Animal experiments were conducted in strict accordance with the ethical requirements of the Laboratory Animal Ethics Committee of Tibet College of Agriculture and Animal Husbandry, China (License No. XZA-2005-012). The study was conducted in accordance with local laws and institutional requirements.

In the present study, 21 weaned Tibetan piglets (initial body weight 5.4 ± 0.2 kg) at 30 days of age were selected. These weaned Tibetan piglets are all from the Tibetan pig breeding base in Zengba Village, and the weaned piglets were clinically healthy, with normal feed intake and feces. They were tested to confirm that they were free of contagious diseases such as swine fever, blue ear disease, and other pathogens such as Salmonella and Escherichia coli. Furthermore, it was ascertained that the piglets had not been treated with any antibiotics or antimicrobial drugs in the past 30 days. The animals were divided into three experimental groups by randomization: the basal diet group (Nor), the lincomycin-treated group (Ant), and the fecal microbial extract-added group (Fec). The control group (Nor) was fed the basal diet, while the Fec and Ant groups were supplemented with 2,000 mL/kg of fecal bacterial extracts and 1,000 mg/kg of lincomycin hydrochloride, respectively, to the basal diet. The nutritional composition of the experimental diets is shown in Table 1 and was formulated according to the National Research Council (NRC) standards to meet or exceed the nutritional requirements for a 30-day trial period.

A total of 10 g of fresh feces was weighed, added with 40 mL of sterile PBS (0.1 mol/L, pH = 7.2) (feces-water ratio of 1:4), placed in a sterile beaker, stirred with a magnetic stirrer at a low speed (100–200 rpm) for 5 min, and mixed well to a paste. The suspension was passed through 2.0, 1.0, and 0.5 mm sterile stainless steel sieves to remove impurities such as crude fiber and fecal residue, and the filtrate was collected. The filtrate was transferred to a 50-mL sterile centrifuge tube and centrifuged at 200×g for 10 min at 4°C. The precipitate was discarded (residual impurities and host cells were removed), while the supernatant was retained. The supernatant was centrifuged at 3,000×g for 15 min at 4°C and discarded; the precipitate was retained. Twenty milliliters of sterile PBS was added to the precipitate, which was resuspended using a vortex shaker and then centrifuged at 3,000×g for 10 min at 4°C. This cycle was repeated twice to obtain the purified bacterial sludge. The precipitate was resuspended with sterile PBS solution, and 25% glycerol was added to prepare a fecal suspension. Sterile centrifuge tubes were dispensed and stored frozen at −80°C, and a 37°C water bath was used for 1 h before use. A fecal suspension with a bacterial load of 5 × 10⁸ cfu/mL was prepared at the time of use.

Fresh fecal samples were obtained from the piglets on days 0 and 30 of the experiment using rectal massage. Animals were fasted for 8 h prior to sampling in order to standardize intestinal contents while maintaining free access to water in order to reduce stress. This sampling method ensures aseptic operation and circumvents environmental contamination. The fecal samples were subjected to freeze-drying and pulverization (60 Hz, 30 s), and a 50-mg sample was placed in a centrifuge tube. Subsequently, 700 µL of precooled (−40°C) extract (methanol–water, 3:1 v/v, containing internal standard) was added, vortexed, and mixed for 30 s. The sample was then homogenized at 35 Hz for 4 min and sonicated in an ice-water bath for 5 min. The above homogenization–sonication steps were then repeated on three occasions. Subsequently, the samples were left to stand at 4°C in a homogenizer overnight. Following centrifugation at 4°C for 15 min at 12,000 rpm (centrifugal force 13,800×g, radius 8.6 cm), the resultant filtrate was filtered through a 0.22-µm microporous membrane and then diluted fivefold with the extract. The mixture was then vortexed for 30 s and combined into a quality control (QC) sample of 40 µL per sample. This sample was stored at −80°C for subsequent testing.

