Nfil3 integrates circadian rhythm and microbial metabolite signaling to maintain gut–liver immune–metabolic homeostasis under high-fat diet stress

Abstract

Background

Disruption of circadian regulation and gut microbial homeostasis is a hallmark of metabolic dysfunction associated with steatotic liver disease (MASLD). Nuclear factor interleukin 3 (Nfil3) integrates circadian and immune signaling; however, how Nfil3 interfaces with microbiota-associated metabolic cues in MASLD remains incompletely understood. We investigated the role of Nfil3 in linking microbial functional states to hepatic metabolic and immune responses under high-fat diet (HFD) stress and assessed the modulatory impact of probiotic VSL#3 intervention.

Methods

We integrated exploratory human peripheral blood mononuclear cell (PBMC) transcriptomic profiling with genetic Nfil3 deletion and probiotic VSL#3 supplementation in HFD-fed mice. Experimental assessments included liver histopathology, metabolic phenotyping, immune flow cytometry, gut epithelial barrier analysis, 16S rRNA microbiome profiling with predictive functional inference, and RT-PCR.

Results

Exploratory PBMC transcriptomic analysis of obese individuals suggested that NFIL3 may function as a candidate transcriptional node associated with circadian-related genes and short-chain fatty acid (SCFA) sensing receptors in inflammatory signaling pathways. In mice, HFD feeding was associated with increased Nfil3 expression, hepatic steatosis, metabolic dysfunction, immune cell expansion, and impaired intestinal epithelial barrier integrity. Probiotic VSL#3 supplementation mitigated several HFD-associated phenotypes, including weight gain, glucose intolerance, dyslipidemia, transaminase elevation, hepatic lipid accumulation, and gut epithelial permeability, while partially normalizing intrahepatic immune cell composition. Nfil3-deficient mice displayed attenuated responses to several HFD-induced metabolic and inflammatory alterations, with partial phenotypic overlap with probiotic-treated wild-type (WT) mice. Microbiome analyses showed that VSL#3 enriched SCFA- and mucin-associated taxa while suppressing endotoxin-associated bacteria (Desulfovibrionaceae, Romboutsia). Predictive functional profiling suggested restoration of microbial pathways related to amino acid, redox, and energy metabolism, alongside reduced representation of lipopolysaccharide and toxin biosynthesis pathways.

Conclusions

These findings support a role for Nfil3 as a regulatory node linking microbial functional potential with immune and metabolic responses in MASLD. Although preclinical in nature, this work provides a mechanistic framework that may inform future translational investigations into how microbiota-associated metabolic reprogramming influences host immune–metabolic homeostasis. Further circadian-resolved and metabolite-level studies, together with human interventional validation, will be required to determine the clinical relevance of the microbiota–Nfil3 axis.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12967-026-08000-9.

Background

Metabolic dysfunction-associated steatotic liver disease (MASLD) and obesity are increasingly recognized as systemic disorders that arise from the disruption of circadian, immune, and metabolic networks [1, 2]. The gut microbiota is a central orchestrator of these interactions, influencing host energy homeostasis, endocrine signaling, and immune tone through microbial metabolites, such as short-chain fatty acids (SCFAs), bile acids, and indole derivatives [3–7]. Disturbance of this gut–liver axis [8, 9], often triggered by high-fat diet (HFD) feeding, leads to chronic low-grade inflammation, hepatic steatosis [10, 11], and impaired gut epithelium barrier integrity [12–16]. Although the mechanisms that synchronize microbial and host circadian rhythms remain poorly defined, accumulating evidence suggests that disruption of temporal coordination between the microbiota and host clock genes contributes to metabolic inflammation and liver injury.

Among the transcriptional regulators linking circadian rhythm to metabolic and immune homeostasis, Nfil3 (nuclear factor, interleukin-3 regulated; also known as E4BP4) has emerged as a critical node [17–19]. Nfil3 is rhythmically expressed under the control of the CLOCK–BMAL1 complex, functioning as a transcriptional repressor of PER2, DBP, and metabolic regulators involved in lipid synthesis and cytokine signaling [17, 18, 20]. Under physiological conditions, Nfil3 oscillation aligns nutrient metabolism and immune responses with the light–dark cycle [21, 22]. However, under metabolic stress such as obesity [23, 24], chronic inflammation [23, 25], or circadian misalignment [26, 27], Nfil3 expression becomes persistently elevated, repressing clock activators and thereby dismantling temporal homeostasis [20, 27]. Elevated Nfil3 has been implicated in hepatic steatosis, insulin resistance, and altered macrophage polarization [28–30], however its role in microbiota-dependent metabolic regulation remains incompletely understood.

Recent studies have illuminated a bidirectional relationship between microbial metabolites and circadian transcriptional programs [30–32]. SCFAs, such as butyrate and propionate, entrain peripheral clocks by activating G-protein–coupled receptors (FFAR2/3, HCAR2) and inhibiting histone deacetylases (HDACs), thereby promoting anti-inflammatory and metabolic flexibility [33–36]. Conversely, gut dysbiosis induced by HFD reduces SCFA production and enhances endotoxin synthesis, triggering TLR4–MYD88–Nfil3 signaling and chronic inflammation [12, 37, 38]. These findings suggest that Nfil3 may act as a transcriptional sensor integrating microbial signals with host circadian and immune pathways.

Probiotic interventions, particularly with multi-strain formulations with pleiotropic effects, such as VSL#3^®, have demonstrated potential to restore metabolic balance by reshaping microbial ecology and metabolite output [39–44]. VSL#3 contains eight live bacterial species with proven capacity to enhance gut epithelial barrier function [45], increase SCFA production, and attenuate hepatic steatosis and fibrosis [39, 46–49]. While VSL#3’s beneficial effects on lipid and glucose metabolism have been documented [48, 50, 51], the transcriptional mechanisms mediating its systemic impact remain elusive. We hypothesized that the probiotic’s metabolic benefits depend on Nfil3-mediated reprogramming of the microbiota–host axis.

In this study, we integrated human peripheral blood mononuclear cell (PBMC) transcriptomics, genetic ablation of Nfil3, and microbiota-targeted intervention with VSL#3 to define the molecular and microbial mechanisms that couple circadian regulation with metabolic homeostasis. We first identified Nfil3 as a transcriptional hub linking disrupted circadian and inflammatory pathways in PBMCs from obese individuals. Using mouse models, we showed that Nfil3 deficiency or probiotic supplementation confers protection against HFD-induced metabolic derangement, hepatic steatosis, and immune activation. Through integrated 16S rRNA and functional metagenomic analyses, we revealed that VSL#3 restores SCFA and amino acid metabolic pathways that feed back to modulate Nfil3 rhythmicity. These findings define a microbiota–metabolite–Nfil3–liver immune axis that mechanistically connects microbial functional reprogramming to circadian–immune–metabolic regulation.

Methods

Patient specimens

PBMC samples were collected within a standardized morning window (08:00–10:00) following overnight fasting to minimize variability related to feeding status and sleep–wake cycles. The samples were obtained from healthy volunteers and individuals with obesity, including both men and women, who were recruited and stratified according to body mass index (BMI). Participants were classified into two groups: a non-obese group (BMI < 25 kg/m²; n = 4) and an overweight/obese group (BMI ≥ 25 kg/m²; n = 4).

Inclusion criteria comprised age between 18 and 45 years, stable body weight, absence of acute infection at the time of sampling, and no history of chronic metabolic, inflammatory, autoimmune, or malignant disease. Participants were not taking regular medications, immunomodulatory agents, or recreational drugs.

To minimize potential confounding related to circadian rhythm and lifestyle, individuals with diagnosed sleep disorders, irregular sleep patterns, or current or recent shift-work schedules were excluded. Lifestyle-related variables, including dietary habits, physical activity level, smoking status, alcohol consumption, and medication use, were assessed at recruitment.

Both groups were balanced for sex distribution (~ 50% male/female) and did not differ significantly in mean age. Although gut microbiota composition was not directly assessed in this cohort, lifestyle- and metabolism-related factors known to influence microbiota variation were controlled as much as during recruitment. Residual confounding related to unmeasured circadian or microbiota factors is acknowledged as a limitation of the study. The sample size was estimated to provide 80% statistical power to detect a 10% difference in BMI between groups at a significance level of P < 0.05.

All procedures involving human participants were conducted in accordance with relevant ethical regulations and the principles of the Declaration of Helsinki. The study protocol was reviewed and approved by the Clinical Research Ethics Committee of National Cheng Kung University Hospital. Approval for the study involving participants with obesity was granted by the Human Ethics Committee (IRB No. B-BR-108-018-T). Written informed consent was obtained from all participants prior to sample collection.

Animals, diet, and intervention

Male wildtype (WT) mice were obtained from the National Laboratory Animal Center (Taipei, Taiwan). Nfil3 Knockout (KO) mice were obtained from the National Laboratory Animal Center and transferred to the Immunology Institute of the National Taiwan University. Seven weeks prior to the experiment, animals were housed in a temperature-controlled environment between 20 °C and 24 °C with a 12 h light/dark cycle. Male C57BL/6 mice were purchased from the National Laboratory Animal Center and kept in the National Yang Ming Chiao Tung University College of Medicine Laboratory Animal Center in specific pathogen-free conditions. The study protocol was designed as per the animal experiment guidelines of the National Yang Ming Chiao Tung University College of Medicine. For all in vivo experiments, mice were maintained under a 12:12 h light–dark cycle, and tissue and blood collection was performed between Zeitgeber time (ZT) 4 and ZT6 for all genotypes, diets, and treatment conditions to ensure sampling at a consistent circadian phase. After one week of acclimatization, mice were divided into two groups based on diet. One group was fed HFD (TD.210600; ENVIGO++++ Inc., USA), hereinafter referred to as the HFD group (n = 10). The other group received the recommended normal chow diet (5010 LabDiet; Purina Mills, USA), hereinafter referred to as the NCD group (n = 10). For the interventions, 22-week-old HFD-fed mice were orally administered VSL#3 (Sigma-Tau Pharmaceutics, Inc., Gaithersburg, MD, USA), containing Lactobacillus paracasei, Lactobacillus plantarum, Lactobacillus acidophilus, Lactobacillus delbrueckii subspecies bulgaricus, Bifidobacterium longum, Bifidobacterium lactis, Bifidobacterium breve, and Streptococcus thermophiles. Meanwhile, another group of mice was continuously fed HFD for an additional 22 weeks. Throughout the experimental period, animal body weight, food intake, and general health status were monitored weekly.

