Sheep-derived probiotics alleviate weaning-induced oxidative stress in lambs via the gut-liver axis

Abstract

Weaning is a critical transition in lamb production that can disrupt intestinal homeostasis, promote oxidative stress, and contribute to liver injury. This study investigated whether sheep-derived probiotics could mitigate weaning-associated hepatic injury through the gut–liver axis. Microbiota profiling of six intestinal segments from healthy Hu sheep led to the isolation of nine lactic acid bacterial strains. After in vitro screening for antimicrobial and antioxidant activities, Lactiplantibacillus plantarum M1 and Limosilactobacillus reuteri K4 were selected and administered daily to early-weaned lambs from birth to day 30. Probiotic supplementation partially restored relative organ weights, alleviated hepatic histopathological damage, and reduced biochemical markers of liver injury. It also modulated the colonic mucosal microbiota, enhanced intestinal barrier integrity, and increased colonic acetate and butyrate levels. Liver metabolomics revealed enrichment of the pentose phosphate pathway and glutathione metabolism, accompanied by increased hepatic spermidine abundance and activation of the Nrf2-Keap1 antioxidant pathway. Cell-based validation further showed that spermidine attenuated oxidative stress, at least in part, through modulation of Nrf2-Keap1 signaling. Together, these findings suggest that sheep-derived probiotics may alleviate weaning-induced liver injury in lambs through coordinated microbial, metabolic, and antioxidant regulation along the gut–liver axis.

Introduction

Mutton is an important component of the human diet because it provides high-quality protein and amino acids, contains relatively low levels of cholesterol, and is a source of vitamins¹. Hu sheep are an important indigenous breed in China, characterized by rapid growth, high prolificacy, and favorable meat production performance^2,3. In large-scale sheep production, lamb survival and growth are key determinants of meat output and flock reproductive efficiency, thereby affecting farm profitability. Weaning is one of the most critical stages of lamb development. During this period, lambs are exposed to multiple stressors, including maternal separation, dietary changes, social reorganization, and environmental adjustment⁴. These stressors can disrupt intestinal microbiota colonization and impair digestive enzyme secretion. As a result, young animals may have difficulty digesting solid diets, which can compromise intestinal barrier function⁵. When this barrier is compromised, harmful bacteria and toxins from the gut can more easily enter the bloodstream, thereby increasing the risk of disease in young animals. Weaning can impair intestinal barrier and immune functions and disturb redox homeostasis, leading to oxidative stress^6,7. The liver is one of the first organs exposed to intestinal microorganisms and their products because it receives portal blood from the gut. Persistent oxidative stress may therefore contribute to liver injury⁸. Therefore, promoting early intestinal development and maturation of the antioxidant system may help reduce oxidative stress-related injury in young animals.

Furthermore, compared with young animals, adult livestock generally have more mature immune and antioxidant systems and a more stable and diverse gut microbiota. The gut microbiota participates in mutualistic interactions that support host metabolism and intestinal development while maintaining microbial community fitness through reciprocal nutrient exchange and metabolite production^9,10. Through host–microbe metabolic interactions, gut microbes generate bioactive metabolites, including antioxidants and short-chain fatty acids (SCFAs). SCFAs provide energy for intestinal epithelial cells and also exert antioxidant and anti-inflammatory effects¹¹. Other antioxidants, such as glutathione, vitamins, polyphenols, and polyamines, are important for maintaining redox balance and mitigating oxidative stress^12,13. Moreover, because the portal vein connects the intestine and liver, gut-derived metabolites can influence liver health through the gut–liver axis¹⁴. The relationship between the gut microbiota and oxidative stress has been comprehensively discussed by Marciano et al.¹⁵. Weaning-associated stress may increase intestinal permeability, allowing pathogens and microbial toxins to reach the liver via the portal vein. This process may promote reactive oxygen species (ROS) accumulation and hepatic oxidative stress^16,17. Therefore, interventions targeting the gut microbiota during critical developmental periods may help improve microbial colonization and alleviate oxidative stress^18–20.

Recently, probiotics have attracted increasing attention because of their potential to regulate the gut microbiota, inhibit pathogen colonization, and exert antioxidant effects. In vitro studies have shown that probiotics can suppress harmful microbes and modulate their metabolic activities²¹. Among probiotic candidates, lactic acid bacteria (LAB) have become one of the most extensively studied groups because they produce functional metabolites, such as organic acids, antioxidants, and bacteriocins²². These metabolites can enhance tight junction protein expression, support intestinal barrier integrity^23,24, reduce intestinal inflammation²⁵, and increase antioxidant enzyme activity²⁶. For example, L. plantarum J26 can increase intestinal SCFA levels. SCFAs can regulate the hepatic Nrf2 signaling pathway and enhance antioxidant capacity²⁷. Similarly, Lactobacillus reuteri has been reported to reduce hepatic ischemia-reperfusion injury by activating the Nrf2/HO-1 pathway, maintaining metabolic balance, and modulating the gut microbiota²⁸. Therefore, identifying LAB strains with both antibacterial and antioxidant properties is important. Based on these findings, we hypothesized that selected LAB strains may protect against weaning-associated liver injury by improving intestinal barrier function and modulating microbiota-related metabolites. In particular, increased SCFA levels may help maintain intestinal homeostasis and reduce endotoxin leakage. Spermidine, identified by hepatic metabolomic analysis and further examined in vitro, may contribute to hepatic antioxidant defense by modulating the Nrf2-Keap1 pathway. Because liver metabolomics alone cannot determine the origin of hepatic spermidine, the spermidine-related findings were interpreted as evidence of a candidate metabolic association and were further examined in vitro for their potential antioxidant relevance.

