers, including blood lipid contents—total cholesterol (TC), triglycerides (TG), high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C)—as well as blood glucose, lipopolysaccharide (LPS), tumour necrosis factor-alpha (TNF-α), interleukin-1β (IL-1β), IL-10, and IL-6.

Liver tissues were randomly selected, fixed in 10% formalin for 24 h, dehydrated, embedded in paraffin, sectioned into 5-μm slices, and stained with haematoxylin and eosin (H&E). Histological observations were performed using a light microscope, and the adipocyte area fraction was quantified (relative quantification) using Image-Pro Plus 6.0 (Media Cybernetics, Rockville, MD, United States).

The fresh, collected faeces were stored at −80°C. Gut microorganisms were measured and analysed by Shanghai Applied Protein Technology Co., Ltd. Briefly, the 16S rRNA V3–V4 region was amplified using primers 341F (CCTAYGGGRBGCASCAG) and 806R (TAATCTATGGGGNNCATAAGG). The Illumina Novaseq 6000 was used to conduct pyrosequencing of the polymerase chain reaction, followed by a rigorous quality assessment of the raw data, and merged paired-end reads.

The QIIME 2 package in R was used to select representative reads of each Amplicon Sequence Variant and perform alpha and beta diversity index analysis and visualisation. Finally, linear discriminant analysis effect size (LEfSe) was used to assess the importance of variations in species abundance and identify the different microbial biomarkers. The Kyoto Encyclopaedia of Genes and Genomes (KEGG) was used to predict the microflora metabolic function.

The freshly-collected faeces were stored at −80°C, and SCFAs were measured and analysed by Shanghai Applied Protein Technology Co., Ltd. Briefly, the faeces were re-suspended in 50 μL of 20% phosphoric acid solution, with 4-methylpentanoic acid added as an internal standard. The mixtures were thoroughly agitated and then centrifuged, and the supernatant was carefully transferred into the gas chromatography-mass spectrometry (GC-MS) equipment (GC 7890 B, 5977 B, Agilent Technologies Co. Ltd., CA, United States) to determine SCFAs concentrations.

The samples were separated using a GC system equipped with an Agilent DB-FFAP capillary column (30 m × 250 μm, 0.25 μm). A 1-μL injection was performed using a 10:1 split ratio, with the column temperature initially set at 90°C. The temperature was increased to 160°C at a rate of 10°C/min, then to 240°C at 40°C/min, and held steady for 5 min. Helium was used as the carrier gas at a flow rate of 1.0 mL/min. MS followed, with inlet, ion source, and transmission line temperatures set at 250°C, 230°C, and 250°C, respectively, and a quadrupole temperature of 150°C. The electron ionisation source operated at 70 eV electron energy.

The freshly collected faeces were stored at −80°C, and untargeted metabolomics were measured and analysed by Shanghai Applied Protein Technology Co., Ltd. Briefly, faecal samples were extracted with a cold methanol-acetonitrile-water (2:2:1, v/v), followed by vortexing and sonication. Finally, the supernatant was vacuum-dried, reconstituted in acetonitrile-water (1:1 v/v), centrifuged, and analysed using ultra-high-performance liquid chromatography coupled with a quadrupole time-of-flight system (UHPLC-QTOF) (Agilent 1290 Infinity/AB Sciex Triple TOF 6600).

The mobile phase consisted of two components: A (25 mM ammonium acetate/hydroxide in water) and B (acetonitrile). The gradient ran from 95% B (0.5 min) to 65% (6.5 min), then to 40% (1 min), held (1 min), returned to 95% (0.1 min), and re-equilibrated (3 min). Electrospray ionisation settings: ion source gas (Gas), curtain gas = 30, ion spray voltage = ±5,500 V. Primary m/z detection: 60–1,000 Da; secondary: 25–1,000 Da; collision energy = 35 ± 15 eV.

After sum-normalisation, the datasets were analysed using the ropls package in R and assessed using principal component analysis (PCA). The t-test was used to identify statistically significant differences between two independent sample groups. Significant differential metabolites were selected for KEGG pathway enrichment analysis.¹

The collected liver tissue samples were maintained at −80°C, and transcriptome was analysed by Shanghai Applied Protein Technology Co., Ltd. Briefly, liver RNA was quantified (A260/A280) using a Nanodrop ND-2000 system (Thermo Fisher Scientific, MA, United States). Sequencing was conducted using an Illumina Novaseq 6000. Using the HISAT2 package (version 2.2.1) in orientation mode, clean reads were individually aligned to the reference genome to acquire the mapped reads. The DESeq2 package (version 1.47.1) was analysed for differential expression genes (|log₂ FC| >1 and p_adj <0.05). Finally, Gene Set Enrichment Analysis (GSEA) was conducted by calculating the enrichment scores, assessing their significance levels, and correcting for multiple hypothesis testing.

GraphPad Prism 10.1.2 was used for statistical analyses. A two-way analysis of variance followed by Dunnett’s test was used to analyse body weight. One-way analysis of variance (ANOVA) followed by Tukey’s test was used to analyse the differences in blood glucose and lipid content, inflammatory factors, proportion of SCFAs, and F/B ratio. Correlations between gut microbes and body fat ratio, blood lipid content, inflammatory factors, and differential metabolites were evaluated using Spearman’s correlation, with significance at p < 0.05.

