Thirty-two Hu sheep with similar nutritional status (average body weight of 27.28 ± 0.83 kg, approximately 120 days old) were selected and divided into two groups. The control group (RC) received corn grain as the primary concentrate feed. The experimental group (RO) was fed a mixed concentrate diet with oat grain, maintaining a ratio of 35:65 (DM basis). In the RO group, unsalable oat grain partially replaced maize, resulting in an oat grain content of 35% (DM basis). The nutrient composition of the diets is detailed in Table 1. Throughout the experiment, sheep had ad libitum to roughage and water. And they were raised under the same environmental conditions. During the experiment, both groups of sheep were kept under the same temperature and humidity conditions to ensure consistency in the experimental conditions. The study spanned 100 days, comprising a 10-day adaptation period followed by a 90-day experimental phase. After the end of the experiment, six Hu lambs from each group were randomly selected for slaughter. The experimental protocol was approved by the professional committee of the Academy of Agricultural and Animal Husbandry Sciences of Inner Mongolia Autonomous Region and the Animal Welfare Association (No. IMAAAHS-2023-21).

Each lamb was weighed every 30 days before morning feeding to calculate the average daily gain (ADG). Throughout the experiment, both the feed offered and any surplus feed were precisely weighed and recorded daily to estimate the average daily feed intake (ADFI).

Prior to slaughter, veterinary examinations confirmed the lamb’s good health. Post-slaughter, rumen contents were collected consistently from all lambs, ensuring uniform sampling locations. To minimize contamination, samples were handled aseptically whenever possible. Rumen contents were placed in 50 ml aseptic, enzyme-free centrifuge tubes for subsequent analysis of the gut microbiota, SCFAs, and FAs composition. Samples from both the RC group (RC1-RC6) and the RO group (RO1-RO6) were immediately frozen in liquid nitrogen and stored at −80°C.

The hexadecyl trimethyl ammonium bromide (CTAB) method was used to extract total microbial DNA from the samples. The concentration and purity of the extracted DNA were assessed using electrophoresis on 1% agarose gels. The DNA was then diluted to a concentration of 1 ng/μl with sterile water.

Specific regions of the 16S rRNA genes, including 16S V4, 16S V3, 16S V3-V4, and 16S V4-V5, were amplified using designated primers (16S V4, 515F-806R) with barcodes. Each PCR reaction included 15 μl of Phusion® High-Fidelity PCR Master Mix (New England Biolabs), 0.2 μM of each primer, and 10 ng of target DNA. The PCR involved 30 cycles of denaturation at 98°C for 10 s, annealing at 50°C for 30 s, and extension at 72°C for 30 s, followed by a final extension at 72°C for 5 min. For DNA detection, an equal volume of 1 × loading buffer (containing SYB green) was combined with the PCR products and subjected to electrophoresis on a 2% agarose gel. The PCR products were then mixed in equal proportions and purified using a Qiagen Gel Extraction Kit (Qiagen, Germany). Libraries for sequencing were constructed with NEBNext® Ultra™ IIDNA Library Prep Kit (Cat No. E7645). Library quality was assessed using the Qubit@2.0 Fluorometer (Thermo Scientific) and Agilent Bioanalyzer 2100 system. Sequencing was performed on the Illumina NovaSeq platform, generating 250 bp paired-end reads.