In this study, we characterized the microbiota of different intestinal segments in Hu sheep and isolated potential probiotic strains. Candidate isolates were screened in vitro for antimicrobial activity against Escherichia coli K99, Staphylococcus aureus, and Salmonella Typhimurium, as well as for antioxidant capacity using 2,2-diphenyl-1-picrylhydrazyl (DPPH), hydroxyl radical, and superoxide anion scavenging assays. Based on these in vitro results, L. plantarum M1 and L. reuteri K4 were selected and orally administered to early-weaned lambs once daily from day 1 to day 30 at a dose of 1 × 10⁹ CFU/g and 2 g/day to evaluate their potential to mitigate weaning-induced oxidative stress. The effects of probiotic supplementation on intestinal microbial communities were analyzed using 16S rRNA sequencing. Untargeted metabolomic analysis of liver tissues was performed to explore metabolic changes associated with the antioxidant effects of probiotic supplementation in lambs. These findings provide supportive evidence that early dietary probiotic intervention may help improve the health of young animals.

Results

Gut microbiota sequencing results in different intestinal segments of healthy Hu sheep

Venn analysis was used to assess shared OTUs among intestinal segments. The foregut contained 281 shared OTUs (Fig. 1A), whereas the hindgut contained 964 shared OTUs (Fig. 1B), suggesting greater microbial richness in the hindgut. Compared with the foregut, the hindgut microbiota showed significantly higher richness and diversity. No significant differences in richness or diversity were observed among hindgut segments. In the foregut, the ileum showed lower diversity than the duodenum and jejunum, whereas richness was similar across all segments. Good’s coverage indices exceeded 0.99 in all samples, indicating sufficient sequencing depth for microbial community analysis (Fig. 1C). At the phylum level, Firmicutes (70.58%) and Actinobacteria (22.25%) dominated the foregut, whereas Firmicutes (48.52%) and Bacteroidetes (34.43%) were predominant in the hindgut (Fig. 1D, E). At the genus level, Olsenella, Lachnospiraceae_NK3A20_group, Roseburia, and Turicibacter were prevalent in the foregut, whereas Eubacterium_coprostanoligenes_group, UCG-005, and Treponema were dominant in the hindgut. Overall, the hindgut exhibited greater annotated bacterial diversity than the foregut.

Fig. 1. Intestinal microbes of six intestinal segments of sheep.

A, B Venn diagram showing the OTUs shared among the foregut segments and hindgut segments. C Alpha diversity index. Chao1 and ACE indices characterize richness, while diversity is assessed using Shannon indices, and Good’s coverage reflects community coverage. D, E The phylum-level and genus-level microbial composition of each intestinal segment. F Principal coordinate analysis (PCoA) based on all samples. G Histogram of the LEfSe analysis for bacterial taxa differentially abundant among different intestinal segments. Data are expressed as mean ± SEM (C) (n = 3) and one-way ANOVA was performed, followed by Fisher’s LSD test (C) (n = 3). Different lowercase letters in the same figure denote significant differences (P < 0.05).

Community structure analysis showed that the ileal microbiota was distinct from that of the other intestinal segments. Principal coordinate analysis (PCoA) showed that PC1 and PC2 explained 60.56% and 18.05% of the variation, respectively (Fig. 1F). ANOSIM confirmed significant differences in microbial community structure among the six segments (P = 0.001). The duodenum and jejunum clustered together, as did the cecum, colon, and rectum, whereas the ileum was separated from the other segments. LEfSe analysis (LDA > 3, P < 0.05) identified segment-specific taxa in the foregut. The duodenum was enriched in Lachnospiraceae_NK3A20_group and Acetitomaculum, the jejunum in Olsenella, Syntrophococcus, Howardella, Bifidobacterium, and Lactobacillus, and the ileum in Aeriscardovia. In the hindgut, the cecum showed higher abundances of UCG-005, GCA-900066575, Phascolarctobacterium, and Christensenellaceae_R-7_group. Oscillibacter was predominant in the colon, whereas Treponema and Prevotella were enriched in the rectum (Fig. 1G).

