Abstract
Obesogenic diets induce systemic and gut luminal metabolic perturbations, but whether these alterations occur in parallel across biospecimens remains unclear. In particular, the extent to which high-fat diet (HFD) and Western diet (WD) produce shared or compartment-specific metabolic responses in circulation and feces has not been systematically compared. In this study, targeted LC-MS/MS-based metabolite profiling was performed using serum and fecal samples from mice fed a normal diet (ND), HFD, or WD. Serum samples were analyzed at the individual-animal level, whereas fecal samples were analyzed as cage-level pooled specimens and interpreted as exploratory. Group differences were assessed using non-parametric statistics with Benjamini-Hochberg false discovery rate correction, followed by cross-compartment comparison of HFD-versus-ND and WD-versus-ND directional changes among metabolites detected in both matrices. In serum, obesogenic diets were associated with significant alterations in branched-chain amino acid-related metabolites, phenylalanine, serotonin, butyrylcarnitine, and taurocholic acid. In exploratory fecal metabolomics, significant diet-associated differences were observed mainly in amino acid-related metabolites, cholic acid, and 3-indolepropionic acid. Cross-compartment comparison of HFD-versus-ND and WD-versus-ND responses showed that several amino acid-related metabolites, including valine, leucine, and phenylalanine, were decreased in serum but increased in feces. WD also showed fecal bile acid- and indole-related changes in the exploratory fecal dataset under the present conditions. These findings suggest that HFD and WD are associated with distinct and compartment-specific metabolic remodeling across circulating and luminal compartments and support the value of multi-compartment metabolomics in studies of diet-associated metabolic dysfunction.
Supplementary Information
The online version contains supplementary material available at https://doi.org/10.1186/s12986-026-01169-2.
Keywords: Metabolomics, High-fat diet, Western diet, Bile acids, Indole metabolism, Branched-chain amino acids
Introduction
High-fat diet (HFD) and Western diet (WD) are widely used experimental paradigms for modeling diet-induced obesity and metabolic dysfunction in rodents [1–4]. Although both diets promote adverse metabolic phenotypes, they differ in macronutrient composition, cholesterol content, carbohydrate quality, and overall dietary complexity, and therefore should not be considered interchangeable models of obesogenic exposure. These compositional differences may differentially affect host intermediary metabolism, bile acid homeostasis, and the intestinal metabolic environment. Characterizing the metabolomic signatures associated with HFD and WD may therefore improve our understanding of how dietary composition remodels both systemic and gut luminal metabolism.
This distinction is also relevant to human health. High-fat dietary patterns are commonly associated with excess energy intake, adiposity, insulin resistance, and dyslipidemia, whereas Western-style dietary exposure has been linked not only to high fat intake but also to increased consumption of refined carbohydrates, sucrose, and cholesterol, together with gut dysbiosis and broader metabolic dysfunction [1, 2]. In the context of the present study, the HFD used here provided 60 kcal% fat, whereas the WD model was characterized by a Western-style formulation containing high fat together with high sucrose and cholesterol (Supplementary Table S1). These compositional differences are biologically relevant because they may differentially influence circulating metabolites, bile acid homeostasis, and host-microbiota metabolic interactions [5].
Metabolomics has emerged as a useful approach for the comprehensive evaluation of diet-associated metabolic alterations [6]. Prior studies have described changes in amino acids, bile acids, acylcarnitines, and microbiota-related metabolites in obesity-associated settings [3, 4, 7, 8]. These metabolite classes are of particular interest because they reflect multiple layers of metabolic regulation, including nutrient utilization [9], host intermediary metabolism [10], bile acid homeostasis [11], and host-microbiota metabolic interactions. Accordingly, metabolomic profiling can provide pathway-level insight into how obesogenic diets reshape metabolic physiology.
Despite these advances, most previous studies have focused on a single biospecimen type, typically plasma or serum, whereas fecal metabolites have often been evaluated independently. As a consequence, it remains uncertain whether obesogenic diets induce parallel metabolic changes across circulation and gut lumen or whether these compartments exhibit distinct responses. This distinction is important because serum and feces reflect different metabolic environments. Serum primarily reflects systemic host metabolism, whereas feces represent a luminal compartment influenced by dietary substrates, host-derived secretions, intestinal processing, and microbial transformation [8, 12]. Integrated analysis of these compartments may therefore reveal features that would not be apparent from single-compartment profiling alone.
This issue may be particularly important for metabolite groups linked to diet-microbiota interactions [5]. For example, bile acid composition is shaped by both host metabolism and microbial transformation, while tryptophan-derived indole metabolites represent a major interface between intestinal microbial activity and host metabolic signalling [8, 12]. Likewise, branched-chain and aromatic amino acid-related metabolites have been repeatedly implicated in obesity-associated metabolic dysfunction [7]. Examining these pathways in both serum and feces may therefore help clarify whether obesogenic diets are associated with coordinated or divergent metabolic responses across host and luminal compartments.
