The 6-week-old male SD rats (Vitalriver laboratory animal technology, Beijing, China) were randomly divided into 3 groups after 1 week of adaptive feeding: (I) normal chow diet group (NCD; 70% carbohydrate, 20% protein, 10% fat by energy); (II) HFD-fed group (HFD; 35% carbohydrate, 20% protein, 45% fat by energy); (III) HFD with 1.5% betaine-treated HFD-fed group (HBE). After fed with HFD for 16 weeks, group (III) were changed to add 1.5% betaine administration in drinking water (13328B, Adamas-beta®, Shanghai, China) once daily for another 8 weeks, while continuing to consume the HFD. At the end of experiment, all rats’ epididymal white adipose tissues (eWATs) were collected and stored at −80°C. Animal experimental ethics review committee of Henan TCM University granted approval for the animal experiments (approval number: IACUC-202307016).

Blood samples were obtained from fasted rats at 4°C for serum separation, which involved centrifugation at 3000 g for 15 min at the same temperature. The serum was subsequently stored at −20°C. Servicebio Biotechnology Co., Ltd. (Wuhan, China) conducted the measurements of serum liver transhepatic enzyme levels and lipid profile indicators. The concentrations of triglycerides (TG) in eWATs were quantified with a triglyceride assay kit (Rayto life and analytical sciences Co., Ltd., Shenzhen, China). SCFAs content in feces was analyzed using gas chromatography–mass spectrometry (8890B, Agilent, United States). Toward the end of the experiment, after rats that had fasted for the whole night received an intraperitoneal injection of 2 g glucose/kg body weight, the blood glucose level measured from the tail vein was determined after 0, 30, 60, 90, and 120 min of injection with the glucometer (Roche Diagnostics, Mannheim, Germany), respectively.

After fixation with 10% neutral formalin, epididymal white adipose tissue (eWAT) and liver tissue were embedded in paraffin and subsequently sectioned into 5-μm-thick slices. These slices were then stained using hematoxylin and eosin (H&E) and examined under the upright microscope (Eclipse, Niokon, Japan).

Genomic DNA was promptly extracted from fresh stool samples employing a fecal DNA isolation kit (MJYH, Shanghai, China), adhering to the manufacturer’s instructions, and subsequently stored at −80°C. PCR amplification cycling conditions were set as follows: initial denaturation at 95°C for 3 min, followed by 27 cycles of 95°C for 30 s (denaturation), 55°C for 30 s (annealing), and 72°C for 45 s (extension). A final extension at 72°C lasted for 10 min, ending at 4°C. Subsequently, high-throughput pyrosequencing of the PCR products was performed on an Illumina NextSeq 2000 PE300 platform (Illumina, San Diego, United States). Using the OTUs data, Mothur v1.30.2 was employed to compute rarefaction curves as well as alpha diversity metrics, such as the Chao and Ace indices. Principal coordinate analysis (PCoA) was conducted to assess the similarity between microbial communities across various samples and the linear discriminant analysis effect size (LEfSe) was utilized to identify bacterial taxa with significant abundance, defined by an LDA score > 3.5 and a p-value < 0.05.

LC–MS/MS analysis was conducted on a Thermo UHPLC-Q Exactive HF-X Vanquish Horizon system equipped with an Accucore C30 column (100 mm × 2.1 mm i.d., 2.6 μm; Thermo, United States). Mass spectrometric data was acquired employing a Benchtop Orbitrap Mass Spectrometer (UHPLC-Q-Exactive HF-X, Thermo, USA) which was furnished with a heated-electrospray ionization (HESI) source capable of operating in both positive and negative ion modes. The data matrix obtained from the preprocessing results encompassed information on lipid class, retention time (RT), mass-to-charge ratio (m/z) values, along with the corresponding peak intensities.

Total RNA was extracted from six individual adipose tissue samples of each group. The transcripts per million reads (TPM) method was employed to calculate the expression level of each transcript and identify the differential expression genes (DEGs). RSEM was used to quantify gene abundances (27). Essentially, the DESeq2 package was utilized to conduct the differential expression analysis. Genes were deemed significantly differentially expressed if they exhibited a |log2FC| ≥ 1 and a p-value < 0.05, according to DESeq2 criteria. Furthermore, functional-enrichment analysis was conducted to annotate DEGs that were significantly enriched in metabolic pathways, with a p-value threshold of <0.05, relative to the whole-transcriptome background.