Screening of probiotic candidates

A total of nine putative probiotic strains were isolated and identified by sequence analysis and comparison against the NCBI database. A phylogenetic tree was then constructed using MEGA 10 (Supplementary Fig. 1A). Their antimicrobial activity against Escherichia coli, Salmonella Typhimurium, and Staphylococcus aureus was assessed using an agar diffusion assay (Fig. 2A). Lactiplantibacillus plantarum M1 and Limosilactobacillus reuteri K4 showed the strongest inhibitory effects against all three pathogens. Pediococcus pentosaceus M2 also showed marked activity, particularly against E. coli and S. Typhimurium. Other isolates, including Enterococcus faecium Z4, Enterococcus hirae J1, and Limosilactobacillus mucosae M3, showed relatively weaker antimicrobial activity.

Fig. 2. Screening of potential probiotic bacterial strains.

A Antagonistic activity against pathogens in vitro. B DPPH scavenging rate of LAB in vitro. C Hydroxyl radical (·OH) scavenging rate of LAB in vitro. D Superoxide anion radical (O₂·-) scavenging rate of LAB in vitro. E Comprehensive evaluation of LAB by principal component analysis. CFS: cell-free supernatant, IC: intact cell. Data were expressed as the mean ± SEM (B, D), and one-way ANOVA was performed, followed by Fisher’s LSD test (B, D). Different lowercase letters indicate significant differences among the same components of different strains (P < 0.05).

For antioxidant evaluation, cell-free supernatants (CFS) and intact cells (IC) from the nine strains (1.0 × 10⁹ CFU/mL) were examined in vitro (Fig. 2B–D). DPPH radical scavenging activity in CFS was uniformly high across all strains (>90%), with no significant differences among isolates (Fig. 2B). In contrast, DPPH scavenging activity in IC was substantially lower and varied among strains. Notably, L. reuteri K4 showed the highest DPPH scavenging activity among IC preparations (50.83%), followed closely by L. plantarum M1. For hydroxyl radical scavenging, L. plantarum K1 showed the strongest activity in both CFS (65.85%) and IC (84.81%) (Fig. 2C). For superoxide anion scavenging, L. plantarum M1 showed the highest activity among IC preparations (48.49%), whereas E. faecium Z4 and E. hirae J3 showed significantly higher activity in CFS, with values of approximately 80% (Fig. 2D).

Principal component analysis (PCA) integrated the antimicrobial and antioxidant indicators into two principal components, with PC1 and PC2 explaining 53.05% and 20.32% of the total variance, respectively (Fig. 2E). The PCA scatter plot showed that L. plantarum M1 and L. reuteri K4 had the highest composite scores and were therefore selected as the most promising candidates for further application. Their positions in the PCA space further supported their favorable antimicrobial and antioxidant profiles. No hemolytic activity was observed in these candidate strains, supporting their suitability for subsequent studies (Supplementary Fig. 1B, C).

Genome-based functional characterization of L. reuteri K4 and L. plantarum M1

Whole-genome sequencing and comparative genomic analyses were performed to characterize the functional potential of L. reuteri K4 and L. plantarum M1. L. plantarum M1 had a larger genome than L. reuteri K4 and contained more genes associated with transcription, carbohydrate and amino acid metabolism, translation, and replication/repair, suggesting broader metabolic and adaptive potential (Fig. 3A–F). Consistently, CAZyme annotation showed that L. plantarum M1 encoded more carbohydrate-active enzymes, particularly glycoside hydrolases, suggesting greater carbohydrate utilization potential (Supplementary Table 4). Both strains also harbored multiple antioxidant-related genes associated with ROS clearance, the thioredoxin system, and glutathione metabolism, with L. plantarum M1 showing a broader repertoire of oxidative stress-related genes (Supplementary Table 5). In addition, bacteriocin- and plantaricin-related genes were identified in L. plantarum M1, whereas L. reuteri K4 contained a putative reuterin biosynthesis gene cluster, suggesting distinct antimicrobial and metabolic features (Supplementary Table 6 and Supplementary Fig. 2).

Fig. 3. Whole-genome sequencing and functional annotation of L. reuteri K4 and L. plantarum M1.

A Circular genome map of L. reuteri K4. B Circular genome map of L. plantarum M1. C COG functional classification of genes identified in L. reuteri K4. D COG functional classification of genes identified in L. plantarum M1. E KEGG pathway annotation of genes identified in L. reuteri K4. F KEGG pathway annotation of genes identified in L. plantarum M1.

Comparative genomic analysis further showed high synteny between L. reuteri K4 and L. reuteri RGW1, as well as between L. plantarum M1 and L. plantarum WCFS1, and revealed strain-specific genomic regions (Fig. 4A–E). Shared gene cluster analysis indicated that the two tested strains retained the core genomic features of their respective species while also possessing unique gene clusters. To further explore strain-specific functional potential, these unique gene clusters were annotated. In L. reuteri K4, many unique genes were associated with mobile genetic elements, including transposases, phage-related proteins, integrases, conjugative transfer proteins, and partition proteins, suggesting relatively high genomic plasticity (Supplementary Table 7). In addition, several unique genes were related to stress adaptation or cellular maintenance, including ClpS, glyoxalase, arsC, and transcriptional regulators. In L. plantarum M1, unique genes included transposases, integrases, transcriptional regulators, glycoside hydrolase domain-containing proteins, ABC transporter-associated proteins, and stress response-related genes such as pinR. These features may indicate enhanced environmental adaptability and metabolic diversity (Supplementary Table 8). Moreover, both strains contained a T3PKS biosynthetic gene cluster, whereas L. plantarum M1 additionally harbored RiPP-like and cyclic-lactone-autoinducer-related clusters (Fig. 4F, G). Together, these results suggest that L. reuteri K4 and L. plantarum M1 possess shared and complementary genomic features that may contribute to their probiotic potential and possible roles in gut–liver axis regulation.

