Lactobacillus fermentum supplementation modulates jejunal microbiota, metabolome, and morphology in yaks under high-energy feeding

Abstract

Concentrated supplementary feeding is an effective strategy for addressing nutritional deficiencies in yaks during the cold-season grazing period. However, limited research has investigated whether long-term implementation may alter the gut microbiota and metabolite profiles of yaks, potentially posing health risks to the host. This study investigated the regulatory role of Lactobacillus fermentum in host metabolism through the gut microbiota, employing a “nutrition-gut-metabolism” perspective. Eighty healthy male Pamir yaks with similar body conditions were randomly divided into four groups (n = 20 per group), including Control (Con), Medium Energy (ME), High Energy (HE), and Medium Energy plus Lactobacillus fermentum (MEJ). Body weight was recorded every 30 days during the 150-day trial. Jejunal tissues and contents from nine yaks per group were collected for subsequent analyses. Key findings revealed that the HE group showed significantly increased crypt depth and elevated relative abundance of Clostridium_sensu_stricto_1 (P < 0.05). In contrast, the MEJ group reversed the decline in villus height and width observed in the ME group, while significantly enhancing muscularis thickness. Notably, MEJ yaks exhibited higher abundance of Paeniclostridium, Romboutsia and Treponema (P < 0.05) and markedly increased short-chain fatty acids concentrations (P < 0.001). Metabolomic analysis identified upregulated pathways, including D-amino acid metabolism and Neuroactive ligand-receptor interaction. Furthermore, comparative analyses pinpointed five critical microbial taxa (e.g., vadinBE97, Lachnospiraceae_UCG-008) and 18 key metabolites (e.g., 15-Deoxyprostaglandin J2, Lasalocid A) associated with these effects. In conclusion, Lactobacillus fermentum supplementation improved yak growth performance and mitigates metabolic risks linked to excessive energy intake. These findings provided a theoretical foundation for optimizing yak feeding strategies and advancing precision nutrition in ruminant production.

Supplementary Information

The online version contains supplementary material available at 10.1186/s42523-025-00479-9.

Background

The yak (Bos grunniens), uniquely adapted to high-altitude environments (>3,000 m), exhibits remarkable tolerance to cold, hypoxia, and low atmospheric pressure. This species constitutes a vital economic and subsistence resource for Tibetan Plateau and alpine communities [1, 2]. To address the inadequate nutrient intake in yaks caused by the significant decline in forage quality and yield during winter in high-altitude regions, strategic supplementary feeding has become a key approach to improving their nutritional status [3]. While increased dietary concentrate proportions enhance ruminant growth and productivity [4–6], studies in monogastric models reveal that high-energy diets disrupt gut microbial equilibrium, compromise intestinal barrier integrity, and impair host health [7–10]. The gut microbiota plays a vital role in regulating host metabolic processes, defending against pathogenic microorganisms, and modulating the immune system [11, 12]. Co-evolved with the host, it establishes a mutualistic relationship and serves as a key mediator for optimizing animal health and production performance [13–15]. Nevertheless, the mechanistic response of the ruminant gut microbiota to high-energy diets remains insufficiently characterized. Therefore, elucidating yak intestinal adaptations to concentrated feeding and exploring the potential role of Lactobacillus fermentum in gut repair, is of great significance for improving yak productivity and supporting the development of plateau-region economies.

Lactobacillus spp. have been extensively utilized in animal nutrition research owing to their robust environmental adaptability and bioactive properties. These bacteria exhibit high tolerance to gastric acidity and bile salts, facilitating effective intestinal colonization and sustained viability. Furthermore, they metabolically synthesize lactic acid, short-chain fatty acids (SCFAs), and antimicrobial compounds (e.g., bacteriocins and hydrogen peroxide) [16–18]. These metabolites function both as signaling molecules and energy substrates, not only reducing intestinal pH and suppressing pathogenic bacterial colonization but also mediating host immune regulation and nutritional metabolism. Collectively, they modulate intestinal microbiota homeostasis [19–21]. Currently, the application of Lactobacillus fermentum in ruminants primarily focuses on enhancing production performance, regulating rumen metabolism potential, and increasing milk yield in dairy cows. For instance, in vitro fermentation trials with dairy cows have demonstrated that Lactobacillus plantarum can enhance the fermentation efficiency and fiber degradation capacity of high-yield dairy cow diets, supporting its feasibility as a direct-fed microbial (DFM) [22]. Moreover, Lactobacillus-based formulations have been applied for udder care in dairy cows, modulating the udder microbiota and reducing mastitis risk [23, 24]. Furthermore, supplementation of milk replacer for pre-weaning calves with a probiotic mixture containing Lactobacillus and Bacillus spp. significantly enhanced average daily gain and feed conversion efficiency, without compromising daily feed intake or skeletal growth [25]. In a study on Sunite sheep, supplementation with Lactobacillus activated the AMPK/PGC-1α signaling pathway. This activation promotes mitochondrial biogenesis and induces a shift toward oxidative muscle fiber types. Consequently, it significantly enhanced meat quality and energy metabolism efficiency. Furthermore, supplementation with Lactobacillus fermentum optimized the rumen microbial community structure and modulated signaling pathways associated with muscle protein synthesis. Additionally, it elevated the content of intramuscular aromatic compounds, thereby further improving meat tenderness and flavor [26, 27]. Recent studies demonstrate that long-term supplementation with Lactobacillus-based DFMs in dairy cows significantly optimizes rumen microbial composition [28]. This intervention increases the abundance of beneficial genera such as Eubacterium_Q and Sharpea, promoting rumen ecosystem homeostasis. These findings provide new perspectives for microbial strategies to enhance dairy cow performance and health management, highlighting the feasibility and broad potential of Lactobacillus fermentum as a dietary modulator.