Biochemistry

Blood samples were tested using Contour^®Plus glucometer (Ascensia Diabetes Care, Basel, Switzerland) immediately after blood collection. Hepatocellular disintegration and necrosis were assessed spectrophotometrically in serum samples in terms of aspartate aminotransferase (AST) and alanine transaminase (ALT) activity using a Cobas^®c111Analyzer (Roche Diagnostics GmbH, Penzberg, Germany) at 340/378 nm extinction. Plasma tryglyceride (TG) levels were determined using a TG Colorimetric Assay Kit (Nr.:10010303, Cayman Chemical Company, Hamburg, Germany), according to the manufacturer’s instructions.

Tissue collection

The ileum, colon, liver, and adipose tissue of the mice were collected after sacrifice and preserved in 4% formaldehyde. Tissues were paraffined, sectioned, and stained with hematoxylin and eosin (H&E, Abcam, UK). Immunohistochemistry was carried out as previously reported [52].

H&E staining

H&E staining was used to examine tissue nuclei and cytoplasm. Tissue samples were stained with hematoxylin for 1–2 min before washing. Next, they were soaked in eosin dye for 30–60 s before being rinsed under running water (the staining duration varied per tissue). Pure ethanol was then used to lighten the samples, which were consequently observed by microscopy.

Picrosirius red staining

Picrosirius Red Staining was used to observe fibrosis. Picrosirius red (Micro-Sirius Red, Abcam, ab150681) was applied to tissue samples for 60 min. These were rinsed twice with acetic acid and again with pure alcohol and microscopically observed.

Alcian blue-periodic acid-Schiff staining (Ab/PAS)

Alcian Blue 8GX (Sigma, A5268-10G) dye was used to clean tissue samples for 15 min. After 5 min, periodic acid (Abcam, ab150680) was removed. Schiff’s solution was used to rinse the samples for 5–10 min. Hematoxylin-stained cell nuclei were washed after 1 min and observed by microscopy (Nexcope NE 910, U.S.A).

Fasting blood glucose and oral glucose tolerance test (OGTT)

Fasting glucose levels were measured every four weeks. After a 14 h fast, blood glucose levels were determined from blood collected from a tail incision (0 min data) using test strips and a Taiwanese Super OK meter. The OGTT was performed at 20 weeks. A 25% glucose solution (1 g/kg) was supplied to the mice via a gavage catheter. Blood glucose levels were monitored at 15, 30, 60, and 120 min using incision-site blood glucose strips.

Ussing chamber intestinal permeability assessment

After longitudinal dilatation of the intestinal section, the waste was carefully removed. The white tank of the instrument contained intestinal segment with the villous side facing upwards, whereas the second transparent tank enclosed it. Both serosal and mucosal chambers were equipped with two sets of electrodes. The mucosal chamber contained mannitol, whereas the serosal chamber contained glucose. After 800 µL was transported, an equivalent volume of serosal solution was aspirated. The mucosal chamber was then incubated with fluorescein isothiocyanate (FITC) dextran 4000.

Paracellular permeability assay

The mucosal-to-serosal flux rate of FITC dextran (average mol wt 4,000; Sigma Aldrich, UK) (catalog number: 46944) was used in this assay to determine gut epithelial permeability. After equilibration, 300 µL of 4 kDa FITC dextran was added to the mannitol-Krebs solution. A FITC dextran starting concentration of 1 mg/mL was used. Initial tests revealed that FITC dextran of 4 kDa was not readily detectable in healthy murine tissues. At 0, 30, 60, and 90 min post-luminal addition of FITC dextran, 800 µL samples of serosal buffer were collected and replaced with equal volumes of Krebs buffer/glucose, and fluorescence was measured using TECAN Infinite 200/200 PRO, TECAN Spark. The read emission and fluorescence absorption wavelengths were 488 nm and 520 nm, respectively. The serosal concentrations were determined using a standard curve derived from known concentrations. The paracellular flux was expressed as pmol/cm²/h.

Bacterial translocation

After euthanasia, the spleen and mesenteric lymphoid tissues of the mice were removed and deposited in a microcentrifuge tube with homogenized particles. The net weight of each tissue sample was determined; for every 0.01 g of tissue weight, 100 µL of sterile phosphate-buffered saline (PBS) solution was added to the tube. The tissues were homogenized, and tissue lysate (200 µL) was distributed on a Scientific Biotech Corporation sheep blood agar plate using a sterile glass rod; the plate was incubated at 37 °C overnight. On the following day, bacterial colonies were evaluated.

RNA extraction

TRIzol (Invitrogen, Carlsbad, CA, USA) (100 µL/0.01 g) was added to intestinal or hepatic tissue. After shearing, a pulverizing device was used to homogenize the tissues and carefully combine them with 100% ethanol. The sample was then centrifuged for 1 min at 12,000 ×g and 4 °C and the supernatant transferred to a column. RNA was isolated using Direct-zol RNA MiniPrep (ZYMO RESEARCH, R2050). A Nanodrop was used to measure RNA concentration, and an agarose gel was run at 90 V for 20 min. The RNA was then maintained at − 80 °C.

Complementary DNA (cDNA) transcription

Nuclease-free water was used to modify and normalize RNA concentrations in the samples. A reverse transcription buffer was prepared using high-capacity cDNA reverse transcription reagent kit (Thermo Fisher, 4374966). A Bio-Rad T100TM thermal cycler was used to facilitate cDNA transcription. Sample temperature was maintained at − 20 °C.

Quantitative reverse transcription polymerase chain reaction (qRT-PCR)

Total RNA isolation and RT-PCR were conducted in accordance with the manufacturer’s instructions. The Ct values were used to calculate the results, which were then normalized to GAPDH mRNA levels. The primer sequences for Nfil3 were forward: CAGGACTACCAGACATCCAAGG, reverse: AGGACACCTCTGACACATCGGA. The calculated Ct values were employed to execute statistical analyses following the experimental procedure.

Full-length 16S rRNA gene sequencing and microbiota analysis

The concentration and purity of fecal DNA were first assessed using a NanoDrop spectrophotometer. Following quality validation by Taiwan’s BIOTOOLS Co., Ltd, samples underwent full-length 16S rRNA gene sequencing using third-generation metagenomic platforms. Universal primers (27 F and 1492R) were employed to amplify the complete 16S rRNA gene. After quality control, library preparation, DNA purification, and damage repair, sequencing was conducted using the PacBio platform. Amplicon sequence variants (ASVs) were generated and taxonomically annotated by comparison with reference databases, including NCBI, GreenGenes, and SILVA. Microbial diversity, taxonomic composition, functional prediction, and correlation analyses were conducted. Data processing and bioinformatic analyses were performed via the cloud-based platform of Taiwan’s BIOTOOLS Co., Ltd. To identify key microbial features, linear discriminant analysis effect size (LEfSe) was used to detect statistically and biologically relevant biomarkers across groups. LEfSe enabled identification of taxa with significant differential abundance across multiple taxonomic levels (phylum, class, order, family, and genus), emphasizing both statistical significance and effect size. Additionally, Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt) was applied to predict metagenomic functional profiles. This approach infers gene content and potential functional capabilities of microbial communities based on marker genes, such as 16S rRNA. For statistical analysis, Welch’s t-test was performed to identify significantly different taxa between groups at various taxonomic ranks (phylum to species). Taxa with P < 0.05 were considered significantly different and visualized using bar plots representing intergroup differences [53–70]. Functional prediction was performed using PICRUSt, and all Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway annotations were interpreted as inferred functional potential rather than experimentally measured metabolic activity.

Flow cytometry and liver-infiltrating leukocyte isolation

The liver was perfused with Hank’s balanced salt solution (HBSS [II]) until blood was completely cleared from the portal vein. An infusion of 0.05% collagenase (SI-C5138; Sigma Aldrich) was administered to soften the liver. After removing the liver, the tissue was shredded and submerged in a collagenase solution. A 50 mL centrifuge tube and a 70 μm filter were used to grind the liver and fluids. The tube was then filled with 50 mL PBS. Centrifugation was performed at 50 × g for 5 min to separate non-parenchymal cells from hepatocytes. The supernatant containing non-parenchymal cells was centrifuged at 3000 g for 10 min and discarded, and 1 mL of RBC Lysing Buffer Hybri-MaxTM (Sigma Aldrich, #R7757-100ML) was used to lyse blood cells for 1 min. The lysate was neutralized with PBS in a flow tube. To count the cells, 10 µL of the supernatant was centrifuged at 600 g for 5 min. The samples were then incubated for 10 min with 100 µL of the Fc block and Golgi stop combination. APC bead buffer (65 µL) was added to 100 µL of anti-mouse CD45 APC antibody and incubated for 30 min. iMeg buffer (BD Biosciences, 552362) (1 mL) was added, and the sample was incubated for 8 min. After 4 min of incubation, another 1 mL of iMag buffer was added to the non-brown region that adhered to the magnetic stand. Using a magnetic support, the liquid was removed from the non-brown zone. Cell subdivision using targeted antibody detection was performed. Before fixing with 2% PFA, 100 µL of the produced antibody was incubated for 30 min. The supernatant was removed, 200 µL PBS was added, and flow cytometry and gating strategy were performed [71]. Fluorochrome-conjugated monoclonal antibodies against CD45, F4/80, CD11b, Ly6C, Ly6G, MHC II (I-A/I-E), CD11c, CD19, Thy1.2 (CD90.2), CD4, and CD8 (all from BioLegend) were used as detailed in Table 1. The same gating strategy was applied consistently across all experimental groups. Representative gating plots are shown in Supplementary Fig. 4.

Table 1.