In the present study, we compared serum and fecal metabolite profiles in mice fed a normal diet (ND), HFD, or WD. By integrating individual-level serum metabolomics with exploratory cage-level fecal metabolomics, we aimed to identify shared and compartment-specific metabolic signatures associated with obesogenic diets. We further sought to determine whether WD was associated with a metabolomic profile distinguishable from that of HFD, with particular attention to amino acid-, bile acid-, and indole-related pathways. Rather than focusing on the novelty of individual metabolites per se, the study was designed to provide an integrated comparison of serum and fecal metabolic responses within the same dietary framework.
Materials and methods
Animals and dietary intervention
Six-week-old male C57BL/6J mice were obtained from Orient Bio (Seoul, Republic of Korea). Animals were housed under controlled conditions (23 ± 2 °C, 50 ± 10% relative humidity, and a 12-h light/dark cycle) with free access to sterile water. After arrival, mice were acclimated for at least 1 week on a normal chow diet (Teklad certified and irradiated 18% protein rodent diet, 2918 C; Envigo RMS, Indianapolis, IN, USA). All animal procedures were approved by the Korea Food Research Institutional Animal Care and Use Committee (approval numbers: KFRI-M-22038, KFRI-M-23009, KFRI-M-23035, KFRI-M-24034, and KFRI-M-24039) and were conducted in accordance with institutional guidelines. Samples were derived from multiple experiments conducted under comparable diet protocols, and only samples processed under the same analytical framework were included in the present metabolomics analysis.
After acclimation, mice were assigned to one of three dietary groups: ND, HFD, or WD. Mice in the ND group received a normal chow diet (2918 C, Envigo/Inotiv, Indianapolis, IN, USA), mice in the HFD group received a purified high-fat diet providing 60 kcal% fat (D12492, Research Diets, New Brunswick, NJ, USA), and mice in the WD group received an adjusted-calorie Western diet (TD.88137, Envigo/Inotiv). Detailed composition and key characteristics of the experimental diets are provided in Supplementary Table S1.
Sample collection
At the end of the 8-week dietary intervention, mice were fasted for 16 h and anesthetized with isoflurane before sample collection. Blood was collected from the abdominal vena cava, and serum was isolated by centrifugation and stored at -80 °C until analysis. Fecal samples were collected at the cage level immediately before sacrifice, frozen on dry ice, and stored at -80 °C until extraction. Freshly excreted fecal samples were collected from each cage immediately before euthanasia; aged fecal material remaining in the cage was not used.
For serum metabolomics, samples from individual animals were analyzed (ND, n = 34; HFD, n = 27; WD, n = 9). For fecal metabolomics, pooled cage-level samples were analyzed, with feces from three mice combined per cage before extraction (ND, n = 15 cages; HFD, n = 12 cages; WD, n = 3 cages). Because fecal samples were analyzed as pooled cage-level specimens, the cage was treated as the experimental unit for fecal statistical analyses to avoid pseudoreplication.
Serum sample preparation
Serum aliquots (20 µL) were transferred into microcentrifuge tubes and mixed with 20 µL of an internal standard solution, consisting of a mixture of isotope-labeled compounds (L-phenyl-¹³C₆-alanine, L-leucine-¹³C₆, and cholic acid-d₄), and 160 µL of acetonitrile for protein precipitation. After protein precipitation, samples were vortexed and centrifuged at 10,000 × g for 10 min at 4 °C. The resulting supernatants were collected and subjected to liquid chromatography-tandem mass (LC-MS/MS) analysis.
Fecal sample preparation
Frozen fecal samples were thawed on ice, and 50 mg of each pooled cage-level fecal specimen was extracted with 750 µL of 50% methanol. Samples were homogenized and subjected to three cycles of sonication for 10 min each in an ice bath. After centrifugation at 13,000 × g for 10 min at 4 °C, the supernatants were collected and filtered through a PTFE syringe filter (0.45 μm) before LC-MS/MS analysis.
LC-MS/MS analysis
Metabolite analysis was performed using a Xevo TQ MS instrument (Waters, Manchester, UK) coupled to an ACQUITY UPLC system equipped with a BEH C18 column (1.7 μm, 2.1 × 100 mm; Waters). The flow rate was set at 0.3 mL/min. Solvent A consisted of water containing 0.1% formic acid (FA), and solvent B consisted of acetonitrile (ACN) containing 0.1% FA. The gradient elution program was as follows: 0–17.5 min, a linear increase from 5% to 95% B; 17.5–19.0 min, 95% to 89% B; 19.0–19.5 min, a linear decrease from 89% to 5% B; and 19.5–22.0 min, 5% B. The column oven temperature was maintained at 40 °C, and the injection volume was 5 µL. The capillary, cone, and extractor voltages were set at 3 kV, 16 V, and 3 V, respectively. The source temperature, desolvation temperature, and desolvation gas flow rate were maintained at 150 °C, 400 °C, and 800 L/h, respectively.
The targeted panel comprised 36 metabolites, of which 29 and 36 were detected in serum and feces, respectively. Quantitative calibration was performed using authentic reference standards. Stock solutions of the standards were prepared and serially diluted to generate working calibration solutions spanning the concentration ranges indicated in Supplementary Tables S2 and S3. Analytical information for the targeted assay, including MRM transitions, retention times, and calibration/quantification parameters, is provided in Supplementary Tables S2 and S3. Metabolites were identified using predefined multiple reaction monitoring (MRM) transitions and retention time matching against authentic standards. Dedicated pooled QC samples and repeated QC injections for run-level drift assessment were not included in the present workflow.