Paired-end reads were assigned to samples based on their unique barcodes and truncated to remove the barcodes and primer sequences. Reads were merged using FLASH (Version 1.2.11), and splicing sequences were termed Raw Tags. Fastp (Version 0.20.0) ensured high-quality data and clean tags were aligned against the Silva database via Vsearch (Version 2.15.0) to identify and remove chimeric sequences, yielding effective tags. We conducted α-diversity analysis on normalized amplicon sequence variants (ASVs) using five indices (observed ASVs, Chao1, Simpson, Shannon, and Good’s coverage) in QIIME2 to evaluate species diversity complexity. Also, β-diversity analysis involved calculating unweighted unifrac distances in QIIME2. Principal coordinate analysis (PCoA) was performed using the ade4 and ggplot2 packages in R (version 2.15.3) to reduce dimensionality. A T-test analysis in Version 3.5.3 identifies species with significant differences at each taxonomic level (Phylum, Class, Order, Family, Genus). LEfSe analysis, using LEfSe software (Version 1.0) with an LDA score threshold of 4, identified potential biomarkers. Tax4Fun software was used for function annotation analysis. Spearman correlation analysis in R (Version 2.15.3) was conducted using the psych and pheatmap packages. SPSS software (IBM SPSS 26.0) analyzed growth performance and FAs data, with significance reported at p < 0.05. T-tests assessed significant differences.

Rumen content samples were thawed on ice to minimize degradation. A 20 μl of rumen content samples were added to a 96-well plate and transferred to the Eppendorf epMotion Workstation. Then, 120 μl of ice-cold methanol containing internal standards was automatically added to each sample and vortexed for 5 min. The plate was centrifuged at 4,000 g for 30 min. After centrifugation, 30 μl of supernatant was transferred to a clean 96-well plate, and 20 μl of freshly prepared derivatization reagent was added to each well. Derivatization was performed at 30°C for 60 min. Following this, 330 μl of ice-cold 50% methanol solution was added to dilute the samples. The plate was stored at −20°C for 20 min and then centrifuged at 4,000 g for 30 min at 4°C. Finally, 135 μl of supernatant was transferred to a new 96-well plate, ensuring each well contained 10 μl of internal standard. Serial dilutions of derivatized stock standards were added to the left wells. The plate was sealed for LC–MS analysis.

An ultra-performance liquid chromatography coupled to tandem mass spectrometry (UPLC-MS/MS) system (ACQUITY UPLC-Xevo TQ-S, Waters Corp., Milford, MA, USA) was utilized to quantitate all targeted metabolites in Novogene Co., Ltd. (Beijing, China). Samples were injected onto an ACQUITY UPLC BEH C18 1.7 μM VanGuard pre-column (2.1 × 5 mm) and an ACQUITY UPLC BEH C18 1.7 μM analytical column (2.1 × 100 mm) using an 18-min linear gradient at a flow rate of 0.4 ml/min for the positive/negative polarity mode. The mobile phases consisted of eluent A (0.1% Formic acid in water) and eluent B (acetonitrile: IPA = 70:30). The gradient program was as follows: 0–1 min (5% B), 1-11 min (5–78% B), 11–13.5 min (78–95% B), 13.5–14 min (95–100% B), 14–16 min (100% B), 16–16.1 min (100–5% B), and 16.1–18 min (5% B). The Xevo TQ-S mass spectrometer was operated in positive (negative) polarity mode with the following settings: capillary voltage of 1.5 (2.0) KV, source temperature of 150°C, desolvation temperature of 550°C, and desolvation gas flow of 1,000 L/Hr.

The detection of the experimental samples via MRM (Multiple Reaction Monitoring) was based on the novogene self-built method. The Q1, Q3, RT (retention time), DP (declustering potential), and CE (collision energy) were used for the metabolite identification. The ratio of the Q3 peak area of the compound to that of the internal standard was brought into the standard curve. The concentration of the compound was calculated from the known internal standard concentration. The data files generated by UPLC-MS/MS were processed using the MassLynx Version 4.1 for peak integration and correction.