The target compounds were separated using a Waters UPLC column coupled with an EXIONLC system (SCIEX) ultra performance liquid chromatograph (UPLC). The acquisition and quantitative analysis of target compounds were performed using the SCIEX Analyst WorkStation software (v1.6.3). The raw mass spectrometry data were converted to TXT format by the MS converter software (v16.2.0.9), and then the peaks were extracted and annotated by using a self-developed R package in combination with a self-constructed database. Finally, orthogonal partial least squares discriminant analysis (OPLS-DA) was carried out. The validity of the model was evaluated by the R²Y and Q² parameters, and the metabolites were screened for differences based on the projected importance values (VIP > 1) of the variables analyzed by OPLS-DA between groups and the p-values of univariate tests. In order to enhance the metabolite coverage and detection efficacy, dual ionization modes of positive (POS) and negative (NEG) ions were utilized for metabolite detection, with the data from these two modes subsequently integrated for analysis using OPLS-DA.

Alpha diversity is defined as the abundance and isolation of species in a given habitat. The Chao1 and ACE indices were utilized to evaluate the species richness of the samples, with higher values denoting more significant diversity. The Simpson and Shannon indices were utilized to integrate species richness and evenness. The beta diversity analysis was employed to compare the species diversity among samples. This analysis reflected the distance relationship between samples and revealed the degree of differentiation of bacterial colonies. By quantifying the differences in colony composition among Nor, Ant, and Fec samples, the relative values of distances among samples were calculated, and the distance relationships were then visualized by downscaling the data. The principal coordinate analysis (PCoA) of the species composition of the three groups of samples was carried out by using the R software and QIIME software to characterize differences in species composition among samples based on the colony distance matrix.

LC-MS/MS analyses were performed using a SCIEX 6500 QTRAP + Triple Quadrupole Mass Spectrometer (equipped with an IonDriveTurbo VESI ion source) in multiple reaction monitoring (MRM) mode. The ion source parameters were configured as follows: ion spray voltage set to ±4,500 V, curtain gas pressure set to 35 psi, temperature was set to 400°C, ion source gas 1 and gas 2 pressures were set to 60 psi, and de-cluster voltage was set to ±100 V. These parameters were maintained consistently throughout the study.

The constructed libraries were then subjected to whole-genome sequencing on the Illumina NovaSeq high-throughput sequencing platform, employing a strategy that has been previously described in the literature as the “birdshot” approach. In order to evaluate the stability of the analysis system and to identify variables that demonstrate significant variation during the analysis process, it is imperative that all samples are incorporated within the QC sample category. During the instrumental analysis phase, QC samples were inserted at intervals of 8–10 samples to monitor system performance. In consideration of the multifaceted character of metabolomic data and the substantial interrelation between variables, conventional univariate analysis is challenging in facilitating expeditious and exhaustive extraction of pertinent information from the data. To address this challenge, OPLS-DA, a multivariate statistical methodology, was employed to perform dimensionality reduction and classification, thereby enabling the systematic analysis of the multidimensional data collected.

The subsequent analysis of high-throughput sequencing data or metabolomics data yielded a list of differentially expressed genes (DEGs), differential flora, or differential metabolites. Subsequently, the metabolites were functionally annotated through the KEGG database (https://www.genome.jp/kegg/), and their corresponding KEGG Orthology (KO) numbers and the pathways to which they belong (e.g., metabolic pathways, signaling pathways) were obtained. The enrichment degree of metabolites in a specific pathway was analyzed using statistical methods, such as the hypergeometric test and Fisher’s exact test. Statistical methods (e.g., hypergeometric test, Fisher’s exact test) were used to analyze the degree of metabolite enrichment in a particular pathway and to determine whether the pathway was significantly associated with the study phenotype.