Therefore, we hypothesized that Lactobacillus fermentum supplementation modulates jejunal microbiota composition, thereby mediating host energy metabolism and maintaining intestinal homeostasis. We measured the body weight of 80 healthy Pamir male yaks in different groups for 150-day trial and collected jejunal contents and tissues from 36 randomly selected individuals. We employed an integrated approach combining histomorphology, 16 S rRNA sequencing, LC-MS-based untargeted metabolomics, and SCFAs profiling to systematically investigate how Lactobacillus fermentum restructures jejunal microbiota and metabolite profiles, and the underlying regulatory mechanisms. This work provides novel insights into microbiota-host dynamics following Lactobacillus fermentum supplementation in ruminants. Our findings establish a theoretical foundation for optimizing nutritional interventions and developing microbiota-directed yak supplementation strategies, ultimately enhancing livestock productivity and supporting sustainable high-altitude pastoral economies.

Methods

Animals, feeding regime, and body weight determination

Eighty healthy male yaks with comparable body conditions were selected from the Mahong Livestock Farm in Kashgar, Xinjiang, China, and were randomly assigned to four groups, each consisting of 20 yaks. All experimental yaks were subjected to one of the following dietary treatments. The control group’s yaks grazed on the pasture throughout the entire experimental period (Con: Neg 0.11 MJ/kg; the blank control group), a medium-energy diet group (ME: Neg 2.12 MJ/kg, with 5 kg of the mixed diet per capita per day), a high-energy diet group (HE: Neg 2.69 MJ/kg, with 5 kg of the mixed diet per capita per day), and a medium-energy diet with Lactobacillus fermentum supplementation group(MEJ: Neg 2.12 MJ/kg, with 5 kg of the mixed diet per capita per day and administered 0.02% of the Lactobacillus fermentum powder (0.2 g/kg feed) throughout the entire experimental period). The Lactobacillus fermentum utilized was a commercial live bacterial preparation, procured from Shandong Weier Bioengineering Co., Ltd, with a viable count of ≥ 1 × 10⁹ CFU/g, meeting the national standards for active probiotics. The trial commenced on November 1, 2023, and concluded on April 18, 2024, spanning a total of 170 days. A 20-day adaptation period was implemented, during which the proportion of concentrate feed was gradually increased. The subsequent 150 days constituted the data collection phase. Prior to the commencement of the trial, all test animals were dewormed, and their body weights were recorded every 30 days. Feeding and grazing were conducted according to a pre-established schedule, with initial and final body weights meticulously recorded, and the total weight calculated. Throughout the entire experimental period, all captive yaks were allowed to freely consume the designated feed twice daily, at 10:00 AM and 6:00 PM. Fresh drinking water was continuously provided during the trial. The feeding strategy was formulated based on the measurement results of feed intake of grazing yaks in the Pamir region, and complies with the recommendations of the “Chinese Beef Cattle Feeding Standard” (NY/T 815–2004). The concentrate feed required for the experiment was custom-produced by Tianjian (Shandong) Agricultural Technology Co., Ltd. The experimental feeding plan involved mixing concentrate feed and roughage (wheat straw) in equal proportions. The nutritional composition of the mixed feed was determined in the laboratory of Qilu Normal University using the“Feed Analysis and Quality Inspection Technology”. The specific component composition and nutritional levels were detailed in Table S1.

Jejunal sample collection and measurement

Following the trial, all yaks underwent a 24 h fasting period with 8 h water restriction. Nine yaks per group were then randomly selected, euthanized via captive bolt stunning, and exsanguinated via carotid artery transection. Jejunal contents and tissues were immediately collected, placed in sterile 5 mL centrifuge tubes, snap-frozen in liquid nitrogen, and stored at -80 °C. For histology, jejunal tissues were fixed in 4% paraformaldehyde, dehydrated, trimmed, paraffin-embedded, sectioned (5 μm), and stained (H&E). Morphological assessments were performed using light microscopy. Villus height, villus width, crypt depth, and muscularis thickness were quantified using Motic Images Advanced 3.2 software.

16 S rRNA gene amplification and Miseq sequencing

Jejunal content samples were immediately snap-frozen and stored at -80°C. Bacterial DNA was extracted using the DNeasy PowerSoil Kit (Qiagen, Germany) according to the manufacturer’s protocol. DNA concentration and purity were determined using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, USA) and 1% agarose gel electrophoresis, with subsequent storage at -20°C. The hypervariable V3-V4 region of the bacterial 16S rRNA gene was amplified via polymerase chain reaction (PCR) in 25 µL reactions using universal primers 343F (5’-TACGRAGGCAGCAG-3’) and 798R (5’-AGGGTATCTAATCCT-3’). Resulting amplicons were purified using AMPure XP beads (Beckman Coulter, USA), quantified with the Qubit dsDNA HS Assay Kit on a Qubit 4.0 fluorometer, and pooled at equimolar concentrations. The pooled library was sequenced on an Illumina NovaSeq 6000 platform (250-bp paired-end reads) by Shanghai OE Biotech Co., Ltd. (China).

Sequence and processing of jejunal flora

Library preparation, sequencing, and subsequent data analysis by OE Biotech Co., Ltd. (Shanghai, China). Raw FASTQ files were processed using Cutadapt (v3.5) for primer removal. Paired-end reads were then quality-filtered, denoised, merged, and subjected to chimera removal via DADA2 (v1.24.0) in QIIME 2 (v2022.8) [29, 30], generating amplicon sequence variants (ASVs) and representative sequences. Taxonomic assignment of ASVs was conducted against the SILVA 138 database using q2-feature-classifier (default parameters). Alpha diversity (Chao1 [31], Shannon [32]) and beta diversity (unweighted UniFrac distances) were calculated in QIIME 2. Principal Coordinates Analysis (PCoA) of distance matrices was performed in R (v4.2.1). Differential abundance analysis employed Kruskal-Wallis tests for multi-group comparisons and Student’s t-tests for pairwise contrasts, with false discovery rate (FDR) correction.

