Cultivated Enterococcus faecium B6 from children with obesity promotes nonalcoholic fatty liver disease by the bioactive metabolite tyramine

ABSTRACT

Gut microbiota plays an essential role in nonalcoholic fatty liver disease (NAFLD). However, the contribution of individual bacterial strains and their metabolites to childhood NAFLD pathogenesis remains poorly understood. Herein, the critical bacteria in children with obesity accompanied by NAFLD were identified by microbiome analysis. Bacteria abundant in the NAFLD group were systematically assessed for their lipogenic effects. The underlying mechanisms and microbial-derived metabolites in NAFLD pathogenesis were investigated using multi-omics and LC-MS/MS analysis. The roles of the crucial metabolite in NAFLD were validated in vitro and in vivo as well as in an additional cohort. The results showed that Enterococcus spp. was enriched in children with obesity and NAFLD. The patient-derived Enterococcus faecium B6 (E. faecium B6) significantly contributed to NAFLD symptoms in mice. E. faecium B6 produced a crucial bioactive metabolite, tyramine, which probably activated PPAR-γ, leading to lipid accumulation, inflammation, and fibrosis in the liver. Moreover, these findings were successfully validated in an additional cohort. This pioneering study elucidated the important functions of cultivated E. faecium B6 and its bioactive metabolite (tyramine) in exacerbating NAFLD. These findings advance the comprehensive understanding of NAFLD pathogenesis and provide new insights for the development of microbe/metabolite-based therapeutic strategies.

Graphical abstract

KEYWORDS:

• Gut microbiota
• Enterococcus
• non-alcoholic fatty liver disease
• metabolites
• tyramine

Introduction

The global obesity pandemic has led to nonalcoholic fatty liver disease (NAFLD) becoming the most prevalent chronic liver disease worldwide.^1,2 NAFLD encompasses a spectrum of phenotypic manifestations ranging from simple steatosis to nonalcoholic steatohepatitis, liver fibrosis, cirrhosis, or hepatocellular carcinoma.³ Alarmingly, there has been a rising incidence of NAFLD in younger individuals, with approximately 2.6–11.3% of children and 40–70% of children with obesity worldwide suffering from NAFLD.⁴ The increasing prevalence of childhood NAFLD foreshadows a significant burden on social care and healthcare systems,^5,6 making it a pressing public health concern that demands greater attention.

The pathogenesis of NAFLD is now recognized to be more complex than the traditional “two-hit hypothesis”, and the precise mechanisms underlying NAFLD remain largely unknown.^7–9 Accumulating evidence suggests that gut microbiota plays a crucial role in the pathogenesis of childhood NAFLD, although the data are not always consistent.^10–14 Notably, obesity may be a confounding factor that hinders the real relationship between gut microbiota and NAFLD, since it is both related to NAFLD and gut microbiota.^15,16 However, there is a lack of published research focusing on the gut microbiota in children with obesity, accompanied with or without NAFLD, which hinders our understanding of the actual role of the gut microbiota in this population.¹⁷ Moreover, the taxonomic and functional resolution of gut microbiota has been limited to the 16S ribosomal RNA (16S rRNA) gene sequencing. Most of these studies relied on observational data without biological validation, thereby precluding causal inference. The contribution of individual bacterial strains to NAFLD pathogenesis remains largely unknown.

Advancing from correlation to causation is essential in elucidating the interaction between the gut microbiota and the host. Typically, the gut microbiota interacts with host metabolism by producing bioactive metabolites. Recent studies have highlighted the role of gut microbiota in exacerbating the progression of fatty liver disease in adults by generating endogenous alcohol.^13,18 Moreover, alterations in gut microbiota have been associated with changes in circulatory metabolite levels, including trimethylamine-N-oxide, phenylacetic acid, 3-(4-hydroxyphenyl) lactate, branched-chain amino acids, aromatic amino acids, bile acids, and short-chain fatty acids, which were closely associated with metabolic disease.^19–23 Some gut microbiota could generate aromatic amines from aromatic amino acids using aromatic acid decarboxylase enzymes.^24,25 For example, tyramine is derived from tyrosine through the action of tyrosine decarboxylase found in several microbial taxa.^26,27 The genes encoding tyrosine decarboxylase, such as tyrDC, tyrP, mfnA, adcA, are typically organized in an operon.^28–30 Recent studies suggested that tyramine concentrations were significantly correlated with biomarkers of cardiometabolic risk factors and the pro-inflammatory state of metabolic syndrome.³¹ In addition, a preclinical study indicated that tyramine might induce liver damage and adversely affect life conservation.³² However, the causal relationship between specific species/strains and their derived metabolites in NAFLD pathogenesis has not been fully elucidated, especially in children.

In this study, we aimed to investigate the roles of gut microbiota and derived metabolites in NAFLD development among children with obesity. Firstly, we identified and cultivated key taxa associated with NAFLD development. Then, in vitro and in vivo experiments were performed to elucidate the pathogenic effects and underlying mechanisms linking the natural Enterococcus faecium B6 (E. faecium B6) to NAFLD. Furthermore, the crucial bioactive metabolite of E. faecium B6 was identified, and its pathogenic mechanisms in NAFLD progression were investigated in vitro and in vivo. Furthermore, the results were validated in an independent cohort. Hence, we identified a novel gut commensal strain and its bioactive metabolite contributing to NAFLD, as well as elucidating the underlying pathogenic mechanisms of these compositional factors involved in the disease. These findings present critical evidence for host-microbe interactions and provide promising intervention targets for addressing NAFLD in children.

Results

Microbiome signatures of children with obesity accompanied by NAFLD

A total of 156 children with obesity aged 6–18 years were enrolled as the discovery cohort, including 78 children with obesity accompanied by NAFLD (the NAFLD group) and 78 children with simple obesity (the control group). No significant differences were observed between the two groups, except for several NAFLD-associated clinical phenomics parameters (Table 1). The sequencing depth and quality control of each sample met the expected standards, providing high-quality data for microbiome analysis (Figure S1). The total number of operational taxonomic units (OTUs) was significantly decreased in the NAFLD group compared to the control group (Figure S1). A significantly lower microbiota α-diversity (observed, chao1, shannon, and ACE diversity indexes) was observed in the NAFLD group than that of the control group (Figure 1a). The weighted UniFrac distances in principal coordinate analysis (PCoA) results showed a significant compositional difference in the microbiota β-diversity between the two groups (Figure 1b).

Figure 1. Microbial alterations in the control and NAFLD groups. (a) Alpha diversity analysis between the two groups. (b) Beta-diversity analysis showing different taxonomic compositions. (c) LEfSe depicting taxonomic association at the genus level. (d) Comparison of the relative abundance for the main detected genera. (e) Correlation analysis and hierarchical clustering of genus abundance with NAFLD-clinical indicators. (f) Linear regression analysis of the Enterococcus abundance and NAFLD-clinical indicators. Control (n = 78), NAFLD (n=78), error bars represented the mean ± SD, *p < .05, **p < .01.

Table 1. Characteristics of the participants.

