The DGs used in this study were produced by Kweichow Moutai Group in Moutai Town, Renhuai City, Guizhou Province, China. The main ingredients of the fermentation substrate include 92% fresh DG, 3% corn flour, 3% rapeseed meal, and 2% wheat bran. After the fermentation substrate was evenly mixed with a feed mixer, 50 kg of microbial starter and water mixture were added, which was prepared by mixing 100 g of biological fermentation agent with 5 kg of water containing 0.1% brown sugar. The mixture is added while stirring until it is completely added. After activation for 3–5 days, the DGs were bagged and sealed and then fermented for 9 days to obtain the FDGs (the actual cost of FDGs in either the 15% or 30% FDG group is approximately 0.4 RMB/kg) to be used for subsequent experiments. The components of the microbial starter included lactic acid bacteria, yeast, Bacillus, Bifidobacterium, Clostridium butyricum, amylase, protease, cellulase, and lipase.

This study was conducted at the Yueyuan cattle farm located in Guanling County, Anshun City, Guizhou Province (105° 58′ E, 25° 98′ N, altitude, approximately 1863 m). The experimental period was from September 2020 to November 2020, during which the average minimum outdoor temperature was 16.9°C and the maximum temperature was 25.6°C. Nine Guanling crossbred cattle (Guizhou Guanling Yellow cattle × Simmental cattle) in good health and of similar age (18 months old), weighing 300 ± 25 kg, were provided by the beef cattle fattening farm of Guanling County Yellow Cattle Group. Laboratory tests showed that these cattle were negative for infection with Brucella, Mycobacterium tuberculosis, foot-and-mouth disease virus, and lumpy skin disease virus. These cattle were ear-marked and randomly divided into a basal diet (BD) group, a Case A group, and a Case B group (containing 0%, 15%, and 30% FDG, respectively, to replace part of the concentrates), with three cattle in each group. The composition and nutrient content of the diets are shown in Supplementary Table S1 (21). The diets were thoroughly mixed to form a complete mixed diet and fed twice daily (9:00 a.m. and 4:30 p.m.). Cattle in each group were housed in separate pens and fed free water under identical housing conditions throughout the duration of the experiment.

On the 60th day of the formal feeding period, the animals underwent overnight fasting, and then tissues were collected from the jejunum, namely jejunum-BD (J-BD), jejunum-FDG-Case A (J-FDG-Case A), and jejunum-FDG-Case B (J-FDG-Case B), and from the cecum, namely cecum-BD (C-BD), cecum-FDG-Case A (C-FDG-Case A), and cecum-FDG-Case B (C-FDG-Case B). The collected tissues were then rinsed with RNase-free saline. QC samples were obtained by mixing equal volumes of the extracts from all collected samples. Prior to the metabolomics analysis, 30 mg of intestinal tissue samples were added to a 1.5-mL tube with 20 μL of L-2-chlorophenylalanine (0.3 mg/mL) dissolved in methanol as internal standard, and then 400 μL mixture of methanol/water (1/4, v/v) was added and pre-chilled at −20°C for 2 min and then ground in a grinder for 2 min. All the samples were subjected to ultrasonic extraction for 10 min, stored at −20°C for 30 min, and then centrifuged at 13,000 × g for 10 min at 4°C. A total of 300 μL of supernatants were dried in a freeze-concentration centrifugal dryer and resolubilized by 200 μL of methanol/water (1/4), vortexed for 30 s, and subjected to ultrasonic extraction for 3 min. After placement for 2 h at −20°C, the samples were centrifuged at 4°C (13,000 × g) for 10 min, and 150 μL of supernatants were filtered through 0.22-μm microfilters and then transferred to liquid chromatography (LC) vials (22).

The LC-MS analysis of intestinal samples in both ESI positive and ESI negative ion modes was conducted using an ACQUITY UPLC I Class system (Waters Corporation Milford, USA), coupled with a VION IMS Q-TOF mass spectrometer (Waters Corporation Milford, USA). A 1-μL aliquot of intestinal tissue samples was injected into an ACQUITY UPLC™ BEH C18 column (50 × 2.1 mm i.d., 1.7 μm; Waters Corporation, Milford, MA, USA). Mobile phase A consisted of 0.1% formic acid in water, and mobile phase B comprised a mixture of acetonitrile/methanol (2/3) containing 0.1% formic acid (v/v). The flow rate was 0.4 mL/min. The conditions for UPLC separation and ESI-VIONIMS Q-TOF detection are shown in Supplementary Tables S2, S3. The QC samples were utilized to evaluate the reproducibility and reliability of the LC-MS system, which contains metabolic information that enables the assessment of the stability of the mass spectrometry system. The ion peaks with a relative standard deviation of >0.4 for the QC group samples were removed. The stability of the system was then assessed using principal component analysis (PCA). The PCA model plot was obtained after seven cycles of cross-validation to verify whether the QC samples were closely clustered together, thus judging the stability of the instrument's detection.