Metabolite extraction and quality control

A 100 mg aliquot of each sample was homogenized in a 1.5 mL microcentrifuge tube with a 6 mm grinding bead and 400 µL of ice-cold methanol: water (4:1, v/v) containing 4 µg/mL L-2-chlorophenylalanine (internal standard). After incubation at -40 °C for 2 min, samples were bead-beaten at 60 Hz for 2 min. The homogenate subsequently underwent ultrasonic extraction in an ice-water bath (10 min), followed by incubation at -40 °C for 30 min. After centrifugation (12,000×g, 10 min, 4 °C), 300 µL of supernatant was transferred to an LC-MS vial and evaporated to dryness under nitrogen. The residue was reconstituted in 300 µL methanol: water (1:4, v/v), vortex-mixed (30 s), and sonicated in an ice-water bath (3 min). Following incubation at -40 °C (2 h) and centrifugation (12,000×g, 10 min, 4 °C), 150 µL of supernatant was filtered through a 0.22 μm PVDF syringe filter. Filtrates were stored at -80 °C pending LC-MS analysis. Quality control (QC) samples were prepared by pooling equal aliquots from all samples and processed identically to monitor analytical stability.

LC-MS detection

Untargeted metabolomic analysis was performed using an ACQUITY UPLC I-Class + system coupled to a Q-Exactive + HF-X mass spectrometer (Thermo Fisher Scientific) with heated electrospray ionization (H-ESI). Chromatographic separation was achieved on an ACQUITY UPLC HSS T3 column (1.8 μm, 2.1 × 100 mm) maintained at 45 °C. The gradient elution system consisted of (A) water with 0.1% formic acid (v/v) and (B) acetonitrile with 0.1% formic acid (v/v), with the following gradient: 0.01 min, 5% B; 2 min, 5% B; 4 min, 30% B; 8 min, 50% B; 10 min, 80% B; 14 min, 100% B; 15 min, 100% B; 15.1 min, 5% B; and 16 min, 5% B. The flow rate was 0.35 mL/min, and the column temperature was maintained at 45 °C. All samples were maintained at 10 °C during the analysis. The injection volume was 2 µL. The mass range was from 100 m/z to 1200 m/z. The resolution of MS1 scan was 70,000, and the resolution of MS2 scan was 17,500. The collision energies were 10, 20, and 40 eV, respectively. The mass spectrometer was operated as follows: spray voltage, 3800 V (+) and 3200 V (−); sheath gas flow rate, 35 arbitrary units; auxiliary gas flow rate, 8 arbitrary units; capillary temperature: 320 °C.

Metabolome statistical analysis

LC-MS data processing was performed using Progenesis QI V2.3 software (Nonlinear Dynamics, Newcastle, UK). This included baseline filtering, peak detection, peak integration, retention time correction, peak alignment, and normalization. Key processing parameters were set as follows: precursor mass tolerance of 5 ppm, fragment ion mass tolerance of 10 ppm, and a fragment ion match threshold of 5%. Compounds were identified based on accurate mass-to-charge ratio (m/z), MS/MS fragmentation spectra, and isotopic distribution. Annotation was achieved by matching against the Human Metabolome Database (HMDB), LipidMaps (V2.3), Metlin, and an in-house database. Following feature extraction, peaks exhibiting missing values (ion intensity = 0) in more than 50% of samples within any experimental group were removed. Remaining zero values were imputed with half of the minimum detectable intensity value observed in the dataset. Features were further filtered based on identification confidence; metabolites with a database matching score below 36 (out of a maximum of 80) were deemed unreliable and excluded. Finally, data acquired in positive and negative ionization modes were combined into a single unified data matrix.

The processed data matrix was imported into R for multivariate statistical analysis. Principal component analysis (PCA) was first performed to assess overall sample distribution and analytical process stability. To identify metabolic differences between experimental groups, orthogonal partial least squares-discriminant analysis (OPLS-DA) and partial least squares-discriminant analysis (PLS-DA) were subsequently applied. Model quality and overfitting prevention were evaluated using seven-fold cross-validation and 200 response permutation tests (RPT). Variable importance in projection (VIP) scores derived from the OPLS-DA model ranked the contribution of individual metabolites to group discrimination. Additionally, two-tailed Student’s t-tests were conducted to assess the statistical significance of inter-group metabolic differences. Differential metabolites were defined by a VIP score > 1 and a P-value < 0.05. Metabolic pathway enrichment analysis for these differential metabolites was performed using the KEGG database, with significance defined as a P-value < 0.05.

SCFAs determination

The concentrations of five SCFAs, including acetic acid, propionic acid, butyric acid, isobutyric acid, and isovaleric acid, were quantified using ultra-performance liquid chromatography coupled with electrospray ionization tandem mass spectrometry (UPLC-ESI-MS/MS). Quantitative analysis was performed in multiple reaction monitoring (MRM) mode using a triple quadrupole mass spectrometer. Concurrently, bacterial genomic DNA was extracted from gastrointestinal contents using the DNeasy PowerSoil DNA Extraction Kit (Qiagen, Hilden, Germany) according to the manufacturer’s protocol. DNA concentration and purity were assessed using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, USA) and integrity was verified by 1% agarose gel electrophoresis. Purified DNA was stored at -20°C for subsequent analysis. The hypervariable V3-V4 region of the bacterial 16S rRNA gene was amplified via PCR in 25 µL reactions using universal primers 343F (5’-TACGRAGGCAGCAG-3’) and 798R (5’-AGGGTATCTAATCCT-3’). Amplification products were purified using AMPure XP beads (Beckman Coulter, USA), quantified with the Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific, USA), and pooled in equimolar amounts. Paired-end sequencing libraries were constructed from the pooled amplicons and sequenced (2 × 250 bp) on an Illumina NovaSeq 6000 platform.