The rumen content samples were thawed on ice, and 200 μl was pipetted into 2 ml glass centrifuge tubes. Subsequently, 900 μl of 0.5% phosphoric acid was added, and the mixture was shaken and vortexed for 2 min. The tubes were centrifuged at 14,000 g for 10 min. Afterward, 800 μl of the supernatant was transferred to a new tube, and an equal volume of ethyl acetate was added. This mixture was again shaken and centrifuged at 14,000 g for 10 min. Then, 600 μl of the resulting supernatant was collected, and 4-methylpentanoic acid was added to achieve a final concentration of 500 μM as an internal standard. The solution was mixed and transferred to a sample vial for Gas chromatography–mass spectrometry (GC–MS) analysis. The injection volume was set at 1 μl with a split ratio of 10:1. Standard substances of acetic acid, propionic acid, butyric acid, isobutyric acid, valeric acid, isovaleric acid, and hexanoic acid were prepared with concentration gradients ranging from 0.1 to 100 μg/ml in ethyl acetate. Specifically, the concentrations were set at 0.1, 0.5, 1, 5, 10, 20, 50, and 100 μg/ml. A volume of 600 μl of each standard solution was combined with 25 μl of 4-methylpentanoic acid as the internal standard. The mixture was then transferred into a sample vial for subsequent analysis by GC–MS.

The samples were separated on an Agilent DB-WAX capillary column (30 m × 0.25 mm ID × 0.25 μm) within a GC system. The heating process started at 90°C and increased to 120°C at a rate of 10°C/min. Subsequently, the temperature rose to 150°C at 5°C/min. Finally, it escalated to 250°C at 25°C/min and remained for 2 min. The carrier gas (helium) flow rate was 1.0 ml/min. Quality control (QC) samples were interspersed within the sample queue at regular intervals to ensure system stability and repeatability. Mass spectrometry analysis was performed using the Agilent 7890A/5975C GC–MS system.

The initial body weights of Hu sheep were not significantly different between the RC and RO groups. However, the RO group exhibited higher final body weights compared to the RC group. The average final weight of Hu sheep in the RC group was 38.83 kg, whereas that in the RO group was 42.97 kg, making a 10.66% increase over the control group (p < 0.001). Additionally, the average daily gain (ADG) was greater in the RO group (0.2 kg/d) compared to the RC group (0.14 kg/d; p < 0.001). Although the average daily feed intake (ADFI) of Hu sheep in the RO group tended to be higher than in the RC group, this difference was not statistically significant (p > 0.05; Figure 1A).

In this study, 16S rRNA gene sequencing was conducted on 12 Hu sheep samples, generating a total of 964,519 sequences. After chimera checks and filtering, we obtained 776,482 high-quality sequences across all samples, averaging 64,706 sequences per sample. Additional details are provided in Supplementary Table S1. Taxonomic assignment revealed 5,009 ASVs at 100% nucleotide sequence similarity. Specifically, the RC and RO groups produced 3,192 and 3,852 ASVs, respectively (Figure 1B). Notably, all samples in both groups exhibited Good’s Coverage index values of 1 (Figure 1C). The rarefaction curves stabilized when the number of effective sequences exceeded 5,000, indicating adequate sequencing depth and quantity (Figure 1D). Furthermore, the species accumulation curves for each sample group displayed a relatively flat trend at the extremity, suggesting an adequate sample size (Figure 1E). These findings collectively validate the comprehensiveness and representativeness of our sequencing data.

To evaluate the α-diversity of the microbial community, we used the Chao1, Shannon, and Simpson indices to assess species richness and community diversity. The average Chao1 and Shannon indices for the RC group were 1,257.09 and 8.59, respectively, while the RO group showed significantly higher values at 1476.63 and 8.92 (p < 0.05, Table 2). Specifically, both indices were markedly higher in the RO group compared to the RC group. Although the Simpson index averaged 0.991 in the RC group and 0.994 in the RO group, this difference was not statistically significant. Additional details are provided in Supplementary Table S2. The PCoA (Figure 1F), based on unweighted Unifrac distances, reflected the species composition and relative abundances of ASVs in the samples. A clear separation between the RC and RO groups was observed, suggesting that incorporating oat grain into the diet led to a significant divergence in the structure of the gut microbiota.