	Control (n=78)	NAFLD(n=78)	P value
Age, y	10.49 ± 1.97	10.49 ± 2.05	1.000
Males, n (%)	63 (80.7%)	63 (80.7%)	1.000
Weight (kg)	60.35 ± 9.46	62.06 ± 17.47	0.449
BMI (kg/m^Citation2	28.26 ± 3.15	27.88 ± 3.84	0.792
SBP (mmHg)	119.21 ± 11.96	121.28 ± 14.68	0.336
DBP (mmHg)	71.93 ± 8.89	72.15 ± 10.54	0.886
Neckline (cm)	33.96 ± 2.01	35.32 ± 3.23	0.002
Waistline (cm)	88.64 ± 6.41	91.36 ± 10.70	0.056
Hipline (cm)	96.06 ± 6.17	95.76 ± 10.54	0.831
Body fat (%)	39.95 ± 5.08	39.81 ± 4.63	0.858
Steatosis	Absent	Present
TG (mmol/L)	1.33 ± 0.94	1.67 ± 0.88	0.020
TC (mmol/L)	1.15 (0.81, 1.52)	1.43 (1.02, 2.13)	0.180
HDL-c (mmol/L)	1.20 ± 0.25	1.09 ± 0.21	0.004
LDL-c (mmol/L)	2.05 (1.81, 2.54)	2.39 (1.88, 2.85)	0.042
ALT (IU/L)	20.7 (16.70, 30.04)	(84.30, (61.3, 106.35)	<0.001
AST (IU/L)	21.22 (17.50, 26.93)	45.90 (31.58, 58.55)	<0.001
AST/ALT	0.97 (0.71, 1.18)	0.53 (0.47, 0.65)	<0.001
Total bile acid (μmol/L)	2.65 (1.50, 3.70)	2.85 (2.00, 4.85)	0.096
Total bilirubin (μmol/L)	10.18 ± 3.81	11.34 ± 4.64	0.089
Direct bilirubin (μmol/L)	3.08 ± 1.48	3.51 ± 1.44	0.071
Indirect bilirubin (μmol/L)	7.09 ± 2.80	7.83 ± 3.57	0.151
Total protein (g/L)	71.41 ± 4.22	71.35 ± 3.42	0.927
Albumin (g/L)	42.36 ± 3.08	42.23 ± 2.42	0.763
Globulin(g/L)	29.05 ± 2.77	29.13 ± 2.84	0.864
Albumin/Globulin	1.47 ± 0.18	1.46 ± 0.18	0.801
FBG (mmol/L)	4.51 ± 0.42	4.63 ± 1.10	0.352
HbA1c (g/dl)	0.77 (0.71, 0.74)	0.93 (0.79, 4.69)	<0.001
HbA1c/Hb	4.73 (4.44, 5.07)	4.57 (0.39, 4.98)	0.033
Insulin (μU/mL)	21.36 (15.19, 29.90)	29.06 (18.53, 40.57)	0.004
C-peptide (ng/mL)	3.00 (2.36, 3.78)	3.75 (2.64, 4.87)	0.001

The phylum level analysis detected a lower relative abundance of Firmicutes, and a higher abundance of Proteobacteria in the NAFLD group compared to the control group (Figure S2). The linear discriminant analysis effect size (LEfSe) analysis suggested that Enterococcus, Escherichia, Klebsiella, Dialister, and Enterobacter have higher abundance in the NAFLD group, while Faecalibacterium, Eubacterium_eligens_group, Roseburia, Fusicatenibacter, Clostridium, Coprococcus, and parasutterella have higher abundance in the control group (Figure 1c). The relative abundance of these genera consistently showed a significant difference between the NAFLD group and the control group (Figure 1d, S2). Moreover, the abundance of these genera, except for Eubacterium_eligens_group, was correlated with at least one of the typical clinical indicators of NAFLD (Figure 1e). Correlation analysis between the differential taxa and NAFLD clinical indicators also emphasized the significant association of Enterococcus with serum TC, TG, HDL-c, LDL-c, AST, and ALT levels (Figure 1e). More specifically, Enterococcus appeared to have linear correlations with the levels of serum ALT, AST, TG, and TC (Figure 1f). Therefore, we speculated that the Enterococcus genus might serve as a driver in the progression of NAFLD.

Data were expressed as mean ± standard deviation (SD) or median (interquartile range), and P Value was determined by t-test or mann-whitney U. BMI: body mass index; SBP, systolic blood pressure; DBP, diastolic blood pressure; TG: Triglycerides; TC: Total cholesterol; HDL-c, high-density lipoprotein cholesterol; LDL-c, low-density lipoprotein cholesterol; AST, aspartate aminotransferase; ALT, alanine aminotransferase; FPG, fasting plasma glucose; HbA1c, glycated hemoglobin; Hb, hemoglobin.

E. faecium B6 colonization promoted NAFLD development in mice

Following the clues illustrated by the microbiome discoveries from the case-control study, we further explored the specific strains and causal relationship of the Enterococcus genus in inducting NAFLD. We isolated a series of bacterial strains from children with obesity accompanied by NAFLD using culturomics, as depicted in Figure 2a. Briefly, we screened the lipogenic effects of each bacterial strain from these isolates belonging to the Enterococcus genus in vitro. We noted E. faecium B6 showed prominent lipid-increasing activities, which significantly promoted lipid accumulation in human hepatocellular carcinoma cell lines (HepG2) (Figure S3).

Figure 2. Colonization of E. faecium B6 promoted NAFLD symptoms in mice. (a) Schematic diagram of the isolation, identification, and screening of gut bacterial strain with lipogenic effects. (b) The procedure of mice experiments. (c) Changes in body weight among the NCD, NCD+B6, HFD, and HFD+B6 groups during 12 weeks. (d) Quantification of the liver index in mice. (e) Quantification of the steatosis score, inflammation score, ballooning score, and NAFLD activity score in mice. (f) Representative images of fixed liver sections with H&E, ORO, and masson staining, respectively. Magnification, × 200, the scale bar, 150 μm. (g-h) the concentrations of (g) serum and (h) hepatic ALT, AST, TG, TC, LDL-c, and HDL-c. N=8 per group, error bars represented the mean ± SD, *p < .05, **p < .01.

