Prebiotic xylooligosaccharides ameliorate metabolic dysfunction-associated steatotic liver disease via modulating gut microbiota and metabolites

Abstract

The study investigated the dose-dependent protective effects of xylooligosaccharides (XOS) against metabolic dysfunction-associated steatotic liver disease (MASLD) in mice, focusing on gut microbiota and microbial metabolites. Male C57BL/6 mice were fed either a low-fat diet(LFD), a high-fat diet (HFD), HFD supplemented with low-dose XOS (0.38 g/kg), or HFD supplemented with high-dose XOS (1.0 g/kg) for 16 weeks. Hepatic histology, serum biochemistry, ileal morphology, gut microbiota composition (16 S rRNA gene sequencing), and fecal metabolites profiles were analyzed. High-dose XOS, but not low-dose, significantly reduced serum Alanine aminotransferase (ALT) and Triglycerides (TG) levels, alleviated hepatic steatosis, and improved the villus height–to–crypt depth ratio. It attenuated HFD-induced enrichment of Ruminococcaceae and Lachnospiraceae, while restoring the abundance of Bacteroidaceae, Bifidobacteriaceae and Bacillaceae. Metabolomic analysis revealed rebalancing of the tryptophan–indole and arachidonic-acid pathways, with reduced pro-inflammatory eicosanoids and kynurenine derivatives, and increased hepatoprotective indole derivatives and omega-3 polyunsaturated fatty acids. High-dose XOS reconstructs the gut microbiota-metabolite network toward an anti-inflammatory, hepatoprotective state in HFD-fed mice, underscoring the importance of dose optimization in XOS-based interventions for MASLD.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-48643-8.

Introduction

Metabolic dysfunction-associated steatotic liver disease (MASLD) is among the most prevalent chronic liver disorders worldwide. It comprises a continuum of hepatic conditions, ranging from simple steatosis to metabolic dysfunction-associated steatohepatitis (MASH), progressive fibrosis, cirrhosis and eventually hepatocellular carcinoma^1,2. Globally, the prevalence has been rising, reaching approximately 39.7% in men and 25.6% in women, and this upward trend imposes a considerable public-health and socioeconomic burden³.

Recent population-based analyses indicate that patients with MASLD exhibit increased all-cause mortality, with a 5.6-fold higher risk of liver-related death, a 1.30-fold higher risk of cardiovascular mortality, and a 1.17-fold higher risk of cancer-related mortality compared with individuals without steatotic liver disease⁴. Lifestyle modification, including healthy eating and regular physical activity, remains the cornerstone of MASLD management; however, maintaining long-term adherence to these behavioral changes is often challenging⁵. Despite the recent U.S. Food and Drug Administration (FDA) approval of Rezdiffra (resmetirom), a thyroid hormone receptor beta (THR-β) agonist, for a subset of MASH patients with moderate to advanced liver fibrosis (F2–F3 stages)⁶, substantial unmet therapeutic needs persist, particularly for individuals with early-stage disease or those ineligible for this drug. These limitations underscore the importance of further exploring alternative pharmacological and non-pharmacological strategies.

Notably, pioneering fecal microbiota transplantation (FMT) studies in gnotobiotic mice have demonstrated that microbial dysbiosis can serve as a causal driver of MASLD progression⁷. This has fueled increasing interest in microbiota-targeted interventions to modulate disease development. Among these, prebiotics such as xylooligosaccharides (XOS) have emerged as promising candidates, given their capacity to selectively stimulate beneficial gut bacteria, re-establish microbial homeostasis, and modulate host metabolic and inflammatory pathways. Consequently, investigating whether XOS supplementation can ameliorate HFD-induced MASLD through remodeling of gut microbiota composition and function represents a question of considerable scientific and therapeutic interest.

XOS constitute a prebiotic class that has attracted increasing attention in the functional food sector. They are typically derived from corncobs (Zea mays subsp. mays) through xylanase-mediated hydrolysis followed by purification, are resistant to digestion by human gastrointestinal enzymes, and are fermented by colonic microbiota⁸. Chemically, XOS consist primarily of xylose residues linked by β-1,4-glycosidic bonds, with a typical degree of polymerization ranging from two to seven; arabinose substitutions may occur, and such structural variations can confer distinct biological activities⁹. In terms of safety, the European Food Safety Authority Panel on Dietetic Products, Nutrition and Allergies has reported no evidence of genotoxicity, and available human studies likewise indicate no safety concerns for XOS consumption¹⁰.

Beyond their prebiotic properties, XOS derived from rice bran exhibited free-radical-scavenging activity against DPPH, hydroxyl (•OH), and ABTS•⁺ radicals. At a concentration of 1.0 g/L, the respective scavenging rates reached 65.76% (DPPH), 62.10% (•OH), and 97.70% (ABTS•⁺), values comparable to those of vitamin C. These preparations also enhanced the growth of the probiotic strains, including Lactobacillus plantarum, Lactobacillus reuteri, Lactobacillus acidophilus, and Lactobacillus rhamnosus¹¹. Similarly, XOS derived from rice husks and rice straw significantly enhanced the growth of lactic acid-producing bacteria such as Lactobacillus sakei and Lactobacillus brevis¹². In vivo, XOS supplementation increased the relative abundance of Bifidobacteria in the murine intestine, particularly in the ileum, accompanied by elevated short-chain fatty acids (SCFAs), especially propionate. These metabolites can enter systemic circulation and are associated with downregulation of low-grade inflammatory cytokines, including IL-1β and IFN-γ¹³. Consistent with these biological effects, XOS exhibit acid-stable and a favorable safety profile^14–16. In vitro, they are preferentially fermented by Bifidobacterium species¹⁷, and comparative studies indicate superior bifidogenic activity relative to other prebiotics, such as fructo-oligosaccharides (FOS)^18–20. Experimental studies further suggests that dietary XOS supplementation attenuates HFD-induced hepatic steatosis in rodents²¹. Nevertheless, important gaps remain: the extent to which XOS alleviate MASLD through targeted modulation of gut microbiota and their metabolites is not well defined, and optimal dosing parameters have not been established.