Statistical analysis
Group-wise metabolic separation was assessed using multivariate visualization methods. For univariate analyses, overall differences among ND, HFD, and WD were evaluated using the Kruskal-Wallis test, and pairwise comparisons were performed using the Mann-Whitney U test. False discovery rate (FDR) was applied using the Benjamini-Hochberg method.
Because fecal samples were analyzed as cage-level pooled specimens rather than as individual-animal samples, fecal metabolomic results were interpreted at the cage level and regarded as exploratory [13]. Because serum and fecal metabolomes were not obtained as matched within-animal pairs and were generated at different biological resolutions (individual-animal serum vs. cage-level feces), cross-compartment comparisons were restricted to descriptive directional summaries and were not interpreted as within-animal paired, correlational, or causal analyses. PLS-DA and PCA were performed using MetaboAnalyst 6.0. Prior to multivariate analysis, data were log2-transformed. Because PLS-DA is a supervised visualization method, model interpretation was restricted to exploratory visualization, whereas statistical inference was based on univariate analysis with FDR correction.
Integrated serum-feces comparison
Metabolites detected in both serum and feces were compared across matrices to summarize the directionality of HFD-versus-ND and WD-versus-ND changes for shared metabolites. Hierarchical clustering heatmaps were generated using row-wise z-scores of log2-transformed metabolite abundances. Cross-compartment comparisons were based on median-centered log2 fold changes and were used to identify concordant or discordant response patterns between circulation and gut lumen. Because serum and fecal datasets were not analyzed as matched within-animal pairs and differed in experimental unit (individual animal vs. cage), these cross-compartment comparisons were interpreted descriptively to summarize directional concordance or discordance and not as direct biological correlation or paired-association analyses.
Results
Serum metabolomics showed diet-associated differences in circulating metabolites
Exploratory multivariate visualization suggested diet-associated differences in the serum metabolome (Fig. 1A). Unsupervised PCA of serum samples provided a complementary, unsupervised view of the dataset and showed partial overlap among dietary groups rather than complete separation (Supplementary Fig. S1A). Unsupervised hierarchical clustering of individual serum metabolomic profiles further suggested heterogeneity within the HFD group, as HFD samples did not form a single uniform cluster and showed partial overlap with ND-like and WD-like patterns (Supplementary Fig. S2). Hierarchical clustering heatmaps likewise indicated substantial restructuring of circulating metabolite profiles across dietary conditions (Fig. 2A and B). Because PLS-DA was used primarily for visualization, the principal statistical interpretation was based on univariate analysis with FDR correction. A summary of differential serum metabolites is provided in Table 1, and detailed quantitative values are provided in Supplementary Table S4.
Fig. 1.
Partial least squares-discriminant analysis (PLS-DA) score plots of serum and fecal metabolomes across dietary groups. (A) PLS-DA score plot of serum samples from ND, HFD, and WD groups. (B) PLS-DA score plot of fecal samples from ND, HFD, and WD groups. PLS-DA was used for exploratory visualization of group-wise metabolic separation. Fecal samples were analyzed as cage-level pooled specimens and should therefore be interpreted as exploratory. ND, normal diet; HFD, high-fat diet; WD, Western diet
Fig. 2.
Hierarchical clustering heatmaps of serum metabolites across dietary groups. (A) Heatmap showing the top 10 differential serum metabolites ranked by overall false discovery rate (FDR). (B) Heatmap of all analyzed serum metabolites. Heatmaps were generated using row-wise z-scores of log2-transformed metabolite abundances. Red indicates relatively higher abundance (2) and blue indicates relatively lower abundance based on row-wise z-scores of log2-transformed values (-2). ND, normal diet; HFD, high-fat diet; WD, Western diet
Table 1.