To ascertain the role of E. faecium B6 in NAFLD development under natural non-disease conditions, we conducted animal experiments as illustrated in Figure 2b. Briefly, mice were randomly assigned to four groups: the normal chow diet with saline gavage (the NCD group), the normal chow diet with E. faecium B6 gavage (the NCD+B6 group), the high-fat diet with saline gavage (the HFD group), and the high-fat diet with E. faecium B6 gavage (the HFD+B6 group), and the duration was 12 weeks. As expected, the E. faecium B6 was successfully colonized in mice (Figure S4). Body weights were significantly higher in the HFD group than in the NCD group, but no significant difference was observed between the NCD+B6 group and the NCD group, as well as between the HFD+B6 group and the HFD group (Figure 2c). E. faecium B6 seemed to have no effect on body weight, which was consistent with microbiome analysis that the association of Enterococcus and BMI was not statistically significant (Table 1, Figure 1e). The liver index, histological evaluations (steatosis, inflammation, and ballooning), and NAFLD activity score (NAS) were significantly increased in the normal chow diet with E. faecium B6 gavage mice (the NCD+B6 group) compared to the NCD group, and in the high-fat diet with E. faecium B6 gavage (the HFD+B6 group) compared to the HFD group (Figure 2d, e). Meanwhile, histopathological analysis using hematoxylin and eosin (H&E), oil red O (ORO), and masson staining showed that colonization of E. faecium B6 promoted hepatic steatosis, inflammation, and fibrosis, demonstrating experimental NAFLD symptoms (Figure 2f). Additionally, serum and hepatic ALT, AST, TG, TC, and LDL-c levels were significantly elevated in mice administered E. faecium B6 than those control mice, while the HDL-c level showed an opposite trend (Figure 2g, h). Together, these results demonstrated that colonization of E. faecium B6 promoted NAFLD development in mice.

E. faecium B6 colonization affected the hepatic transcriptomes

To elucidate the underlying mechanisms of E. faecium B6-induced NAFLD, we performed a transcriptomic analysis on the liver tissues of mice. A total of 209 differentially expressed genes (DEGs) were identified in the NCD+B6 group compared to the NCD group, with 125 genes up-regulated and 84 genes down-regulated (Figure 3a). Similarly, in the HFD+B6 group compared to the HFD group, 459 DEGs were identified, including 370 up-regulated and 89 down-regulated genes (Figure 3a). Hierarchical cluster analysis of DEGs showed roughly consistent patterns among the individual samples within each group (Figure 3b). When focusing on the specific DEGs involved in lipid metabolism, a clear pattern emerged. Genes related to hepatic synthetic, transportation, and excretory functions, such as Acsl1, Acsl4, Fasn, Fmo3, Ugt1a9, Plin4, Plin5, Cyp7a1, Cyp2b10, were significantly changed under the influence of E. faecium B6 (Figure 3c). The qPCR analysis successfully confirmed the expression of these genes in liver tissues (Figure 3d), thus confirming the reliability of the transcriptomic data. Based on the DEGs profiles, we then explored the gene ontology (GO) database to gain further insights into the biological processes associated with these DEGs (Figure S5). Furthermore, the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis identified several significantly altered signaling pathways (Figure 3e, f). E. faecium B6 colonization may induce transcriptional reprogramming in the liver. In particular, common alternations in known pathways related to lipid metabolism, inflammation, and fibrosis, such as PPAR signaling pathway, Chemokine signaling pathway, NF-kappa B signaling pathway, TGF-beta signaling pathway, Linoleic acid metabolism, might emerge as the major roles in this process.

Figure 3. Transcriptomic analysis of liver tissues from the NCD, NCD+B6, HFD, and HFD+B6 groups. (a) Volcano plot of the DEGs distribution. Each dot represented a DEG. (b) Clustering and heatmap analysis of the DEGs between different groups. Colored bars indicate standardized log₂ expression intensities (FPKM) of each sample in the NCD, NCD+B6, HFD, and HFD+B6 groups. Red and blue represent up-regulation and down-regulation, respectively. (c) Heatmap displaying the expression patterns of DEGs based on the transcriptomic data. Up-regulated or down-regulated genes are plotted in red or blue, respectively. Scale bars indicate a log₂ fold change of gene expression in the HFD group compared with the NCD group, the NCD+B6 group compared with the NCD group, and the HFD+B6 group compared with the HFD group. (d) Validation of selected DEGs identified from the transcriptomic analysis using qRT-PCR. (e) The bubble plot illustrated the enriched KEGG pathways for DEGs between the NCD+B6 and NCD groups. (f) The bubble plot illustrated the enriched KEGG pathways for DEGs between the HFD+B6 and HFD groups. The dot’s color represented the P_adj value, while the size of the dots represented the number of genes annotated to each KEGG term. N=8 per group, error bars show the mean ± SD, *p < .05, **p < .01. Scale bars, Z score log2 transformed intensity values.

E. faecium B6 played an impact on the PPAR signaling pathway

We next investigated the expression levels of genes involved in the PPAR signaling pathway based on the clues provided by transcriptomic analysis results. PPAR signaling pathway was mainly involved in lipid metabolism (KEGG map03320). Transcriptomic analysis suggested that the expression of genes encoding peroxisome proliferator-activated receptor γ (PPAR-γ) and the cluster of differentiation 36 (CD36) was significantly up-regulated in the E. faecium B6 colonization group compared to the control group, while the expression of gene encoding carnitine palmitoyltransferase 1A (CPT-1α) was significantly down-regulated. The expression of these genes was successfully validated by qRT-PCR results (Figure 4a). Furthermore, western blot analysis confirmed the up-regulation of PPAR-γ and its target CD36 and the down-regulation of CPT-1α at the protein level (Figure 4b, c). In addition, PPARs are a family of ligand-regulated nuclear receptors, including PPAR-α, PPAR-β, and PPAR-γ, which also function as transcription factors. We found that the Ppar-α and Ppar-β mRNA levels had no significant changes between the E. faecium B6 colonization group and the control group (Figure S6). Moreover, the nuclear receptor superfamily members involved in the regulation of metabolic enzymes, such as liver X receptor (LXR), farnesoid X receptor (FXR), aryl hydrocarbon receptor (AHR), and pregnane X receptor (PXR), which acted as the downstream target of CD36, were also examined (Figure S6). We noted that PPAR-γ and its interacting protein retinoid X receptor (RXR) exhibited the most significant alterations (Figure S6). Furthermore, IHC staining confirmed that liver PPAR-γ expression increased markedly in the E. faecium B6 colonization group compared to the control group (Figure 4d). These findings suggested that E. faecium B6 colonization probably affected the PPAR signaling pathway. The activation of PPAR-γ, instead of LXR, FXR, AHR, and PXR, was more likely to play a crucial impact on the pathogenic role of E. faecium B6 in NAFLD.

Figure 4. Expression analysis of mRNA and proteins related to the key signaling pathways. (a) The expression levels of key genes involved in the PPAR signaling pathway, TGF-beta signaling pathway, and inflammatory response using qRT-PCR analysis. (b) Western blot analyzed the expression of the key proteins. (c) Grey-scale quantitative analysis of western blotting. (d) Representative IHC staining of PPAR-γ, α-SMA, and IL-6 in the liver. Scale bar, 150 μm. (e) The concentrations of hepatic TNF-α, IL-6, and IL-1β based on ELISA analysis. (f) PPI network analysis of key genes related to NAFLD. N=8 per group, error bars represented the mean ± SD, *p < .05, **p < .01.