To address these gaps, the present study systematically characterized the dose-response relationships between XOS intervention and HFD-induced metabolic disturbances in murine liver, ileum, and serum. Particular emphasis was placed on alterations in gut microbiota composition and microbiota-derived metabolites, through an integrated multi-omics approach.

Materials and methods

Animals

Four-week-old male C57BL/6 mice were obtained from Jinan Ponyue Laboratory Animal Breeding Co., Ltd. Throughout the experiment, mice were single-housed in standard cages, with ad libitum access to water under a 12 h light-dark cycle, 22 ± 2 °C and 50 ± 5% relative humidity. After a 1-week adaptation period, the mice were randomly assigned to four dietary treatment groups (n = 10/group): (a) LFD, low-fat diet (10% of total calories derived from fat); (b) HFD, high-fat diet (60% of total calories derived from fat); (c) XOS.L, HFD plus low-dose XOS (daily oral gavage of 0.38 g/kg of XOS); and (d) XOS.H, HFD plus high-dose XOS (daily oral gavage of 1.0 g/kg of XOS). Daily food intake (g/day per mouse) was recorded by weighing the remaining chow every 24 h, and average values were calculated for each group.

XOS were donated by Shandong Longlive Biotechnology Co., Ltd. (Shandong, China). The product was extracted from corncob hemicellulose, with an approximate purity of 95%, consisting mainly of xylobiose, xylotriose, and xylotetraose (average degree of polymerization = 2–4). Composition and purity were verified by high-performance liquid chromatography (HPLC) and matched the specifications provided by the manufacturer (CAS No. 87099-0). All irradiated diets were obtained from Beijing Botai Hongda Biotechnology Co., Ltd. In humans, the safety daily dose of XOS ranges from 4.0 to10g (0.057–0.14 g/kg)¹⁰. The mouse dosage was calculated by converting human doses to mouse-equivalent doses based on body-surface-area normalization according to the U.S. Food and Drug Administration (FDA) guidelines²². Consequently, the low dose group received 0.38 g/kg/day, and the high dose group received 1.0 g/kg/day.

Health and behavior were monitored at least once daily throughout the study. Humane endpoints included > 20% body weight loss, persistent inability to eat or drink, severe lethargy, or any signs of distress unrelieved by intervention; no animals reached these endpoints before scheduled euthanasia. This study was approved by the Institutional Animal Care and Use Committee of Qingdao University (N. 16–035) and conducted in accordance with relevant guidelines and regulations. All experimental procedures are reported in accordance with the ARRIVE guidelines (https://arriveguidelines.org).

Necropsy, Blood Analyses, Tissue Collection and Histology

After the 16-week dietary intervention, non-fasted mice were placed in a transparent chamber and gradually exposed to a CO₂–air mixture at a displacement rate of approximately 30–70% of the chamber volume per minute, in accordance with the AVMA Guidelines for the Euthanasia of Animals (2020). Deep anesthesia was confirmed by the loss of pedal reflex and absence of voluntary movement, after which euthanasia was completed by continued CO₂ exposure until respiratory arrest. Cardiac puncture was immediately performed for terminal blood collection. Blood samples were centrifuged at 2,000 rpm for 15 min, and serum was collected for biochemical analysis using a KONELAB 20XTi analyzer (Diagnostic Products Corporation, Los Angeles, CA, USA) to determine triglyceride (TG) and alanine aminotransferase (ALT) levels. The left hepatic lobe and the terminal ileum were fixed in 4% buffered formalin for histological examination, while the remaining tissues were snap‑frozen in liquid nitrogen and stored at − 80 °C. Colonic feces were also collected into sterile 2‑mL tubes and preserved for subsequent gut microbiota and metabolite analysis.

Hepatocyte histology was evaluated by experienced pathologists blinded to the experimental grouping allocation. Histological scoring was based on steatosis–activity–fibrosis system (SAF), the pathological diagnosis scoring system recommended by the American Association for the Study of Liver Diseases (AASLD) guidelines, which independently assesses the steatosis grade (S; S0–S3), activity grade (A; A0–A4; determined by the combined grades of hepatocellular ballooning and lobular inflammation, each scored 0–2) and fibrosis stage (F; F0–F4)^23,24. Morphological parameters of the distal ileum were obtained from digital scanned sections using spot VR 5.0 software, including villus height, crypt depth, and submucosal wall thickness. The villus-height-to-crypt-depth ratio was subsequently calculated.

Gut Contents Gut Contents and DNA Extraction

The contents of the proximal colon and cecum were collected aseptically at necropsy, immediately snap-frozen in liquid nitrogen, and stored at − 80 °C until analysis. Total genomice DNA was extracted using the cetyltrimethylammonium bromide (CTAB)/sodium dodecyl sulfate (SDS) method. Briefly, approximately 200 mg of frozen sample was suspended in CTAB extraction buffer (2% CTAB, 1.4 M NaCl, 100 mM Tris-HCl, 20 mM EDTA, pH 8.0), followed by incubation at 65 °C for 30 min. The lysate was subjected to chloroform: isoamyl alcohol (24:1, v/v) extraction and centrifuged at 12,000 × g for 10 min at 4 °C. The aqueous phase was transferred to a new tube, and DNA was precipitated with isopropanol, washed with 70% ethanol, air-dried, and dissolved in sterile TE buffer. DNA concentration and purity were quantified using a NanoDrop spectrophotometer (Thermo Fisher Scientific). DNA integrity was further verified by electrophoresis on a 1% agarose gel.

16 S rRNA Gene Sequencing and Data Processing

The V3–V4 hypervariable regions of the 16 S rRNA genes were amplified using specific primers with unique barcodes. The same volume of 1X loading buffer (containing SYBR green) was mixed with the PCR products and subjected to electrophoresis on 2% agarose gel for detection. The mixture of PCR products was purified using the Qiagen Gel Extraction Kit (Qiagen, Germany). Sequencing libraries were generated using TruSeq^® DNA PCR-Free Sample Preparation Kit (Illumina, USA) and were sequenced on an Illumina NovaSeq platform. Raw paired-end reads were merged using FLASH (V1.2.7). Quality filtering and chimera removal were conducted using Quantitative Insights Into Microbial Ecology (QIIME, version 1.9.1) according to standard quality-controlled procedures. High-quality sequences were clustered into operational taxonomic units (OTUs) with ≥ 97% similarity using Uparse software (Uparse v7.0.1001). Taxonomic classification of representative sequences was performed using the Silva 16 S rRNA database (Release 138) within the QIIME framework.