The study used the SPSS 18.0 software to implement Duncan’s multiple comparison test. The criteria for determining statistical significance were as follows: a p-value of less than 0.05 indicated a statistically significant difference between the groups; a p-value of less than 0.01 indicated a highly significant difference; and a p-value of greater than 0.05 indicated that the difference between the groups was not statistically significant.

In mass spectrometry-based metabolomics research, QC samples are essential for ensuring reliable, high-quality data. These samples were prepared by pooling equal aliquots of all experimental samples—including those from the Nor, Ant, and Fec groups—and were used to monitor stability across the entire workflow, from sample extraction and instrumental analysis to data acquisition. Though theoretically identical, QC samples often exhibited minor differences due to systematic errors introduced during these processes; smaller discrepancies indicated greater method stability and better data quality. To assess intra- and interbatch variations, QC samples were inserted at regular intervals throughout sequencing and metabolomics runs, helping verify the reliability and consistency of results. In this study, the dense clustering of QC samples in the PCA visualization ( Figure 1A ) further confirmed the robustness of our data.

R2X, R2Y, and Q2 were commonly used metrics for this purpose. In general, R2Y and Q2 values greater than 0.5 indicated the reliability of the model ( Table 2 ). The OPLS-DA model established in this experiment exhibited R2Y and Q2 values exceeding 0.5, thus signifying the stability, reliability, and statistical significance of the model ( Figures 1B–D ).

Differential metabolites were identified based on three parameters: VIP, FC, and p-value. To be considered differential metabolites, the VIP value had to be greater than 1 and the p-value had to be less than 0.01. Furthermore, specific differential metabolites were selected based on the principle of p-value less than 0.01, VIP value greater than 1, and FC value greater than 1 or less than 1. These criteria were determined using VIP values obtained from statistical analyses and OPLS-DA.

At the 42nd day of the trial, in comparison with the control groups “Nor vs. Ant,” “Nor vs. Fec,” and “Fec vs. Ant,” a total of 20, 4, and 40 differential metabolites were identified, respectively, with 5, 0, and 23 being upregulated and 26, 4, and 17 being downregulated ( Tables 3 – 5 ).

Based on untargeted metabolomics data, a bidirectional clustering heatmap of samples and metabolites was constructed using a hierarchical clustering algorithm. The heatmap revealed significant separation of the three groups of samples along the row direction, with tight clustering of intragroup samples. The color gradient (red: upregulation; blue: downregulation) further demonstrated gradient changes in differential metabolites across groups, suggesting their potential role in mediating phenotypic differences through the regulation of signaling pathways ( Figures 1E–G ).

KEGG enrichment bar plots and enrichment circle plots were generated to illustrate the metabolic pathways associated with different feed additives in the Nor and experimental groups. At the end of the experiment, compared with the Nor group (control group), the Ant group exhibited 15 differential metabolic pathways ( Figure 2A ), including tryptophan metabolism, benzoic acid family metabolism, serotonin receptor agonists, cholinergic synapse metabolism, and gap junction metabolism. The Fec group showed 15 metabolic pathways ( Figure 2B ), mainly including gap junction metabolism, riboflavin metabolism, and dopaminergic synapse. Additionally, 15 differential metabolic pathways were identified between the Fec and Ant groups ( Figure 2C ), primarily involving gap junction metabolism, tryptophan metabolism, CAMP signaling pathway, serotonin receptor agonists, and cholinergic synapse.