Fig. 4. Comparative genomic analysis of L. reuteri K4 and L. plantarum M1.

A Circular genome comparison among L. reuteri K4, L. reuteri RGW1, L. plantarum M1, and L. plantarum WCFS1, together with GC content and GC skew. B Venn diagram analysis of shared and strain-specific gene clusters between L. reuteri K4 and L. reuteri RGW1. C Genome collinearity analysis between L. reuteri K4 and L. reuteri RGW1. D Venn diagram of shared and strain-specific gene clusters between L. plantarum M1 and L. plantarum WCFS1. E Genome collinearity analysis between L. plantarum M1 and L. plantarum WCFS1. F Secondary metabolite biosynthetic gene cluster identified in L. reuteri K4. G Secondary metabolite biosynthetic gene clusters identified in L. plantarum M1.

Growth performance and relative organ weights in the lamb-feeding trials

Because L. plantarum M1 and L. reuteri K4 showed favorable antioxidant and antibacterial activities in vitro, three groups were established to assess the effects of a sheep-derived probiotic mixture on early-weaned lambs (Fig. 5A). Growth performance parameters are summarized in Fig. 5B. Initial body weight did not differ significantly among the three groups (P > 0.05), indicating comparable baseline conditions. Compared with the CON group, lambs in the EW group tended to have lower final body weight and ADG, together with higher ADFI, suggesting reduced growth efficiency after early weaning. Probiotic supplementation in the EWP group partially improved these trends. Compared with the EW group, final body weight and ADG were numerically higher, whereas ADFI was numerically lower, although these differences were not significant (P > 0.05).

Fig. 5. Effect of probiotic supplementation on growth performance and organ index of lambs.

A Experimental design flowchart. B Lamb initial body weight, final weight, average daily feed intake and average daily gain (n = 20). C The lamb organs (heart, liver, spleen and kidney) were weighed and the relative weight was calculated (n = 6). Data are expressed as mean ± SEM, and one-way ANOVA was performed, followed by Fisher’s LSD test. Statistical significance is denoted by *P < 0.05, **P < 0.01.

Assessment of organ indices showed that early weaning significantly decreased the relative heart and spleen weights compared with the CON group (P < 0.05; Fig. 5C). The EWP group showed relative heart and spleen weights comparable to those of the CON group, suggesting that probiotic supplementation partially alleviated the effects of early weaning on these organ indices. In contrast, the relative liver weight was significantly lower in the EWP group than in the EW group (P < 0.05), whereas relative kidney weight did not differ among the groups.

Probiotics alleviate weaning-induced liver damage

H&E staining revealed pronounced hepatic injury in the EW group compared with the CON group. Specifically, liver sections from EW lambs showed extensive inflammatory cell infiltration, central vein congestion, hepatocyte vacuolization, localized hemorrhage, necrosis, and increased intercellular spacing, suggesting marked hepatic damage. In contrast, liver tissues from the EWP group showed largely preserved hepatic architecture, with reduced inflammatory infiltration and only occasional mild congestion. These changes suggest that probiotic supplementation alleviated weaning-induced liver pathology (Fig. 6A). To further evaluate liver injury, Masson’s trichrome and Sirius Red staining were performed. Compared with the CON group, the EW group showed stronger positive staining around hepatic vessels and more obvious histological alterations, suggesting aggravated hepatic injury. In contrast, these staining changes were reduced in the EWP group, further supporting the protective effect of probiotic supplementation on hepatic histoarchitecture (Fig. 6A).

Fig. 6. Effect of probiotics on liver morphology and liver function indicators.

A Histopathological analysis of liver tissues was conducted using hematoxylin and eosin (H&E) staining, Masson’s trichrome staining, and Sirius Red staining.H&E staining magnification ×100, upper; partial enlarged pictures ×200, down. Black arrows indicate hepatocellular inflammatory infiltration, yellow arrows indicate central venous congestion, red arrows indicate hepatocyte vacuoles, and green arrows indicate hepatocyte hemorrhage and necrosis. B Serum biochemical indicators related to liver function. TP total protein, ALB albumin, ALP alkaline phosphatase, ALT alanine aminotransferase, AST aspartate transaminase, and LDH lactate dehydrogenase (n = 6). Data are expressed as mean ± SEM, and one-way ANOVA was performed, followed by Fisher’s LSD test. Statistical significance is denoted by *P < 0.05, **P < 0.01.