Extraction, Identification and Analysis of Fecal Metabolites

Approximately 100 mg of each fecal sample was accurately weighed into a 2 mL centrifuge tube. Then, 400 µL of methanol (MeOH, pre-chilled at − 20 °C) containing 2‑amino‑3‑(2‑chloro‑phenyl)‑propionic acid (4 ppm, internal standard) was added, and the mixture was vortexed for 30 s. Subsequently, 100 mg of glass beads were added, and the samples were homogenized in a tissue grinder at 60 Hz for 90 s. The homogenates were then sonicated at room temperature for 10 min, followed by centrifugation at 12,000 rpm and 4 °C for 10 min. The resulting supernatants were filtered through a 0.22 μm membrane and transferred into autosampler vials for Liquid Chromatography–Mass Spectrometry (LC–MS) analysis. LC analysis was performed on a Vanquish UHPLC system (Thermo Fisher Scientific, USA) equipped with an ACQUITY UPLC^® HSS T3 column (150 ⋅ 2.1 mm, 1.8 μm; Waters, Milford, MA, USA). The column temperature was maintained at 40 °C, with a flow rate of 0.25 mL/min and an injection volume of 2 µL. Metabolite detection was carried out on a Q Exactive mass spectrometer (Thermo Fisher Scientific, USA) equipped with an electrospray ionization (ESI) source. Data were acquired in Full MS–ddMS² mode (data‑dependent MS/MS), enabling simultaneous MS¹ and MS/MS acquisition.

Statistical Analyses

Statistical analyses of physiological and biochemical parameters (except for gut microbiota and metabolite data) were performed using Statistical Package for the Social Sciences (SPSS, Version 26.0). Data normality was evaluated by the Shapiro–Wilk test. For normally distributed data, one-way analysis of variance (ANOVA) was used, followed by Tukey’s post-hoc test for multiple comparisons. For non-normally distributed data, the Kruskal–Wallis test (or median test) was applied, with pairwise comparisons adjusted using Benjamini–Hochberg false discovery rate (FDR) correction where applicable. Results are presented as mean ± standard deviation (SD). Associations between quantitative variables were evaluated using partial Spearman’s correlation coefficients, with p < 0.05 considered statistically significant.

Microbiota data processing and statistical analyses were carried out using QIIME (Version 1.9.1) and R packages (Version 4.3.2). Alpha diversity indices, Chao1 and ACE, were calculated using the OTU table in QIIME. Beta diversity analysis was performed using Euclidean and Bray–Curtis distance metrics, with the results visualized by principal component analysis (PCA) and non-metric multidimensional scaling (NMDS, stress = 0.065), respectively. The significance of microbiota structural differences among groups was assessed using permutational multivariate analysis of variance (ADONIS, 9,999 permutations), analysis of similarities (ANOSIM, 9,999 permutations), and multi-response permutation procedures (MRPP, 9,999 permutations) implemented in the R package vegan (v2.6-4). The corresponding effect-size measures are now reported to quantify between-group dissimilarity: R-values for ANOSIM, pseudo-F and R² values for ADONIS, and A-values calculated from MRPP delta statistics A = 1 – (observed-delta / expected-delta). These statistics with associated p values are presented in Table 1. Comparisons of taxa abundances at different taxonomic levels were analyzed using parametric (Student’s t-test) or non-parametric Wilcoxon rank-sum test depending on data normality, with FDR-adjusted p < 0.05 considered significant.

Table 1.

Assessing significant β-diversity variation among sample groups through Anosim, Adonis, and MRPP.

Group	Anosim		Adonis		MRPP
	R-value	P-value	R2	P-value	observed-delta	expected-delta	P-value
HFD-LFD	0.5767	0.001	0.21102(0.78898)	0.001	0.5145	0.5661	0.001
HFD-XOS.L	0.5006	0.001	0.21668(0.78332)	0.001	0.4983	0.55	0.001
HFD-XOS.H	0.3604	0.002	0.16797(0.83203)	0.001	0.5183	0.5555	0.001
XOS.H-XOS.L	0.2566	0.002	0.13338(0.86662)	0.002	0.5138	0.5401	0.002

Metabolomics data were processed using the Ropls software (v1.6.1). Models were were constructed with orthogonal partial least-squares discriminant analysis (OPLS-DA). To avoid model overfitting, 7-fold cross-validation and 200-response permutation testing (RPT) were performed to assess model robustness. The model exhibited high goodness of fit (R²X(cum) = 0.717; R²Y(cum) = 0.995) and strong predictive ability (Q² = 0.849). p-value < 0.05 and variable importance projection values > 1 were considered statistically significant. Differential metabolites were subjected to pathway analysis using MetaboAnalyst v5.0. The identified metabolites in metabolomics were then mapped to the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway^25–27 and visualized using the KEGG Mapper tool.

Inter-omics correlations were evaluated using partial Spearman’s correlation analysis with FDR correction (p < 0.05). The resultant correlation matrices were visualized as clustered heatmaps generated in R.

Power Analysis for the ANOVA design (four groups, n = 10each; an anticipated large effect size Cohen’s f = 0.6; and α = 0.05) was performed in R using the pwr package (v1.3-0). The resulting statistical power of 0.868, exceeding the conventional 0.8 criterion.