Differential serum metabolites associated with high-fat and Western diets relative to the normal diet
| - | Metabolites | HFD/ND | WD/ND | Overall FDR |
|---|---|---|---|---|
| Amino acids | Arginine | 0.78* | 1.04 | 0.028456 |
| Betaine | 1.02 | 0.66## | 0.007346 | |
| Cystine | 1.17 | 0.57 | 0.455703 | |
| Glutamine | 1.04 | 1.28## | 0.016524 | |
| Histidine | 1.01 | 1.22 | 0.264826 | |
| Isoleucine | 0.72** | 0.75# | 0.005962 | |
| Leucine | 0.74*** | 0.78# | 0.000048 | |
| Lysine | 1.03 | 1.16 | 0.424555 | |
| Ornitine | 1.41* | 0.92 | 0.028456 | |
| Phenylalanine | 0.85*** | 0.91# | 0.000000 | |
| Proline | 0.89 | 0.85 | 0.250954 | |
| Serine | 1.76** | 1.50# | 0.006141 | |
| Threonine | 1.00 | 1.04 | 0.973927 | |
| Trans-4-hydroxyproline | 0.72** | 0.81# | 0.002109 | |
| Tryptophan | 1.09 | 0.95 | 0.050046 | |
| Tyrosine | 1.05 | 1.13 | 0.355329 | |
| Valine | 0.68*** | 0.86## | 0.000000 | |
| Bile acids | Cholic acid | 0.64 | 0.81 | 0.062229 |
| Deoxycholic acid | 0.73* | 0.76 | 0.072263 | |
| Taurocholic acid | 0.84 | 3.37# | 0.006277 | |
| Carnitines | Butyrylcarnitine | 0.62*** | 1.00 | 0.000000 |
| Carnitine | 1.21 | 1.25 | 0.194916 | |
| Lauroylcarnitine | 0.67 | 1.00 | 0.206090 | |
| Fatty acids | 2-Hydroxypalmitic acid | 0.53* | 0.85 | 0.028941 |
| Indoles | Indole-3-carboxyaldehyde | 1.03 | 0.95 | 0.455703 |
| Indole-3-lactic acid | 0.94 | 1.21 | 0.072263 | |
| Serotonin | 0.78*** | 0.50### | 0.000001 | |
| Purines | Hypoxanthine | 1.02 | 0.08# | 0.016141 |
| Steroids | Cortisone | 0.79 | 1.03 | 0.429877 |
Values indicate pairwise significance for HFD vs. ND and WD vs. ND, together with the overall FDR from the Kruskal-Wallis test
*FDR-adjusted p < 0.05, **FDR-adjusted p < 0.01, ***FDR-adjusted p < 0.001 for HFD vs. ND; #FDR-adjusted p < 0.05, ##FDR-adjusted p < 0.01, ###FDR-adjusted p < 0.001 for WD vs. ND
ND, normal diet; HFD, high-fat diet; WD, Western diet; FDR, false discovery rate
Serum metabolomics identified significant diet-associated differences in multiple circulating metabolites, with the strongest effects observed for branched-chain amino acid-related metabolites, phenylalanine, serotonin, butyrylcarnitine, and taurocholic acid (Table 1). In general, valine, leucine, isoleucine, and phenylalanine were lower in HFD and WD than in ND, whereas serotonin also showed a diet-associated decrease under obesogenic conditions. By contrast, taurocholic acid showed a diet-dependent bile acid-related pattern, with relative elevation in WD, and butyrylcarnitine was particularly reduced in HFD. Together, these results showed diet-associated differences in circulating metabolites, including lower branched-chain amino acid-related metabolites and phenylalanine, lower serotonin, reduced butyrylcarnitine in HFD, and relatively higher taurocholic acid in WD.
Exploratory cage-pooled fecal metabolomics showed diet-associated differences in fecal metabolites
Exploratory PLS-DA analysis of cage-pooled fecal samples suggested diet-associated differences in luminal metabolite composition (Fig. 1B). Unsupervised PCA of fecal samples likewise suggested diet-associated structure, although overlap among groups remained and the results should therefore be interpreted cautiously given the exploratory cage-level design (Supplementary Fig. S1B). Hierarchical clustering heatmaps likewise indicated diet-associated variation in fecal metabolite profiles (Fig. 3A and B), although these findings should be interpreted cautiously because fecal samples were analyzed at the cage level and the WD group included only three cages. A summary of differential fecal metabolites is provided in Table 2, and detailed quantitative values are provided in Supplementary Table S5.
Fig. 3.
Hierarchical clustering heatmaps of fecal metabolites across dietary groups. (A) Heatmap showing the top 13 differential fecal metabolites ranked by overall false discovery rate (FDR). (B) Heatmap of all analyzed fecal metabolites. Heatmaps were generated using row-wise z-scores of log2-transformed metabolite abundances. Fecal samples were analyzed as cage-level pooled specimens and should therefore be interpreted as exploratory. Red indicates relatively higher abundance (2) and blue indicates relatively lower abundance based on row-wise z-scores of log2-transformed values (-2). ND, normal diet; HFD, high-fat diet; WD, Western diet
Table 2.