Antibodies used for flow cytometric analysis

Marker	Fluorochrome	Clone	Catalog No.	Company
CD45	APC	30-F11	103112	BioLegend
F4/80	BB700	T45-2342	746070	BD
CD11b	FITC	M1/70	101206	BioLegend
Ly6C	BV421	HK1.4	128032	BioLegend
Ly6G	PE	1A8	127608	BioLegend
MHC II (I-A/I-E)	BV510	M5/114.15.2	107636	BioLegend
CD11c	PE	N418	117308	BioLegend
CD19	PE	6D5	115508	BioLegend
Thy1.2 (CD90.2)	PerCP/Cy5.5	30-H12	105338	BioLegend
CD4	FITC	GK1.5	100406	BioLegend
CD8	PE-Cy7	53-6.7	100722	BioLegend

Flow Cytometry Reagents: Surface markers were stained using fluorochrome-conjugated monoclonal antibodies listed in Table. All antibodies were obtained from BioLegend (San Diego, CA, USA) with clones and catalog numbers as indicated. Cells were stained for 30 minutes at 4°C in PBS with 2% FBS, washed twice, and analyzed on a flow cytometer

Statistical analysis

Data are presented as mean ± standard error of the mean (SEM) unless otherwise indicated. Statistical analyses were performed using GraphPad Prism (version 10.6.1; GraphPad, San Diego, CA, USA). For experiments involving two independent variables (genotype and treatment), two-way analysis of variance (ANOVA) was used, followed by Tukey’s multiple comparisons test for post hoc analysis. For comparisons involving a single factor, one-way ANOVA with Tukey’s post hoc test was applied. To evaluate the interactive effects of genotype (WT vs. Nfil3⁻/⁻) and probiotic treatment (HFD vs. HFD + VSL#3), two-way ANOVA was conducted, and genotype × treatment interaction terms were assessed. Phenotypic responses were classified as phenocopy or partial overlap based on the statistical equivalence between WT-HFD + VSL#3 and Nfil3⁻/⁻ HFD groups, combined with significant improvement relative to WT-HFD controls and the presence of a significant interaction effect.

Results

Comprehensive transcriptomic analysis of PBMCs revealed a distinct Nfil3-associated transcriptional reprogramming pattern between non-obese and obese individuals

Transcriptomic profiling of PBMCs revealed differential expression of Nfil3-associated genes between non-obese and overweight/obese participants. Gene set enrichment analysis volcano plot (Fig. 1A) identified a clear divergence between non-obese and obese subjects, with immune activation, TNF, IL-17, and cytokine–cytokine receptor interaction pathways markedly enriched in obesity, whereas circadian rhythm and microbial metabolite–sensing pathways were significantly downregulated. Consistently, the MA plot (Fig. 1B) displayed widespread transcriptomic remodeling, characterized by increased expression of inflammatory mediators and decreased metabolic regulators. Heatmap profiling (Fig. 1C) demonstrated strong upregulation of Nfil3, accompanied by repression of canonical circadian rhythm gene (NFIL3, CLOCK, DBP, PER2, PER3, NR1D1, NR1D2, CRY1, RORA). This expression pattern signifies a circadian inversion, consistent with stress-induced or metabolically driven Nfil3 activation. Moreover, Nfil3 expression positively correlated with TLR4, MYD88, CD14 and TICAM1, indicating engagement of innate immune signaling. SCFA-sensing transport genes, including SLC16A11, SLC16A3, FFAR3, HCAR2, and FFAR2, were consistently increased in obese PBMCs (Fig. 1C). This transcriptional repression coincided with activation of the TLR4–Nfil3 pathways, implying that increased SCFA sensing may release inhibitory control over pro-inflammatory transcriptional programs. KEGG pathway analysis (Fig. 1D) highlighted significant enrichment in cytokine receptor interaction, lipid metabolism, hematopoietic cell lineage, and TNF/IL-17 signaling cascades. The clustering of these immune-related pathways reinforces the notion that Nfil3 upregulation accompanies a systemic inflammatory bias rather than an adaptive metabolic response. Integration with KEGG circadian rhythm maps (Fig. 1E) positioned Nfil3 as a central repressor within the CLOCK–BMAL1–PER–CRY loop. Obese PBMCs showed reduced expression of PER2 and DBP, but increased expression of Nfil3, indicating a shift toward clock repression and loss of temporal homeostasis. STRING-based protein–protein interaction (PPI) network analysis (Fig. 1F) identified Nfil3 as a hub linking transcriptional regulators (NFkBIA, IL1B, CXCL2, PTGS2), bridging metabolic stress and inflammation. These interactions reveal that Nfil3 not only governs rhythmic transcription but also integrates metabolic cues and innate immune signals. Given the limited sample size and the use of peripheral immune cells, these findings reflect associative transcriptional patterns rather than definitive evidence of tissue-specific circadian or microbiota-driven regulation. Accordingly, human PBMC data were interpreted as exploratory and hypothesis-generating, providing a translational context for subsequent mechanistic analyses in mouse models.

Fig. 1.

Transcriptomic landscape reveals Nfil3-associated circadian and microbial metabolite–sensing pathway reprogramming in PBMCs from obesity-related subjects. (A) A gene set enrichment analysis (GSEA) volcano plot shows the distribution of significantly enriched pathways between non-obese (blue) and obese (red) groups. Obese subjects exhibited upregulation of inflammation-related and metabolic stress pathways, whereas circadian and SCFA-sensing pathways were downregulated. (B) MA plot displaying the global transcriptomic shift, with differentially expressed genes (DEGs; red) significantly altered in obesity. (C) Heatmap depicting representative genes involved in circadian rhythm (Nfil3, CLOCK, DBP, PER2, NR1D1, CRY1), SCFA sensing (SLC16A1, FFAR2, FFAR3, HCAR2), and TLR signaling (TLR4, MYD88, TICAM 1). Notably, Nfil3 and TLR4–MYD88 were upregulated, whereas SCFA receptors (FFAR2/3, HCAR2) and circadian activators (CLOCK, PER2) were reduced in obese PBMCs. (D) KEGG pathway enrichment of DEGs indicates activation of cytokine–cytokine receptor interaction, IL-17, and TNF signaling, highlighting immune-inflammatory skewing. (E) The KEGG circadian rhythm pathway map shows that Nfil3 is selectively activated and CLOCK–PER modules are turned off. (F) STRING-based protein–protein interaction (PPI) network analysis linking Nfil3 to immune and inflammatory nodes, including NFKBIA, IL1B, CXCL2, and PTGS2, illustrating Nfil3’s central role in coordinating inflammation and metabolic adaptation.

Microbiota-derived probiotic VSL#3 modulates host metabolic and immune phenotypes in association with Nfil3 signaling

To verify Nfil3 gene deletion, total RNA was extracted from liver tissues of WT and Nfil3^−/− mice. qRT–PCR revealed that Nfil3 transcripts were readily detected in WT mice but completely absent in Nfil3^−/− mice (Supplemental Fig. 1), confirming the effective knockout of the Nfil3 locus. The Nfil3^−/− line used in this study was generously provided by Prof. Ping-Ning Hsu, who originally established and characterized this strain. The colony has been maintained and genotyped in our facility to ensure homozygous knockout integrity.

To confirm successful administration and colonization of probiotic strains, 16S rRNA sequencing data were analyzed for canonical species contained in the VSL#3 formulation (Bifidobacterium animalis, Lactobacillus paracasei, L. acidophilus, and Streptococcus thermophilus). As shown in Supplemental Fig. 2A–B, Bifidobacterium animalis and Lactobacillus paracasei were nearly undetectable in the HFD + Water group but became clearly enriched following VSL#3 supplementation. The abundance of B. animalis increased more than 40-fold, while that of L. paracasei increased approximately 20-fold (p < 0.0001 and p < 0.05, respectively). This confirms that VSL#3-derived probiotic species successfully colonized the gut ecosystem of treated mice and were detectable in fecal microbial profiles, validating the fidelity of probiotic intake throughout the experiment.

To determine whether the probiotic VSL#3 mitigates HFD-induced metabolic dysfunction via an Nfil3-dependent mechanism, both WT and Nfil3^-/- mice were fed NCD or HFD with or without daily oral VSL#3 administration for 22 weeks (Fig. 2A–J). Body weight trajectories demonstrated a robust genotype × treatment interaction (two-way ANOVA, P = 0.0001), indicating that the effect of HFD and VSL#3 supplementation was strongly dependent on Nfil3 status (Fig. 2A–C). WT mice fed an HFD exhibited progressive and significant weight gain, which was markedly attenuated by VSL#3 supplementation. In contrast, Nfil3^-/- mice showed reduced weight gain under HFD compared with WT controls, although HFD still induced a significant increase relative to NCD-fed Nfil3^-/- mice. Notably, baseline body weight in Nfil3^-/- mice under NCD was comparable to that of VSL#3-treated WT mice, suggesting a partial phenocopy of probiotic-mediated metabolic protection.

Fig. 2.

Nfil3 deficiency partially overlaps with probiotic-mediated protection against HFD-induced metabolic dysfunction. (A) Longitudinal body weight trajectories of WT and Nfil3-/- mice fed a NCD or HFD and receiving water or VSL#3 supplementation over 22 weeks. (B) Cumulative body weight gain during dietary intervention. (C) Representative images of WT and Nfil3-/- mice under the indicated dietary and treatment conditions. (D) Daily food intake across experimental groups. (E) Fasting blood glucose levels measured at indicated time points during diet exposure. (F) Oral glucose tolerance test (OGTT) curves. (G) Quantification of the area under the curve (AUC) for OGTT. (H) Representative hematoxylin and eosin (H&E)-stained sections of white adipose tissue (WAT) and brown adipose tissue (BAT). (I) Quantification of WAT mass. (J) Quantification of BAT mass. Data are presented as mean ± SEM. n = 5–8 mice per group; each data point represents an individual mouse. Statistical significance was assessed using two-way ANOVA with genotype and treatment as factors, followed by Tukey’s multiple comparisons test. Significant genotype × treatment interactions were detected for several metabolic outcomes (see ANOVA tables). Exact p-values are reported where available; threshold notation (e.g., p < 0.0001) is used when values fall below the reporting limit

Food intake did not differ significantly across groups (Fig. 2D), indicating that observed differences in body weight were not attributable to altered caloric intake.