To further assess liver function, key serum biochemical indices were measured (Fig. 6B). TP levels were significantly lower in the EW group than in the CON group (P < 0.05). No significant differences in ALB or ALP were observed among groups (P > 0.05), although ALP tended to be higher in EW lambs. Early weaning significantly increased serum ALT, AST, and LDH activities (P < 0.05), suggesting hepatocellular injury. Probiotic supplementation significantly decreased ALT, AST, and LDH activities compared with the EW group (P < 0.05), but had no significant effect on TP. Collectively, these findings suggest that probiotic supplementation alleviated early weaning-induced liver dysfunction and histopathological damage.

Probiotics activate the Nrf2-Keap1 pathway to enhance the antioxidant capacity of weaned lambs

Oxidative stress is a major factor contributing to liver injury after early weaning. To characterize the hepatic antioxidant response, we evaluated ROS accumulation, oxidative stress-related biochemical parameters, and key components of the Nrf2-Keap1 signaling pathway at the mRNA and protein levels.

Early weaning significantly increased hepatic oxidative stress, as shown by enhanced ROS fluorescence, increased MDA levels, and decreased CAT, GSH-Px, T-AOC, and SOD activities (Fig. 7A, C, E; P < 0.05). Probiotic supplementation reduced hepatic ROS and MDA levels and increased antioxidant enzyme activities (P < 0.05), suggesting improved hepatic redox status.

Fig. 7. Effect of probiotic supplementation on the antioxidation capacity and activation of the Nrf2 pathway in lambs.

A Representative fluorescence image of ROS staining in liver tissue. Red fluorescence indicates ROS, and blue fluorescence indicates nuclei stained with DAPI. B Heatmap of relative mRNA expression of Nrf2-keap1 in the liver tissue (Nrf2 = Nuclear factor-erythroid 2 related factor 2, Keap1 = Kelch-like ECH-associated protein 1, HO-1 = heme oxygenase-1, NQO1 = NAD(P)H quinone oxidoreductase 1) (n = 3). C Quantitative analysis of hepatic ROS levels. D Protein levels of Nrf2, Keap1, HO-1, and NQO1 in liver tissue, determined by ELISA and normalized to total protein concentration measured by the BCA assay. E The enzyme activities of (MDA = malondialdehyde, CAT = catalase, SOD = superoxide dismutase, GSH-PX = glutathione peroxidase and T-AOC = total antioxidant capacity) were measured in liver lysates (n = 6). Data are expressed as mean ± SEM, and one-way ANOVA was performed, followed by Fisher’s LSD test. Statistical significance is denoted by *P < 0.05, **P < 0.01.

To explore the potential mechanism, the hepatic Nrf2-Keap1 pathway was examined at both the mRNA and protein levels. Early weaning downregulated Nrf2, HO-1, and NQO1 and upregulated Keap1 (P < 0.05), whereas probiotic supplementation partially reversed these changes (P < 0.05; Fig. 7B). Consistently, ELISA results normalized to total protein content showed that weaning reduced the protein levels of Nrf2, HO-1, and NQO1 and increased Keap1 levels (P < 0.05). Probiotic supplementation significantly attenuated these changes (P < 0.05; Fig. 7D). Together, these findings suggest that probiotic supplementation alleviated weaning-induced hepatic oxidative damage, possibly through modulation of the Nrf2-Keap1 pathway.

Overall, early weaning induced hepatic oxidative stress, whereas probiotic supplementation alleviated oxidative damage and improved hepatic antioxidant capacity. These effects were associated with reduced ROS accumulation, increased antioxidant enzyme activities, and modulation of the hepatic Nrf2-Keap1 pathway at both the mRNA and protein levels.

Probiotics improve intestinal barrier structure and metabolites in weaned lambs

Early weaning markedly disrupted intestinal barrier homeostasis. Intestinal ROS fluorescence was significantly increased after weaning, whereas probiotic supplementation reduced ROS accumulation (P < 0.01), suggesting alleviation of intestinal oxidative stress (Fig. 8A, C). Consistently, serum DAO and LPS levels were elevated in the weaned group (P < 0.05) but were significantly decreased after probiotic treatment (P < 0.05; Fig. 8D), suggesting improved intestinal permeability and barrier integrity.

Fig. 8. Effect of probiotic supplementation on the colonic intestinal barrier structure and short-chain fatty acid content.

A Representative ROS fluorescence image of intestinal tissue (n = 6). B Immunofluorescence staining of intestinal tight junction proteins, including ZO-1, Claudin-1, and Occludin (n = 6). C Quantification of intestinal ROS levels (n = 6). D LPS lipopolysaccharide and DAO = diamine oxidase in plasma (n = 6). E Relative fluorescence intensity of ZO-1, Claudin-1, and Occludin. F Heatmap of intestinal ZO-1, Occludin, and Claudin-1 mRNA expression (n = 3). G The levels of SCFAs (acetate, propionate, butyrate, valerate, isobutyrate and isovalerate) in colon contents (n = 6). Data are expressed as mean ± SEM, and one-way ANOVA was performed, followed by Fisher’s LSD test. Statistical significance is denoted by *P < 0.05, **P < 0.01, ***P < 0.001.