For metabolomics data processing, the obtained LC-MS raw data were initially analyzed using the Progenesis QI v2.3 software (Non-linear Dynamics, Newcastle, UK). The metabolites were qualitatively analyzed by utilizing the Human Metabolome Database (HMDB), Lipid Maps (v2.3), the METLIN database, and a self-developed database. For the extracted data, ion peaks with missing values (0 value)>50% were eliminated, the 0 value with half of the minimum value was replaced, and the qualitatively obtained compounds were screened according to the qualitative result evaluation (score). The qualitative results were screened between 36 and 60 points, below which they were considered inaccurate and deleted. Finally, the positive and negative ion data were combined into a data matrix containing all the information extracted from the original data necessary for analysis. Subsequently, multivariate statistical analyses were carried out using SIMCA 14.0 (Umetrics AB, Umeå, Sweden).

The PCA and orthogonal partial least squares discriminant analysis (OPLS-DA) were performed to reveal the global metabolic changes between the BD and FDG groups using SIMCA 14.0. An unsupervised PCA was first used to examine both the metabolic differences between the groups and the individual metabolic differences within samples. To maximize the differences within the model, a supervised OPLS-DA was performed on metabolites between groups. In addition, to prevent overfitting of the model, the quality of the model was examined by response permutation testing with 200 responses. Then, univariate analyses were performed on the sample data, which consisted of a t-test and fold change analysis. The fold change (FC) value was calculated through a fold analysis of variance to assess differences in metabolite expression between the two groups. The p-values were obtained using a t-test, and then volcano plotting was performed using Origin 2020 (Origin Lab, Northampton, MA, USA) to visualize the p-values and FC to facilitate the rapid identification of metabolites with significant differences. Multivariate and univariate statistical analyses were combined to screen the differential metabolites between groups.

To reveal the mechanism of metabolic pathway variation in intestinal samples, an analysis of differential metabolites was carried out with metabolic pathway enrichment analysis based on the Kyoto Encyclopedia of Genes and Genomes (KEGG) database (http://www.kegg.jp/kegg/pathway.html). Their KEGG IDs and pathways were obtained, and then the number of metabolites enriched in each corresponding pathway was calculated. The pathway with a p < 0.05 was selected as an enriched pathway. The formula for calculating p-values is provided as follows:

where N: the total number of metabolites, n: the number of differential metabolites, M: the number of metabolites annotated as a specific pathway, and m: the number of differential metabolites annotated as a specific pathway. A lower p-value indicates a more significant difference within the metabolic pathway.

All continuous variables were expressed as the mean ± standard deviation (M ± SD). The differences between the groups were analyzed using Student's t-test, where the significant difference was set at a p < 0.05. All statistical analyses were performed using SPSS Statistics 26.0 (Chicago, IL, USA).

An LC-MS full-scan detection method was used to illustrate the metabolic alterations in the jejunum and cecum tissues of Guanling crossbred cattle fed the FDG diets. In total, 8,464 features in ESI⁺ mode and 2,564 ion features in ESI⁻ mode were detectable, and 1,399 metabolites were identified. PCA scoring plots were initially generated using the processed data obtained from the metabolomics analysis of the jejunum and cecum to evaluate systematic errors in the samples and the trends within each comparison. As shown in Supplementary Figure S1, the quality control (QC) samples clustered in the center of the PCA score plots suggest that the analyses were robust and reproducible. Clear separation trends were observed in the FDG-Case A vs. BD, FDG-Case B vs. BD, and FDG-Case B vs. FDG-Case A comparisons, respectively (Figures 1A–F). To maximize the identification of differential metabolites between BD and FDG groups, OPLS-DA was used to observe the tendency for separation on the metabolic spectrum (Figures 2A–F). The validity of the model was evaluated through permutation testing (n = 200), with R² values close to 1 and intercepts between the regression line and the vertical axis of Q² <0. Moreover, Q² gradually decreased with reduced displacement retention, affirming the absence of overfitting in the original model (Figures 2G, H). These results provide assurance for subsequent tests and analyses.

To visualize the fold change (FC) analysis of 1,399 metabolites for quick identification of significantly differential metabolites, a volcano plot was created by transferring the FC values of each metabolite to log2(FC) and by transferring the p-value (p = 0.05) of Student's t-test to –log (p) while simultaneously meeting the variable importance in projection (VIP) values of > 1 for the first principal component. As shown in Figures 3A–F, a total of 178 metabolites [78 metabolites upregulated with log2(FC) > 0 and 100 metabolites downregulated with log2(FC) <0], 227 metabolites (121 upregulated and 106 downregulated), and 179 metabolites (174 upregulated and 5 downregulated) were identified in the J-FDG-Case A vs. J-BD, J-FDG-Case B vs. J-BD, and J-FDG-Case B vs. J-FDG-Case A comparisons, respectively, meeting the criteria of p < 0.05 and VIP value of >1. In addition, 12 metabolites (10 upregulated and 2 downregulated), 58 metabolites (57 upregulated and 1 downregulated), and 55 metabolites (43 upregulated and 12 downregulated) were identified in the C-FDG-Case A vs. C-BD, C-FDG-Case B vs. C-BD, and C-FDG-Case B vs. C-FDG-Case A comparisons, respectively.