Consistent with these findings, fasting glucose levels and glucose tolerance tests revealed significant genotype × treatment interactions (Fig. 2E–G). WT mice developed fasting hyperglycemia and impaired glucose tolerance in response to HFD, both of which were significantly ameliorated by VSL#3 supplementation. Nfil3^−/− mice exhibited improved baseline glucose homeostasis relative to WT-HFD mice; however, HFD still induced significant deterioration in glucose tolerance in the absence of Nfil3, demonstrating incomplete protection.

Histological and quantitative analyses of adipose tissues further supported these observations (Fig. 2H–J). HFD markedly increased white adipose tissue (WAT) mass and adipocyte hypertrophy in WT mice, effects that were significantly attenuated by VSL#3 supplementation. Nfil3^−/− mice displayed reduced WAT expansion compared with WT counterparts under HFD, yet WAT mass remained responsive to dietary fat content. Brown adipose tissue (BAT) mass was also influenced by genotype and treatment, with significant genotype × treatment interactions indicating that BAT regulation involves both Nfil3-dependent and Nfil3-independent mechanisms.

Collectively, these data demonstrate that Nfil3 deficiency partially recapitulates the metabolic benefits of VSL#3 at baseline, while full protection against HFD-induced metabolic derangements requires intact Nfil3 signaling.

Nfil3 modulates the severity of HFD-induced steatohepatitis and hepatic injury

Histological examination revealed that HFD feeding induced marked steatosis, inflammatory infiltration, and collagen deposition in WT mice, as evidenced by H&E and Sirius Red staining (Fig. 3A). These pathological features were significantly attenuated by VSL#3 supplementation in WT mice. In contrast, Nfil3^−/− mice exhibited substantially reduced baseline hepatic injury and fibrosis under HFD conditions, with only modest additional improvement upon VSL#3 treatment.

Fig. 3.

Nfil3 modulates susceptibility to HFD-induced steatohepatitis and hepatic metabolic dysfunction. (A) Representative H&E and Sirius Red–stained liver sections from WT and NKO mice fed NCD or HFD with water or VSL#3 supplementation. Scale bars, 100 μm. (B) NAFLD activity score (NAS). (C) Liver weight. (D–E) Serum ALT and AST levels. (F–I) Serum lipid profiles, including total cholesterol, triglycerides, LDL, and HDL. Data are presented as mean ± SEM. n = 5–8 mice per group; each data point represents an individual mouse. Statistical significance was assessed using two-way ANOVA with genotype and treatment as factors, followed by Tukey’s multiple comparisons test. Genotype × treatment interactions were detected for selected hepatic and metabolic outcomes (see ANOVA tables), indicating endpoint-specific Nfil3-associated modulation rather than global dependence. Exact p-values are reported where available; threshold notation (e.g., p < 0.0001) is used when values fall below the reporting limit

Consistent with the histological findings, NAS scores were significantly increased in WT mice fed an HFD, whereas VSL#3 supplementation markedly reduced disease activity (Fig. 3B). In Nfil3^−/− mice, NAS scores remained low across dietary and treatment conditions, indicating partial phenocopy of probiotic-mediated protection. Two-way ANOVA revealed significant genotype × treatment interactions, supporting a genotype-dependent modulation of disease severity.

Nfil3-dependent effects of probiotic supplementation on liver weight and transaminase levels

HFD feeding significantly increased liver weight in WT mice, an effect that was partially reversed by VSL#3 supplementation (Fig. 3C). In contrast, Nfil3^−/− mice displayed attenuated liver enlargement in response to HFD, with limited responsiveness to probiotic treatment. Two-way ANOVA confirmed significant genotype × treatment interactions, indicating that the hepatoprotective effects of VSL#3 are modulated by Nfil3 status.

Serum ALT and AST levels were markedly elevated in WT mice under HFD conditions but were significantly reduced by VSL#3 administration (Fig. 3D–E). In Nfil3^−/− mice, baseline transaminase levels were substantially lower and exhibited minimal variation across dietary or probiotic conditions, suggesting that Nfil3 deficiency confers intrinsic protection against HFD-induced hepatocellular injury.

Distinct Nfil3-dependent and -independent regulation of lipid metabolic profiles

HFD feeding induced significant dyslipidemia in WT mice, characterized by increased total cholesterol, triglycerides, and LDL levels, alongside altered HDL concentrations (Fig. 3F–I). VSL#3 supplementation partially normalized these lipid parameters in WT mice. In Nfil3^−/− mice, lipid profiles were less severely perturbed by HFD, although residual diet-induced alterations persisted.

Two-way ANOVA analyses demonstrated significant genotype and treatment effects, as well as genotype × treatment interactions for multiple lipid parameters, indicating that while Nfil3 contributes to probiotic-mediated lipid regulation, additional Nfil3-independent mechanisms also participate in dietary lipid sensing and metabolism.

Nfil3-dependent probiotic modulation reshapes hepatic innate and adaptive immune remodeling under metabolic stress

To delineate the impact of probiotic supplementation on hepatic immune remodeling under metabolic stress, we performed multiparameter flow cytometric profiling of liver immune populations in WT and Nfil3^−/− mice subjected to NCD or HFD with or without VSL#3 (Fig. 4).

Fig. 4.

VSL#3 is associated with Nfil3-dependent remodeling of intrahepatic immune cell composition. (A) Flow cytometric analysis of Kupffer cells (KCs; F4/80⁺ CD11b⁻) and monocytes (MoMs; F4/80⁺ CD11b⁺) in the livers of WT and Nfil3-/- mice after 22 weeks of dietary intervention. Representative flow cytometry plots and quantification of KC and MoM frequencies are shown. (B) Analysis of Ly6C^hi (Ly6C⁺ CD11b⁺) and Ly6C^lo (Ly6C⁻ CD11b⁺) MoM subsets in the liver. Representative plots and quantification are shown. (C) Flow cytometric analysis of dendritic cells (DCs; CD11c⁺ MHCII⁺) in the liver. (D) Quantification of total T cells (Thy1.2⁺ CD3⁺) in the liver. (E) Quantification of CD4⁺ T cells (Thy1.2⁺ CD3⁺ CD4⁺) in the liver. (F) Quantification of NK cells in the liver. Data are presented as mean ± SEM. n = 3–5 mice per group; each data point represents an individual mouse. Statistical significance was assessed using two-way ANOVA with genotype and treatment as factors, followed by Tukey’s multiple comparisons test. Exact p-values are reported where available; threshold notation (p < 0.05, p < 0.01, p < 0.001, p < 0.0001) is used when values fall below the reporting limit

Innate immune remodeling

Analysis of intrahepatic macrophage subsets revealed that HFD primarily promoted monocyte-derived macrophages (MoMs; CD45⁺F4/80⁺CD11b⁺) rather than depletion of resident Kupffer cells (KCs; CD45⁺F4/80⁺CD11bˡᵒ) in WT mice (Fig. 4A). Specifically, MoMs were significantly expanded following HFD, whereas KC frequencies remained relatively stable. VSL#3 supplementation markedly suppressed MoM accumulation and partially restored macrophage balance toward a KC-dominant phenotype.

In contrast, Nfil3^−/− mice displayed a distinct macrophage landscape characterized by elevated KC proportions and consistently reduced MoM frequencies across dietary conditions. VSL#3 further augmented KC abundance in Nfil3-deficient livers, indicating that Nfil3 critically regulates macrophage fate and inflammatory infiltration under metabolic stress.

Consistent with these findings, HFD induced a robust expansion of Ly6C⁺ inflammatory monocytes in WT mice, which was significantly attenuated by VSL#3 (Fig. 4B). Ly6C⁺Ly6G⁺ granulocytic populations were concurrently reduced under HFD and partially restored by probiotic treatment. In Nfil3^−/− mice, both Ly6C⁺ monocytes and Ly6C⁺Ly6G⁺ cells were markedly diminished compared with WT controls, with only partial recovery following VSL#3, indicating impaired inflammatory myeloid mobilization in the absence of Nfil3.

We next assessed hepatic dendritic cells (DCs; CD45⁺CD11c⁺MHCII⁺), representing a key antigen-presenting compartment (Fig. 4C). HFD significantly increased DC frequencies in WT mice, whereas VSL#3 attenuated this expansion. By contrast, Nfil3^−/− mice exhibited persistently reduced DC levels across all conditions, suggesting that Nfil3 is required for DC accumulation in response to metabolic challenge and microbiota modulation.

Adaptive immune remodeling

Flow cytometric analysis of hepatic lymphocyte populations revealed that HFD feeding markedly reshaped adaptive immune composition in the liver in a genotype-dependent manner. In WT mice, HFD significantly reduced hepatic T cell frequencies while moderately increasing B cell proportions compared with NCD controls. Notably, VSL#3 supplementation partially restored T cell abundance under HFD conditions, indicating probiotic-mediated preservation of hepatic T cell homeostasis. In contrast, Nfil3^−/− mice exhibited a profoundly altered immune response to HFD, characterized by a pronounced expansion of hepatic B cells and a concomitant reduction in T cells relative to WT counterparts. This HFD-induced B cell accumulation was significantly attenuated by VSL#3 treatment, accompanied by partial recovery of T cell frequencies (Fig. 4D).

Further subset analysis demonstrated that HFD markedly reduced CD4⁺ T cells in WT mice, an effect significantly reversed by VSL#3. In contrast, CD4⁺ T cell frequencies in Nfil3-deficient mice remained blunted regardless of diet or probiotic intervention. CD8⁺ T cells showed modest elevation under HFD in WT mice and normalization following VSL#3, whereas Nfil3^−/− mice exhibited constitutively elevated CD8⁺ T cell proportions independent of treatment (Fig. 4E).

Collectively, these findings demonstrate that Nfil3 orchestrates hepatic immune homeostasis under metabolic stress by coordinating inflammatory myeloid recruitment, antigen-presenting capacity, and adaptive lymphocyte balance. Probiotic VSL#3 modulates this axis by restraining inflammatory innate responses and partially restoring T cell homeostasis, effects that are markedly compromised in the absence of Nfil3.