E. faecium B6 facilitated inflammation and fibrosis progression in liver

Meanwhile, E. faecium B6 colonization also affected the inflammatory response and the TGF-beta signaling pathway. The qPCR analyses confirmed the significant up-regulation of key genes related to the TGF-beta signaling pathway, including genes encoding smooth muscle actin (α-SMA) and transforming growth factor-beta (TGF-β), in the livers of mice administrated with E. faecium B6 compared to the control group (Figure 4a). The protein levels of α-SMA and TGF-β were increased in the liver of the E. faecium B6 colonization group compared to the control group, as demonstrated by western blot and IHC analysis (Figure 4b–d). The expression change of these fibrosis markers (α-SMA and TGF-β) was consistent with the masson staining results, which showed an increased collagen deposition in the liver (Figure 2f). Moreover, the expression of cytokines involved in the inflammatory response, including genes encoding tumor necrosis factor-alpha (TNF-α), interleukin-6 (IL-6), and interleukin-1β (IL-1β), were significantly up-regulated in the liver of mice administrated with E. faecium B6 compared to the control group (Figure 4a). The protein expression of TNF-α, IL-6, and IL-1β was aligned with the mRNA analysis results (Figure 4b–d). Moreover, the liver and serum concentrations of these secretory pro-inflammatory cytokines were significantly increased in the E. faecium B6 colonization group compared to the control group, as estimated by ELISA analysis (Figure 4e, S7). Subsequently, the PPI network analysis showed that PPAR-γ appeared to occupy a pivotal position in linking lipid metabolism to inflammation and fibrogenesis (Figure 4f). Collectively, E. faecium B6 might induce lipid accumulation, inflammation, and fibrogenesis in the liver, and PPAR-γ probably played a vital role in this process.

E. faecium B6 colonization induced specific metabolomic signatures

To further explore the metabolic mechanisms of E. faecium B6 contributed to NAFLD, we performed a widely untargeted metabolomics analysis of serum samples from mice using liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS). A total of 1380 metabolites were identified and categorized into 11 SuperClass categories (Figure S8). Furthermore, the orthogonal partial least-squares discriminant analysis (OPLS-DA) score plots separated the metabolite profiles among the NCD, NCD+B6, HFD, and HFD+B6 groups (Figure 5a). Also, the 7-fold cross-validation tests indicated that the OPLS-DA model was not overfitting and was reliable in screening differential metabolites (Table S1). Furthermore, we observed 30 significantly up-regulated metabolites and 85 significantly down-regulated metabolites in the NCD+B6 group compared to the NCD group, as well as 18 significantly up-regulated metabolites and 45 significantly down-regulated metabolites in the HFD+B6 group compared to the HFD group (Figure 5b). Moreover, pathway analysis showed that E. faecium B6 colonization significantly influenced the metabolic process of the host, especially impacting the amino acid metabolism, carbon metabolism, and fatty acid metabolism (Figure 5c). Furthermore, we found increased concentrations of metabolites such as tyramine, myristic acid, and bergaptol, while decreased concentrations of pentadecanoic acid, g-guanidinobutyrate, and isosafrole in the E. faecium B6 colonization groups compared to the control groups (Figure 5d, e, S9). Notably, serum tyramine exhibited the most significant change, with a higher concentration in the NCD+B6 group and the HFD+B6 group compared to their respective control groups (Figure 5e). Moreover, the targeted quantitative analysis further confirmed the presence and higher concentration of serum and hepatic tyramine in the E. faecium B6 colonization groups (Figure 5f). Therefore, tyramine was closely related to E. faecium B6, and it might be a crucial metabolite in the development of NAFLD.

Figure 5. Metabolomic analysis of the metabolite profiles in mice. (a) OPLS-DA score plots of serum metabolite profiles in the NCD, NCD+B6, HFD, and HFD+B6 groups. (b) Volcano plot illustrating the altered metabolites. Each dot represented a metabolite. (c) The bubble plot of KEGG pathway enrichment analysis. (d) Alterations of common differential metabolites in the NCD+B6 group compared to the NCD group, and HFD+B6 group compared to the HFD group. (e) Alterations of specific metabolite levels in serum. (f) The concentration of tyramine in the serum and liver using the targeted quantitative determination. N=8 per group, error bars show the mean ± SD, *p < .05, **p < .01.

Metabolites worked as messengers in NAFLD development

Microbiota-derived metabolites are vital mediators in host-commensal interactions. To better understand the bridging role of small molecules in the gut-liver axis and investigate the biological processes influenced by E. faecium B6 and its metabolites, we performed an integrative analysis of the metabolomics and transcriptomics data. As shown in Figure S10, tyramine, bergaptol, and isosafrole were the top 3 metabolites most closely associated with various genes, while Il-6, Il-1β, and Tnf-α were the top 3 genes most closely associated with various metabolites. Moreover, network association analysis revealed significantly positive correlations between tyramine and the mRNA level of Ppar-γ, Cd36, α-Sma, Tgf-β, Tnf-α, Il-6, and Il-1β, but negative correlation with the mRNA level of Cpt-1α (Figure S10A). Meanwhile, a dose-response relationship existed between tyramine level and Ppar-γ, Cd36, Tgf-β, Tnf-α, and Il-6 gene expression (Figure S10B). Furthermore, we observed a significant increase in serum TG levels, an important index for NAFLD diagnosis, in response to tyramine content (Figure S10C). Additionally, our findings confirmed a significant elevation in serum tyramine levels in response to the abundance of E. faecium B6 (Figure S10D). Tyramine was likely generated from gut microbial metabolism and probably served as a biomarker for NAFLD. Hence, the NAFLD symptoms induction by E. faecium B6 might be attributed to the effects of tyramine on lipid metabolism and inflammatory response.

Tyramine was an important bioactive metabolite of E. faecium B6

To further investigate if tyramine was derived from E. faecium B6, we performed a fermentation of E. faecium B6 on a preparative scale and successfully identified tyramine (m/z = 138.09176 in positive mode) based on the mass spectrometry analysis (Figure 6a, b). Subsequently, we confirmed that the concentration of tyramine was 100-fold higher in the cell-free supernatant (CFS) from the fermentation extracts compared to E. faecium B6 cells, and 30-fold higher compared to the sterility BHI medium, respectively (Figure 6c). Notably, tyramine (with a retention time of approximately 10.06 min) was not detected in the CFS of E. faecium B28 (Figure 6d). E. faecium B28 was an isolated strain that couldn’t induce NAFLD (Figure S11). To gain further insights into the molecular basis of tyramine production, we assembled the whole-genome sequences of E. faecium B6 and E. faecium B28, and annotated their functions, respectively. We then conducted a comparative genome analysis analysis to identify the specific role of E. faecium B6. The unique genomes of strain B6 were involved in carbohydrate metabolism, membrane transport, and lipid metabolism, implying the specific cellular activities of E. faecium B6 (Figure S12). Moreover, we noted a gene encoding tyrosine decarboxylase (EC:4.1.1.25, mfnA) involved in the tyramine-forming reaction (R00736: C00082 → C00483 + C00011) existed in the genome of E. faecium B6. Additionally, we successfully detected the presence of the mfnA gene in the genome of E. faecium B6, while it was absent in the genome of E. faecium B28 (Figure 6e). Furthermore, we quantitatively measured the abundance of the mfnA gene using qPCR in fecal samples of mice, which indicated that the abundance of the mfnA gene was higher in the E. faecium B6 administration group than in the control group (Figure 6f). Consistent with this, the concentration of fecal tyramine was significantly increased in the E. faecium B6 colonization group compared to the control group (Figure 6g). Importantly, the serum and hepatic tyramine levels were also higher in the E. faecium B6 colonization group than in the control group (Figure 5e, f). Meanwhile, the levels of mfnA gene and tyramine were higher in the HFD group than in the NCD group, suggesting that HFD was likely to induce the overgrowth of E. faecium B6 in gut microbiota (Figure 6f, g, S4). These findings suggested that E. faecium B6 might possess some specific functions that were responsible for tyramine production.