All animal experimental data were expressed as mean ± SD. Before performing statistical analyses, the normality of the data was assessed using the Shapiro–Wilk test. GraphPad Prism 9.5 (GraphPad Software, San Diego, United States) was utilized to plot histograms and line graphs. For data that were normally distributed, multiple comparisons were performed by one-way ANOVA with Tukey’s post-hoc test. For data that did not meet the assumption of normality, the Kruskal-Wallis one-way analysis of variance by ranks was applied. p < 0.05 compared between two groups were considered statistically significant. Using iPath 3.0, a comprehensive transcriptome and metabolome analysis was performed to identify significant alterations in metabolic pathways. Bioinformatics analyses were performed on the online platform.¹

After administering 1.5% betaine in the drinking water and continuing HFD feeding for a duration of 8 weeks (Figure 1A), Lee’s index and the weight of epididymal white adipose tissue (eWAT) were lower in the HBE group than in the HFD group (Figures 1B,C). The lipid profile findings indicated that the levels of TG in the HBE group were markedly decreased compared to those observed in the HFD group, while high-density lipoprotein cholesterol (HDL-C) levels were increased (Figure 1E). Glucose tolerance tests (GTT) showed that the HBE group exhibited higher glucose tolerance compared to HFD rats (Figure 1F). However, no statistically significant differences were observed in fasting insulin levels (Figure 1G), alanine aminotransferase (ALT), or aspartate aminotransferase (AST) between the two groups (Figure 1J).

In addition, another characteristic of diet-induced obesity is hyperleptinemia, and mutations in leptin and leptin receptors can in turn lead to obesity (28). We found that betaine intervention significantly reduced serum leptin levels and body weight gain (Figures 1D,H) in HFD-fed rats when HFD was continuously administered, suggesting that the anti-obesity effect of betaine may be related to leptin signaling and carbohydrate metabolism.

Histological examination with H&E staining (Figure 1I) showed a decreased adipocyte area of the adipose tissue in the HBE group compared to the HFD group (Figure 1K). Together, the current findings collectively highlight the beneficial role of betaine in alleviating obesity-related metabolic disorders resulting from HFD.

Initial investigations have demonstrated the interrelationship between gut microbes and obesity, as well as the contribution of betaine as an edible extract in daily foods to microbiome homeostasis in obese individuals. However, its fundamental regulation mechanism is still mostly unknown. To assess changes in microbial community diversity between the three groups, we measured the ace index (p = 0.041) and the chao index (p = 0.078), which showed a significant reduction in α diversity in the HFD group (Supplementary Figures 1A,B). These findings were consistent with the petal plot analysis, which indicated that betaine supplementation led to an increase in amplicon sequence variants (ASVs) (Figure 2A). Principal coordinates analysis (PCoA), non-metric multidimensional scaling (NMDS) and community heatmap analysis shown that HFD altered the gut microbiota profile significantly but betaine remodelled its profile to that of NCD (Figures 2B,C; Supplementary Figure 1C).

Additionally, an investigation was conducted into the effects of betaine on gut microbiota composition. As depicted in Figure 2D, at the phylum level, Firmicutes predominated with approximately 90%, trailed by Bacteroidetes, Desulfobacterota, and Actinobacteriota. A 16-week HFD led to a notably higher Firmicutes/Bacteroidetes (F/B) ratio in the HFD group compared to the normal control diet (NCD) group. After 8-week betaine treatment, relative abundance of Bacteroidetes was significantly increased while that of Firmicutes was decreased, resulting in a decreased F/B ratio (Figure 2F). At the genus level, the relative abundance of Lactobacillus significantly decreased, while that of Lachnospiraceae increased markedly under the HFD. Betaine administration mitigated this shift, leading to a gut microbiota profile similar to that of the NCD group (Figure 2E). These findings were also in agreement with the results from PCoA and NMDS analyses. The Kruskal-Wallis H test of the relative abundance of top 10 genera at the genus level indicated that the relative abundance of Muribaculaceae, Clostridia_UCG-014, Lachnospiraceae_NK4A136_group, and Ruminococcus was also lowered by HFD, but elevated by betaine. Meanwhile, HFD raised the relative abundance of Blautia and Romboutsia (Figure 2G). Furthermore, LEfSe analysis was conducted to investigate the particular microorganisms affected by betaine, with the results displayed in Figures 2H,I. At the genus level, Lactobacillus, Clostridia_UCG-014, Ruminococcus were the differentially characterized genera for the NCD group, while Staphylococcus and Lachnospiraceae_NK4A136_group were, respectively, for the HFD and HBE groups. In addition, the relative abundance of Oscillospirales and Ruminococcaceae were also significant in the HBE group.