The top 20 enriched pathways were selected to generate KEGG enrichment circle plots ( Figures 2D–F ). Based on the KEGG enrichment circle plot analysis, differentially abundant metabolites between the Nor and Ant groups were significantly enriched in metabolism (e.g., ABC transporters), signal transduction (e.g., MAPK signaling), and disease-related pathways (e.g., cancer pathways), suggesting that experimental interventions may mediate phenotypic differences through regulating efflux transport, cellular stress response, and metabolic reprogramming. The prominent enrichment of the ABC transporter pathway (ko02010) implies its potential role in drug resistance mechanisms, while activation of the MAPK pathway (ko00380) may correlate with inflammatory responses ( Figure 2D ). Differential metabolites between the Nor and Fec groups were significantly enriched in metabolism, organismal systems, and disease-related functional categories. The marked enrichment of metabolic pathways (ko01100) suggests potential involvement in energy metabolism reprogramming, such as enhanced glycolysis or TCA cycle activity, which may support cell proliferation or stress adaptation. The activation of bile secretion (ko04976) may correlate with abnormal lipid metabolism, including hypercholesterolemia or cholestasis. The enrichment of flavonoid biosynthesis (ko01063) likely reflects alterations in host–microbial co-metabolism products, exhibiting antioxidant effects. Additionally, the activation of neuroactive ligand–receptor interactions (ko04080) may involve neuroendocrine regulation, such as appetite modulation ( Figure 2E ). Differential metabolites between the Fec and Ant groups were significantly enriched in organismal systems, metabolism, and disease-related functional categories. The prominent enrichment of the bile secretion pathway (ko04976) suggests its potential involvement in lipid metabolism dysregulation or bile acid homeostasis imbalance, such as cholestasis or impaired cholesterol clearance. The activation of the neuro-metabolic hub (ko04080) may correlate with neuroendocrine regulation of metabolic adaptation, for instance, adrenergic signaling promoting gluconeogenesis or lipolysis during stress responses. Additionally, the significant enrichment of tryptophan metabolism (ko00380) implies its role in gut–brain axis signaling or immune modulation, exemplified by microbiota-derived indole metabolites modulating host inflammatory responses.

Fecal samples from the Nor, Fec, and Ant groups were analyzed using 16S rRNA sequencing technology. The Chao1 ( Figure 3A ), Shannon ( Figure 3B ), and Simpson ( Figure 3C ) indices revealed that the Nor and Ant groups exhibited lower microbial diversity compared to the Fec group. PCoA based on weighted UniFrac distance demonstrated significant differences in beta diversity of the fecal microbiota among the Nor, Fec, and Ant groups ( Figure 3D ).

Analysis of the phylum-level microbial community structure in the Nor, Fec, and Ant groups via 16S rRNA sequencing revealed that Firmicutes and Bacteroidota were the dominant phyla, with changes in their ratios reflecting the regulatory effects of different treatment groups on the gut microbiota ( Figure 3E ). The increased abundance of Bacteroidota in the Fec group might be attributed to enhanced dietary fiber metabolism induced by fecal microbiota transplantation, while the significant reduction of Firmicutes accompanied by an increase in Proteobacteria in the Ant group suggests that antibiotics likely disrupted microbial balance and promoted pathogenic bacterial proliferation.

Analysis at the genus level via 16S rRNA sequencing revealed significant differences in gut microbiota among the Nor, Fec, and Ant groups ( Figure 3F ). The Fec group exhibited significant enrichment of probiotic genera (e.g., Lactobacillus and Prevotella) due to FMT, which may improve intestinal function by enhancing dietary fiber metabolism and short-chain fatty acid (SCFA) production. In the Ant group, a reduction in Firmicutes (e.g., Lactobacillus) and an increase in Proteobacteria (e.g., Streptococcus) were observed, potentially leading to elevated intestinal permeability and triggering endotoxemia, suggesting that microbial dysbiosis may induce inflammation or metabolic disorders. The stable abundance of core genera (e.g., Christensenellaceae_R-7_group, each accounting for ~8%) in the Nor group reflects the homeostatic maintenance function of the core microbiota.