At the epithelial barrier level, early weaning reduced tight junction protein expression, as shown by weaker immunofluorescence signals for ZO-1, Claudin-1, and Occludin and decreased mRNA expression of the corresponding genes (P < 0.05; Fig. 8B, E, F). Probiotic supplementation partially restored the fluorescence signals and transcriptional levels of these tight junction markers (P < 0.05), suggesting protection against weaning-induced epithelial damage.

Changes in colonic SCFAs further supported the beneficial effects of probiotic supplementation (Fig. 8G). Although early weaning had limited effects on several major SCFAs, isobutyrate and isovalerate levels were significantly reduced (P < 0.05). Probiotic supplementation increased acetate and butyrate levels (P < 0.05) and partially restored isobutyrate and isovalerate levels, suggesting improved intestinal metabolic homeostasis. Overall, these findings suggest that probiotic supplementation alleviated weaning-induced intestinal oxidative stress, preserved barrier integrity, and improved colonic SCFA profiles.

Effect of probiotics on the colonic mucosal microbiota of weaned lambs

Weaning can induce marked shifts in the gut microbiota, impair the mucosal barrier, and reduce resistance to pathogen colonization. Such dysbiosis may contribute to mucosal barrier dysfunction and systemic oxidative stress through the gut–liver axis^8,29. To investigate these changes, we examined the colonic mucosa-associated microbiota of lambs using 16S rRNA gene sequencing.

Venn analysis showed that 314 OTUs were shared among the three groups (Fig. 9A). The number of observed OTUs decreased in the EW group but increased in the EWP group. Alpha diversity analysis showed no significant differences among the groups (Fig. 9B). However, PCoA showed clear separation between the CON group and the EW and EWP groups, with partial overlap between the EW and EWP groups (Fig. 9C). At the phylum level, Firmicutes, Bacteroidetes, and Proteobacteria were the dominant taxa (Fig. 9D). The Firmicutes-to-Bacteroidetes (F/B) ratio was significantly lower in both the EW and EWP groups than in the CON group (P < 0.05; Fig. 9F). At the genus level, the CON group showed higher abundances of Lactobacillus and Limosilactobacillus. The EW group was enriched in Prevotella, Streptococcus, and Anaerostipes. The EWP group showed higher abundances of UBA7173, Bariatricus, and Acetatifactor (Fig. 9E, G). Random forest analysis and differential genus abundance analysis identified consistent microbial changes (Fig. 9H, I). Both the EW and EWP groups showed significant reductions in Lactobacillaceae and Limosilactobacillus, possibly associated with weaning. Prevotella, Streptococcus, and Anaerostipes were significantly increased in the EW group, whereas probiotic supplementation attenuated these increases. In the EWP group, UBA7173, Anaerovibrio, Acetatifactor, and Faecalibacterium were significantly increased (Supplementary Fig. 3).

Fig. 9. The effect of probiotic supplementation on the microorganisms of the colonic mucosa.

A The Venn diagram for OTUs. B Alpha diversity index. Shannon and Simpson indices assessed diversity (n = 6). C Principal coordinates analysis (PCoA) was performed to calculate beta diversity on a distance matrix of weighted UniFrac indices (n = 6). D The relative abundances of bacteria on the phylum level (TOP 10). E The relative abundances of bacteria on the genus level (TOP 20). F Relative abundance of Bacteroidetes, Firmicutes, and F/B ratio (n = 6). G Linear discriminant analysis effect size (LEfSe) comparison analysis between the groups. H Random Forests analysis (top 20 at the genus level). I Venn diagram analysis of differential genera. The top 20 differential genera and the important top 20 genera from Random Forest analysis are presented in a Venn diagram, with the coincidence part indicating the potential biomarkers. J Spearman’s correlation analysis of colonic mucosa-associated microbiota and intestinal barrier indicators. Data are expressed as mean ± SEM, and one-way ANOVA was performed, followed by Fisher’s LSD test. Statistical significance is denoted by *P < 0.05, **P < 0.01, *** P < 0.001.

Spearman’s correlation analysis showed that Lactobacillus and Limosilactobacillus were positively correlated with the tight junction markers ZO-1, Occludin, and Claudin-1, suggesting a potential association with barrier integrity (Fig. 9J). In contrast, Prevotella and Streptococcus were negatively correlated with these tight junction markers. Faecalibacterium and Anaerovibrio were positively correlated with ZO-1, whereas Bariatricus was negatively correlated with Claudin-1. DAO and LPS levels were positively correlated with Prevotella and Anaerostipes but negatively correlated with Lactobacillus, Limosilactobacillus, and Faecalibacterium.