We analyzed SCFA composition in fecal samples from NCD, HFD and HBE groups to observe the response of SCFA to betaine treatment. Compared with the NCD group, HFD reduced the concentration of acetic acid,butanoic acid, valeric acid, hexanoic acid, isobutyric acid, isovaleric acid, isohexanoic acid, as well as the straight-chain fatty acids, branched-chain fatty acids, and total SCFAs. However, almost all of the above SCFAs reductions were increased after betaine intervention (Figures 3A–K). To obtain deeper understanding of the effects of SCFAs on obesity and metabolic processes, we conducted a correlation analysis of differential genus (top 20) and SCFAs concentrations. The results indicated that Blautia and Romboutsia were associated with a reduction in multiple SCFAs, whereas the representative genus of HBE group (Lachnospiraceae_NK4A136_group, Ruminococcus, Muribaculaceae, Clostridia_UCG-014) were linked to an increase in several specific SCFAs and total SCFAs. Furthermore, there was a positive correlation between the concentration of Lactobacillus and hexanoic acid (Figure 3L). As shown in Figure 3M, Mantel-test analysis also demeonstrated that straight-chain fatty acids (Mantel’s r = 0.76, p = 0.028) and total SCFAs (Mantel’s r = 0.76, p = 0.027) are the best explanator variable for HBE group and isohexanoic acid is highly correlated to HFD group (Mantel’s r = 0.44, p = 0.026). A more detailed description of the variables is available in Supplementary Table 1.

To further verify the effects of betaine on lipid metabolism in the adipose tissues, a lipidomic analysis was performed using LC–MS/MS. A total of 1,511 metabolites were annotated, with 1,283 ESI + ions and 228 ESI- ions detected, respectively. The characteristic relative standard deviation (RSD) analysis of quality control samples showed that the repeatability and stability of lipidomics analysis were qualified (Supplementary Figures 3A,B). Principal component analysis (PCA) showed that significant differences were observed between HFD-fed and NCD groups (Figures 4A,B), while betaine treatment resulted in characteristics in the HBE group more similar to those in the NCD group. Orthogonal partial least squares discriminant analysis (OPLS-DA) plot exhibited a distinct segregation pattern among the three groups in terms of adipose tissues (Figures 4C,D,F). In hierarchical cluster analysis of differentially altered lipids (DALs), significant changes in lipid profiles were found in rats fed HFD compared with the NCD group, and betaine treatment further modulated HFD-induced dysregulation of lipid metabolism (Figure 4E). Figure 4G; Supplementary Figures 3C,D displayed the enrichment of pathways via KEGG analysis, with regulation of lipolysis in adipocytes, glycerophospholipid (GP) metabolism, alpha-linolenic acid metabolism and thermogenesis pathway were up-regulated (differential abundance score >0.1, p < 0.001) in HBE group than in HFD group. Besides, the DALs observed between the HBE and HFD control groups were predominantly enriched in various lipid metabolic pathways, including choline metabolism in cancer, cholesterol metabolism, fat digestion and absorption, lipid and atherosclerosis, retrograde endocannabinoid signaling, insulin resistance, vitamin digestion and absorption, and arachidonic acid metabolism pathway.