Differential fecal metabolites associated with high-fat and Western diets relative to the normal diet
| - | Metabolites | HFD/ND | WD/ND | Overall FDR |
|---|---|---|---|---|
| Amino acids | Arginine | 2.60** | 3.08# | 0.002501 |
| Aspartate | 1.87** | 1.65 | 0.010869 | |
| Betaine | 2.70* | 1.23 | 0.026534 | |
| Cystine | 7.97 | 6.65## | 0.125078 | |
| Glutamine | 1.74 | 3.03## | 0.012252 | |
| Histidine | 2.92 | 1.32 | 0.125078 | |
| Isoleucine | 2.81 | 3.19## | 0.057443 | |
| Leucine | 2.89** | 3.08## | 0.006776 | |
| Lysine | 1.58 | 2.65## | 0.056354 | |
| N-acetylglycine | 2.39** | 1.73 | 0.014011 | |
| Ornitine | 2.26 | 1.49 | 0.14229 | |
| Phenylalanine | 3.86** | 3.06## | 0.001308 | |
| Proline | 2.37*** | 2.50## | 0.000997 | |
| Serine | 2.45** | 2.72## | 0.000997 | |
| Threonine | 2.07** | 2.50## | 0.003789 | |
| Trans-4-hydroxyproline | 2.49 | 2.99## | 0.046645 | |
| Tryptophan | 4.05** | 4.60## | 0.006776 | |
| Tyrosine | 2.69** | 3.21## | 0.000997 | |
| Valine | 2.54 | 4.06## | 0.027282 | |
| Bile acids | Glycocholic acid | 1.10 | 3.21# | 0.066226 |
| Cholic acid | 6.04 | 54.89## | 0.009432 | |
| Deoxycholic acid | 1.47 | 5.29## | 0.051564 | |
| Carnitines | Butyrylcarnitine | 1.19 | 1.02 | 0.626807 |
| Carnitine | 1.01 | 3.93 | 0.135324 | |
| Lauroylcarnitine | 0.90 | 1.05 | 0.730598 | |
| Fatty acids | 2-Hydroxypalmitic acid | 2.10 | 2.21 | 0.475959 |
| Indoles | 3-Indoleacrylic acid | 1.12 | 0.96 | 0.929451 |
| 3-Indolepropionic acid | 2.60 | 0.07## | 0.045514 | |
| Indole-3-carboxyaldehyde | 3.75* | 4.33## | 0.009432 | |
| Indole-3-lactic acid | 2.35 | 0.89 | 0.730598 | |
| Serotonin | 0.83 | 1.89 | 0.125078 | |
| Purines | Hypoxanthine | 1.13 | 0.46# | 0.14229 |
| Pyridines | Nicotinamide | 0.76 | 0.42 | 0.117012 |
| Pyrimidines | Pseudouridine | 4.02 | 0.22 | 0.209407 |
| Uridine | 0.92 | 1.16 | 0.552078 | |
| Steroids | Cortisone | 0.99 | 0.99 | 0.626807 |
Values indicate pairwise significance for HFD vs. ND and WD vs. ND, together with the overall FDR from the Kruskal-Wallis test
*FDR-adjusted p < 0.05, **FDR-adjusted p < 0.01, ***FDR-adjusted p < 0.001 for HFD vs. ND; #FDR-adjusted p < 0.05, ##FDR-adjusted p < 0.01, ###FDR-adjusted p < 0.001 for WD vs. ND
Fecal samples were analyzed as cage-level pooled specimens and should therefore be interpreted as exploratory
ND, normal diet; HFD, high-fat diet; WD, Western diet; FDR, false discovery rate
Exploratory fecal metabolomics revealed significant diet-associated differences in multiple luminal metabolites, particularly amino acid-related metabolites and selected bile acid- and indole-related compounds (Table 2). Several amino acid-related metabolites, including proline, serine, phenylalanine, leucine, tryptophan, arginine, and threonine, were higher in HFD and WD than in ND, indicating broad luminal accumulation of amino acid-related metabolites under obesogenic dietary conditions. WD was additionally associated with a distinctive fecal bile acid/indole-related pattern, including higher cholic acid and lower 3-indolepropionic acid relative to ND, while deoxycholic acid showed a positive WD-versus-ND pairwise signal despite not reaching the overall FDR threshold in the three-group comparison. Because fecal samples were analyzed as cage-level pooled specimens and the WD group included only three cages, these findings should be interpreted as exploratory. Among other tryptophan-derived fecal metabolites, indole-3-carboxaldehyde showed an overall significant group effect, whereas indole-3-lactic acid did not show a statistically robust overall difference and is therefore interpreted cautiously.
Cross-compartment directional comparison suggested both concordant and discordant responses among shared metabolites
To descriptively assess whether diet-associated changes showed similar or opposing directions across circulation and gut lumen, metabolites detected in both serum and feces were compared across compartments relative to the ND group. Several branched-chain and aromatic amino acid-related metabolites, including valine, leucine, isoleucine, and phenylalanine, showed opposite directions of change, with relative decreases in serum but increases in feces. By contrast, glutamine and serine showed broadly concordant increasing tendencies across the two matrices. These results showed that the direction of diet-associated changes was not always the same in serum and feces.
Integrated log2 fold-change heatmap analysis of shared metabolites (Fig. 4) highlighted both concordant and discordant dietary responses across serum and feces. The heatmap summarizes median-centered log2 fold changes for HFD-versus-ND and WD-versus-ND comparisons in each matrix and shows that several branched-chain and aromatic amino acid-related metabolites, including isoleucine, leucine, valine, and phenylalanine, changed in opposite directions between serum and feces, whereas glutamine and serine showed broadly concordant increasing tendencies. A tabulated summary of these shared-metabolite comparisons is provided in Supplementary Table S6.
Fig. 4.
Integrated log2 fold-change heatmap of shared serum and fecal metabolites across dietary groups. Rows represent shared metabolites detected in both serum and feces, and columns represent compartment-specific pairwise comparisons (serum HFD vs. ND, feces HFD vs. ND, serum WD vs. ND, and feces WD vs. ND). Red indicates relative increase and blue indicates relative decrease compared with ND
Discussion
In the present study, integrated metabolomic profiling of serum and feces identified distinct diet-associated changes across circulating and luminal compartments in mice fed ND, HFD, or WD. Because the ND diet was a natural-ingredient chow diet whereas the HFD and WD were purified/adjusted-calorie experimental diets, differences among groups should be interpreted in the context of both caloric composition and overall diet formulation. A notable finding was that several metabolites did not change in parallel across compartments, indicating that obesogenic dietary exposure was associated with compartment-dependent metabolic remodeling rather than a uniform directional shift across biospecimens.