Figure 6. E. faecium B6-derived tyramine induced lipid accumulation and activated PPAR-γ. (a) Preparation of the metabolites. (b) Mass spectrogram of the identified tyramine in E. faecium B6 fermentation. (c) Relative concentrations of tyramine in the sterile BHI medium (the BHI group), E. faecium B6 cells (the B6 group), and cell-free supernatant of the E. faecium B6 fermentation (the SB6 group). (d) Ultra-high performance liquid chromatography system (UHPLC) analysis of the tyramine from E. faecium B6 and E. faecium B28. (e) Detection of mfnA gene in the genome of E. faecium B6 and E. faecium B28. (f-g) quantitative measurement of the levels of (f) mfnA and (g) tyramine in fecal samples of mice. (h-i) effects of tyramine (0, 5, 25, 50 μM) on intracellular lipid metabolism and liver function indexes in the (h) hepatocellular carcinoma cell lines (HepG2) and (I) human normal hepatocyte cell lines (THLE3). (j-k) effects of tyramine (0, 5, 25, 50 μM) on the expression of PPAR-γ and its target genes in the (j) HepG2 and (k) THLE3 cell lines. (l-m) effects of tyramine (0, 5, 25, 50 μM) on the transcriptional activity of PPAR-γ in the (l) HepG2 cells and (m) THLE3 cells measured with PPAR-γ transcription factor assay kit. Error bars represent the mean ± SD. *p < .05, **p < .01, ***p < .001.

Tyramine up-regulated the transcription of PPAR-γ

Furthermore, we performed in vitro experiments on human hepatocellular carcinoma cell lines (HepG2) and human normal hepatocyte cell lines (THLE3) to investigate the important function and molecular mechanisms of tyramine in NAFLD development. Tyramine didn’t cause severe cytotoxicity at concentrations lower than 100 μM, with an IC₅₀ of 211.8 and 131.3 μM toward HepG2 and THLE3 cell lines, respectively (Figure S13). Then 0 µM, 5 µM, 25 µM, and 50 µM of tyramine were further applied to explore their roles. Tyramine significantly increased the intracellular levels of TG, TC, LDL-c, AST, and ALT, while decreasing the concentration of HDL-c in a dose-dependent manner (Figure 7h,i). To evaluate the transcriptional activity of PPAR-γ, we measured the expression levels of PPAR-γ and its target genes. Tyramine significantly regulated the expression of PPAR-γ and its target genes involved in lipid metabolism (CD36, CPT-1α), inflammatory response (TNF-α, IL-6, IL-1β), and fibrosis (α-SMA, TGF-β) (Figure 7j,k). Moreover, the transcriptional activity of PPAR-γ was confirmed with the PPAR-γ transcriptional activity kit, which suggested that tyramine dose-dependently increased the transcriptional activity of PPAR-γ in cells (Figure L, M). In addition, molecular docking analysis revealed that tyramine bound to the active binding pocket of human PPAR-γ, and many residues of PPAR-γ were involved in the process (Figure S14). These findings confirmed that tyramine could effectively activate PPAR-γ, which potentially affected lipid metabolism, inflammatory response, and fibrosis in the liver, thus promoting NAFLD development.

Figure 7. E. faecium B6-derived bioactive tyramine promoted NAFLD symptoms in mice. (a) Schematic diagram of the tyramine-treatment mice experiments. (b) Changes in body weight among the NCD, NCD+Tyr, HFD, and HFD+Tyr groups during. (c) Quantification of the liver index in mice. (d) Quantification of the steatosis score, inflammation score, ballooning score, and NAFLD activity score in mice. (e) Representative images of fixed liver sections with H&E, ORO, and masson staining, respectively. Magnification, × 200, the scale bar, 150 μm. (f) The concentrations of hepatic ALT, AST, TG, TC, LDL-c, and HDL-c. (g-i) the relative mRNA levels of Ppar-γ and its targeted genes involved in (g) lipid metabolism (Ppar-γ, Cd36, cpt-1α), (h) inflammatory response (tnf-α, il-6, il-1β), and (i) fibrosis (α-sma, tgf-β). (j) The transcriptional activity of PPAR-γ in mice livers was measured with the PPAR-γ transcription factor assay kit. (k-m) the concentrations of hepatic PPAR-γ and proteins involved in (k) lipid metabolism (PPAR-γ, CD36, CPT-1α), (l) inflammatory response (TNF-α, IL-6, IL-1β), and (m) fibrosis (α-SMA, TGF-β) based on ELISA analysis.

Tyramine effectively promoted NAFLD in mice

We further assessed the lipogenic effect of tyramine in NAFLD development through mice experiments (Figure 7a). Consistent with the E. faecium B6 colonization, the body weights were significantly higher in the HFD group than in the NCD group, but showed no significant change among the tyramine-treatment and control groups (Figure 7b). E. faecium B6 and tyramine might not influence the weight of mice. The liver index, steatosis score, inflammation score, and NAFLD activity score were significantly increased in the NCD+Tyr group compared to the NCD group, and in the HFD+Tyr group compared to the HFD group (Figure 7c,d). Histopathological evaluations clearly showed that the degrees of lipid degeneration, inflammation, and fibrosis were significantly aggravated upon tyramine treatment (Figure 7e). Additionally, hepatic ALT, AST, TG, TC, and LDL-c levels were increased, while the HDL-c level was decreased in the tyramine-treatment group than the control group (Figure 7f). All these results indicated that tyramine could effectively facilitate NAFLD development in mice. In addition, the expression of crucial genes related to the PPAR signaling pathway, TGF-beta signaling pathway, and inflammatory response were evaluated. Tyramine remarkably up-regulated the relative mRNA levels of Ppar-γ and it targeted genes involved in lipid metabolism (Cd36, Cpt-1α), inflammatory response (Tnf-α, Il-6, Il-1β), and fibrosis (α-Sma, Tgf-β) (Figure 7g–i). Moreover, we confirmed the transcriptional activity of PPAR-γ with the PPAR-γ transcriptional activity kit, which showed that tyramine promoted the transcriptional activity of PPAR-γ in mice livers (Figure 7j). Consistently, the expressions of PPAR-γ, CD36, CPT-1α, TNF-α, IL-6, IL-1β, α-SMA, and TGF-β protein levels from mice livers were successfully determined by ELISA analysis (Figure 7k–m). Collectively, E. faecium B6-derived the bioactive metabolite, tyramine, could promote lipid accumulation, inflammation, and fibrogenesis in the liver. The stimulation of PPAR-γ might play a crucial impact on the process and finally contribute to NAFLD development.