Effect of probiotics on liver metabolism in weaned lambs

Untargeted metabolomics was used to explore metabolic changes associated with the effects of probiotics on hepatic antioxidant capacity. Liver metabolomic profiles were assessed across the three groups, and 672 metabolites were identified in positive and negative ionization modes. Partial least squares discriminant analysis (PLS-DA) showed clear separation between the CON and EW groups, with the first two components explaining 30.5% and 13.4% of the variance, respectively (Fig. 10A). Clear separation was also observed between the EW and EWP groups, with the first two components explaining 26.0% and 21.7% of the variance, respectively. Differential metabolites were identified using the criteria VIP > 1 and P < 0.05 (Fig. 10B). Compared with the CON group, the EW group showed 96 upregulated and 41 downregulated metabolites (Supplementary Table 9). Compared with the EW group, the EWP group showed 21 upregulated and 49 downregulated metabolites (Supplementary Table 10). These significantly altered metabolites were further analyzed to identify enriched metabolic pathways. Compared with the CON group, the EW group showed primary enrichment in glycerophospholipid metabolism, choline metabolism in cancer, and linoleic acid metabolism (Fig. 10C). Compared with the EW group, differential metabolites in the EWP group were mainly enriched in the pentose phosphate pathway, glutathione metabolism, and central carbon metabolism in cancer.

Fig. 10. Effects of probiotics on hepatic metabolomics.

A PLS-DA score plot for CON and EW, EW and EWP groups (n = 6). B Differentially expressed metabolites (DEMs) selected based on VIP > 1, P < 0.05 are represented by volcanic plots. C The Sankey bubble map displays 11 significantly different pathways and their enriched DEMs. The left side is the Sankey diagram, indicating the metabolites within each pathway. The right side features the bubble diagram. The bubble size represents the number of metabolites that the pathway belongs to, and the bubble color represents the p-value. D Heat map of metabolites involved in glycerophospholipid metabolism. E The integrative metabolism pathway according to the KEGG pathway database. The green represents downregulated metabolites, while the red represents upregulated metabolites. F Correlation analysis among SCFAs, differential microbiota, liver injury markers, hepatic antioxidant indices, Nrf2-Keap1 pathway-related factors, and key liver metabolites. GSSG = Oxidized glutathione; Ru-5-P = Ribulose 5-phosphate; G-6-P = glucose-6-phosphatase; ATP = adenosine triphosphate; ADP = adenosine diphosphate; NAD + = nicotinamide adenine dinucleotide; NADP + = nicotinamide adenine dinucleotide phosphate. The red and green words indicate the metabolites that have significantly increased or decreased in EWP compared to EW. Data are expressed as mean ± SEM, and one-way ANOVA was performed, followed by Fisher’s LSD test. Statistical significance is denoted by *P < 0.05, **P < 0.01, ***P < 0.001.

Weaning notably altered glycerophospholipid metabolism, as shown by increased levels of glycerophosphoric acid, Gpetn(14:0/22:2), and Pe-Nme2(18:1(9Z)/18:1(9Z)) and reduced levels of choline and phosphocholine (Fig. 10D). In the EWP group, metabolites were enriched in pathways related to the pentose phosphate pathway (PPP) and glutathione metabolism (Fig. 10E). Compared with the EW group, the EWP group showed significantly higher levels of D-ribulose 5-phosphate and ADP in the PPP, along with increased NAD+ levels. These changes may be related to NADK-catalyzed phosphorylation to NADP + . Within glutathione metabolism, 5-L-glutamyl-L-alanine, Cys-Gly, oxidized glutathione (GSSG), and spermidine were significantly increased, whereas cysteinyl glutathione disulfide was significantly decreased (Supplementary Fig. 4). Correlation analysis showed significant associations among SCFAs, differential microbiota, hepatic redox indices, and key liver metabolites (Fig. 10F). Metabolites associated with the PPP and glutathione metabolism were correlated with antioxidant enzyme activities, Nrf2-Keap1-related factors, and liver injury indicators. These associations suggest that gut–liver axis-related metabolic changes may contribute to improved hepatic antioxidant status. Given the increased hepatic spermidine abundance after probiotic supplementation and its reported role in redox regulation, we next performed cell-based experiments to examine whether spermidine modulates oxidative stress through Nrf2–Keap1 signaling.

Spermidine alleviated oxidative stress through modulation of the Nrf2-Keap1 pathway

To further explore the potential mechanism, cell-based loss-of-function experiments were performed using siRNA-mediated knockdown of Nrf2 and Keap1. RT-qPCR and Western blotting confirmed efficient silencing of Nrf2 and Keap1, and the most effective siRNAs were selected for subsequent experiments (Fig. 11A, B). Under H₂O₂ challenge, spermidine significantly reduced intracellular ROS accumulation and MDA levels while increasing SOD, GSH-Px, and GSH levels, suggesting improved cellular antioxidant capacity (Fig. 11C, D). Nrf2 silencing weakened these protective effects, whereas Keap1 silencing further enhanced them. Consistently, spermidine increased Nrf2 protein abundance in nuclear extracts, decreased Keap1 expression, and upregulated the downstream antioxidant proteins HO-1 and NQO1. These effects were attenuated by Nrf2 knockdown but enhanced by Keap1 knockdown (Fig. 11E). Together, these results suggest that spermidine alleviated H₂O₂-induced oxidative stress, at least in part, through modulation of the Nrf2-Keap1 pathway.