VSL#3 remodels HFD-induced gut dysbiosis and restores SCFA-associated microbial networks

Diet and VSL#3 distinctly reshape global microbial community structure

Microbiota remodeling by VSL#3 restores the gut–liver axis and reestablishes SCFA-associated microbial networks under HFD conditions. Beta-diversity analysis demonstrated clear separation between NCD and HFD groups along the PC1 axis (48.3%), indicating a dominant dietary effect on microbial community structure (Fig. 5A). Notably, the HFD.VSL group exhibited a partial shift from the HFD.Water (W) cluster, suggesting probiotic-mediated modulation of gut microbiota.

Fig. 5.

VSL#3 reshapes gut microbial community composition under HFD. (A) Principal coordinates analysis (PCoA) based on weighted UniFrac distances showing overall differences in gut microbial community structure among mice fed a normal chow diet (NCD) or high-fat diet (HFD) with water or VSL#3 supplementation. (B) Venn diagram illustrating shared and unique operational taxonomic units (OTUs) among the indicated dietary and treatment groups. (C) Relative abundance of bacterial families across experimental groups. (D) Relative abundance of selected bacterial species across samples. (E) Heatmap depicting the relative abundance of the top 50 bacterial species across individual mice. (F) UpSet plot showing intersections of bacterial genera among different dietary and treatment conditions. (G) Cladogram generated by linear discriminant analysis effect size (LEfSe) analysis identifying taxa with differential relative abundance across groups. (H) LDA score bar plot highlighting taxa with significant differential abundance among groups. n = 5 mice per group

Venn diagram analyses revealed a shared core microbiota comprising 151 ASVs, while the NCD and HFD groups retained distinct microbial signatures, reflecting diet-dependent community restructuring. Venn diagram analysis showed that HFD reduced total ASV richness, whereas VSL#3 increased the number of shared ASVs with the NCD group, reflecting enhanced microbial diversity and stability (Fig. 5B).

VSL#3 reshapes HFD-altered microbial composition and community membership

At the family level, HFD feeding reduced SCFA-associated families including Lachnospiraceae and Ruminococcaceae, while increasing Enterobacteriaceae, a taxon linked to endotoxin production and inflammation (Fig. 5C).

Species-level profiling revealed marked compositional differences across experimental groups, accompanied by substantial inter-individual variability. In NCD.W mice, microbial communities were primarily composed of Kineothrix alysoides together with unclassified taxa (“Other”), while Muribaculum intestinale and Blautia luti were variably represented among individual animals. HFD.W mice exhibited a distinct shift in species composition, characterized by increased relative abundance of Kineothrix alysoides and expansion of Acetatifactor muris in multiple individuals. In addition, Faecalibaculum rodentium emerged in several HFD-fed mice, indicating diet-associated restructuring of species composition with pronounced mouse-to-mouse heterogeneity. In NCD.VSL mice, VSL#3 supplementation consistently reshaped microbial profiles toward Muribaculum intestinale–dominant communities, accompanied by a relative reduction in Kineothrix alysoides and unclassified taxa, demonstrating a reproducible probiotic-induced compositional shift under normal dietary conditions. In contrast, HFD.VSL mice displayed a more heterogeneous response to VSL#3. Most animals retained high proportions of Kineothrix alysoides, while Acetatifactor muris and Oscillibacter valericigenes remained detectable across multiple individuals. Although Muribaculum intestinale increased in some HFD.VSL mice, its enrichment was inconsistent across the cohort (Fig. 5D). Collectively, these data indicate that HFD induces species-level microbial restructuring with increased inter-individual variability, whereas VSL#3 produces a consistent shift toward Muribaculum intestinale dominance under NCD but elicits a more variable compositional response under HFD.

Heatmap-based visualization of diet- and probiotic-associated microbial signatures

Hierarchical clustering of differentially abundant taxa further illustrated diet- and probiotic-dependent microbial patterns (Fig. 5E). Samples segregated primarily according to dietary background, with secondary clustering driven by VSL#3 supplementation.

Figure 5E visualizes row-wise hierarchical clustering of selected taxa and highlights group-associated abundance patterns across samples (columns arranged by experimental groups). Under NCD.VSL, Muribaculaceae-related features (including Muribaculum and Muribaculum intestinale) show broadly elevated signals across most animals, whereas HFD.VSL samples exhibit higher signals for Oscillibacter/Oscillibacter valericigenes and Acetatifactor/Acetatifactor muris compared with other groups. In HFD.W, increased signals are more apparent for several Erysipelotrichia/Romboutsia-related lineages, although substantial inter-individual variability is observed. Overall, the heatmap supports diet- and probiotic-associated shifts in specific taxonomic modules, with consistent enrichment of Muribaculaceae under NCD.VSL and a distinct, more variable pattern under HFD.VSL.

The UpSet plot (Fig. 5F) summarizes the intersections of amplicon sequence variants (ASVs) among the four experimental groups and highlights both the shared core microbiota and group-specific microbial features. A substantial set of ASVs was shared across all groups, indicating the presence of a conserved core community that persists regardless of diet or probiotic supplementation. In addition to this shared fraction, each group retained distinct ASV subsets, reflecting diet- and treatment-associated restructuring of microbial membership. Consistent with the beta-diversity separation observed in Fig. 5A, the intersection patterns in Fig. 5F indicated that HFD feeding was associated with a redistribution of ASV membership, with HFD.W displaying a characteristic set of features that were not broadly shared with NCD groups. Importantly, VSL#3 supplementation under HFD conditions altered these intersection profiles: the HFD.VSL group exhibited increased overlap with VSL-treated controls and/or NCD-associated feature sets while also containing unique ASVs not present in HFD.W. Together, these results suggest that VSL#3 modulates microbial community membership under dietary stress by reshaping the pool of detectable ASVs promoting convergence toward a probiotic-responsive microbiome while retaining treatment-specific features.

Linear Discriminant Analysis Effect Size (LEfSe)-based identification of diet- and probiotic-associated discriminative taxa

LEfSe identified distinct discriminative taxa across groups (Fig. 5G and H). NCD.W mice were characterized by enrichment of multiple Clostridia/Lachnospiraceae-associated taxa, including Blautia (and Blautia luti), Roseburia (and Roseburia faecis), Lacrimispora (including Lacrimispora saccharolytica), Anaerocolumna (including Anaerocolumna cellulosilytica), and Faecalicatena (including Faecalicatena contorta). Under NCD conditions, VSL#3 supplementation was associated with a Bacteroidetes/Muribaculaceae-dominant signature, with Muribaculaceae, Muribaculum, and Muribaculum intestinale showing high LDA scores, together with Tenericutes/Mollicutes- and Spiroplasma-related features and Anaerobacterium chartisolvens. HFD.W mice displayed discriminative enrichment of Kineothrix (including Kineothrix alysoides), Romboutsia (including Romboutsia ilealis), Peptostreptococcaceae, and Parabacteroidetes-associated taxa (including Parabacteroides and Parabacteroides goldsteinii). In contrast, HFD.VSL mice were characterized by enrichment of Firmicutes-associated taxa, including Acetatifactor muris, Oscillibacter valericigenes, Faecalibaculum rodentium, Erysipelotrichia-related lineages, and Peptococcus niger, indicating a diet-dependent microbial signature under probiotic supplementation.

Reinforcement of the gut barrier by VSL#3 completes the microbiota–Nfil3–liver axis, preventing microbial leakage and systemic inflammation in HFD-fed mice

To investigate whether probiotic-mediated hepatic immune remodeling was accompanied by alterations in intestinal integrity, we next evaluated gut morphology, epithelial barrier function, bacterial translocation, and intestinal immune populations in WT and Nfil3^−/− mice subjected to NCD or HFD with or without VSL#3 (Fig. 6). HFD feeding significantly shortened total gut length and colon length in WT mice, whereas VSL#3 supplementation partially restored intestinal length under HFD conditions (Fig. 6A–B). In contrast, Nfil3^−/− mice exhibited attenuated gut shortening responses, and probiotic treatment exerted only modest effects, suggesting Nfil3-dependent regulation of diet-induced intestinal structural alterations. Histological analysis of ileal and colonic tissues revealed that HFD induced villus blunting and architectural disruption in WT mice, accompanied by reduced goblet cell abundance (Fig. 6C–D). These pathological changes were markedly ameliorated by VSL#3 supplementation. Conversely, Nfil3^−/− was associated with preserved villus morphology and goblet cell numbers across dietary conditions, with limited additional benefit from probiotic treatment. Consistent with these morphological findings, HFD markedly increased intestinal permeability in WT mice, as assessed by FITC–dextran flux using Ussing chamber, whereas VSL#3 significantly attenuated this barrier disruption (Fig. 6E). In Nfil3^−/− mice, baseline permeability remained comparatively low and was minimally affected by HFD or probiotic intervention, indicating intrinsic resistance to HFD-induced barrier dysfunction.

Fig. 6.