One of the clearest observations was the divergence in amino acid-related metabolites between serum and feces. Branched-chain and aromatic amino acid-related metabolites, including valine, leucine, isoleucine, and phenylalanine, were lower in serum, whereas several of the same or related metabolites were increased in feces. This pattern may reflect altered amino acid handling at multiple levels, potentially including systemic utilization, intestinal processing, luminal retention, and microbial transformation [7, 14, 15]. Although the present study does not distinguish among these possibilities mechanistically, the opposing directional changes across compartments are consistent with differential responses of circulating and luminal metabolite pools to obesogenic dietary exposure [16]. Not all shared metabolites showed discordant behavior, however, as selected metabolites such as serine and glutamine showed broadly concordant tendencies across compartments, indicating that both shared and compartment-specific responses were present. Notably, the direction of circulating branched-chain amino acid-related changes in the present study differs from that reported in several previous obesity-associated studies, including the work of Newgard and colleagues, in which elevated circulating BCAA levels were linked to obesity and insulin resistance [7]. In the present dataset, however, valine, leucine, and isoleucine were reduced in serum in both the HFD and WD groups relative to ND. Several factors may contribute to this discrepancy, including differences in mouse strain, diet composition, fasting duration, sample collection timing, and analytical platform, as well as differences between systemic circulating pools and luminal metabolite accumulation [14, 15]. In addition, the present study compared two obesogenic diets against a chow-fed ND group within a targeted metabolomics framework, whereas previous studies have often focused on different dietary models or metabolic contexts. Thus, rather than contradicting the broader literature, the present findings may indicate that circulating BCAA-related responses are context-dependent and should be interpreted alongside compartment-specific changes observed in the gut lumen.
The dataset also supports the view that WD was associated with a metabolomic profile that was partially distinct from that of HFD. In serum, taurocholic acid showed a diet-dependent pattern, with relative reduction in HFD and elevation in WD, whereas serotonin was reduced under obesogenic dietary conditions, with the lowest level observed in WD. In feces, WD was associated with higher cholic acid, higher deoxycholic acid in the WD-versus-ND pairwise comparison, and a diet-dependent pattern of 3-indolepropionic acid, with relative elevation in HFD but reduction in WD. Taken together, these findings suggest that WD was associated with distinct bile acid- and indole-related changes in the exploratory fecal dataset under the present experimental conditions, rather than definitive evidence of a stronger biological effect of WD than HFD [8, 17]. This interpretation is broadly consistent with previous literature indicating that distinct obesogenic diets can induce overlapping but non-identical metabolic and intestinal responses. Comparative profiling studies in diet-induced obese mice have shown that while multiple obesogenic diets may produce broadly similar dysbiosis, their effects on plasma, fecal, cecal, and liver metabolomes can differ, including diet-specific changes in bile acid profiles [18]. Given the established roles of bile acids and indole derivatives in gut-liver and host-microbiota signaling, these pathway-level differences may be relevant to the metabolic distinctions observed between the two obesogenic diets [19, 20]. In parallel, HFD has been associated with altered gut microbiota and metabolome profiles in mice, including coordinated changes in amino acid- and bile acid-related pathways [3, 4]. Western-style dietary exposure has likewise been linked to gut dysbiosis and altered host–microbiota metabolic interactions, supporting the view that WD may affect intestinal and luminal metabolic remodeling in ways that are not fully equivalent to HFD [2, 5, 8].
The observed differences in serotonin, butyrylcarnitine, bile acids, and tryptophan-derived indole metabolites may be biologically relevant because these metabolites lie at the interface of host intermediary metabolism, mitochondrial substrate handling, bile acid signaling, and host–microbiota communication. Reduced serum serotonin in the obesogenic diet groups, particularly in WD, may reflect altered tryptophan metabolism and serotonergic signaling under diet-induced metabolic stress [21, 22]. Lower serum butyrylcarnitine, especially in HFD, may indicate altered short-chain acylcarnitine handling and shifts in fatty acid-related mitochondrial metabolism [3, 4]. The bile acid pattern observed here, including diet-dependent serum taurocholic acid and increased fecal cholic acid and deoxycholic acid, is consistent with the view that distinct obesogenic diets can differentially affect bile acid synthesis, transformation, and gut-liver signaling [8, 17, 20]. Among tryptophan-derived microbial metabolites, lower fecal 3-indolepropionic acid in WD may be of particular interest because this metabolite has been linked to intestinal barrier integrity and metabolic homeostasis [12, 23]. Fecal indole-3-carboxaldehyde and indole-3-lactic acid may likewise reflect altered intestinal tryptophan metabolism and microbial activity; however, because not all indole-related metabolites showed equally robust statistical support in the present dataset, these findings should be interpreted cautiously [12, 19, 22]. Although direct extrapolation from mice to humans is not warranted, the more pronounced WD-associated bile acid/indole-related pattern observed here may be relevant to the broader concept that Western-style dietary exposure imposes metabolic effects beyond high fat intake alone, particularly through host–gut metabolic interactions [2, 8]. Because the fecal dataset was exploratory, based on cage-level pooled specimens, and not matched to microbiome profiling or physiological phenotyping, these interpretations should be regarded as hypothesis-generating rather than definitive.