The levels of the top 50 differential metabolites were visualized by a heatmap depicted in Figures 4A–F, in which colors represent increased (red) or decreased (blue) abundance, with the intensity reflecting the corresponding concentration. Of these, the relative richness (referring to the ratio of FDG/BD) of 23, 49, and 50 metabolites were decreased in the J-FDG-Case A vs. J-BD, J-FDG-Case B vs. J-BD, and J-FDG-Case B vs. J-FDG-Case A comparisons, respectively. Conversely, the richness of 7, 1, and 0 metabolites was significantly increased in the aforementioned jejunum comparisons, respectively. As for cecum samples, the relative richness of 3, 26, and 21 metabolites was decreased in the C-FDG-Case A vs. C-BD, C-FDG-Case B vs. C-BD, and C-FDG-Case B vs. C-FDG-Case A comparisons, respectively. However, no upregulated differential metabolites were found in the aforementioned cecum comparisons, respectively.

The classification of all significantly differential metabolites is shown in Figures 5A, B, and Tables 1, 2 offer detailed information on the top 20 differential metabolites. Of these, in the J-FDG-Case A vs. J-BD comparison, they were mainly classified into lipids and lipid-like molecules (fatty acyls, steroids and steroid derivatives, and prenol lipids) and organoheterocyclic compounds (imidazopyrimidines). In the J-FDG-Case B vs. J-BD comparison, the differential metabolites were mainly classified into lipids and lipid-like molecules (fatty acyls, steroids and steroid derivatives, and glycerophospholipids), organic acids and derivatives (carboxylic acids and derivatives), and organic oxygen compounds (organooxygen compounds). In the J-FDG-Case B vs. J-FDG-Case A comparison, the differential metabolites were mainly classified into lipids and lipid-like molecules (glycerophospholipids, fatty acyls, and prenol lipids), organic acids and derivatives (carboxylic acids and derivatives), and organoheterocyclic compounds (tetrahydroisoquinolines). As for cecum, in the C-FDG-Case A vs. C-BD comparison, the differential metabolites were mainly classified into lipids and lipid-like molecules (steroids and steroid derivatives, and prenol lipids) and organoheterocyclic compounds (diazines). In the C-FDG-Case B vs. C-BD, the differential metabolites were mainly classified into lipids and lipid-like molecules (fatty acyls, steroids and steroid derivatives, and glycerophospholipids), organoheterocyclic compounds (imidazopyrimidines and diazines), and benzenoids (phenols, benzene, and substituted derivatives). In C-FDG-Case B vs. C-FDG-Case A, the differential metabolites were mainly classified into lipids and lipid-like molecules (fatty acyls, glycerophospholipids, and steroids and steroid derivatives), organoheterocyclic compounds (imidazopyrimidines), and organic acids and derivatives (carboxylic acids and derivatives). Collectively, these results indicate that the metabolite profiles in the jejunum and cecum are predominantly affected by the 15% FDG and 30% FDG diets.

Next, we investigated potential metabolic pathways responsible for the observed changes in the intestinal metabolic profile associated with FDG diets. The identified metabolites were annotated with KEGG and HMDB. Results showed that a total of 40 metabolic pathways were identified (p < 0.05). Among them, there were 2 (insulin resistance and bile secretion), 14 (including biosynthesis of unsaturated fatty acids and terpenoid backbone biosynthesis), and 18 (including biosynthesis of unsaturated fatty acids, linoleic acid metabolism, and GnRH signaling pathway) significant metabolic pathways (p < 0.05) enriched in the J-FDG-Case A vs. J-BD, J-FDG-Case B vs. J-BD, and J-FDG-Case B vs. J-FDG-Case A comparisons, respectively, and 1 (primary bile acid biosynthesis), 2 (primary bile acid biosynthesis and adrenergic signaling in cardiomyocytes), and 3 (linoleic acid metabolism, biosynthesis of unsaturated fatty acids, and choline metabolism in cancer) significantly enriched metabolic pathways (p < 0.05) were obtained in the C-FDG-Case A vs. C-BD, C-FDG-Case B vs. C-BD, and C-FDG-Case B vs. C-FDG-Case A comparisons, respectively (Figures 6A–F). Furthermore, four significant pathways, including biosynthesis of unsaturated fatty acids, linoleic acid metabolism, insulin resistance, and primary bile acid biosynthesis [p < 0.05, –log(p)>1.3] were significantly enriched in Guanling crossbred cattle fed with FDG diets (Table 3). Among them, the insulin resistance pathway was found to be specific to the jejunum, while the primary bile acid biosynthesis pathway was specific to the cecum. The biosynthesis of unsaturated fatty acids and linoleic acid metabolism pathways were shared between the jejunum and cecum.