We quantified the relative proportions of predominant taxa at both the phylum and genus levels based on the distribution of microbial taxa in different groups. As shown in Figure 2, the top 10 phyla were Bacteroidota, Firmicutes, Acidobacteriota, Proteobacteria, Fibrobacterota, Siprochaetota, Chloroflexi, Patescibacteria, Actinobacteriota, and Euryarchaeota. Notably, Bacteroidota were the most abundant phylum across all samples, with Firmicutes being the second most abundant. In the RC group, Bacteroidota and Firmicutes accounted for 91.43% of the relative abundance, with Bacteroidota at 57.40% and Firmicutes at 34.03%. Similarly, these two phyla retained the highest relative abundance in the RO group, comprising 46.90% (Bacteroidota) and 33.76% (Firmicutes; Supplementary Table S3). Compared to the RC group, the RO group exhibited a significant increase in the abundance of Acidobacteriota, Proteobacteria, Chloroflexi, and Actinobacteriota (p < 0.05). The abundance of Patescibacteria increased, but the difference was not statistically significant (p = 0.287). In contrast, the abundance of Bacteroidota significantly decreased (p < 0.05), while the abundance of Siprochaetota also decreased, though not significantly (p = 0.25).

At the genus level, the top 10 genera were Prevotella, Rikenellaceae_RC9_gut_group, Bacteroidales_BS11_gut_group, F082, Christensenellaceae_R-7_group, Prevotellaceae_UCG-003, Muribaculaceae, Saccharofermentans, Subgroup_2, and Fibrobacter. The core genera of the RC group included Prevotella (15.94%), Rikenellaceae_RC9_gut_group (11.61%), F082 (9.16%), Muribaculaceae (4.13%), Christensenellaceae_R-7_group (4.07%), Saccharofermentans (3.48%), and Bacteroidales_BS11_gut_group (3.41%). In the RO group, the genera with high relative abundance were Prevotella (16.99%), Rikenellaceae_RC9_gut_group (8.42%), F082 (5.61%), Christensenellaceae_R7_group (4.18%), and Saccharofermentans (3.26%; Supplementary Table S4). Compared to the RC group, the RO group exhibited a significant increase in the abundance of Subgroup_2 (p < 0.05). Although the abundances of Prevotella, Bacteroidales_BS11_gut_group, Christensenellaceae_R-7_group, and Prevotellaceae_UCG-003 also increased, these differences were not statistically significant. On the other hand, the abundance of Rikenelaceae_RC9_gut_group, F082, Muribaculaceae, Saccharofermentans, and Fibrobacter was reduced in the RO group, although the difference was not significant.

Differences between the RC and RO groups were examined using LEfSe analysis (LDA > 2) across all taxonomic levels. Seven significant differences were identified, with the RO group showing enrichment in two phyla (Acidobacteriota and Proteobacteria), one class (Acidobacteriae), two orders (Acidobacteriales and Subgroup_2), one family (Subgroup_2), and one genus (Subgroup_2).

In this study, Tax4fun was used to predict and investigate the molecular functions of microbial communities in two sample sets. As shown in Figure 3A, 44 functional genes are associated with various biological pathways, including cellular processes, environmental processing, genetic processing, human diseases, metabolism, and organismal systems pathways. At the Kyoto Encyclopedia of Genes and Genomes (KEGG) level 2, we identified 44 functional genes across the RC and RO groups. These genes predominantly pertained to carbohydrate metabolism, replication and repair, translation, membrane transport, amino acid metabolism, and energy metabolism. Further analysis revealed that the gut microbiota in the RO group mainly focused on metabolic pathways such as cofactor and vitamin metabolism, terpenoid and polyketide metabolism, other amino acid metabolism, the endocrine system, lipid metabolism, and enzyme families. In contrast, intestinal microbes in the RC group primarily focused on nucleotide metabolism, replication and repair, transport and catabolism, drug resistance, and translation (Figures 3B,D).