Results

High-dose XOS attenuates hepatic steatosis, improves ileal morphology, and reduces serum ALT and TG levels in HFD-fed mice

Across all groups, the mean daily food intake remained comparable (approximately 15–17 g/day) with no statistically significant differences among LFD, HFD, XOS.L, and XOS.H treatments (p > 0.05). Thus, XOS administration did not affect food consumption. At the 16‑week endpoint, histopathological analyses revealed that HFD-fed mice developed pronounced hepatic steatosis, characterized by extensive fat accumulation and ballooning degeneration, as indicated by swollen hepatocytes with rarefied cytoplasm. Oil Red O staining further confirmed a significantly increase of lipid droplet deposition in the HFD group compared with the LFD controls. In the small intestine, HFD feeding caused severe morphological damage, including marked villus shortening, blunting, and occasional fusion of villus tips (Fig. 1A).

Fig. 1.

High-dose XOS attenuated hepatic steatosis, improved distal ileum morphology, and reduced serum ALT/TG levels in HFD-fed mice. (A) Representative histological sections of liver and distal ileum from mice fed a low-fat diet (LFD), high-fat diet (HFD), or HFD supplemented with low-dose (XOS.L) or high-dose xylooligosaccharides (XOS.H). Hepatic H&E and Oil-Red-O staining (400×) show severe lipid accumulation in HFD mice. Low-dose XOS (L) produced modest improvement, whereas high-dose XOS (H) markedly reduced hepatic lipid deposition. Distal ileum H&E staining (200×) reveals blunted villi and shallower crypts in HFD mice, partially corrected by XOS.L and substantially normalized by XOS.H. (B, C) Serum alanine aminotransferase (ALT) and triglyceride (TG) levels significantly increased in HFD mice were dose-dependently reduced by XOS treatment (n = 10 per group). (D) Histopathological scoring (SAF system) confirmed the dose-dependent improvement of hepatic steatosis. (E) Quantification of ileal villus-to-crypt ratio showed restoration of intestinal morphology primarily in the high-dose XOS group. Data are shown as mean ± SD (n = 10 per group). Statistical significances are indicated by asterisks: *p < 0.05, **p < 0.01, and ***p < 0.001; ns, not significant.

High-dose XOS supplementation markedly alleviated these histopathological alterations, as evidenced by reduced hepatic fat accumulation, diminished ballooning degeneration, and preserved villus architecture (Fig. 1A). In contrast, low-dose XOS showed only minimal or negligible improvement compared with the HFD group (Fig. 1A).

Additionally, high‑dose XOS attenuated the HFD-induced elevations in serum ALT and TG (Figs. 1B–C; n = 10, p < 0.05), whereas low dose exhibits no improvement (Figs. 1B–C; n = 10, p > 0.05). SAF scores were higher in HFD-fed mice (Fig. 1D; n = 10, p < 0.05) and were significantly reduced by high‑dose XOS (Fig. 1D; n = 10, p < 0.001), but remained unchanged by low dose treatment. Similarly, HFD feeding decreased the distal ileal villus-height-to-crypt-depth (V/C) ratio (Fig. 1E; n = 10, p < 0.001), which was improved by high‑dose XOS supplementation (Fig. 1E; n = 10, p < 0.01) but unaltered in the low-dose group.

High-dose XOS improves gut microbiota dysbiosis in MASLD mice as revealed by 16 S rRNA gene sequencing

Alpha diversity, assessed using Chao1 and ACE indices (Wilcoxon rank–sum test), exhibited no significant differences among LFD, HFD, and XOS.L groups (Fig. 2A–B; n = 10; p > 0.05). Conversely, high-dose XOS treatment significantly increased both indices compared with all other groups (Fig. 2A–B; n = 10; p < 0.05), suggesting a dose-dependent enhancement of microbial richness and community evenness. Beta diversity, evaluated using PCA and NMDS, revealed distinct clustering patterns among all groups (Fig. 2C–D). These observations were supported by statistical tests, including ANOSIM, ADONIS, and MRPP (Table 1). HFD feeding markedly altered microbial community composition, and both high- and low-dose XOS treatments significantly ameliorated these compositional shifts (Fig. 2C–D; Table 1, n = 10, all p < 0.05). Moreover, significant differences in community structure were observed between high- and low-dose XOS groups, indicating dose-specific effects.

Fig. 2.

High-dose XOS ameliorated HFD-induced gut microbiota dysbiosis in MASLD mice. (A–B) Alpha-diversity analysis based on the Chao1 and ACE indices (Wilcoxon rank–sum test) shows that microbial richness was markedly reduced in high-fat-diet (HFD) mice, while high-dose xylooligosaccharide (XOS.H) treatment significantly restored diversity; low-dose XOS (XOS.L) produced no noticeable effect. (C–D) Beta-diversity analysis by principal component analysis (PCA; Euclidean distance) and non-metric multidimensional scaling (NMDS; Bray–Curtis dissimilarity, Stress = 0.065) reveals distinct clustering of HFD microbiota compared with other groups. The XOS.H group exhibited a clear shift in community structure relative to HFD, indicating partial correction of the dysbiosis trend, whereas XOS.L remained close to HFD. (E–F) Taxonomic composition profiles at the family and genus levels show marked alterations in major microbial taxa following HFD feeding and partial normalization with high-dose XOS supplementation. Data are presented as mean ± SD (n = 10 per group). Statistical significances are indicated by asterisks: *p < 0.05, **p < 0.01, and ***p < 0.001; ns, not significant.

Taxonomic composition and relative abundance at the family and genus levels are summarized in Fig. 2E–F. At the family level, HFD feeding significantly altered the gut microbial composition compared with the LFD controls, characterized by increased relative abundances of Ruminococcaceae, Erysipelatotrichaceae, and Lachnospiraceae, and a decreased relative abundance of Bacteroidaceae (Figs. 3A–D; n = 10; p < 0.05). High-dose XOS treatment significantly reversed these HFD-induced changes in all four families (Figs. 3A–D; n = 10; p < 0.05), whereas low-dose XOS only increased Erysipelatotrichaceae abundance (Fig. 3B, n = 10, p < 0.05) without significantly modifying Ruminococcaceae, Lachnospiraceae, and Bacteroidaceae levels. At the genus level, HFD feeding markedly increased several MASLD-associated genera with potential pathogenic features including Alistipes, Erysipelatoclostridium, Faecalibaculum, Fusobacterium, Lachnoclostridium, and Oscillospira (Figs. 3E–J; n = 10; p < 0.05), while decreasing the abundance of the putatively beneficial genus Bacteroides (Fig. 3K, n = 10, p < 0.05). High-dose XOS supplementation significantly attenuated the HFD-induced alterations in Alistipes, Bacteroides, Erysipelatoclostridium, Fusobacterium, and Lachnoclostridium (Figs. 3E, F, H, I, and K; n = 10, p < 0.05), whereas low-dose XOS treatment reduced only Erysipelatoclostridium and Fusobacterium abundance (Figs. 3F, and H; n = 10, p < 0.05).