HFD feeding for 6 w substantially increased the total intake, total calories, weight gain, total body weight, body fat weight, body fat ratio, and feed intake of the rats (p < 0.05). In intervention groups with different doses of probiotics, there was no significant effect on the feed intake of rats, but the three parameters of body weight (p < 0.01), body fat weight, and body fat ratio (p < 0.05) showed dose-dependent improvements, indicating that supplementation with L. reuteri HM108 can significantly reduce high-fat diet induced weight gain and decrease body fat weight and body fat ratio (Figures 1B–E). Other indices were reduced to varying degrees, although not statistically (Supplementary Figures 1A–E).

Prolonged exposure to an HFD may result in hepatic lipid accumulation and a deterioration of liver function. Obesity elevates the risk of non-alcoholic fatty liver disease and associated conditions, which are characterised by enlarged lipid droplets and decreased interstitial spaces, which compress nearby cells (19). Analysis of the liver histopathological morphology using H&E staining indicates that adipose tissue stores excess energy by increasing the size and number of adipocytes (20). In the current study, L. reuteri HM108 inhibited HFD-driven adipocyte growth (Figure 1F).

Changes in lipid content in rats are shown in Figures 2A–D. After 6 w of intervention, glucose content in the HFD group was similar to that in the other groups (Supplementary Figure 1F). However, TC, TG, and LDL-C levels were higher, and HDL-C levels were lower than those in the control and ND groups (p < 0.05). TG and TC levels were significantly reduced in the intervention group, indicating that L. reuteri HM108 can improve blood lipid levels.

GSP reflects mean glycemia over the preceding 2–3 w and serves as a reliable indicator for short-term blood glucose monitoring. The GSP content in HFD rats was elevated compared to both the ND and the intervention groups with L. reuteri HM108, particularly the HFD-L group (Figure 2E).

These results indicate that L. reuteri HM108 alleviats obesity by reducing lipid and decreasing GSP contents in the blood of rats fed an HFD.

HFD induces inflammation and dysregulation of adipocytokines as adipose tissue releases adipokines and biologically active molecules such as leptin, which affect systemic homeostasis. Most adipokines promote inflammation (21). Serum leptin, LPS, IL-6, IL-10, IL-1β, and TNF-α were quantified using ELISA (Supplementary Figures 1G,H). The findings indicated no marked alteration in TNF-α serum levels, whereas LPS, IL-6, IL-10, and IL-1β contents increased substantially in response to HFD. With L. reuteri HM108 treatment, the contents of LPS, IL-10, and IL-1β in HFD-induced rats were markedly reduced and restored to the contents of the ND group (Figures 2F–H). Leptin, an inflammatory factor that regulates body intake and energy expenditure, tended to decline (p = 0.06; Figure 2I).

The results suggest that L. reuteri HM108 can reduce chronic inflammation linked to obesity by decreasing serum inflammatory factor contents in a dose-dependent manner.

To determine whether L. reuteri HM108 affected the gut microbiota and reduced obesity, 16S rRNA sequencing was used to investigate gut microbiota composition. The α diversity ACE index indicated that L. reuteri HM108 affected the diversity of intestinal microflora (Figure 3A), whereas the PCA of β diversity demonstrated no marked variation in gut microbiota across all HFD groups (Figure 3B). The relative abundance of the predominant intestinal flora was evaluated at the phylum level. All faecal samples were found to have a similar community structure; Firmicutes and Bacteroidetes were dominant, representing over 90% of the relative abundance. The F/B ratio rose to 3.46 ± 1.23 in obese rats on the HFD versus the ND group but dropped to 2.25 ± 0.94 in the intervention group with L. reuteri HM108 (Figures 3C,D).

LEfSe analysis was used to identify the microbiota altered by L. reuteri HM108 in HFD rats (Figures 3E,F). Limosilactobacillus reuteri HM108 increased the abundance of Bacteroides, Prevotellaceae, Eubacterium, Anaerovoracaceae, Holdemanella, Enterococcaceae, Jeotgalicoccus, and Ruminococcaceae while reducing the abundance of harmful bacteria such as Lachnospiraceae, Clostridia_UCG-014, Butyricimonas, Staphylococcaceae, and Rothia. The abundance of Lactobacillus showed no marked variation across the groups.

Functional changes in the gut microbiota with L. reuteri HM108 intervention were predicted using the KEGG database (Supplementary Figure 2I). Compared to the ND group, HFD consumption enriched the metabolism-related KEGG pathways, including the pentose phosphate pathway, ABC transporters, pyruvate metabolism, TCA cycle, PPAR signalling pathway, LPS biosynthesis, carbohydrate digestion, and absorption, all of which were reversed by L. reuteri HM108. Moreover, L. reuteri HM108 substantially enhanced the synthesis of several small molecules and metabolic pathways, including amino acid and polyamine metabolism; these small-molecule metabolites act as growth signals and nutrients for microbes and the host gut epithelium (22).