Statistical analysis

Prior to statistical analysis, data distributions were examined for outliers using box plots and assessed for normality using the Shapiro-Wilk test. For growth performance and jejunal development indices, outliers identified by the Grubbs test were removed. Intergroup differences for these parameters were then evaluated using independent samples t-tests in SPSS 20.0 (IBM, USA). Results were presented as means ± standard deviation (SD). For omics data, P-values were adjusted for multiple testing using the false discovery rate (FDR) method. Standard bioinformatics pipelines were applied to process 16S rRNA sequencing data for microbial taxa identification. Univariate analyses (independent samples t-tests or analysis of variance, as appropriate) were used to compare the relative abundance of microbial taxa between groups and identify composition differences. Metabolites meeting the criteria of VIP >1 and FDR-adjusted P-value < 0.05 were defined as differential. These differential metabolites were further assessed for significance between groups using univariate t-tests. Statistical significance for omics analyses was defined as FDR < 0.05. Given the high diversity, dimensionality, and discrete nature of omics data, relationships between key jejunal microbiota (identified as differentially abundant) and differential metabolites were explored using Spearman’s rank correlation analysis. Correlation coefficients and their significance (P < 0.05) were visualized using a clustered heatmap generated with the heatmap package in R [33, 34].

Result

Growth performance and jejunum development of yaks

Growth performance characteristics of yaks were summarized in Tables 1 and 2. During the fattening period, the HE group exhibited significantly higher final body weight, total weight gain, and average daily gain compared to both the ME and control groups (P < 0.001; Table 1). These results indicated that higher dietary energy intake enhances yak growth performance. Supplementation with Lactobacillus fermentum significantly increased final body weight relative to the ME group (P < 0.05; Table 2).

Table 1.

Growth performance of yaks fed diets differing in energy levels

Items	Group (mean ± SD)
	Con	ME	HE	P-value
Initial weight (kg)	252.20 ± 9.57	245.30 ± 7.57	252.10 ± 12.53
Final weight (kg)	302.10 ± 13.247c	323.30 ± 19.23b	345.00 ± 22.40a	P<0.001
Total weight gain (kg)	49.9 ± 9.81b	78.00 ± 19.44a	92.9 ± 19.22a	P<0.001
Average daily weight gain (g/d)	415.83 ± 81.79b	650.00 ± 162.02a	774.17 ± 160.22a	P<0.001

Table 2.

Growth performance of yaks in the ME group supplemented with Lactobacillus fermentum

Items	Group (mean ± SD)
	ME	MEJ	P-value
Initial weight (kg)	245.30 ± 7.57	258.20 ± 9.60
Final weight (kg)	323.30 ± 19.23b	351.80 ± 19.13a	P<0.004
Total weight gain (kg)	78.00 ± 19.44	93.60 ± 15.06	P = 0.06
Average daily weight gain (g/d)	650.00 ± 162.02	780.00 ± 125.54	P = 0.06

Jejunal morphology measurements were presented in Fig. 1. Villus length, villus width, and muscularis thickness were significantly greater in the Con group compared to both the ME and HE groups (P < 0.001; Fig. 1A, B). Conversely, crypt depth was significantly higher in the HE group than in the Con and ME groups (P < 0.001; Fig. 1A, B). Supplementation with Lactobacillus fermentum significantly increased villus length, villus width, and muscularis thickness relative to the ME group (P < 0.001; Fig. 1C, D).

Fig. 1.

Jejunal morphology in yaks under different dietary treatments. (A) Representative histological sections of jejunal tissues from Con, ME, and HE groups. Key morphometric parameters: villus length (Vl), villus width (Vw), crypt depth (Cd), muscularis thickness (MI). Scale bar: 200 μm. (B) Quantitative comparison of jejunal morphometric parameters among Con, ME, and HE groups. Data = mean ± SD; ***P < 0.001 (one-way ANOVA). (C) Representative histological sections of jejunal tissues from ME and MEJ groups. Scale bar: 200 μm. (D) Quantitative comparison of jejunal morphometric parameters between ME and MEJ groups. Data = mean ± SD; ***P < 0.001 (Student’s t-test)

Composition and diversity of jejunal microbiota assessed by 16 S rRNA sequencing

Jejunal chyme samples (n = 36) from four treatment groups underwent 16 S rRNA gene sequencing, yielding 28,798,721 raw sequences. Following quality filtering, denoising, and chimera removal, 2,336,405 high-quality sequences were retained (mean ± SD: 64,900 per sample; Table S2). Good’s coverage exceeded 99% for all samples (Fig. S1), confirming adequate sequencing depth for microbial diversity analysis. High-quality sequences were clustered into 26,957 ASVs. ASV counts for each group were Con (5,977), ME (6,920), HE (6,585), and MEJ (7,511) (Table S2). The MEJ group exhibited the highest ASV richness, suggesting that Lactobacillus fermentum supplementation enhanced microbial richness in the yak jejunum.

The gut microbiota diversity among different treatment groups was evaluated using the Shannon index, Simpson index, and Chao1 index. Significant overall differences in Shannon index were observed among Con, ME, and HE groups (P < 0.001), with MEJ also differing significantly from ME (P < 0.05; Fig. 2A). Simpson index differed significantly between Con and HE (P < 0.001) and between ME and MEJ (P < 0.05; Fig. 2B). Chao1 index showed significant variation among Con, ME, and HE groups (P < 0.001), while MEJ exhibited significantly higher richness than ME (P < 0.001; Fig. 2C). PCoA based on Bray-Curtis distances revealed distinct clustering patterns (Fig. 2D-E). Con and ME groups showed partial overlap, while HE formed a discrete cluster separate from both. ME and MEJ groups exhibited clear separation (Fig. 2E), indicating that both dietary energy levels and Lactobacillus fermentum supplementation significantly restructure jejunal microbial communities.

Fig. 2.