Validation of E. faecium B6 and tyramine levels in an additional independent cohort

The above results showed that E. faecium B6 and its bioactive metabolite tyramine were important in NAFLD development. We further corroborated our findings in an additional independent validation cohort consisting of 123 children with obesity accompanied by NAFLD and 123 controls. The characteristics of these 246 subjects are shown in Table S2. The levels of E. faecium B6, tyramine-producing gene (mfnA), and tyramine were higher in the NAFLD groups than in the control groups (Figure 8a). Moreover, the concentrations of inflammatory factors (TNF-α, IL-6, IL-1β) were significantly higher in the NAFLD groups than in the control groups (Figure 8b). Furthermore, the correlation analysis showed that the levels of E. faecium B6, mfnA gene, and tyramine were positively correlated to fatty liver indexes (including TG, TC, LDL, AST, and ALT levels) and the inflammatory cytokines (including TNF-α, IL-6, IL-1β), but negatively correlated to HDL-c level (Figure 8c). These results were consistent with observations in the discovery cohort, cell experiments, as well as the two animal experiments, further inferring the involvement of E. faecium B6 and derived tyramine in NAFLD development.

Figure 8. The association of E. faecium B6 and its bioactive metabolite tyramine with NAFLD in the validation cohort. (a) The levels of E. faecium B6, tyramine-producing gene (mfnA), and tyramine. (b) The concentration of inflammatory factors (TNF-α, IL-6, IL-1β). (c) The correlation analysis of the E. faecium B6, mfnA, tyramine, inflammatory cytokines, and NAFLD-clinical indicators. Control (n = 123), NAFLD (n=123), error bars represented the mean ± SD, *p < .05, **p < .01, ***p < .001.

Discussion

In recent years, major advances have been made in understanding the association between gut dysbiosis and NAFLD. However, the mechanistic role of the individual gut bacterial strain and its metabolite in triggering NAFLD, especially in children, remains poorly understood. In this study, we observed a higher abundance of the Enterococcus genus in fecal samples of children in the NAFLD group than in the control group. We further identified a specific strain E. faecium B6, capable of inducing NAFLD-related symptoms in mice, providing the first evidence of its causal role in NAFLD development. Moreover, E. faecium B6 might contribute to NAFLD by producing the bioactive metabolite tyramine. Tyramine probably interacted with PPAR-γ, leading to lipid metabolism disorder, inflammatory injury, and fibrogenesis in the liver. Furthermore, the findings were validated in an additional cohort. This pioneering study sheds light on the causal roles and molecular mechanisms underlying the increased abundance of E. faecium B6 in NAFLD development.

Gut microbiome analysis in this study showed that the relative abundance of the Enterococcus (p_Firmicutes; f_Enterococcaceae; g_Enterococcus) was significantly up-regulated in children with obesity accompanied by NAFLD, suggesting a positive association between Enterococcus and NAFLD phenotype. However, it is worth noting that previous reports on the association between Enterococcus and NAFLD development have yielded inconsistent results.^33–36 These discrepancies can be attributed, at least in part, to factors such as age, sex, geography, race/ethnicity, genetics, local dietary habits, and lifestyle.^37–39 Besides, the evidence linking gut microbiota and childhood NAFLD is limited, and often characterized by small sample sizes and a lack of well-characterized participants.^40–42 Notably, obesity is both related to NAFLD and gut dysbiosis, making it challenging to determine whether altered microbial species or obesity itself promotes the exacerbation of NAFLD.^43,44 To address this issue, our study focused specifically on investigating the fecal microbial profiles of children with obesity accompanied with or without NAFLD. By ensuring that participants had comparable characteristics, we minimized confounding factors and attained precise identification of specific taxa associated with NAFLD symptoms.

To further explore the association and causality between Enterococcus and NAFLD, we conducted a batch screening of Enterococcus species/strains with lipogenic effects according to microbiome analysis. We found E. faecium B6 significantly contributed to lipid accumulation in cells and mice, consistent with the characteristic feature of NAFLD.^45,46 Recently studies have identified several bacteria capable of inducing NAFLD in mice/rats, including Klebsiella pneumonia W14 and TH1, Enterobacter cloacae B29, Escherichia fergusonii strain ATCC 35,469, Bilophila wadsworthia ATCC 49,260, and Porphyromonas gingivalis.^18,47–50 However, few strains have been precisely isolated from humans with NAFLD, especially in children. Moreover, we found the lipogenic effects were specific to the E. faecium B6 strain, since other Enterococcus species isolated from the NAFLD group did not exhibit the same effects. It has been recognized that even strains of the same genus or species might exhibit different impacts and mechanisms on the liver.⁵¹ For instance, some strains belonging to Enterococcus have been shown to significantly reduce obesity, improve hepatic steatosis, and alleviate liver inflammation,^33,52,53 but some strains were regarded as opportunistic pathogens contributing to liver damage and fibrosis.^54–56 The heterogeneity of these results suggested that the metabolic capabilities of strains within the Enterococcus genus largely determined their roles. Thus, studies on the role of gut microbiota in disease should narrow their focus to the strain level. This study is the first to investigate the roles of E. faecium B6 isolated from children with obesity accompanied by NAFLD in liver disease.

To gain further insights into the molecular mechanisms underlying E. faecium B6-induced NAFLD, we integratively analyzed the transcriptomic and metabolomic data. We recognized the involvement of hepatic PPAR-γ and its target genes in the E. faecium B6-induced NAFLD. PPAR-γ, a vital ligand-activated receptor in the PPAR signaling pathway, emerges as a critical player in metabolism and hepatic steatosis.^57,58 The expression of PPAR-γ and its downstream genes are closely related to the progression of NAFLD. In both human and experimental animal models of NAFLD, PPAR-γ was reported to exhibit increased expression in the hepatocytes, causing hepatocyte steatosis and activating hepatic stellate cells.^59–61 Moreover, CD36 and CPT-1α were crucial targets of PPAR-γ that mediate lipid metabolism, and abnormal expression of them might result in hepatic steatosis.^62,63 In addition to its powerful lipogenic effect, PPAR-γ had a crucial role in exacerbating inflammation and fibrosis.⁶⁴ We demonstrated that the genes and proteins related to pro-inflammation and fibrosis were significantly up-regulated by E. faecium B6. To our knowledge, inflammation was a key driver of NAFLD progression and fibrosis development.^8,65 Inflammatory mechanisms were involved along the entire spectrum of NAFLD but particularly at more advanced disease stages, including fibrosis and cirrhosis development.^66,67 In addition, the activation of PPAR-γ might also impact the progression of inflammation and fibrosis.⁵⁸ These might emphasize the pathogenic functions of E. faecium B6 and PPAR-γ in inflammation and fibrosis, which further aggravated NAFLD. Conversely, previous research indicated that heat-killed Enterococcus faecalis EF-2001 administration might prevent lipid accumulation by inhibiting adipose PPAR-γ in the Sprague-Dawley rats.⁶⁸ The inconsistent results might be due to the tissue specificity, modeling method, and the different genetic backgrounds of animals, as well as the specific function of bacterial strains.^69,70 The important roles of Enterococcus species in PPAR-γ of mice and other models have not been explored, which could be investigated more in-depth in the future. In addition, we noted that the extent of pathological manifestations was more pronounced in the HFD+B6 group than in the NCD+B6 group, indicating that E. faecium B6 might facilitate the occurrence of NAFLD in the context of an unhealthy diet. Future studies should explore the interactions among diet, gut microbiota, and NAFLD to fully unravel the complexity of disease development.