These results suggest that L. reuteri HM108 can alter the gut flora composition of HFD rats to promote the expansion of bacteria that are beneficial in the context of obesity.

SCFAs are key regulators of lipid metabolism. Propionic, acetic, and butyric acids account for >95% of intestinal SCFAs (23). GC-MS analysis of rat faeces indicated that the HFD group reduced most SCFAs (p < 0.05), except for butyric acid (p = 0.15), compared to the ND group, suggesting that HFD inhibited SCFAs production. Limosilactobacillus reuteri HM108 tended to increase butyric acid contents in rat faeces (p = 0.097) (Supplementary Figures 2A–H). These results suggest that HFD inhibits SCFAs production, which can be reversed by L. reuteri HM108.

To assess the metabolic alterations caused by L. reuteri HM108 in HFD-induced rats, faecal samples were subjected to untargeted metabolomic profiling using UHPLC-Q-TOF MS. After combining the positive and negative ions, PCA indicated that the HFD diet altered metabolites and was screened for differential metabolites (|log₂FC| >1 and p < 0.05) (Figure 4A). Compared to the HFD, 61 differential metabolites were present in the HFD-L group, 27 in the HFD-M group, and 73 in the HFD-H group. Lipids and lipid molecules with substantial changes in abundance across the five groups, including sn-glycero-3-phosphocholine, butyric acid, hetisine, and linoleic acid, were analysed using heat maps (Figure 4B).

PICRUSt2 was used to forecast the gut microbiota’s KEGG function before and after the L. reuteri HM108 intervention (Figures 4C–E). Relative to the HFD group, the HFD-L group exhibited 11 pathways associated with obesity: arginine and proline metabolism, ABC transporters, taurine and hypotaurine metabolism, metabolic pathways, glutathione metabolism, d-amino acid metabolism, histidine metabolism, protein digestion and absorption; and aminoacyl-tRNA biosynthesis. The HFD-M group also changed mineral absorption and glycine, serine, and threonine metabolism. The HFD-H group increased nucleotide and pyrimidine metabolism and valine, leucine, and isoleucine biosynthesis.

These results suggest that L. reuteri HM108 can alter the metabolites in faeces and affect pathways associated with obesity, such as energy metabolism, protein synthesis, and amino acids, thereby alleviating obesity in rats.

After identifying the obesity-reducing phenotype of L. reuteri HM108, transcriptome analysis was conducted on liver tissues from HFD and HFD-H rats to elucidate the association between changes in gene expression related to liver tissue and obesity resulting from the HFD. A total of 81 differentially expressed genes were identified, with 33 upregulated and 48 downregulated genes. The upregulated genes included Map3k5, Reg3b, G6pc, Ndst2, Flrt1, Reg3g, Plcd3, and Ndufa1; downregulated genes included Rgs3, Dact1, Hmgcr, Socs3, Cyp7a1, Foxo3, Lpin1, Cish, Gck, and Insig1 (Figure 5A). KEGG pathway enrichment analysis results are shown in Supplementary Table S1.

GSEA was examined using actual overall trends rather than specific differential gene thresholds. In this research, compared to the HFD group, the expression of JAK-STAT signalling pathway in liver tissue decreased in the intervention group (Figure 5B). The main genes involved in this pathway were Myc (ENSRNOG00000004500), Cish (ENSRNOG00000029543), and Osmr (ENSRNOG00000033192).

In this study, compared to the HFD group, the expression of JAK-STAT signalling pathway in liver tissue decreased in the intervention group of Lactobacillus rhamnosus HM108.

The results indicate that intervention with L. reuteri HM108 can effectively inhibit the JAK-STAT signalling pathway, with the important downregulation of Myc, Cish, and Osmr genes.

Obesity and metabolic disorders induced by an HFD largely stem from a gut microbiota imbalance. To study the effect of L. reuteri HM108 on obesity via the gut microbiota, 20 gut microbiota with marked changes in genus-level abundance before and after the intervention were selected for Spearman’s correlation analysis (Figure 5C). Although Lachnospiraceae are SCFAs producers (24), they are more abundant in obese and diabetic mice and humans (25). The correlation analysis indicated that Lachnospiraceae_GCA-900066575 exhibited a positive association with several markers, including body fat ratio, body weight, GSP, glucose, low-density lipoprotein, and LPS. This suggests that it plays a key role in the development of obesity.

Limosilactobacillus reuteri HM108 affected the composition of gut microbes and metabolite profiles. Therefore, Spearman’s correlation analysis was conducted for these 20 gut microorganisms and 20 significantly different metabolites associated with obesity (Figure 5D). The important intestinal bacterial genus, Prevotellaceae_UCG-001 was positively correlated with valeric acid-1, β-sitosterol, and γ-muricholic acid and negatively correlated with oleic acid, oleoyl-l-carnitine, and saikosaponin b2. Prevotella was correlated with 1-hydroxyvitamin D, 11-deoxy-16,16-dimethylprostaglandin e2, ergocalciferol (vitamin D2), and saikosaponin b2.