To further determine the preferences of betaine on regulating lipid metabolism in adipose tissues, we explored the changes in content and composition of different lipid classes. Glycerolipids (GLs), comprising triglycerides (TG), diglycerides (DG), and free fatty acids (FFA), constitute the primary constituents of adipose tissue and serve as significant substrates for lipid oxidation. The total abundance of TG and DG did not change significantly between HFD and HBE (Supplementary Figures 2A,B), however, further analysis of the composition of glycerolipids showed that the species of DG and TG had changed. After betaine treatment, the most downregulated TG species were chain lengths of 50 carbons or less (C < 50) with 2–3 double bonds, whereas chain lengths of 55 carbons or more (C > 55) with more than 4 double bonds were the most upregulated, including TG(18:1/18:2/22:4), TG(16:0/18:2/22:6), TG(18:2/18:2/20:4), TG(11:0/18:2/20:4) (Figure 5C). When there is an energy demand, the TGs containing very long chain fatty acids in lipid droplets can be mobilized and rapidly transported to the peroxisome for fatty acid oxidation. However, these TG species were significantly lower in HFD group than in NCD group. Notably, species analyses of DG as a precursor to TG showed similar differences between the three groups. Compared with the NCD group, DG species with more than 4 double bonds were downregulated in HFD group and relieved after the betaine intervention. Interestingly, of the 11 detected DGs that produced the above changes, 6 were DG species that contained docosahexaenoic acid (DHA, 22:6), including DG (18:2/22:6), DG (16:0/22:6), DG (20:4/22:6), DG (18:0/22:6), and DG (18:3/22:6) (Figure 5D). It is generally accepted that DHA modulates energy metabolism and exhibits beneficial properties in anti-inflammatory processes, as well as enhancing insulin secretion. Consistent with DG, all 3 fatty acids (FA) species that contained more than 4 double bonds were increased, while 2 FA species that contained of 2–3 double bonds were decreased (Figure 5H).

Although glycerophospholipids (GPs) represent a small fraction of the lipidome, they are not only the main components of the cell membrane, but also participate in regulating the dynamics and homeostasis of the cell membrane. We observed that multiple phospholipid species, such as phosphatidylethanolamine (PE), phosphatidylcholine (PC), and phosphatidylinositol (PI), decreased after 12-week of HFD feeding and relieved after 8-week of betaine intervention. (Supplementary Figures 3D–F). GPs are the main components of mitochondrial membranes, and an increase in their content in adipose tissue suggests an enhancement in mitochondrial function and energy metabolism levels of adipocytes. Interestingly, in top 20 differential species of the GP subclasses, PE, PC and PI species with 20:4/22:5/22:6 chain also increased in the NCD group and HBE group, whereas in the HFD group we observed a decline pattern (Figures 5E–G). We presented the concentration levels of specific GP species featuring a chain of 22:6, 22:5, and 20:4 in Figure 5I. Most of GP with chain 22:6, 22:5, and 20:4 was higher in NCD group and HBE group, especially chain 22:6. Additionally, volcano plots also demonstrated that most of the lipid species with chain 22:6 and 22:5 were increased in response to betaine (Figures 5A,B). Our study suggests that betaine have a beneficial effect on adipose tissue by regulating lipid composition and promoting lipolysis.

To further explore the underlying mechanisms of betaine in regulating lipid metabolism, RNA-seq-based transcriptomic analysis was conducted to identify potential molecular targets. Supplementary Figures 4A–C showed that the sequencing saturation was close to 1, indicating high-quality transcriptomic data. Analysis of differential expression (Figures 6A–C) revealed 1,133 DEGs between the HFD control and NCD groups, with 544 genes being upregulated and 589 genes downregulated. PCA analysis showed that betaine separated HBE group from HFD group, which accounted for 29.86 and 11.41% of total variation, respectively (Supplementary Figure 4D). The RNA-seq data further highlighted that betaine markedly altered the transcriptomic landscape in the adipose tissues of HFD-fed rats compared to betaine-treated rats. A total of 3,998 DEGs were identified in HBE group vs. HFD group (Figure 6B). These DEGs were predominantly enriched in pathways linked to lipid metabolism, including fatty acid degradation, glycerophospholipid metabolism, biosynthesis of unsaturated fatty acids, and MAPK signaling pathway, aligning with our histopathological and lipidomic analyses (Figure 6D). Furthermore, KEGG and GO pathway analysis indicated that DEGs in betaine-treated rats versus HFD-fed rats were also enriched in processes related to carbohydrate metabolism and thermogenesis, such as thyroid hormone, insulin, and AGE-RAGE signaling pathway in diabetic complications (Figure 6D; Supplementary Figure 4E). Collectively, these results implied that betaine modulated intra-adipose lipid profiles through extensive regulation of genes involved in energy metabolism.