Fig. 3.

High-dose XOS modulated gut microbial taxa associated with HFD-induced MASLD. (A–P) Relative abundance of differentially represented taxa at the family (A–D, L, M, P) and genus (E–K, N, O) levels. High-dose XOS increased the abundance of Bacteroidaceae, Bifidobacteriaceae, and Bacillaceae, while reducing the abundance of Lachnospiraceae, Erysipelotrichaceae, Lactobacillaceae, Fusobacterium, and Lachnoclostridium). In contrast, low-dose XOS induced minimal changes. Data are presented as mean ± SD (n = 10 per group). Statistical significances are indicated by asterisks: *p < 0.05, **p < 0.01, and ***p < 0.001; ns, not significant.

Importantly, high-dose XOS markedly increased the abundance of probiotic- associated taxa, including Bacillaceae, Bifidobacteriaceae, and Bifidobacterium, compared with the HFD group (Figs. 3L–N; n = 10, p < 0.05). Unexpectedly, high-dose XOS supplementation reduced the abundance of Lactobacillaceae and Lactobacillus (Figs. 3O–P; n = 10, p < 0.05), taxa that are generally considered as beneficial probiotics.

High-dose XOS ameliorates intestinal microbial metabolite alterations in MASLD mice

OPLS-DA revealed a clear separation among the four experimental groups (Fig. 4A). Compared with the LFD group, HFD feeding markedly altered the relative intensities of multiple potential metabolites associated with arachidonic acid (AA) and tryptophan metabolism, as well as peroxisome proliferator-activated receptor (PPAR) and cAMP signaling, insulin secretion, and adipocyte lipolysis (Fig. 4B). This study primarily focused on AA and tryptophan metabolic pathways.

Fig. 4.

High-dose XOS ameliorated intestinal microbial metabolite alterations in HFD-induced MASLD mice. (A) OPLS-DA score plot based on untargeted metabolomic profiling shows distinct separations among the LFD, HFD, and XOS-treated groups. Both XOS.L and XOS.H groups display metabolite profiles clearly differentiated from the HFD group, indicating marked alterations in intestinal metabolic composition upon XOS supplementation. (B) KEGG pathway enrichment analysis identifies multiple metabolic pathways altered in HFD mice compared with the LFD group, including Arachidonic acid metabolism (mmu00590) and Tryptophan metabolism (mmu00380), mapped using the Kyoto Encyclopedia of Genes and Genomes (KEGG) database. KEGG pathway imagery reproduced with permission from Kanehisa Laboratory (CC BY 4.0 license; Ref: 253373). (C–M) Changes in pathway-specific metabolites derived from arachidonic acid (AA) and tryptophan catabolism. HFD feeding induced pronounced alterations in both pathways, including elevated proinflammatory AA-derived lipids and disturbed tryptophan-kynurenine turnover, which were partially normalized by high-dose XOS. (N–O) Changes in long-chain polyunsaturated fatty acids (DHA and DPA), showing restoration toward normal levels mainly in the XOS.H group. Data are presented as mean ± SD (n = 10 per group). Statistical significances are indicated by asterisks:

*p < 0.05, **p < 0.01, and ***p < 0.001; ns, not significant.

In the AA pathway, HFD feeding significantly increased fecal levels of the pro-inflammatory mediators 12-keto-tetrahydro-leukotriene B₄ and Δ¹²-prostaglandin J₂ (Figs. 4C–D; n = 10; p < 0.05) compared with the LFD group. These increases were reversed by high-dose XOS supplementation (Figs. 4C–D, n = 10, p < 0.05), whereas the low-dose XOS group exhibited no significant changes (p > 0.05). Additionally, arachidic acid levels were significantly reduced in both high- and low-dose XOS groups relative to the HFD group (Fig. 4E; n = 10; p < 0.05).

In the tryptophan metabolism pathway, HFD feeding elevated the pro-inflammatory metabolites kynurenic acid and xanthurenic acid, while reducing the anti-inflammatory metabolite 3-hydroxyanthranilic acid (3-HAA) (Figs. 4F–H; n = 10; p < 0.05). High-dose XOS supplementation restored these metabolites levels to those observed in the LFD group (p < 0.05), but low-dose XOS failed to produce significant changes (p > 0.05). Moreover, several MASLD-protective metabolites, including indole, indole-3-acetate (IAA), melatonin, N-acetylserotonin, and indolepyruvate, were significantly elevated in the high-dose XOS group compared with the HFD group (Figs. 4I–M; n = 10; p < 0.05). These responses were dose-dependent, as the low-dose XOS group showed no significant alterations (Figs. 4I–M, n = 10, p > 0.05).

Interestingly, two anti-inflammatory omega-3 polyunsaturated fatty acids (PUFAs) in cecal contents, docosahexaenoic acid (DHA) and docosapentaenoic acid (DPA), were significantly higher in the high-dose XOS group than in the HFD group (Figs. 4N–O; n = 10; p < 0.05). In the low-dose XOS group, only DPA levels were elevated (Fig. 4N, n = 10, p < 0.05), whereas DHA levels remained unchanged.