VSL#3 modulates intestinal barrier–associated phenotypes and bacterial translocation in an Nfil3-associated manner. (A) Total gut length measured in WT and Nfil3-/- mice after 22 weeks of dietary intervention. (B) Colon length measured in WT and Nfil3-/- mice. (C–D) Representative Alcian Blue/Periodic Acid–Schiff (AB/PAS)–stained sections of the ileum and colon from WT and NKO mice, illustrating diet- and treatment-associated differences in mucus-producing goblet cells and epithelial morphology. (E) Intestinal permeability assessed by FITC–dextran assay, expressed as serum FITC–dextran levels. (F) Bacterial translocation to mesenteric lymph nodes (MLNs) and spleen, quantified as colony-forming units per gram of tissue (CFU/g). Representative agar plates and quantification are shown. (G) Flow cytometric analysis of dendritic cells (DCs; CD11c⁺ MHCII⁺) in the ileum and colon. Data are presented as mean ± SEM. n = 5–6 mice per group; each data point represents an individual mouse. Statistical significance was assessed using two-way ANOVA with genotype and treatment as factors, followed by Tukey’s multiple comparisons test. Exact p-values are reported where available; threshold notation (P < 0.05, P < 0.01, P < 0.001, P < 0.0001) is used when values fall below the reporting limit

To determine whether barrier disruption translated into systemic bacterial dissemination, we quantified bacterial colony-forming units (CFUs) in mesenteric lymph nodes (MLNs) and spleen (Fig. 6F). WT mice exhibited pronounced bacterial translocation under HFD, which was substantially reduced by VSL#3. In contrast, Nfil3^−/− mice showed markedly lower CFU levels across all conditions, with minimal probiotic responsiveness. Finally, we examined intestinal dendritic cells (DCs) in the ileum and colon. HFD significantly increased DC frequencies in WT mice, whereas VSL#3 normalized this expansion (Fig. 6G–H). In Nfil3^−/− mice, intestinal DC populations remained consistently low and largely unresponsive to dietary or probiotic modulation. Collectively, these data demonstrate that Nfil3 critically governs intestinal barrier vulnerability, bacterial translocation, and mucosal immune activation under metabolic stress. Probiotic VSL#3 preserves epithelial integrity and suppresses microbial dissemination in an Nfil3-dependent manner, linking gut barrier protection to downstream hepatic immune remodeling.

Functional restoration of the gut microbiota by VSL#3 reestablishes the microbiota–metabolite–Nfil3–liver axis disrupted by HFD

High-fat diet profoundly reprograms microbial metabolic functions toward lipid utilization and inflammation

Consistent with the diet-driven functional remodeling of the gut microbiome, pathway-level prediction revealed a clear separation between NCD Water and HFD Water mice (Fig. 7A). Under NCD, multiple core metabolic programs were significantly enriched, including porphyrin and chlorophyll metabolism (strongest positive effect, P = 9.65 × 10⁻⁵), pyruvate metabolism (P = 1.70 × 10⁻³), pentose and glucuronate interconversions (P = 2.62 × 10⁻³), methane metabolism (P = 1.32 × 10⁻⁴), and phenylalanine/tyrosine/tryptophan biosynthesis (P = 3.43 × 10⁻⁴). NCD mice also showed higher representation of pathways linked to biosynthetic capacity and cellular maintenance, including the pentose phosphate pathway (P = 7.69 × 10⁻³), fatty acid biosynthesis (p = 0.020), lipid biosynthesis proteins (P = 7.46 × 10⁻⁵), components related to chromosome/cytoskeleton-associated functions (P = 0.021 and 0.049, respectively). Collectively, these signatures suggest that the NCD-associated microbiome maintains relatively higher predicted metabolic versatility and anabolic potential.

Fig. 7.

Predictive functional profiling indicates diet- and probiotic-associated shifts in gut microbial functional potential. (A) Differentially enriched KEGG level 3 pathways predicted by PICRUSt2 in gut microbiota from mice fed a normal chow diet with water (NCD.W) or a high-fat diet with water (HFD.W). (B) Changes in predicted microbial functional pathways following VSL#3 supplementation in HFD-fed mice (HFD.W vs. HFD.VSL), as inferred by PICRUSt2 analysis. Pathways are grouped according to functional categories associated with amino acid metabolism, SCFA-related pathways, and antioxidant-associated processes. These results reflect predicted functional potential inferred from microbial community composition, rather than direct measurements of metabolic activity or metabolite production. Error bars represent 95% confidence intervals of mean relative pathway abundance. Statistical significance was assessed using Welch’s t-test with Benjamini–Hochberg false discovery rate correction (q < 0.05). n = 5 mice per group

In contrast, HFD Water mice exhibited enrichment of functions typically associated with carbohydrate uptake/processing and inflammatory potential, including the phosphotransferase system (PTS) (P = 0.037), amino sugar and nucleotide sugar metabolism (p = 0.011), fructose and mannose metabolism (P = 0.038), and lipopolysaccharide (LPS) biosynthesis proteins (P = 0.038). Additional HFD-enriched signatures included glycosyltransferases (P = 2.80 × 10⁻⁵), inorganic ion transport and metabolism (P = 4.83 × 10⁻⁴), and broad amino acid metabolism (P = 1.93 × 10⁻⁴), together indicating a functional shift toward nutrient acquisition and envelope/glycan-related processes under HFD feeding.

Notably, several KEGG “disease” and host-associated annotation categories (e.g., RIG-I-like receptor signaling, influenza A, viral myocarditis, p53 signaling, and cancer-related terms) were also detected as differentially represented. These categories should be interpreted cautiously as annotation proxies within predictive functional mapping rather than as direct evidence of host disease pathway activation, but they nonetheless support a broad HFD-associated remodeling of predicted microbial functional potential.

Within the HFD background, probiotic supplementation induced a focused but detectable shift in predicted microbial functional capacity (Fig. 7B). Several pathways exhibited negative effect sizes (HFD Water − HFD VSL#3 < 0), indicating higher representation in HFD VSL#3 mice. These included porphyrin and chlorophyll metabolism (P = 0.045), cysteine and methionine metabolism (P = 0.025), and translation proteins (P = 0.040), suggesting that VSL#3 is associated with increased predicted biosynthetic/housekeeping functions and amino acid–related metabolic potential under HFD conditions.

Conversely, pathways showing positive effect sizes were relatively enriched in HFD Water mice, including glycosyltransferases (P = 0.047) and inorganic ion transport and metabolism (P = 0.041). This pattern is consistent with a partial attenuation of HFD-associated envelope/glycan-related functional signatures following probiotic intervention.

Notably, several KEGG “disease” or host-associated annotation categories (e.g., toxoplasmosis, small cell lung cancer, p53 signaling, influenza A, viral myocarditis, colorectal cancer) appeared among nominally significant features (all P ≈ 0.047) but displayed effect sizes close to zero. These terms should be interpreted cautiously as annotation proxies within predictive functional mapping rather than direct evidence of host pathway activation; nonetheless, their detection is compatible with broader functional reweighting of predicted microbial gene content under VSL#3 supplementation.

Collectively, these results indicate predicted alterations in microbial functional potential associated with dietary and probiotic interventions, rather than direct evidence of altered metabolic output or metabolite production.

Dietary and probiotic interventions reshape predicted gut microbial metabolic pathways

To further explore functional alterations in gut microbial communities, we performed predictive metagenomic pathway analysis based on 16S rRNA sequencing data (Supplemental Fig. 3).

A comprehensive comparison across 130 KEGG modules (Supplemental Fig. 3A–B) demonstrated that VSL#3 not only reactivated carbohydrate-fermentation and amino-acid pathways but also elevated energy-linked electron-transfer carriers, pantothenate and CoA biosynthesis, and peroxisome-related oxidation modules—metabolic signatures tied to enhanced oxidative resilience. In contrast, HFD-associated increases in nitrotoluene degradation, xylene/dioxin detoxification, and RIG-I-like receptor signaling—indicative of xenobiotic stress and immune activation—were markedly diminished following probiotic intervention. Together, these extended data confirm that VSL#3 globally rewires the gut microbial metabolic network from HFD stress-induced catabolism to biosynthetic and antioxidant activity.

Discussion

Nfil3 has previously been shown to play a functionally relevant role in the development of MASLD, as demonstrated in our recent study [24]. These observations suggest phenotype-specific Nfil3-associated regulation rather than uniform dependence across all metabolic and immune outcomes. Building upon these findings, the present study identified Nfil3 as an important transcriptional integrator linking circadian timing, microbial–metabolite signaling, and immune–metabolic homeostasis. By integrating exploratory human PBMC transcriptomics with complementary mouse models involving genetic deletion and probiotic intervention, we provide evidence that Nfil3 participates in coupling gut microbial activity to host circadian and metabolic regulation. Collectively, our data support the existence of a microbiota–Nfil3–liver regulatory axis that coordinates intestinal epithial barrier integrity, hepatic lipid metabolism, and systemic inflammation under HFD stress.

Nfil3 as a circadian–immune integrator in metabolic inflammation

Transcriptomic reprogramming in PBMCs from obese individuals highlighted Nfil3 as a pivotal mediator of metabolic inflammation. Normally, under oscillatory CLOCK–BMAL1 control, Nfil3 becomes persistently elevated during metabolic or inflammatory stress, thereby suppressing canonical circadian activators (CLOCK, DBP, PER2) and disturbing temporal regulation of energy use, hormone secretion, and immune rhythms [17, 72]. The inverse Nfil3–CLOCK/PER relationship observed here reflects a “circadian inversion”, a molecular hallmark of chronic metabolic disturbance [17, 72].

Concordantly, we observed downregulation of SCFA-sensing/transport genes (FFAR2, FFAR3, HCAR2, SLC16A1), indicating attenuated host–microbiota metabolic communication. SCFAs especially butyrate and propionate normally act as GPCR ligands and epigenetic modulators to enforce immune tolerance and phase coherence [30, 31]. Reduced SCFA signaling likely removes inhibitory control over the TLR4–MYD88–Nfil3 axis, promoting low-grade inflammation. KEGG and PPI analyses (Nfil3–NFKB1A-IL1B-JUN-CXCL2) supported a feed-forward circuit whereby metabolic stress activates cytokine/TLR pathways to induce Nfil3, which then represses PER/DBP programs and sustains inflammatory transcription—effectively locking circulating immune cells into a rhythm-independent inflammatory state.

Nfil3 determines the metabolic efficacy of probiotic intervention

Using Nfil3^−/− mice, we examined the contribution of Nfil3 to the metabolic benefits associated with probiotic VSL#3 supplementation. In wild-type mice, VSL#3 significantly reduced HFD-induced obesity, improved glucose tolerance, enhanced thermogenic remodeling of adipose tissue, and ameliorated hepatic steatosis. These effects were partially attenuated in Nfil3^−/− mice, indicating that probiotic efficacy is significantly influenced by Nfil3 signaling but not exclusively dependent upon it.