Gut microbiota characteristics in yaks under dietary interventions. (A-C) α-Diversity indices: Shannon index, Simpson index, and Chao1 index. Data = mean ± SD; *P < 0.05, **P < 0.01, ***P < 0.001 (one-way ANOVA with Tukey’s post-hoc test; Student’s t-test for ME vs. MEJ). (D) PCoA of Con, ME, and HE groups based on Bray-Curtis dissimilarity (PC1/PC2 shown). (E) PCoA comparing ME and MEJ groups. (F) Phylum-level community composition (top 15 taxa). (G) Spirochaetota abundance: *P < 0.05 MEJ vs. HE (t-test). (H) Genus-level community composition (top 15 genera). (I) Treponema abundance: *P < 0.05 MEJ vs. HE (t-test)

We further characterized microbial community shifts under dietary interventions through taxonomic profiling. After classifying the ASV representative sequences of each sample, it was found that a total of 22 bacterial phyla were shared among the four treatment groups at the phylum level. Further analysis of the top 15 most abundant phyla revealed that Firmicutes, Proteobacteria, and Bacteroidetes were the dominant phyla in the four treatment groups. The relative abundance of the dominant phyla in each treatment group were as follows: Con group (Firmicutes: 39.52%, Proteobacteria: 26.97%, Bacteroidetes: 23.77%), ME group (45.52%, 23.95%, 22.86%), HE group (54.99%, 20.25%, 15.35%), and MEJ group (56.58%, 14.27%, 21.44%) (Fig. 2F). Compared with the Con group, the relative abundance of Firmicutes in the ME and HE groups showed a gradual increasing trend, while the relative abundance of Bacteroidetes showed a decreasing trend, indicating that energy intake level significantly affects the balance between these two major bacterial phyla. The abundance of Firmicutes further increased in the MEJ group, while Bacteroidetes partially recovered. The abundance of Proteobacteria was higher in the Con group and lowest in the MEJ group. Notably, the abundance of Actinobacteria increased in the HE group but decreased after intervention with Lactobacillus. In addition, Spirochaetota gradually decreased in the Con, ME, and HE groups but significantly increased in the MEJ group (P < 0.05; Fig. 2G).

Genus-level analysis showed differences in microbial community composition among the treatment groups, with a total of 442 genera identified across the 4 treatment groups. As shown in Fig. 2H, Haemophilus, Paeniclostridium, and Romboutsia were the predominant genera. Among the top 15 identified dominant genera, it was found that with the increase in dietary energy levels, the abundance of Clostridium_sensu_stricto_1 increased from 3.37% to 10.09%, while the abundance of Streptococcus gradually decreased, indicating that energy levels have a regulatory effect on the abundance of these genera. Under Lactobacillus fermentum supplementation, the abundance of Haemophilus was the lowest in the MEJ group, whereas the abundance of Paeniclostridium and Romboutsia reached their highest levels in this group. Notably, the abundance of Treponema in the MEJ group significantly increased to 3.92%, which was markedly higher than in the other groups (P < 0.05; Fig. 2I).

LEfSe analysis identified differentially abundant bacterial taxa across treatment groups (Fig. 3). As shown in Fig. 3A and 14 bacterial taxa were significantly enriched in the Con group, including f_Lactobacillaceae, g_Lactobacillus, f_Muribaculaceae, f_Burkholderiaceae, g_Muribaculaceae, and g_Lautropia. In the ME group, g_Bacteroides, f_Bacteroidaceae, and g_Ruminococcus_torques_group were significantly enriched, while in the HE group, only g_Pseudobutyrivibrio was significantly enriched. Figure 3B showed that in the ME group, three bacterial taxa including g_Mitsuaria were significantly enriched, whereas in the MEJ group, 17 bacterial taxa including f_Lactobacillaceae, g_Lactobacillus, o_Mycoplasmatales, and f_Mycoplasmataceae were significantly enriched.

Fig. 3.

LEfSe identification of discriminant microbial biomarkers at genus level. (A) Differentially abundant genera across Con, ME, and HE groups (LDA score > 3); (B) Differential genera between ME and MEJ groups (LDA score > 3)

Differential bacterial genera were identified in genus-level comparisons in two comparisons, including ME vs. Con vs. HE groups and ME vs. MEJ groups. Five key differential genera emerged, including Lactobacillus, Pseudobutyrivibrio, vadinBE97, Lachnospiraceae_UCG-008, and JGI_0000069-P22. Lactobacillus and JGI_0000069-P22 exhibited the most pronounced differences (P < 0.001). In addition, JGI_0000069-P22, Lactobacillus, and Pseudobutyrivibrio were identified as key microbial markers in the LEfSE analysis, further confirming their important roles under different treatment conditions. These genera exhibited significant abundance shifts in response to combined dietary energy modifications and Lactobacillus fermentum supplementation, suggesting their potential role as key bacterial mediators in the ileal microbial environment of yaks, where probiotic intervention and energy intake jointly exert regulatory effects.

Effects of dietary energy levels and Lactobacillus fermentum supplementation on yak jejunal metabolism

PCA score plots showed a significant separation between the ME group and the HE group, a certain degree of overlap between the ME group and the Con group, and a clear separation between the ME group and the MEJ group (Fig. 4A and B). The R² and Q² values of the response permutation tests for the two comparison groups were (0.0, 0.307) and (0.0, -0.762), (0.0, 0.566) and (0.0, -0.64), respectively. All samples in the score plots fell within the 95% Hotelling T² ellipse, which indicated that the model had good predictability and reproducibility, with no overfitting observed. This made it suitable for subsequent data analysis (Fig. 4C and D).

Fig. 4.