Associations between gut microbial composition and metabolic profiles

Consistent with earlier observations, family-level analysis confirmed that HFD feeding enriched Lachnospiraceae, Ruminococcaceae, and Erysipelotrichaceae compared with the LFD group (Figs. 3A–C). Notably, Lachnospiraceae abundance was positively correlated with pro-inflammatory metabolites, including Δ¹²-prostaglandin J₂, leukotriene E₄, and kynurenic acid (Fig. 5A; n = 10; p < 0.05). Ruminococcaceae abundance was positively associated with Δ¹²-prostaglandin J₂, xanthurenic acid, and kynurenic acid, but negatively associated with 3-HAA and indolepyruvate (Fig. 5A, n = 10, p < 0.05). High-dose XOS markedly suppressed the HFD-induced expansion of both Lachnospiraceae and Ruminococcaceae, whereas no such effect was observed in the low-dose XOS group (Figs. 3A and C). Conversely, Bacteroidaceae abundance was inversely correlated with all measured pro-inflammatory metabolites (Δ¹²-prostaglandin J₂, xanthurenic acid, and kynurenic acid) (Fig. 5A, n = 10, p < 0.05). High-dose XOS treatment significantly increased Bacteroidaceae abundance, whereas low-dose XOS had no significant effect (Fig. 3D). Among probiotic-associated families, Bacillaceae abundance correlated positively with hepatoprotective indole derivatives, including IAA, melatonin, N-acetylserotonin, and indolelactic acid. Bifidobacteriaceae abundance was inversely correlated with arachidic acid (Fig. 5A, n = 10, p < 0.05). Both families were exclusively enriched in the high-dose XOS group (Figs. 3L–M), suggesting complementary metabolic benefits.

Fig. 5.

Integrated correlation analysis of gut microbiota and metabolic profiles in HFD-induced MASLD mice. (A) Partial Spearman’s correlation analysis showing associations between fecal bacterial families and metabolites involved in the arachidonic acid metabolic pathway. (B) Partial Spearman’s correlation analysis depicting associations between fecal bacterial genera and metabolites involved in the tryptophan metabolic pathway. Analyses were performed across all samples with “group” included as a covariate to adjust for inter-group mean differences. The Y-axis denotes differential metabolites, and the X-axis denotes bacterial taxa. The color scale represents correlation coefficients (red = positive, blue = negative; color intensity indicates correlation strength). Data are for n = 10 per group. Statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001.

At the genus level, as previously indicated, Fusobacterium, Lachnoclostridium, and Alistipes abundances were significantly higher in the HFD group than in the LFD group (Figs. 3E and H–I). Correlation analysis revealed that all three genera were positively associated with pro-inflammatory metabolites. Specifically, Fusobacterium correlated positively with 12-keto-tetrahydro-leukotriene B₄ and arachidic acid; Lachnoclostridium correlated with 12-keto-tetrahydro-leukotriene B₄ and Δ¹²-prostaglandin J₂; and Alistipes exhibited negative correlations with MASLD-protective metabolites, including DHA, DPA, IAA, indolepyruvate, melatonin, and N-acetylserotonin (Fig. 5B, n = 10, p < 0.05). High-dose XOS intervention effectively attenuated HFD-induced overgrowth of all three genera, whereas low-dose XOS reduced only Fusobacterium abundance (Figs. 3E and H–I).

Associations between gut microbiota composition, biochemical profiles, and histopathological features

At the family level, Lachnospiraceae and Ruminococcaceae abundance were positively correlated with serum ALT levels and liver SAF scores and were negatively associated with small intestinal V/C ratios (Fig. 6A; n = 10; p < 0.05). Conversely, Bacteroidaceae abundance displayed the opposite pattern, showing negative correlations with serum ALT and SAF scores, and a positive correlation with intestinal V/C ratios (Fig. 6A; n = 10; p < 0.05). Aforementioned, Lachnospiraceae and Ruminococcaceae abundances were significantly elevated, whereas Bacteroidaceae abundance was markedly reduced in the HFD group compared with the LFD group. High-dose XOS supplementation effectively reversed HFD-induced microbiota alterations in a dose-dependent manner, whereas low-dose XOS had no significant modulatory effects (Figs. 3A and C–D).

Fig. 6.

Correlation analysis of gut microbiota composition, serum biochemical parameters, and histopathological features in HFD-induced MASLD mice. (A) Partial Spearman’s correlation analysis between serum biochemical parameters (ALT, serum TG), histopathological markers (villus height/crypt depth ratio, SAF score), and gut microbiota composition at the family level. (B) Same analysis at the genus level. Analyses were performed across all samples with “group” included as a covariate to adjust for inter-group mean differences. The Y-axis denotes bacterial taxa, and the X-axis denotes serum biochemical and histopathological parameters. The color scale represents correlation coefficients (red = positive, blue = negative; color intensity indicates correlation strength). Data are for n = 10 per group. Statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001.

At the genus level, Lachnoclostridium abundance was positively correlated with serum ALT and TG, and liver SAF scores but was negatively associated with intestinal V/C ratios (Fig. 6B; n = 10; p < 0.05). In contrast, Bacteroides abundance demonstrated the inverse pattern, showing negative correlations with serum ALT and SAF scores and a positive correlation with intestinal V/C ratio (Fig. 6B; n = 10; p < 0.05). Fusobacterium was uniquely associated with positive correlations with both serum ALT and liver SAF scores (Fig. 6B; n = 10; p < 0.05). High-dose XOS administration significantly reduced the HFD-induced enrichment of Lachnoclostridium and Fusobacterium, and restoring Bacteroides abundance; these effects were absent in the low-dose XOS group (Figs. 3H–I and K).

Collectively, these findings provide further evidence that high-dose XOS exerts MASLD-protective effects by reshaping gut microbiota composition and alleviating related biochemical and histopathological abnormalities.

Discussion

This study integrated histopathology evaluation, serum biochemistry analysis, 16 S rRNA profiling, and untargeted metabolomics to demonstrate that XOS supplementation confers dose-dependent benefits in HFD-induced MASLD. High-dose XOS markedly alleviated hepatic injury and improved ileal morphology while remodeling the gut microbiota and its metabolite milieu toward an anti-inflammatory, hepatoprotective state.