Subsequently, our analysis concentrated on the notable alterations in gene expression, particularly those regulating important enzymes involved in lipid transport, glycerophospholipid metabolism, fatty acid elongation processes, fatty acid oxidation, and adipose tissue thermogenesis during HFD and betaine administration intervention. The major genes involved in lipid digestion and fatty acid transportation increased in the betaine-treated adipose tissues, including Lpl, a gene encoding a crucial hydrolytic enzyme that catalyzes the hydrolysis of triglycerides in lipoproteins into fatty acids and monoglycerides; Slc27a3, a gene encoding a protein with acetyl-coenzyme A (acyl-CoA) synthetase activity that catalyzes the first step of fatty acid beta oxidation by linking fatty acids to CoA to form fatty acyl-CoA esters. In addition, genes that were crucial for lipolysis and oxidation were upregulated, including Pnpla3, a gene encoding a key enzyme involved in lipolysis which catalyzes the release of fatty acids from stored triglycerides (TG) within adipocytes; Acsl1, a gene encoding a rate-limiting enzyme in lipid metabolism, specifically involved in the activation of long-chain fatty acids (LCFAs) for their subsequent utilization in cellular lipid synthesis or degradation via beta-oxidation. Moreover, betaine treatment also significantly increased the expression of phospholipid metabolism genes such as Pld1, Pld4, Plpp3, Plpp4, Lpcat3, thermogenic genes such as Pparg, Prdm16, Ndufaf1, and beige marker genes such as Dio2, Tmem26, Klhl13 and Sla27a3 (Figure 6E).

In order to explore the combined role of genes and metabolites in the defense of betaine against HFD, we used spearman correlation and O2PLS to integrate transcriptomes and lipidomic. The findings indicated that the overall expression trend of the transcriptome and metabolic group samples was the same (Figures 7A,B) and spearman correlation hierarchical clustering analysis can cluster genes and metabolites with similar expression patterns. All these indicate a well correlation between transcriptomics and metabolomics (Figure 7D). The jointly analysis of KEGG pathway showed that DEGs and differentially expressed metabolites were mainly involved in two pathways, including regulation of lipolysis in adipocytes and insulin resistance (Figure 7C; Supplementary Table 2). At the same time, three co-enriched pathways, thermogenesis, retrograde endocannabinoid signaling, and choline metabolism in cancer were significantly changed (Figure 7E). To further verify this observation, a comprehensive analysis was conducted by integrating significant differential metabolites (VIP > 1, p < 0.05) with DEGs (FC ≥ 1.0, p < 0.05) within iPath 3.0, aiming to further elucidate the metabolic pathways altered by betaine intervention (Figure 8). When comparing HFD to HBE, the primary enriched pathways were lipid metabolism and energy metabolism. Based on the combined transcriptome and metabolome analysis, it is hypothesized that betaine may mitigate obesity resulting from HFD through its modulation of energy production and lipid metabolic pathways.

To explore the effects of betaine on the levels of gastrointestinal hormones, the levels of serum glucagon-like peptide-1(GLP-1), cholecystokinin (CCK), motilin (MTL), gastrin (GAS) and peptide YY (PYY) were compared between the three groups. Compared with the NCD group, the HFD reduced the serum CCK, PYY and MTL levels, while the 8-week betaine treatment increased the serum GLP-1, CCK and PYY levels in the HBE group, and the differences in CCK and PYY levels between the HBE group and the HFD group were statistically significant as shown in Figure 9.

Spearman’s correlation analysis was employed to examine the relationships between obesity-related biochemical indexes, microbial genera, SCFA, gastrointestinal hormones, differential metabolites, and genes (Figure 10A; Supplementary Tables 3, 4). Moreover, the correlation network highlighted significant and strong associations among these six components (|R| > 0.5, p < 0.01, as shown in Figure 10B. Overall, several lipid species with chain 22:6, including TG(18:4/18:2/22:6), TG(20:5/18:2/22:6), DG(18:3/22:6), DG(20:4/22:6), OAHFA(22:6/22:5), FA(22:6), PE(18:2e/22:6), PE(18:1e/22:6), PE(16:1e/22:6), PC(18:1/22:6), and PC(16:1/22:6), displayed significantly positive correlations with HDL, PYY, CKK, Lactobacillus, Ruminococcus, Oscillospira, hexanoic acid, genes related to thermogenesis, and significantly negative correlation with body weight and TG. In short, correlation analysis results revealed that these genera, SCFA, hormones, genes, and lipid species with chain 22:6 could serve as potential biomarkers for obesity alleviation.