(A and B) PCA score plots of metabolite data between the two comparison groups. (C and D) Corresponding orthogonal partial least squares discriminant analysis model evaluation parameters (R² and Q²), used to assess the model’s fitting and predictive ability

Volcano plots revealed significant metabolic alterations between groups (Fig. 5). Volcano plots were used to visualize the significant features between groups to identify key differences (Fig. 5). A total of 3,329 metabolites were identified in the jejunal contents of the ME group and the Con group, among which 669 differential metabolites were identified (416 downregulated and 253 upregulated) (Fig. 5A). Among them, Lactacystin, Enkephalin L, and 15-Lipoxygenase Inhibitor 1 were significantly downregulated, while Tetrofosmin and 8-Deoxylactucin were significantly upregulated. Between the ME group and the HE group (Fig. 5B), a total of 1,135 differential metabolites were identified (302 downregulated and 833 upregulated). Difloxacin and Spermidine were significantly downregulated, while Brosimacutin C and Tris(2-butoxyethyl) phosphate were significantly upregulated. Between the ME group and the MEJ group (Fig. 5C), a total of 1,559 differential metabolites were identified (597 downregulated and 962 upregulated), with Crocin 3 and Androsta-1,4,6-triene-3,17-dione significantly downregulated, and Zymosterol and Pentalinonsterol identified as upregulated metabolites.

Fig. 5.

LC-MS metabolomic profiling of yak jejunal contents. (A-C) Volcano plots of differential metabolites: (A) ME vs. Con; (B) ME vs. HE; (C) ME vs. MEJ; Red dots represent significantly upregulated metabolites, blue dots represent significantly downregulated metabolites, and gray dots represent metabolites with no significant difference, with screening criteria of P<0.05 and VIP>1.0; (D–F) KEGG pathway enrichment bubble charts (top 20 pathways): (A) ME vs. Con; (B) ME vs. HE; (C) ME vs. MEJ; Y-axis: names of enriched metabolic pathways; X-axis: enrichment factor; bubble size: number of metabolites involved; color gradient: -log₁₀(P-value)

KEGG analysis results showed that the differential metabolites between the ME group and the Con group were involved in 44 metabolic pathways, among which significant enrichment was observed in the purine metabolism pathway (P < 0.05) (Fig. 5D). Between the ME group and the HE group, differential metabolites were significantly enriched in 20 metabolic pathways, including primary bile acid biosynthesis, arginine and proline metabolism, and D-amino acid metabolism (P < 0.001) (Fig. 5E). Between the ME and MEJ groups, differential metabolites were mainly enriched in pathways such as D-amino acid metabolism, Neuroactive ligand-receptor interaction, protein digestion and absorption, and aminoacyl-tRNA biosynthesis (P < 0.05) (Fig. 5F). These results indicated that Lactobacillus fermentum supplementation had a significant impact on the jejunal metabolome of yaks.

To further identify key metabolites, we conducted a cross-analysis of differential metabolites across the four groups. As illustrated in Fig. 6A, we identified a total of 18 representative differential metabolites, which were classified into five categories: lipids and lipid-like molecules, phenylpropanoids and polyketides, organic acids and derivatives, organophosphorus compounds, and benzenoids (Table S3). Cluster analysis revealed distinct group-specific clustering patterns (Fig. 6B), with conserved directional trends in metabolite abundance across interventions. These key metabolites might therefore mediate metabolic adaptation of the yak jejunum to dietary modifications and gut microbiota restructuring.

Fig. 6.

(A) Venn diagram of shared metabolites across treatment groups; (B) Clustered heatmap of 18 key metabolites; color gradient indicates relative abundance levels across groups

Effects of dietary energy levels and Lactobacillus fermentum supplementation on yak jejunal SCFAs

The HE group exhibited substantially elevated concentrations of total acid, acetic acid, propionic acid, butyric acid, isovaleric acid, and isobutyric acid compared to Con and ME groups (P < 0.001). Furthermore, Lactobacillus fermentum supplementation significantly enhanced SCFA production versus the ME group, with pronounced increases in acetic acid, propionic acid, butyric acid (P < 0.001) (Fig. 7).

Fig. 7.

Effects on SCFAs contents in jejunal chyme of yaks in each group. (A) Total volatile fatty acids; (B) Acetic acid; (C) Propionic acid; (D) Butyric acid; (E) Isovaleric acid; (F) Isobutyric acid. Data were presented as mean ± standard deviation (n = 9). Significance level: *P < 0.05, **P < 0.01, ***P < 0.001

Correlation analysis of key differential microbial communities with jejunal development Indicators, SCFAs, and differential metabolites

Spearman correlation analysis revealed associations between key jejunal genera and intestinal development markers, SCFAs, and differential metabolites (Fig. 8). The results showed that vadinBE97 generally exhibited an opposite trend to JGI_0000069-P22, Lachnospiraceae_UCG-008, Lactobacillus, and Pseudobutyrivibrio. Specifically, JGI_0000069-P22 and Lachnospiraceae_UCG-008 showed strong positive correlations with muscularis thickness and villus width (P < 0.05), but negative correlations with acetic acid and propionic acid (P < 0.05) (Fig. 8A). vadinBE97 demonstrated significant negative correlations with lsobutyric acid and lsovaleric acid (P < 0.001) and positive correlation with Propionic acid (P < 0.001) (Fig. 8B). Metabolite-level correlations (Fig. 8C) revealed vadinBE97 positively correlated with Medicinine and negatively with Lasalocid A and Tuberoside (P < 0.001). JGI_0000069-P22 showed significant positive correlations with multiple metabolites (P < 0.001). These correlations reflected coordinated regulation of intestinal ecology and function through combined Lactobacillus fermentum supplementation and dietary energy modulation, which provided mechanistic insights into microbe-metabolite-host interactions.

Fig. 8.