The concordant improvements in ALT, steatosis scores, and V/C ratios indicate that high-dose XOS alleviates both hepatic and small-intestinal injury, supporting its primary modulatory role in the gut–liver axis. Importantly, XOS supplementation did not influence appetite or overall energy intake, confirming that the metabolic and histological improvements observed were independent of changes in caloric consumption. The absence of benefit in the low-dose group suggests that a threshold level of fermentable substrate is required to overcome HFD-conditioned colonization resistance, support the growth of primary XOS utilizers, and establish cross-feeding networks that yield physiologically relevant metabolites. In practical terms, the prebiotic dose should be considered a critical design parameter rather than assuming that “more is always better”. The optimal dosage range is likely influenced by diet composition, baseline microbiota profile, gut transit time, and luminal pH.

High-dose XOS partially ameliorated HFD-induced gut microbial alterations, as reflected by reduced relative abundances of Ruminococcaceae, Lachnospiraceae, and Erysipelotrichaceae, together with increased Bacteroidaceae and enrichment of Bifidobacteriaceae/Bifidobacterium and Bacillaceae. Previous studies have reported that increased Ruminococcaceae abundance is associated with NAFLD severity, as evidenced by its positive correlation with the NAS in patients consuming a high-carbohydrate diet²⁸. Members of the Lachnospiraceae family have also been linked to altered bile acid metabolism, which may exacerbate NAFLD by influencing hepatic bile acid synthesis and reabsorption²⁹. In contrast, increased Bacteroidaceae abundance has been associated with NAFLD improvement; for example, pure total flavonoids from citrus significantly increased Bacteroidaceae while reducing toxic bile acids³⁰. Notably, however, Ruminococcaceae and Lachnospiraceae are functionally heterogeneous families, and their effects may vary depending on species composition and metabolic output^31,32. Therefore, these family-level changes should be interpreted with caution. By comparison, members of Bifidobacteriaceae, particularly Bifidobacterium, are commonly associated with beneficial effects on gut homeostasis, and some are used as probiotics. Consistent with this notion, Bifidobacterium adolescentis has been shown to alleviate hepatic steatosis and steatohepatitis by enhancing sensitivity to fibroblast growth factor 21 (FGF21)³³, whereas Bifidobacterium bifidum may suppress hepatic inflammation and lipid accumulation by increasing intestinal short-chain fatty acid levels, including propionate and butyrate, and by modulating lipid metabolism and gut permeability³⁴. Taken together, these findings suggest that XOS may beneficially modulate the gut microbial ecosystem and thereby contribute to improved gut–liver axis homeostasis in MASLD.

An unexpected observation was the reduction in Lactobacillaceae/Lactobacillus abundance in the high-dose group. Possible explanations include: (1) competitive exclusion by rapidly expanding primary XOS utilizers (e.g., Bifidobacterium), which may limit available ecological niches³⁵; (2) metabolic feedback, whereby early lactate accumulation favors the lactate-utilizing SCFA producers, thereby shifting equilibrium away from lactobacilli³⁶; and (3) potential sampling bias, as fecal 16 S profiles may underrepresent small-intestinal lactobacilli or strain-level heterogeneity³⁷. The absence of physiological detriment despite this reduction suggests that overall community function, and its metabolic outputs, are more important than the abundance of any single genus.

Our data demonstrate that XOS-driven microbiota remodeling is closely coupled to alterations in two key metabolic axes: the tryptophan–indole and the AA pathways. These pathways were selected further investigation because they were among the most significantly altered after XOS treatment in metabolomics pathway enrichment analysis (p < 0.05), possess well‑documented roles in inflammation and host metabolic regulation, and offer plausible mechanistic links between gut microbiota modulation and hepatic improvement. Dysbiosis‑induced perturbations in tryptophan metabolism, and AA‑derived inflammatory mediators, are increasingly recognized as key molecular bridges in the gut–liver axis in MASLD pathogenesis.

Mechanistically, indole and its derivatives are hallmark microbial metabolites of dietary tryptophan catabolism, possessing well-documented anti-inflammatory and barrier-protective effects. Indole activates nuclear receptors, such as the aryl hydrocarbon receptor (AhR) and pregnane X receptor (PXR), thereby regulating intestinal hormone secretion, maintaining epithelial integrity, and modulating hepatic immunity and metabolism³⁸. For example, Gao et al.. showed that indole attenuates MASLD in an angiotensin‑converting enzyme 2 (ACE2)-dependent manner by reducing reactive oxygen species, preserving mitochondrial membrane potential, and upregulating sirtuin‑3 (SIRT3) expression³⁹. IAA protects the liver by enhancing intestinal mucus sulfation though upregulation of 3′‑phosphoadenosine 5′‑phosphosulfate synthase 2 (PAPSS2) and solute carrier family 35 member B3 (SLC35B3), and by promoting hepatic macrophage polarization toward the M2 phenotype. Comparable effects have been validated in human cohorts after sleeve gastrectomy^40,41.

Conversely, AA and its derivatives play complex, context-dependent roles in hepatic diseases. In MASLD, AA-derived eicosanoids, such as leukotriene B₄ and prostaglandin E₂, produced via 5-lipoxygenase (5-LOX) and cyclooxygenase pathways, are upregulated and pro-inflammatory^42,43. LTE₄, a terminal 5-LOX product, activates P2Y purinoceptor 12 (P2Y₁₂) and cysteinyl leukotriene receptor 1 (CysLT₁), triggering extracellular signal‑regulated kinase (ERK) phosphorylation, inducing chemokine induction, stimulating prostaglandin D₂ (PGD₂) production, and broadly reprogramming inflammatory gene expression^44,45. However, under certain conditions, AA may exert hepatoprotective effects by upregulating organic anion transporting polypeptide 1 (OATP1), thereby promoting hepatic bile-acid clearance during cholestasis⁴⁶.

In our study, high-dose XOS reduced pro-inflammatory AA derivatives (Δ¹²-prostaglandin J₂, leukotriene species, kynurenic acid, and xanthurenic acid) while restoring protective tryptophan–indole metabolites (indole, IAA, melatonin, N-acetylserotonin, and indolepyruvate) and enhancing anti-inflammatory omega-3 PUFAs (DHA and DPA). These biochemical improvements coincided with favorable histological changes and improved barrier integrity, supporting the hypothesis that XOS attenuates MASLD partly by rebalancing the tryptophan-indole and arachidonic acid metabolic pathways.