We identified significant changes in KEGG pathways in the gut microbiota between the two groups using the T-test. Specifically, the RO group exhibited a higher abundance of genes related to cofactor and vitamin metabolism, lipid metabolism, amino acid metabolism, terpenoids, and polyketide metabolism compared to the RC group (p < 0.05). However, the RO group had significantly fewer genes involved in replication and repair, cellular processes, and signaling (p < 0.05; Figure 3C).

Table 3 details the impact of oat grain on the fatty acids composition in the rumen of Hu lambs. A total of 25 FAs were identified, including 24 saturated fatty acids (SFAs), 10 monounsaturated fatty acids (MUFAs), and one polyunsaturated fatty acids (PUFAs). The predominant FAs were azelaic acid (C9:0), oleic acid (C18:1), and caprylic acid (C8:0). The RO group showed lower levels of SFAs and MUFAs compared to the RC group, while the PUFAs content was higher, though the difference was not statistically significant. Among SFAs, azelaic acid (C9:0) was significantly different between the two groups (p < 0.05), with concentrations of 38.94 μmol/L in the RC group and 54.28 μmol/L in the RO group. Indicating a significantly higher level in the RO group. For MUFAs, citraconic acid (C5:1) showed a significant difference (p < 0.05), with concentrations of 0.02 μmol/L in the RC group and 0.01 μmol/L in the RO group, indicating a significantly lower level in the RO group. Additionally, pentadecanoic acid (C15:0) was significantly lower in the RO group compared to the RC group. Further details are provided in Supplementary Table S5.

The concentration of isobutyric acid was significantly lower in the RO group (35.88 μg/ml) compared to the RC group (47.80 μg/ml; p < 0.05). In contrast, the levels of butyric acid and valeric acid were higher in the RO group, although the differences were not statistically significant (p > 0.05). Additionally, the levels of acetic acid, propionic acid, isovaleric acid, and hexanoic acid were lower in the RO group compared to the RC group, but these differences were not significant (Table 4). Further details are provided in Supplementary Table S5.

Diets containing oat grain positively influenced growth performance, SCFAs, FAs metabolism, and the gut microbiota of lambs. We conducted a Spearman’s rank correlation analysis to investigate the microbiota potentially linked to SCFAs, FAs, and growth performance. At the phylum level (Figure 3E), Bacteroidota exhibited a negative correlation with valeric acid (p < 0.05).

Acidobacteriota and Proteobacteria were positively correlated with propionic, butyric, valeric, and hexanoic acids but negatively correlated with isobutyric and isovaleric acids (p < 0.05). Acidobacteriota also showed a negative correlation with C5:1 (p < 0.05). Euryarchaeota was positively associated with acetic, propionic, isobutyric, butyric, isovaleric, and hexanoic acids, and positively correlated with valeric acid (p < 0.05). Gemmatimonadota and GAL15 were positively correlated with C9:0 (p < 0.05). Spirochaetota and Synergistota showed the strongest positive correlation with C5:1 (p < 0.01). Acidobacteriota, Proteobacteria, and Chloroflexi had the strongest positive correlation with ADG (p < 0.01), while Actinobacteriota was also positively correlated with ADG (p < 0.05). At the genus level (Figure 3F), Bacteroidales_BS11_gut_group was negatively correlated with propionic, valeric, and hexanoic acids (p < 0.05), whereas Christensenellaceae_R.7_group was positively correlated with hexanoic acid (p < 0.01). Prevotellaceae_UCG.003 was negatively correlated with acetic, propionic, and hexanoic acids (p < 0.05) and had a highly significant negative correlation with valeric acid (p < 0.01). Bacteroides was negatively correlated with C9:0 (p < 0.05). Lastly, Subgroup_2 was negatively correlated with C5:1 and C15:0 (p < 0.05) and positively correlated with ADG (p < 0.01).