These findings also have methodological implications. If the analysis had been restricted to serum, the principal interpretation would likely have emphasized reduced circulating branched-chain and aromatic amino acid-related metabolites together with changes in serotonin, butyrylcarnitine, and taurocholic acid [21, 22]. By contrast, exploratory fecal analysis suggested additional luminal features that were not apparent from serum alone, including bile acid- and indole-related changes under WD conditions [23]. The integrated comparison therefore provided information that would not have been readily captured from a single biospecimen and supports the use of multi-compartment metabolomics in dietary intervention studies [18].
Several limitations warrant consideration. First, fecal metabolomic profiling was conducted using cage-level pooled specimens rather than individual-animal samples, and the fecal findings should therefore be interpreted at the cage level and regarded as exploratory [3, 12, 24]. Second, the limited number of WD fecal cages (n = 3) constrains statistical robustness and limits the strength of inference regarding WD-associated luminal alterations. Third, because serum and fecal datasets were not obtained as matched within-animal pairs and were generated at different biological resolutions (individual-animal serum vs. cage-level feces), cross-compartment comparisons are appropriately interpreted as directional rather than correlational and do not support matched biological inference. Finally, the absence of matched physiological phenotyping data (such as body weight gain or glucose homeostasis measures), microbiome profiling data (including 16 S rRNA sequencing), flux analysis, and functional validation limited biological contextualization of the observed metabolomic profiles and precluded mechanistic attribution of the observed metabolite changes. In addition, because the analytical workflow did not include dedicated pooled QC samples or repeated QC injections, run-level analytical drift could not be formally assessed within the present dataset. Despite these limitations, the study has notable strengths, including the integration of serum and fecal metabolomics within the same ND/HFD/WD experimental framework, the use of FDR-controlled univariate analysis, and the explicit separation of individual-animal serum data from exploratory cage-level fecal data. Although many individual metabolite alterations have been reported previously, the present study provides an integrated cross-compartment description of shared, concordant, and discordant directional responses across circulation and gut lumen.
Overall, the present findings support the view that HFD and WD are associated with both shared and compartment-specific metabolic remodeling. In particular, the discordant behavior of several amino acid-related metabolites across serum and feces suggests that host and luminal compartments should not be assumed to respond in parallel to obesogenic dietary exposure. Future studies that integrate matched serum, feces, microbiome composition, and metabolic phenotyping at the individual-animal level will be important for clarifying the mechanisms underlying these compartment-specific responses.
Conclusion
Integrated analysis of serum and fecal metabolomes suggested that HFD and WD were associated with distinct and compartment-specific metabolic remodeling in mice. Serum metabolomics highlighted alterations in branched-chain amino acid-related metabolites, phenylalanine, serotonin, butyrylcarnitine, and taurocholic acid, whereas exploratory fecal metabolomics suggested diet-associated changes in amino acid-related metabolites together with a WD-associated bile acid- and indole-related pattern under the present conditions. These findings support the value of multi-compartment metabolomics for investigating diet-associated metabolic dysfunction and provide an integrated description of how host circulation and gut lumen may respond differently to obesogenic dietary exposure.
Supplementary Information
Acknowledgements
This research was supported by the Main Research Program (E0210602) of the Korea Food Research Institute (KFRI) funded by the Ministry of Science and ICT.
Author contributions
Lee YR: Methodology, Formal analysis, Data curation, Writing - original draft. Lee HB: Formal analysis, Writing - review and editing. Kim HJ: Formal analysis. Park JH: Investigation, Data curation. Park HY: Conceptualization, Supervision, Writing - review and editing.
Data availability
The datasets supporting the conclusions of this article are included within the article and its additional files.