Moreover, this study identified E. faecium B6 as a producer of tyramine, and we first discovered that tyramine might interact with PPAR-γ, contributing to hepatic steatosis, inflammation, and fibrosis. Notably, the tyramine concentration was higher in the cell-free suspension, which indicated that the tyramine was probably released to the intestines from E. faecium B6. Generally, gut microbiota may regulate metabolism in the host through the gut-liver axis, and microbial metabolites are crucial factors in the process.⁷¹ We speculated that E. faecium B6 and its tyramine might affect liver metabolism through the gut-liver axis. Gut microbiota has the potential to generate a wide array of metabolites, while only a limited number of these metabolites have been exacted from microbiota and linked to disease pathogenesis.^18,72,73 Our findings added to the growing body of evidence highlighting the significant role of microbial metabolites as key mediators that interacted with host physiology and contributed to disease development. Tyramine, a biogenic amine primarily produced by decarboxylase-producing spoilage microorganisms, has been previously associated with the regulation of blood pressure.^74,75 Likewise, we noted that the blood pressure was higher in the NAFLD group than in the control group (Table 1, S2). Additionally, we validated that the concentration of tyramine was higher in the NAFLD group than in the control group. Consistently, Gu et al., 2022 also suggested a significant increase of microbial-related tyramine in the cecal samples of NAFLD mice based on the metabolomic analysis, but the underlying mechanisms have not been extensively studied.⁷⁶ Notably, although the tyramine was easily oxidized by monoamine oxidase (MAO), the increased production of tyramine and impaired degradation by MAO probably resulted in excessive tyramine and hepatic cirrhosis.^77,78 Future studies should explore the metabolic kinetics of tyramine in the livers, especially in the context of abnormal metabolic states such as obesity. In short, tyramine and PPAR-γ activation provided a potential mechanistic explanation for the disruption of liver metabolism associated with E. faecium B6 colonization.

The relationship between gut microbiota and childhood NAFLD is complex and has not been fully characterized. Most previous studies have focused on the association between gut microbiota and NAFLD in adults. Evidence thoroughly supporting the hypothesis that specific patient-derived strains contribute to NAFLD has been limited. In this work, we systematically explored the functions of specific microbes in childhood NAFLD development from multiple aspects, including the discovery and validated cohorts, E. faecium B6 and tyramine treatment animal experiments, different hepatocyte experiments, as well as multi-omic analysis and molecular docking. We identified E. faecium B6 as a novel pathogenic microbe that probably promoted NAFLD development, representing the first report of its kind. Moreover, we first demonstrated that E. faecium B6 probably triggered NAFLD via tyramine, which led to the activation of the PPAR-γ. Furthermore, similar findings were validated in a validation cohort. This study comprehensively uncovered a causal link between gut E. faecium B6 and NAFLD, and provided a critical basis for the underlying mechanisms of gut microbiota in NAFLD. There were several limitations in this work. We found that there were some bacteria decreased in the NAFLD group and negatively correlated with NAFLD parameters, such as Faecalibacterium, Roseburia, and Coprococcus, which were consistent with previous reports in adults.^79–81 The beneficial bacteria that decreased in children with NAFLD might also contribute to the pathogenesis of NAFLD, and the roles of these bacteria in NAFLD should be explored in the future. Moreover, the role of E. faecium B6 and tyramine in NAFLD development could be evaluated with additional animal models, allowing for a more comprehensive understanding of their effects. We only investigate the roles of bioactive tyramine in the present study, the impacts of other metabolites and factors on such disease warrant systematic investigations. In addition, future longitudinal and longer-term follow-up studies are required to strengthen our claims and provide a more comprehensive understanding of the disease.

Conclusions

In summary, we identified E. faecium B6 as a potentially novel pathogenic strain promoting NAFLD development in children. E. faecium B6-derived bioactive metabolite, tyramine, could play an impact on hepatic steatosis, inflammation, and fibrosis progression, which promote NAFLD development. These results suggested a causal association of E. faecium B6 and tyramine with NAFLD. This study highlighted the host-microbe interactions and provided new insights for gut microbiota management in NAFLD or related metabolic disorders.

Materials and methods

Human studies

Children with obesity aged 6–18 years were recruited in Hunan Children’s Hospital (Hunan, China), from July 2019 to October 2020 (discovery cohort) and January 2021 to April 2022 (validation cohort). The diagnostic criteria for NAFLD were evaluated according to the expert consensus on the diagnosis and treatment of NAFLD in Children (2018) issued by the Endocrine Genetics and Metabolism Group, Pediatrics Branch of the Chinese Medical Association.⁸² The details of inclusion and exclusion criteria were described in Supplementary materials. Children with obesity meeting the criteria for NAFLD were categorized into the NAFLD group, and children without NAFLD matched by age, sex, and BMI were set as the control group. The studies were approved by the Xiangya School of Public Health Central South University Ethics Research Committee (XYGW-2018-04, XYGW-2021-19) and the Hunan Children’s Hospital Ethics Research Committee (HCHLL-2019-12). All procedures were undertaken in accordance with the ethical standards of the Helsinki Declaration. Written informed consents were obtained from all participants or their legal guardians/next of kin. The details of anthropometric and demographic measurements, sample collection, and clinical laboratory measurements were described in Supplementary materials. In total, 156 children with obesity were enrolled in the discovery cohort for microbiome analysis, and 246 children with obesity were enrolled in the independent validation cohort. Rigorous quality control procedures resulted in final analyses of the discovery and validation cohorts.

Bacteria culture

Bacterial strains were isolated and cultured from fecal samples of children with obesity accompanied by NAFLD using culturomic analysis, as previously reported.⁸³ In brief, the fresh fecal samples were suspended and homogenized in a sterile phosphate-buffered solution (PBS, pH 7.0), and then the mixtures were diluted and seeded in a series of agar plates with different mediums for anaerobic culturing. A single colony was picked out, and the purified isolates were identified by sequencing the 16S rRNA genes using the 27F/1492 R primers. The isolates classified as Enterococcus spp. (such as E. faecium, E. faecalis, E. durans, E. lactis, and E. huaxiensis) were selected for further screening of their lipogenic effects, and we found E. faecium B6 showed great potential in promoting lipid accumulation in cells. Unless otherwise stated, Enterococcus strains were cultured in Brian Heart Infusion medium (BHI, Oxoid, USA) for 24 h, within an anaerobic station (Longyue LAI-3T, Shanghai, China) maintained under an atmosphere consisting of 5% hydrogen, 10% carbon dioxide, and 85% nitrogen. Furthermore, whole genome sequencing and comparative genomic analysis were performed to identify the strains and their functions, and the detailed methods are shown in the Supplementary materials.