Nfil3 integrates hormonal, microbial, and inflammatory inputs to regulate lipid and glucose metabolism. Under HFD stress, sustained Nfil3 activation is associated with repression of PER2 and DBP and upregulation of lipogenic regulators such as SREBP1c and PPARγ, contributing to hepatic steatosis and insulin resistance [23]. VSL#3 supplementation reduced Nfil3 overexpression and was associated with reactivation of metabolic clock components and improved insulin sensitivity, consistent with SCFA-mediated modulation of peripheral clock gene expression through FFAR2/3 and HCAR2 [24, 30, 73]. Taken together, these findings support a model in which Nfil3 functions as a context-dependent regulatory integrator that modulates host metabolic responses under dietary and microbial perturbations.

Nfil3 integrates dietary and microbiota-derived signals to modulate metabolic homeostasis

Across genotypes, HFD markedly increased body weight despite comparable food intake; VSL#3 significantly blunted weight gain in both WT and Nfil3^−/− mice, indicating effects on energy expenditure and/or nutrient handling rather than caloric intake per se. Fasting glycemia and glucose tolerance improved with VSL#3 (Fig. 2), consistent with probiotic modulation of insulin sensitivity via gut microbial signaling and systemic inflammation control [39, 49, 74, 75]. These data align with our Nfil3-centric model: restoring microbial metabolites that repress Nfil3 re-entrains peripheral metabolic rhythms, thereby improving whole-body glucose homeostasis.

The present study identifies Nfil3 as a central transcriptional regulator that integrates dietary fat exposure and microbiota-derived signals to shape host metabolic responses. Using a factorial experimental design, we demonstrate that both HFD and probiotic supplementation exert genotype-dependent effects, highlighting Nfil3 as a key determinant of metabolic adaptability.

Although Nfil3^-/- mice exhibited baseline metabolic phenotypes comparable to those of VSL#3-treated WT mice, this similarity was not universal across all endpoints. Our data indicate that Nfil3 deficiency partially phenocopies probiotic-mediated protection, particularly with respect to body weight gain, adiposity, and glucose homeostasis, while failing to confer complete resistance to HFD-induced metabolic stress.

By applying two-way ANOVA across major metabolic parameters, we uncovered significant genotype × treatment interactions, underscoring that the metabolic benefits of VSL#3 are contingent upon intact Nfil3 signaling. These interaction effects provide quantitative support for Nfil3-dependent probiotic responsiveness, while also delineating phenotypes that retain Nfil3-independent dietary sensitivity.

Despite the protective baseline phenotype observed in Nfil3^-/- mice, HFD feeding continued to induce metabolic deterioration, indicating the presence of Nfil3-independent regulatory mechanisms. This residual responsiveness suggests that parallel transcriptional or metabolic pathways remain capable of sensing dietary lipid excess independently of Nfil3.

It should be noted that VSL#3 is a multi-strain probiotic formulation with pleiotropic biological effects, and the beneficial outcomes observed in this study cannot be attributed to individual microbial taxa or specific metabolites. Rather than serving as a strain-specific therapeutic agent, VSL#3 was employed here as a consortium-level perturbation to interrogate host responsiveness to complex microbiota modulation under metabolic stress.

Accordingly, the mechanistic insights presented in this study are best interpreted as reflecting host regulatory pathway engagement, particularly Nfil3-associated immune–metabolic integration, in response to coordinated microbial community shifts. These findings should not be directly generalized to other probiotic formulations or single-strain interventions, as differences in microbial composition, dosing, and host context may yield distinct biological outcomes. Future studies employing defined microbial consortia or metabolite-level manipulation will be required to refine strain- or metabolite-specific mechanisms underlying probiotic–host interactions.

These findings have important implications for understanding inter-individual variability in response to probiotic interventions. Our data support a model in which Nfil3 acts as a host-intrinsic gatekeeper that modulates the efficacy of microbiota-based therapies, providing a mechanistic framework for personalized approaches to metabolic disease intervention.

Hepatic injury, lipid metabolism, and fibrosis

HFD provoked hepatomegaly, lipid accumulation, fibrosis, and elevations of ALT/AST; VSL#3 ameliorated these readouts and improved lipid profiles (↓cholesterol, TG, LDL; ↑HDL), consistent with enhanced lipid catabolism, reduced de novo lipogenesis, and/or improved bile acid turnover (Figs. 2 and 3). These findings mirror prior reports of VSL#3-mediated lipid lowering and attenuation of fibrosis in diet-induced MASH models [49, 50, 76–78]. Within our framework, probiotic re-establishment of SCFA/indole signaling reduces Nfil3 hyperactivation in the liver, relieving steatogenic transcriptional pressure and dampening inflammatory tone.

Intrahepatic immune cell remodeling under metabolic stress

HFD feeding was associated with marked remodeling of intrahepatic immune cell composition, consistent with a shift toward an inflammatory immune landscape. Using flow cytometric profiling, we identified multiple myeloid populations corresponding to previously described liver-resident macrophages and monocyte-derived subsets [79, 80]. Among these, Ly6C⁺ monocytes, which are commonly expanded during inflammatory conditions [81, 82], as well as CD8⁺ T cells, were increased following HFD exposure, in line with immune recruitment patterns reported in metabolic and inflammatory liver injury models [83–85].

VSL#3 supplementation was associated with a reduction in the abundance of Ly6C⁺ monocytes and CD8⁺ T cells, alongside improvements in metabolic and hepatic parameters. These findings suggest that probiotic intervention is accompanied by broad immune compositional changes across innate and adaptive compartments, rather than providing direct evidence of immune cell functional polarization.

Consistent with this interpretation, our flow cytometric analyses demonstrated that HFD increased the relative abundance of Kupffer cells (KCs), monocyte-derived macrophages (MoMs), Ly6C^hi/lo monocytes, dendritic cells, total T cells, and CD4⁺ T cells in both genotypes. VSL#3 treatment was associated with a general attenuation of these HFD-associated immune expansions, although the magnitude of probiotic-associated changes differed between WT and Nfil3^−/− mice (Fig. 4). Importantly, these data reflect immune compositional remodeling rather than direct assessment of macrophage or T-cell functional states, such as polarization or cytokine production.

Together, these results support a role for Nfil3-associated regulation in shaping the immune cellular landscape under metabolic stress, while underscoring the need for future functional studies to define the precise immunological mechanisms through which probiotic interventions influence metabolic liver disease progression.

Probiotic modulation of hepatic metabolism through the Nfil3 axis

At the hepatic level, VSL#3 markedly alleviated steatosis, hepatocyte ballooning, and inflammatory infiltration in WT but not Nfil3^−/− mice. Nfil3 directly binds to the promoters of ApoB, Cyp7a1, Fasn, and Acc1, thereby regulating lipid synthesis, β-oxidation, and bile acid metabolism [86]. Persistent Nfil3 activation under metabolic stress promotes TG accumulation, whereas VSL#3-mediated attenuation of Nfil3 reinstated lipid oxidation and bile acid turnover, improving hepatic transaminases and lipid profiles. Furthermore, Nfil3 interacts with glucocorticoid and TLR4–MYD88 signaling [17, 87], pathways sensitive to microbial endotoxin load. By reducing intestinal epithelial permeability and bacteria translocation (Fig. 6), VSL#3 indirectly suppressed hepatic TLR4–Nfil3 signaling, restoring KC homeostasis. These findings highlight a gut–liver immune–metabolic circuit, where probiotic-induced microbiota rebalancing stabilizes hepatic Nfil3 transcriptional activity and prevents metabolic inflammation.

Microbiota reconfiguration establishes a metabolically favorable ecosystem

HFD feeding caused a profound shift in gut microbial composition, characterized by loss of beneficial commensals (Faecalibaculum rodentium, Akkermansia muciniphila, Muribaculum intestinale) and expansion of endotoxin-producing taxa (Desulfovibrionaceae, Romboutsia). VSL#3 supplementation reversed this dysbiosis, enriching SCFA- and mucin-associated bacteria while suppressing LPS-producing species [88–90]. These compositional improvements translated into structural and functional recovery of the intestinal epithelial barrier (Fig. 6). VSL#3 restored mucosal length, and goblet cell abundance, consistent with reduced epithelial permeability and bacterial translocation. Mechanistically, Nfil3 acts as a transcriptional repressor of Atoh1 and Muc2, modulating goblet cell differentiation and mucus renewal [28]. The preserved goblet cells and reduced intestinal epithelial permeability in Nfil3^−/− mice confirm that Nfil3 overactivation disrupts gut barrier function. Therefore, VSL#3 exerts gut barrier protection by rebalancing Nfil3 activity through SCFA-mediated signaling, preventing endotoxin leakage and subsequent hepatic inflammation.

Predicted microbial functional shifts associated with the SCFA–Nfil3–energy regulatory framework

Predictive functional profiling was performed to explore how VSL#3 supplementation is associated with changes in microbial functional potential under HFD conditions. Using PICRUSt2-based inference, HFD feeding was associated with a predicted enrichment of pathways related to lipid biosynthesis, glycosyltransferase activity, and LPS biosynthesis, alongside suppression of pathways linked to pyruvate metabolism, the pentose phosphate pathway, tryptophan metabolism, and branched-chain amino acid metabolism. These predicted shifts are consistent with prior reports describing HFD-associated microbial communities enriched in inflammatory and nutrient-adaptive functional signatures, although they do not represent direct measurements of metabolic activity or metabolite production [30, 91].

VSL#3 supplementation was associated with partial normalization of predicted microbial functional profiles, including enrichment of pathways related to cysteine–methionine metabolism, porphyrin-related pathways, and translational processes, which are functionally linked to redox balance and cellular metabolic efficiency [34]. Concurrently, pathways associated with glycosyltransferases, secretion systems, and toxin-related functions were predicted to be reduced, whereas pathways related to energy utilization and antioxidant-associated processes (e.g., pantothenate–CoA biosynthesis and electron carrier metabolism) showed relative enrichment (Supplementary Fig. 3).