Correlation analyses of key microbial taxa with jejunal development indicators, SCFAs and metabolites. (A) Correlation heatmap between key identified microorganisms and jejunal developmental parameters. (B) Correlation heatmap between key microorganisms and identified SCFAs. (C) Correlation heatmap between key microorganisms and identified metabolites. Rows and columns represent microbial genera and metabolites (or jejunal indicators), respectively, with cell colors indicating Pearson correlation coefficients. Significance levels: *P < 0.05, **P < 0.01, ***P < 0.001

Discussion

During cold seasons, supplementary feeding of ruminants could effectively alleviate the problem of insufficient nutrient intake caused by the decline in forage quality. However, although long-term feeding with high-energy diets improved growth performance, it could also exert adverse effects on intestinal health. Currently, practical and effective microecological strategies to mitigate these effects remain were lacking. To this end, we conducted an integrated omics-based study in Pamir yaks to systematically evaluate the effects of Lactobacillus fermentum supplementation on jejunal microecology and metabolic homeostasis. Our results suggested that Lactobacillus fermentum might mediate the regulation of host energy metabolism pathways by modulating key differential microbiota in the jejunum. In this study, HE group increased significantly, thereby confirming the beneficial impact of elevated energy levels on growth promotion. Notably, MEJ group also demonstrated a significant increase compared to the ME group, suggesting that Lactobacillus supplementation could effectively promote growth independent of energy level alterations. Furthermore, given that this experiment utilized a fixed-point quantitative feeding method, which controlled for feeding amounts and other confounding factors, the observed weight differences can primarily be attributed to the effects of energy levels and the intervention of fermentation lactobacillus. Therefore, weight gain can be primarily attributed to the increased dietary energy level and the synergistic enhancement of nutrient utilization efficiency mediated by Lactobacillus fermentum. Furthermore, the intervention with Lactobacillus fermentum not only reversed the reduction in the height and width of the jejunal villi but also promoted an increase in the muscularis thickness. Therefore, we speculated that Lactobacillus fermentum might synergistically improve intestinal morphological structure by enhancing barrier function, thereby enhancing its absorption and defense capabilities. These combined effects could ultimately lead to enhanced production performance in ruminants.

Subsequently, we found that Bacteroidota, Firmicutes, and Proteobacteria constituted the core gut microbiota of yaks, which was consistent with previous studies [35, 36]. Proteobacteria has been proven to play an important role in soluble carbohydrate degradation, biofilm formation, and fermentation of ruminal contents [37]. In this study, the abundance of this phylum was higher in free-range yaks, suggesting that it played a key role not only in energy acquisition but also in maintaining intestinal homeostasis and microecological balance. Even under barn-feeding fattening conditions, attention should be given to the ecological niche maintenance and proper regulation of functional microbial communities, including Proteobacteria. Additionally, we found that Clostridium_sensu_stricto_1 was significantly enriched in the HE group. Previous studies had reported that increased abundance of Clostridium_sensu_stricto_1 is closely associated with inflammatory responses and severe intestinal infections in polycystic ovary syndrome-insulin resistance (PCOS-IR) models [38, 39], suggesting that a high-energy diet might induce dysbiosis of the yak gut microbiota. These findings underscored the importance of maintaining intestinal microbial homeostasis and nutritional regulation balance during high-energy fattening management. Paeniclostridium was significantly enriched in the MEJ group, suggesting that Lactobacillus fermentum supplementation might promote the colonization and activity of this acid-producing functional bacterium. Considering that it belongs to the Clostridium group with probiotic potential and plays a key role in SCFA synthesis [40, 41], its increase abundance could improve the intestinal metabolic environment and enhance immune regulation by elevating SCFA levels in the gut. This observation supported the notion that Lactobacillus fermentum indirectly regulates host intestinal homeostasis through “metabolic pathway remodeling”. We also found that the abundance of Spirochaetota and its representative genus Treponema was significantly higher than that in other treatment groups. These microbes have been shown to be closely associated with cellulose degradation, SCFA production, and pectin breakdown, providing a stable energy source for the host [42–44]. Combined with the SCFA results, this data suggested that Lactobacillus fermentum might enhance the colonization and metabolic activity of key acid-producing bacteria such as Paeniclostridium and Treponema, thereby supplying a stable energy source to the host and improving the intestinal metabolic environment and immune status.