Animal experiments

The animal experiments were approved by the Laboratory Animal Welfare and Ethics Committee and adhered to the Animal Ethics Statement of Central South University (XMSB-2022–0027). Healthy specific-pathogen-free (SPF) C57BL/6J mice were purchased from Hunan SJA Laboratory Animal Co. Ltd. (Changsha, China). Male mice aged 3–4 weeks at the beginning of the study were housed in an SPF environment with controlled conditions (12-h light/12-h dark cycle at 24 ± 2°C and 40–60% humidity). Throughout the study, mice were given ad libitum access to food and filtered water.

To investigate the functions of gut bacterial E. faecium B6 in vivo, we randomly assigned mice to four groups: the normal chow diet with saline gavage (the NCD group), the normal chow diet with E. faecium B6 gavage (the NCD+B6 group), the high-fat diet with saline gavage (the HFD group), and the high-fat diet with E. faecium B6 gavage (the HFD+B6 group). The high-fat diet contained 60% kcal from fat, 20% kcal from carbohydrates, and 20% kcal from proteins. In the NCD+B6 and HFD+B6 groups, mice were orally administered with 200 μL of saline containing 3 × 10⁹ CFU/mL of E. faecium B6 every two days by gavage. Correspondingly, mice in the NCD and HFD groups received 200 μL of sterile saline. Body weight and food intake were monitored weekly. At the end of the 12 weeks, all mice were fasted overnight and euthanized, The fecal samples, serum samples, and liver tissues were collected for further analysis.

To assess the lipogenic effect of tyramine in vivo, male C57BL/6J mice were randomly divided into four groups: the normal chow diet with saline gavage (the NCD group), the normal chow diet with tyramine gavage (the NCD+Tyr group), the high-fat diet with saline gavage (the HFD group), and the high-fat diet with tyramine gavage (the HFD+Tyr group). The high-fat diet contained 60% kcal from fat, 20% kcal from carbohydrates, and 20% kcal from proteins. In the NCD+Tyr group and HFD+Tyr group, mice were gavaged with tyramine (2 mg/kg, Macklin, Shanghai, China) every two days, while the NCD and HFD groups were given an equal volume of saline for 12 weeks. At the end of the experiment, mice were fasted overnight and euthanized. The serum samples and liver tissues were collected for further analysis.

A detailed description of the histopathology, biochemical assays, transcriptomic analysis, real-time quantitative polymerase chain reaction (qRT-PCR), western blotting, immunohistochemistry (IHC) examination, metabolomic analysis, liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS) analysis, PPAR-γ transcriptional activity analysis, enzyme-linked immunosorbent assay (ELISA) analysis, and molecular docking analysis is provided in the Supplemental materials.

Cell culture and treatments

Hepatocellular carcinoma cell lines (HepG2) and normal liver cell lines (THLE3) were cultured as our previous report.⁸⁴ For lipid accumulation assay of candidate bacteria, cells were seeded in 96-well or 6-well plates containing 100 µL or 2 mL cell growth medium, respectively. When the confluence reached 80%, the medium was replaced with the 90% fresh cell growth medium supplemented with 10% spent culture broth filtrate or BHI medium. After incubation for 24 h, lipid accumulation was evaluated by oil red O staining (Solarbio, Beijing, China) or TG/TC quantification kit (Jiancheng, Nanjing, China) as described previously.⁸³ For the microbial metabolites (tyramine) treatment experiments, cells were seeded in well plates until the confluence reached 80%, and different concentrations of tyramine were treated. The cell viability of tyramine-treatment cells was examined by a CCK8 assay kit (Vazyme, Nanjing, China). After that, the cells were treated with different concentrations of tyramine (0, 5, 25, 50 µM) for 24 h, then the cells were collected for gene and protein expression determination. All experimental conditions were performed in three biological replicates.

Statistical analysis

The Shapiro-Wilk test was used to determine the distribution type of the clinical parameters in the subjects. Data were reported as mean ± SD or median (interquartile range). Statistical significance was determined by t-test or Mann-Whitney U test in SPSS software (version 24.0, IBM Corp, USA). All experiments were repeated at least three times in animal and cell experiments, and the data were presented as mean ± SD. Statistical significance was determined using one-way ANOVA in GraphPad Prism software (version 9.0, GraphPad Software Inc., San Diego, USA). A post hoc Tukey test was used to conduct multiple comparisons. A significance level of p < 0.05 was considered statistically significant. Multiple testing was corrected for multi-omics analysis using the Benjamini-Hochberg method to control the false discovery rate (FDR), and a P_adj <0.05 was considered statistically significant.

Author contributions

The research was designed by Jia Wei and Miyang Luo; The clinical information and samples from patients were collected by Jia Wei, Wen Dai, Xiongfeng Pan, Yamei Li, Yamei Duan, Xiang Xiao, Ping Ye, Yan Zhong, and Ningan Xu. The animal and cell experiments were performed by Jia Wei, Wen Dai, Yue Yang, Zhenzhen Yao, Yixu Liu, Zhihang Huang, and Jiajia Zhang. The multi-omics analysis was performed by Jia Wei, Wen Dai, and Xiongfeng Pan. The bacterial experiments were performed by Jia Wei and Zhihang Huang. Jiayou Luo, Fei Yang, Xiangling Feng, Ming Zeng, and Miyang Luo provided critical reagents and technical support, and assisted in revising the manuscript. Jia Wei and Miyang Luo wrote and revised the manuscript. All authors reviewed and approved the manuscript.

Ethics statement

The studies were approved by the Xiangya School of Public Health Central South University Ethics Research Committee (XYGW-2018-04, XYGW-2021-19) and the Hunan Children’s Hospital Ethics Research Committee (HCHLL-2019-12). Written informed consent was obtained from all participants or their legal guardians/next of kin. The animal experiments were approved by the Laboratory Animal Welfare and Ethics Committee and adhered to the Animal Ethics Statement of Central South University (XMSB-2022–0027).

Acknowledgments

We thank the Novogene Bioinformatics Technology Co., Ltd. (Beijing, China) for excellent technical assistance in the sequencing of the 16S rRNA gene, Applied Protein Technology Co. Ltd. (Shanghai, China) for excellent technical assistance in transcriptomic and untargeted metabolomics detection, and Personalbio Technology Co., Ltd. (Shanghai, China) for excellent technical assistance in the whole genome sequencing.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

All of the sequencing data that support the findings of the study are openly available in the National Center for Biotechnology Information (NCBI) BioProjects (https://www.ncbi.nlm.nih.gov/bioproject/) with the accession numbers PRJNA838778 and PRJNA1032872. Additional data that support the findings of this study are available on request from the corresponding author.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/19490976.2024.2351620

Additional information

Funding

This work was supported by grants from the National Natural Science Foundation of China (82304171), the Hunan Province Natural Science Foundation (2022JJ40668, 2021JJ30901), and the Fundamental Research Funds for the Central Universities of Central South University (2022ZZTS0846).