The Venn diagram of OTU distribution ( Figure 4A ) demonstrated distinct microbial community structures among the Nor (control), Fec (fecal microbiota transplantation), and Ant (antibiotic-treated) groups. A total of 2,284 OTUs were identified, with 81 OTUs shared across all groups, representing a conserved core microbiota. Unique OTUs were the highest in the Nor group (374), followed by Fec (145) and Ant (132), indicating reduced microbial diversity under antibiotic intervention. Pairwise overlaps revealed 919 OTUs shared between Fec and Nor, suggesting partial restoration of native microbiota through transplantation, while Ant exhibited limited overlap with both Fec (341 OTUs) and Nor (292 OTUs), consistent with antibiotic-induced suppression. Data derived from 16S rRNA sequencing (V3–V4 region) and analyzed via QIIME2 (v2021.4) highlighted the resilience of Nor’s baseline diversity and the functional recalibration achieved by fecal transplantation in the Fec group.

This study compared the differences in microbial abundance at the phylum level among the Nor, Fec, and Ant groups using Tukey’s HSD test (Tukey’s honestly significant difference) ( Figure 4B ). The Tukey’s HSD test is a post-hoc multiple comparison method applied after analysis of variance (ANOVA) to determine specific intergroup differences. The abundance of Firmicutes in the Ant group was significantly lower than in the Nor group (p < 0.05), possibly due to the selective inhibition of Gram-positive bacteria (as Firmicutes are predominantly Gram-positive) by antibiotics (e.g., vancomycin), leading to reduced microbial diversity. The abundance of Firmicutes in the Ant group was also significantly lower than in the Fec group (p < 0.01), suggesting that the inhibitory effect of antibiotics on Firmicutes outweighed the probiotic benefits of fecal microbiota transplantation. No significant difference was observed between the Nor and Fec groups, indicating that fecal microbiota transplantation did not markedly alter the baseline proportion of Firmicutes, likely due to their stable functional role in a healthy gut. The abundance of Bacteroidota in the Fec group was significantly higher than in the Nor group (p < 0.01), which might be attributed to the introduction of donor microbiota rich in Bacteroidota (e.g., Bacteroides succinogenes) via fecal microbiota transplantation, enhancing dietary fiber degradation capacity. The abundance of Bacteroidota in the Fec group was also significantly higher than in the Ant group (p < 0.01), reflecting the substantial suppression of Bacteroidota growth by antibiotics, potentially compromising mucus layer integrity. Antibiotic treatment (Ant group) significantly suppressed beneficial phyla (Firmicutes and Bacteroidota) while promoting the proliferation of potentially pathogenic phyla (Spirochaetota), which may impair host metabolic functions or trigger inflammatory responses. Fecal microbiota transplantation (Fec group) restored Bacteroidota abundance by introducing exogenous microbiota, likely enhancing fiber metabolism to increase SCFA production, lower intestinal pH, inhibit pathogen growth, and reduce opportunistic pathogen colonization (e.g., Spirochaetota) through competitive exclusion.

Through the Tukey’s HSD test analysis, significant differences were observed in the composition of the microbiota at the genus level among the three groups ( Figure 4C ). Specifically, the relative abundance of Streptococcus was significantly higher in the Nor group. In the Fec group, the abundance of Treponema was significantly increased, possibly related to fecal microbiota transplantation. In the Ant group, the abundance of Lactobacillus was significantly reduced, possibly due to antibiotic treatment or other external intervention factors. These differences not only reflect the ecological adaptability changes in the gut microbiota among the groups but may also have significant impacts on the health and survival of the host. Overall, the composition and abundance of the fecal microbiota exhibit distinct group-specific characteristics, revealing the important regulatory effects of environmental, dietary, or intervention factors on the gut microbiota.