Furthermore, we speculated that vadinBE97 might act as a key metabolic responder involved in the regulation of host energy metabolism and intestinal development. Previous studies have shown that vadinBE97 exhibits distinct abundance patterns in the rumen microbiota of dairy cows throughout the entire lactation cycle, with a marked decrease during the lactation stage. Its abundance is closely associated with the activity of energy metabolism pathways in the rumen, such as the tricarboxylic acid (TCA) cycle and SCFAs synthesis [45]. In this study, vadinBE97 showed a highly significant positive correlation with propionate, which is a key precursor of the TCA cycle and directly participates in host energy metabolism pathways. This suggested that vadinBE97 might enhance the host’s energy supply capacity by promoting propionate synthesis or the metabolism of related substrates. Meanwhile, vadinBE97 was significantly negatively correlated with branched-chain SCFAs such as isovaleric acid and isobutyric acid. These metabolites were primarily derived from the fermentation of branched-chain amino acids and are often associated with excessive protein fermentation and intestinal barrier impairment. Therefore, this bacterium might also have a potential role in regulating the degree of protein fermentation and maintaining intestinal homeostasis. The above results suggested that vadinBE97 might occupy a central position in microbial community stability or signaling regulation. We speculated that vadinBE97, under Lactobacillus fermentum supplementation, participates in the regulation of host energy metabolism and might synergistically influence intestinal development and microecological stability. Further investigation into its functional mechanisms and application potential should be warranted. The variation in Lachnospiraceae_UCG-008 was particularly noteworthy. Previous studies have shown that the abundance of this genus is significantly higher in wild Sichuan takins during spring than in winter, which may be closely related to the host’s metabolic adaptation to seasonal nutritional changes and improved nutrient absorption efficiency [46]. In our study, this genus was significantly positively correlated with muscularis thickness and villus width. Combined with its characteristics as a dominant SCFAs-producing bacterium [47], we speculated that it might improve intestinal development by enhancing intestinal barrier function and promoting epithelial cell repair. Furthermore, the positive correlations between Lachnospiraceae_UCG-008 and isovalerate and isobutyrate further supported its functional role in carbon source fermentation and energy metabolism. At the metabolic level, Lachnospiraceae_UCG-008 was also significantly positively correlated with 15-Deoxyprostaglandin J2 (15d-PGJ2). 15d-PGJ2 has been shown to exert anti-inflammatory and anti-tumor effects by inhibiting the PI3K/Akt signaling pathway, suggesting that this genus might participate in host immune homeostasis by modulating lipid metabolism-related signaling pathways. We speculated that Lachnospiraceae_UCG-008 might serve as a metabolic signaling mediator linking the gut microbiota to host metabolic responses, and play a multifaceted role in maintaining intestinal health and enhancing immune regulation. Notably, JGI_0000069-P22 exhibited highly significant positive correlations with multiple intestinal development and metabolic indicators in the correlation analysis, suggesting its potential biological role in regulating host intestinal functions. However, research on this genus remains limited. Previous studies have shown that JGI_0000069-P22 belongs to the phylum Patescibacteria, which is predominantly composed of uncultured microorganisms characterized by small genomes, limited metabolic capabilities, and a strong dependence on hosts. Members of this phylum have been proposed to function as symbiotic or epibiotic microbes in various ecosystems [48, 49]. Based on its strong correlations observed in this study and the potential symbiotic ecological role of Patescibacteria, we speculated that JGI_0000069-P22 might participate in the dynamic balance of the yak small intestinal microecology by modulating metabolic networks, influencing microbial interactions, or forming specific metabolic complementarity with the host. Although clear functional annotations for this genus are currently lacking, its stable and significant response characteristics indicated that it might serve as a sensitive indicator microbial group under Lactobacillus fermentum supplementation or dietary energy level changes, warranting further exploration of its ecological functions and host interaction mechanisms in future studies. In the metabolic pathway analysis, we found that the differential metabolites between the ME and MEJ groups were significantly enriched in key pathways such as D-amino acid metabolism and Neuroactive ligand-receptor interaction. Studies have shown that mammals primarily utilize L-amino acids, whereas gut microbiota can synthesize or utilize D-amino acids (such as D-alanine and D-glutamate). These metabolites not only participate in the formation of bacterial cell walls (e.g., peptidoglycan) but also regulate microbial interactions, metabolic activity, and the stability of the intestinal environment [50]. Some D-amino acids can even enter the host system and participate in physiological processes such as neural regulation, immune responses, and redox balance [51, 52]. Notably, we found that Paeniclostridium and Treponema might be potentially associated with the D-amino acid metabolism pathway. These two bacterial groups are not only important producers of SCFAs but also possess certain amino acid metabolic capabilities. We speculated that the increased abundance of Paeniclostridium might promote the accumulation of D-amino acids in the intestinal environment, thereby driving the enrichment of related metabolic pathways and indirectly influencing the structure and function of the gut microbiota. Moreover, the enrichment of the Neuroactive ligand-receptor interaction pathway might be closely associated with the regulatory mechanisms of the gut-brain axis. This pathway encompasses signal transduction involving various neurotransmitters, hormones, and their receptors, constituting a crucial signaling hub through which the gut microbiota influences host neural function and behavior [53]. Previous studies have shown that gut microbes could modulate the synthesis, release, and transport of serotonin (5-HT) by secreting active metabolites such as SCFAs, tryptophan metabolites, and γ-aminobutyric acid (GABA), thereby affecting intestinal motility, mucosal immunity, and central nervous system function [54]. The results of pathway enrichment in our study suggested that Lactobacillus fermentum might enhance the production of specific metabolites to activate a multi-level neuroendocrine-immune feedback mechanism in the host, which not only improved the local intestinal environment but also enhanced overall metabolic status and stress adaptability.

Although this study systematically elucidated the effects of Lactobacillus fermentum supplementation on the yak gut microecology and metabolic network, and highlighted the potential roles of key microbes in regulating energy metabolism, some limitations should be acknowledged. Our sample size needs to be further expanded and combined with more omics methods to elucidate the specific mechanisms of key microbes within metabolic pathways. Additionally, the activity of Lactobacillus fermentum in the intestine should be further evaluated in future studies to verify their role.

Conclusion

This study investigated the mechanisms through which Lactobacillus fermentum modulates host energy metabolism by reshaping key intestinal microbial communities in yaks during winter shed-feeding. We demonstrated that dietary supplementation with 0.02% fermented Lactobacillus fermentum in a medium-energy diet significantly enhances intestinal morphology, including villus length, villus width, and muscularis thickness. Furthermore, this intervention promoted the proliferation of beneficial gut bacteria, such as Paeniclostridium, Romboutsia, and Treponema, while concurrently altering microbial-derived metabolites. These microbial and metabolic shifts collectively improved nutrient digestion and absorption efficiency, thereby optimizing energy utilization and significantly enhancing yak production performance. Our findings provide novel insights into the role of probiotic-driven microecological regulation in host metabolism and health. Additionally, this study offers practical implications for refining nutritional strategies and management practices during hay-feeding periods, with potential benefits for animal welfare and herder profitability.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

PY and JJL were the overall project leaders who provided financial support and experimental conception. YQR and HC were involved in data analyses, statistical analyses, language revisions, journal selection, and manuscript submissions and revisions. BPJ and LNi contributed to the supervision and assistance of students in managing animals and collecting and analyzing samples. YQR, WT, MCF, DJY and CXY contributed to the experimental design, implementation and sample collection. All authors contributed to the article and approved the submitted version.

Funding

This research was funded by the National Natural Science Foundation of China (U23A20227), Xinjiang Tianchi Talent Research Program (TCYC-TP2023) and Shandong Natural Science Foundation (2022KJ137 and ZR2023MC164).

Data availability

The datasets presented in this study can be found in online repositories. All raw sequence data were deposited in the NCBI Sequence Read Archive database under accession number PRJNA1233467.

Declarations

Ethical approval

All animal experiments should comply with ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines. All animal procedures were approved by the Lanzhou Institute of Husbandry and Pharmaceutical Sciences, Chinese Academy of Agricultural Sciences (Approval No. 20230813-031).

Competing interests

The authors declare no competing interests.