The differences in microbial diversity, community composition, and metabolite profiles between high- and low-dose XOS suggest that insufficient substrate fails to reconfigure dysbiotic communities. Potential underlying mechanisms may involve elevated substrate availability shifting the competitive balance toward primary XOS utilizers, thereby sustaining cross-feeding chains that produce SCFAs and indole derivatives at physiologically relevant levels, and modifying luminal biochemistry environment to favor anti-inflammatory consortia. The observation that low-dose XOS affects only a few taxa without systemic improvement underscores the need for adequate prebiotic dosing to achieve effective microbiota remodeling.

The novelty of this work lies in revealing dose-resolved effects of XOS and a distinct gut-microbiota configuration that differs from previously reported XOS responses, alongside methodological innovations advancing research on its hepatic protective action. For the first time, we demonstrated that different physiological doses of XOS exert distinct effects on hepatic protection in HFD-induced MASLD. This dose-dependent observation extends previous work by revealing dose-specific responses that were not captured in earlier single-dose studies. In contrast to previous studies reporting that XOS concurrently stimulates the growth of Bifidobacteriaceae and Lactobacillaceae^47,48, we found that low-dose XOS had no appreciable effect on either group, whereas high-dose XOS markedly increased Bifidobacteriaceae abundance while suppressing Lactobacillaceae, revealing divergent and opposing dose-dependent trends not previously documented in XOS research. Moreover, high-dose XOS suppressed the inflammation-associated family Lachnospiraceae, indicating a compositional pattern distinct from prior findings. Methodologically, our study further advances XOS-related investigations through several key innovations. First, instead of relying on exploratory clustering⁴⁹, we applied partial Spearman’s correlation with covariate adjustment to identify statistically robust microbe–metabolite associations, uncovering key pathways involving indole and arachidonic acid derivatives. Second, we integrated microbiota–metabolite relationships with serological and histopathological outcomes, providing a systems‑level view of the gut–liver axis. Finally, our use of LC–MS‑based metabolomics enabled broader metabolite coverage and higher sensitivity compared with the NMR or targeted methods used previously⁵⁰. These methodological and conceptual advances yield a more mechanistic and clinically relevant understanding of how XOS mitigates MASLD.

Several limitations of this study should be acknowledged. Our findings are primarily correlative, and causality remains to be verified through fecal microbiota transplantation or targeted manipulation of key taxa and host pathways. Only male mice, a single XOS source, and one intervention time point were used, and microbiome analysis relied on fecal rather than mucosal samples. Although high-dose XOS showed the strongest benefits, its long-term safety and potential dose-related trade-offs remain to be clarified. Moreover, these results were obtained in an animal model and should be extrapolated to humans with caution. Future work should include both sexes, multiple doses, longer interventions, and multi-omic approaches to strengthen causal inference and translational relevance.

Conclusion

High-dose XOS reprograms the gut microbiota–metabolite network toward an anti-inflammatory and hepatoprotective configuration in HFD-fed mice in a dose-dependent manner. This study provides preliminary mechanistic evidence and testable hypotheses supporting the development of prebiotic-based therapeutic strategies for MASLD, and highlights the critical importance of dose optimization for the success of such interventions.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

3‑HAA

3‑hydroxyanthranilic acid

5‑LOX

5‑lipoxygenase

AASLD

American Association for the Study of Liver Diseases

ACE2

Angiotensin‑converting enzyme 2

ADONIS

Analysis of dissimilarities of non‑parametric statistics

AhR

Aryl hydrocarbon receptor

ALT

Alanine aminotransferase

ANOSIM

Analysis of similarities

COX

Cyclooxygenase

CysLT₁

Cysteinyl leukotriene receptor 1

DHA

Docosahexaenoic acid

DPA

Docosapentaenoic acid

ERK

Extracellular signal‑regulated kinase

FMT

Fecal microbiota transplantation

FDA

U.S. Food and Drug Administration

FGF21

fibroblast growth factor 21

FOS

Fructo‑oligosaccharides

H&E

Hematoxylin and eosin

HFD

High‑fat diet

IAA

Indole‑3‑acetate

LC–MS

Liquid chromatography–mass spectrometry

LFD

Low‑fat diet

LTE₄

Leukotriene E₄

LTB₄

Leukotriene B₄

MASLD

Metabolic dysfunction‑associated steatotic liver disease

MRPP

Multiresponse permutation procedures

NMDS

Non‑metric multi‑dimensional scaling

OATP1

Organic anion transporting polypeptide 1

OPLS–DA

Orthogonal partial least squares discriminant analysis

ORO

Oil Red O

P2Y₁₂

P2Y purinoceptor 12

PAPSS2

3′‑phosphoadenosine 5′‑phosphosulfate synthase 2

PCA

Principal component analysis

PGE₂

Prostaglandin E₂

PGD₂

Prostaglandin D₂

PUFAs

Polyunsaturated fatty acids

PXR

Pregnane X receptor

ROS

Reactive oxygen species

SAF

Steatosis-activity-fibrosis system

SIRT3

Sirtuin (silent mating type information regulation 2 homolog) 3

SLC35B3

Solute carrier family 35 member B3

TG

Triglycerides

V/C ratio

Height‑to‑crypt depth ratio

XOS

Xylooligosaccharides

Author contributions

Li Chen and Ti-Dong Shan drafted and edited the manuscript. Both authors confirmed and approved this study.

Funding

This study was supported by the hospital-level research project of The Affiliated Hospital of Qingdao University (grant no. 3393).

Data availability

The 16 S rRNA sequencing data are publicly available in the NCBI SRA under BioProject accession number PRJNA1314982 and SRA Study accession SRP617915. The metabolomics data are available in the Metabolomics Workbench repository under Study ID ST004245 (DataTrack ID 6383) and can be accessed via the project DOI: [https://doi.org/10.21228/M8RG29](https:/doi.org/10.21228/M8RG29) .

Declarations

Competing interests

The authors declare no competing interests.