Importantly, these results reflect predicted microbial functional potential inferred from community composition, rather than experimentally validated metabolic flux or metabolite availability. Accordingly, proposed links between microbial metabolites—such as SCFA or tryptophan-derived indoles—and host Nfil3 regulation should be interpreted as conceptual and associative. Prior studies have demonstrated that SCFAs can modulate host gene expression through histone deacetylase inhibition and signaling via FFAR2/3 and HCAR2, and that indole derivatives may engage AHR-dependent pathways [28, 31]. However, SCFA concentrations were not directly quantified in the present study, and circadian rhythmicity was not assessed across multiple ZTs.

Within these constraints, our data support a model in which diet- and probiotic-associated shifts in microbial functional potential coincide with altered Nfil3-associated immune–metabolic states, assessed at a defined circadian phase. In Nfil3^−/− mice, several HFD-associated metabolic and inflammatory phenotypes were attenuated independently of probiotic supplementation, consistent with a role for Nfil3 as an integrative regulatory node rather than a sole causal driver. Together, these findings support a framework in which microbiota-associated functional cues and Nfil3-dependent host pathways interact at the level of metabolic and immune regulation, without directly establishing metabolite-level causality or circadian oscillatory control.

An integrated microbiota–Nfil3–liver regulatory framework

Integrating findings across human PBMC transcriptomics, microbiota profiling, and genetic mouse models, our data support a conceptual regulatory framework linking diet-associated microbial functional shifts to Nfil3-dependent host immune–metabolic states. Under HFD conditions, microbial community alterations were associated with predicted changes in functional pathways related to SCFA- and tryptophan-associated metabolism, as well as pathways linked to LPS biosynthesis, coinciding with increased intestinal epithelial permeability and hepatic inflammation. These associations are consistent with prior models of diet-induced microbiota–host inflammatory crosstalk, but do not establish direct metabolite-level causality.

VSL#3 supplementation was associated with partial normalization of microbial composition and predicted functional potential, alongside attenuation of gut barrier dysfunction, immune activation, and hepatic injury in WT mice. In this context, Nfil3 emerges as an integrative regulatory node, modulating host transcriptional and physiological responses to combined dietary, microbial, and circadian-associated cues assessed at a defined circadian phase. Rather than acting as a sole upstream driver, Nfil3 likely functions in a permissive and context-dependent manner, shaping immune and metabolic responsiveness to microbiota-associated signals.

Importantly, the proposed microbiota–metabolite–Nfil3–liver axis should be interpreted as a conceptual framework, in which predicted microbial functional cues and host transcriptional programs converge to influence epithelial integrity and hepatic immune–metabolic homeostasis. Direct validation of specific microbial metabolites, causal signaling pathways, and circadian oscillatory dynamics will require future studies incorporating targeted metabolomics and multi–ZT-point analyses.

Limitations

Several limitations of the present study warrant consideration. First, the human PBMC transcriptomic analysis was based on a relatively small cohort, which limits statistical power and increases the risk of overinterpretation of pathway-level inferences. Accordingly, the human data are presented as exploratory and hypothesis-generating, rather than as definitive evidence of Nfil3 dysregulation in human obesity.

Second, the PBMC transcriptomic profiling was performed at a single, standardized morning time point, which precludes formal assessment of circadian rhythmicity or phase-dependent oscillations in gene expression. As such, circadian-related interpretations in this study are based on static, time-point–specific transcriptional differences rather than direct measurements of dynamic rhythmic regulation.

Third, PBMCs reflect systemic immune-associated transcriptional states and cannot be directly extrapolated to tissue-specific regulation in the liver or intestine. Therefore, mechanistic conclusions regarding gut–liver metabolic homeostasis and circadian–metabolic interactions are primarily derived from mouse models, in which organ-level analyses were performed.

Fourth, although Nfil3 deficiency and probiotic intervention were associated with broad metabolic and immune remodeling, Nfil3 is best interpreted as a regulatory node integrating dietary, microbial, and circadian-associated signals, rather than as a sole causal driver of disease. Future studies incorporating time-resolved sampling across multiple circadian phases and tissue-specific functional analyses will be required to further delineate the dynamic regulatory role of Nfil3. Together, our findings support a model in which Nfil3 integrates circadian-associated signals with microbial and metabolic inputs at a defined circadian phase, without directly addressing circadian rhythmicity.

Future studies employing ex vivo manipulation of primary hepatocytes or immune cells with defined SCFA species, in combination with Nfil3 perturbation, will be essential to establish causal links between microbial metabolites and Nfil3-mediated metabolic regulation.

Finally, although sex-dependent effects of Nfil3 have been reported previously [24], the present study was not designed to resolve sex-specific differences in probiotic responsiveness or circadian regulation, representing an important direction for future investigations.

Translational implications

These findings provide a conceptual and mechanistic framework for understanding how Nfil3 integrates microbial metabolic cues with circadian transcriptional regulation in metabolic disease contexts. Rather than establishing Nfil3 as a direct therapeutic target, our data suggest that microbiota-associated modulation of Nfil3 rhythmicity represents a tractable biological axis through which host metabolic homeostasis may be influenced.

In this regard, probiotic interventions that are predicted to enrich SCFA- and tryptophan-related microbial pathways may contribute to the realignment of Nfil3-associated transcriptional programs, particularly under conditions of dietary or circadian stress. Likewise, chrononutritional approaches such as time-restricted feeding or controlled light–dark exposure may interact with microbial signals to modulate Nfil3 responsiveness, although such interactions remain to be formally tested.

Importantly, the present study is preclinical and does not establish causality at the metabolite or receptor level. Future investigations integrating circadian-phase–resolved sampling, direct quantification of microbial metabolites, and cell-type–specific mapping of the Nfil3 cistrome in gut and hepatic tissues will be required to evaluate the translational relevance of the microbiota–Nfil3 axis. Controlled human studies will ultimately be necessary to determine whether modulation of this pathway can be safely and effectively leveraged in metabolic disease settings.

Conclusions

This study supports a role for Nfil3 as an important regulatory integrator linking microbial metabolic programs to host circadian-associated immune and metabolic responses. HFD-induced dysbiosis disrupts microbiota–host communication and is associated with altered Nfil3-related transcriptional patterns, circadian misalignment, impaired gut epithial barrier integrity, and metabolic inflammation.

Using VSL#3 as a mechanistic probe, we show that modulation of gut microbial functional potential, particularly pathways related to SCFA and redox metabolism, is associated with partial restoration of Nfil3-regulated host homeostasis and attenuation of diet-induced hepatic steatosis in mice.

Importantly, these findings support a conceptual microbiota–Nfil3–liver regulatory axis that integrates microbial functional states with host transcriptional and immune regulation, rather than establishing direct therapeutic causality. While our data provide mechanistic insight into how microbial metabolic reprogramming may influence circadian–immune–metabolic coupling, future studies incorporating circadian-phase–resolved analyses, direct metabolite quantification, and human interventional validation will be required to determine the translational relevance of this axis in metabolic disease.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

MASLD

Metabolic dysfunction-associated steatotic liver disease

MASH

Metabolic-dysfunction-associated steatohepatitis

GM

Gut microbiota

HFD

High-fat diet

SCFAs

Short-chain fatty acids

KCs

Kupffer cells

MoMs

Monocyte-derived macrophages

FITC

Fluorescein isothiocyanate

OTUs

Operational taxonomic unit

LEfSe

Linear discriminant analysis effect size

TG

Triglyceride

ALT

Alanine aminotransferase

AST

Aspartate aminotransferase

ZO-1

Zonula occludens-1

AHR

Aryl hydrocarbon receptor

CLOCK

Circadian locomotor output cycles kaput

DBP

D-box binding protein

FFAR2 and FFAR3

Free fatty acid receptors 2 and 3

GPR41 and GPR43

G protein-coupled receptors 41 and 43

HDAC

Histone deacetylase

ILC3

Group 3 innate lymphoid cell

KO

Knockout

LPS

Lipopolysaccharide

PBMC

Peripheral blood mononuclear cell

PCoA

Principal coordinates analysis

PER2

Period circadian regulator 2

PICRUSt2

Phylogenetic Investigation of Communities by Reconstruction of Unobserved States 2

PPARγ

Peroxisome proliferator-activated receptor gamma

SCFA

Short-chain fatty acids

SREBP1c

Sterol regulatory element-binding protein 1c

TLR4

Toll-like receptor 4

TNF

Tumor necrosis factor

VSL#3

Multi-strain probiotic mixture containing Bifidobacterium, Lactobacillus, and Streptococcus species

WT

Wild type

FITC

Fluorescein isothiocyanate

LEfSe

Linear discriminant analysis effect size

Author contributions

Yung-Ni Lin, Yu-Chin Huang, Chi-Yu Lai, Yi-Chen Huang, Jzy-Yu Wang, Chih-Lin Wang, Jia-Rou Hsu and Li-Ling Wu conducted experiments. Yung-Ni Lin, Chi-Yu Lai analyzed the data. Li-Ling Wu wrote this manuscript. Li-Ling Wu revised the manuscript accordingly. All authors reviewed and approved the final version of the manuscript.

Funding

This study was supported by grants from the National Science and Technology Council (NSTC), Taiwan (110-2320-B-002-080-MY3, 111-2314-B-A49-072, 112-2314-B-A49-028-MY3, 113-2321-B-A49-014, 113-2740-B-A49-003, 114-2321-B-A49-004 -, 114-2740-B-A49-003-), Yen Tjing Ling Medical Foundation (CI-115-33), and the TYGH-NYCU Joint Research Program (PTH110001), and Ministry of Health and Welfare (11210, 11310).

Data availability

Data that support the findings of this study are available on request from L.L.W.

Declarations

Ethics approval and consent to participate

This project represents an ethically continuous extension of an IRB-approved clinical investigation (IRB No: B-BR-108-018-T) under the supervision of Professor Yi-Ching Yang, Director of the Department of Geriatric Medicine, National Cheng Kung University Hospital (NCKUH) and Jzy-Yu Wang. All participants provided written informed consent prior to enrollment. The Institutional Animal Care and Use Committee of the National Yang Ming Chiao Tung University approved protocol number 1090914 and 1100902. All relevant ethical regulations for animal use were complied with.

Consent for publication

Not applicable.

Competing interests

The authors declare no conflicts of interest.