LEfSE analysis revealed that the fecal microbiota of the Fec group exhibited significant advantages in metabolic functionality and gut homeostasis regulation ( Figure 4D ). Key enriched taxa included the Eubacterium hallii group (LDA = 4.5) and the Christensenellaceae_R-7_group (LDA = 4.0), which may enhance intestinal barrier integrity and mitigate inflammation through efficient SCFA synthesis and lipid metabolism modulation. Compared to the Ant group, the Fec group lacked pro-inflammatory taxa (e.g., Deferribacteraceae) and demonstrated markedly higher abundance of butyrate producers. Relative to the Nor group, the Fec group displayed a more balanced fiber-degrading capacity and substrate utilization efficiency, driven by synergistic interactions between Clostridium leptum (LDA = 3.5) and Fibrobacter. Additionally, the enrichment of Methanobrevibacter smithii (LDA = 2.8) suggested optimized hydrogen metabolism, supporting microbial ecological stability. Collectively, these features highlight the Fec group’s microbiota as a potential paradigm for metabolic health, anti-inflammatory activity, and functional diversity, aligning closely with an idealized gut microbial profile.

Comparative functional analysis of the gut microbiota among the Nor, Fec, and Ant groups using PICRUSt2 revealed distinct metabolic and adaptive responses ( Figures 5A–E ). The Ant group exhibited hyperactivation of carbohydrate and amino acid metabolism (KEGG units: 273,962.02 vs. 250,860.09 in the Nor group) alongside enhanced DNA replication and repair (129,789.82), indicative of compensatory survival strategies under antibiotic-induced stress, such as upregulated efflux pump activity and antioxidant synthesis. In contrast, the Fec group demonstrated partial metabolic recalibration (e.g., reduced carbohydrate metabolism compared to Ant) and enrichment of fiber-degrading functions (e.g., Bacteroidota), likely attributable to the introduction of exogenous microbiota. The Nor group maintained metabolic homeostasis with minimal pathogenic activity (e.g., infectious diseases: 4,717.11). Notably, while antibiotics suppressed pathogenic pathways (e.g., cancer-related functions: 0 in Ant), they disrupted ecological equilibrium. Conversely, FMT partially restored microbial functionality through enhanced SCFA synthesis and competitive exclusion of pathogens. These findings underscore the dual role of interventions: antibiotics as a double-edged sword and FMT as a partial restorative agent, emphasizing the need for balanced therapeutic strategies to preserve microbial resilience and host health.

In this study, the correlation analysis results of metabolites and 16S rRNA are shown. As shown in Figures 6A, B , the analysis of correlations between metabolites and gut microbiota at the phylum level reveals that Firmicutes contributes to intestinal barrier protection and dietary utilization through tryptophan metabolites (e.g., the anti-inflammatory molecule indole-3-acrylic acid) and degradation of plant-derived compounds (e.g., caryophyllene oxide), while arecoline may exacerbate microbial dysbiosis by inhibiting its proliferation. Bacteroidota shows a positive correlation with secondary bile acids, suggesting its potential role in lipid metabolism regulation, whereas Proteobacteria is linked to oxidative stress markers, indicating inflammatory risks. Although the study uncovers functional modules of microbiota–metabolite interactions (anti-inflammatory and oxidative stress), experimental validation of causal relationships and optimization of statistical methods (e.g., multiple testing correction) are required to enhance reliability. Future efforts should integrate multi-omics data and clinical interventions to develop targeted microbial modulation strategies (e.g., probiotics or dietary interventions) for improving gut health and preventing metabolic diseases.

This study analyzed the correlations between Streptococcus and metabolites, revealing significant associations: positively correlated metabolites included (R)-equol (ρ = +0.583), a compound with estrogen-like activity; xanthurenic acid (ρ = +0.607), a tryptophan-derived metabolite; and panaxtriol (ρ = +0.533), a ginsenoside derivative, suggesting the potential roles of Streptococcus in host anti-inflammatory regulation, oxidative stress modulation, and plant-derived compound metabolism. In contrast, significant negative correlations with acetophenone (ρ = −0.647), tyramine (ρ = −0.649), and an unidentified metabolite RMK (ρ = −0.658) implied its involvement in metabolic balance regulation through degradation or inhibitory pathways. These findings highlight the dual functionality of Streptococcus in host-microbial metabolic interactions (promoting synthesis and substrate clearance).