Declarations
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Uranga RM, Bruce-Keller AJ, Morrison CD, Fernandez-Kim SO, Ebenezer PJ, Zhang L, Dasuri K, Keller JN. Intersection between metabolic dysfunction, high fat diet consumption, and brain aging. J Neurochem. 2010;114:344–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Martinez KB, Leone V, Chang EB. Western diets, gut dysbiosis, and metabolic diseases: Are they linked? Gut Microbes. 2017;8:130–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Jo JK, Seo SH, Park SE, Kim HW, Kim EJ, Kim JS, Pyo JY, Cho KM, Kwon SJ, Park DH, Son HS. Gut Microbiome and Metabolome Profiles Associated with High-Fat Diet in Mice. Metabolites. 2021;11:482. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Cai H, Wen Z, Meng K, Yang P. Metabolomic signatures for liver tissue and cecum contents in high-fat diet-induced obese mice based on UHPLC-Q-TOF/MS. Nutr Metab (Lond). 2021;18:69. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Sonnenburg JL, Backhed F. Diet-microbiota interactions as moderators of human metabolism. Nature. 2016;535:56–64. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Guasch-Ferre M, Bhupathiraju SN, Hu FB. Use of Metabolomics in Improving Assessment of Dietary Intake. Clin Chem. 2018;64:82–98. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Newgard CB, An J, Bain JR, Muehlbauer MJ, Stevens RD, Lien LF, Haqq AM, Shah SH, Arlotto M, Slentz CA, et al. A branched-chain amino acid-related metabolic signature that differentiates obese and lean humans and contributes to insulin resistance. Cell Metab. 2009;9:311–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Collins SL, Stine JG, Bisanz JE, Okafor CD, Patterson AD. Bile acids and the gut microbiota: metabolic interactions and impacts on disease. Nat Rev Microbiol. 2023;21:236–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Noerman S, Landberg R. Blood metabolite profiles linking dietary patterns with health-Toward precision nutrition. J Intern Med. 2023;293:408–32. [DOI] [PubMed] [Google Scholar]
- 10.Ursell LK, Haiser HJ, Van Treuren W, Garg N, Reddivari L, Vanamala J, Dorrestein PC, Turnbaugh PJ, Knight R. The intestinal metabolome: an intersection between microbiota and host. Gastroenterology. 2014;146:1470–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Li F, Patterson AD, Krausz KW, Tanaka N, Gonzalez FJ. Metabolomics reveals an essential role for peroxisome proliferator-activated receptor alpha in bile acid homeostasis. J Lipid Res. 2012;53:1625–35. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Konopelski P, Mogilnicka I. Biological Effects of Indole-3-Propionic Acid, a Gut Microbiota-Derived Metabolite, and Its Precursor Tryptophan in Mammals’ Health and Disease. Int J Mol Sci 2022, 23:1222. [DOI] [PMC free article] [PubMed]
- 13.Festing MF, Altman DG. Guidelines for the design and statistical analysis of experiments using laboratory animals. ILAR J. 2002;43:244–58. [DOI] [PubMed] [Google Scholar]
- 14.Li N, Cen Z, Zhao Z, Li Z, Chen S. BCAA dysmetabolism in the host and gut microbiome, a key player in the development of obesity and T2DM. Med Microecology. 2023;16:100078. [Google Scholar]
- 15.De Bandt JP, Coumoul X, Barouki R. Branched-Chain Amino Acids and Insulin Resistance, from Protein Supply to Diet-Induced Obesity. Nutrients. 2022;15:68. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Li H, Perino A, Huang Q, Von Alvensleben GVG, Banaei-Esfahani A, Velazquez-Villegas LA, Gariani K, Korbelius M, Bou Sleiman M, Imbach J, et al. Integrative systems analysis identifies genetic and dietary modulators of bile acid homeostasis. Cell Metab. 2022;34:1594–e16101594. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Devkota S, Wang Y, Musch MW, Leone V, Fehlner-Peach H, Nadimpalli A, Antonopoulos DA, Jabri B, Chang EB. Dietary-fat-induced taurocholic acid promotes pathobiont expansion and colitis in Il10-/- mice. Nature. 2012;487:104–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Aggarwal H, Gautam J, Kumari D, Gupta SK, Bajpai S, Chaturvedi K, Kumar Y, Dikshit M. Comparative profiling of gut microbiota and metabolome in diet-induced obese and insulin-resistant C57BL/6J mice. Biochim Biophys Acta Mol Cell Res. 2024;1871:119643. [DOI] [PubMed] [Google Scholar]
- 19.Shimi G. Tryptophan Metabolism in Obesity: Pathways, Mechanisms, and Therapeutic Perspectives. Curr Med Sci. 2025;45:1304–18. [DOI] [PubMed] [Google Scholar]
- 20.Won TH, Arifuzzaman M, Parkhurst CN, Miranda IC, Zhang B, Hu E, Kashyap S, Letourneau J, Jin WB, Fu Y, et al. Host metabolism balances microbial regulation of bile acid signalling. Nature. 2025;638:216–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Li G, Dong S, Liu C, Yang J, Rensen PCN, Wang Y. Serotonin signaling to regulate energy metabolism: a gut microbiota perspective. Life Metab. 2025;4:loae039. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Cussotto S, Delgado I, Anesi A, Dexpert S, Aubert A, Beau C, Forestier D, Ledaguenel P, Magne E, Mattivi F, Capuron L. Tryptophan Metabolic Pathways Are Altered in Obesity and Are Associated With Systemic Inflammation. Front Immunol. 2020;11:557. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Chen L, Yang Y, Sun S, Xie Y, Pan C, Li M, Li C, Liu Y, Xu Z, Liu W, Ji M. Indolepropionic acid reduces obesity-induced metabolic dysfunction through colonic barrier restoration mediated via tuft cell-derived IL-25. FEBS J. 2022;289:5985–6004. [DOI] [PubMed] [Google Scholar]
- 24.McKinnon Reish H, Dewey L, Kirschman LJ. A host of issues: pseudoreplication in host-microbiota studies. Appl Environ Microbiol. 2024;90:e0103324. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The datasets supporting the conclusions of this article are included within the article and its additional files.