Fig. 11. Spermidine exerts antioxidant stress protection by activating the Nrf2-Keap1 pathway.

A RT-qPCR validation of the silencing efficiency of different siRNAs targeting Nrf2 and Keap1. B Western blot validation of Nrf2 and Keap1 protein levels after siRNA transfection. C Representative flow cytometry histograms of intracellular ROS levels. D Oxidative stress-related parameters, including MDA, SOD, GSH-Px, and GSH, in different treatment groups. E Nrf2 protein abundance in nuclear extracts was determined by Western blotting and normalized to Histone H3. Keap1, HO-1, and NQO1 were detected in total protein lysates and normalized to GAPDH. Full Western blot images for all three biological replicates are provided in Supplementary Fig. 5. Data are expressed as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001.

Discussion

The gut functions not only as a digestive organ but also as a central hub for immune, metabolic, and neuroregulatory activities, which are strongly influenced by the resident microbial community. The intestinal microbiota plays important roles in host energy supply³⁰, immune regulation³¹, and stress modulation^32,33. In this study, microbial diversity and richness were significantly lower in the foregut than in the hindgut, which is consistent with previous reports of reduced microbial diversity in the small intestine^34,35. The ileal microbiota formed a distinct cluster, likely because of its unique anatomical position and physiological environment. In the small intestine, the dominant microbial genera included Olsenella, Lactobacillus, Bifidobacterium, and Turicibacter. These bacteria can ferment dietary carbohydrates to produce metabolites such as lactic acid, acetate, and butyrate^36,37. These metabolites help regulate intestinal pH, inhibit pathogenic bacteria, provide energy to the host, and support intestinal barrier function³⁸. In addition, Turicibacter has been associated with lipid and bile acid metabolism³⁹. In the large intestine, the dominant bacterial groups included UCG-005, Christensenellaceae R-7 group, and Prevotella. These bacteria are mainly involved in the degradation of proteins and crude fiber, thereby contributing to SCFA production^40–43. SCFAs play important roles in nutrient metabolism and immune regulation. The Christensenellaceae R-7 group has also been associated with lipid metabolism⁴⁴. In summary, these findings provide a basis for subsequent probiotic screening. Lactobacillus may serve as a potential probiotic candidate for improving intestinal health and stress resilience.

LABs exert probiotic effects through multiple mechanisms⁴⁵. To identify suitable candidates for application in weaned lambs, we assessed the ability of LAB isolates to inhibit common ovine pathogens, including E. coli, S. Typhimurium, and S. aureus⁴⁶. We then evaluated antioxidant capacity by measuring scavenging activities against DPPH radicals, superoxide anions, and hydroxyl radicals²¹. This antioxidant-focused screening was included because weaning is associated with increased ROS production and oxidative stress in lambs^47,48. Antioxidant readouts were collected from both cell-free supernatants and intact cells to capture the complementary contributions of secreted metabolites and cell-associated redox activity. Consistent with previous reports, LAB can mitigate oxidative stress by scavenging free radicals and suppressing harmful bacteria^21,26,49. Therefore, PCA integrating antimicrobial and antioxidant datasets identified L. plantarum M1 and L. reuteri K4 as the main candidates for subsequent lamb trials. Whole-genome sequencing further provided genome-level support for the functional potential of these two probiotic strains. In particular, L. plantarum M1 appeared to have greater potential for carbohydrate utilization and oxidative stress resistance, whereas L. reuteri K4 harbored a reuterin biosynthesis-related gene cluster, suggesting distinct antimicrobial potential. Comparative genomic analysis further suggested that, in addition to conserved probiotic-related traits, L. reuteri K4 and L. plantarum M1 harbor distinct strain-specific genes and biosynthetic gene clusters. These complementary genomic traits may contribute to their functional potential. Future probiotic screening should incorporate metabolite-related analyses, WGS-based functional evaluation, and intestinal or hepatic cell-based validation to better identify strains with host-protective antioxidant potential.

Because in vitro experiments cannot fully represent in vivo conditions, the screened probiotic mixture was administered to weaned lambs. Weaning stress is known to increase ROS production, which can compromise hepatocyte function and contribute to inflammation, edema, and collagen deposition^50,51. In our study, early weaning altered organ indices and induced hepatic histopathological alterations, including hemorrhagic necrosis and inflammatory infiltration. These histological changes were accompanied by significantly increased plasma ALT, AST, and LDH levels. Probiotic treatment alleviated these changes, reduced hepatic inflammation, and improved plasma markers of liver injury, suggesting protection against weaning-associated hepatic damage. Previous studies have shown that L. plantarum J26 can reduce alcohol-induced oxidative liver damage²⁷ and that L. reuteri SLZX19-12 can protect against LPS-induced liver tissue damage⁵². Consistent with these findings, the probiotic combination alleviated weaning-induced liver damage in the present study.

As a major target organ of oxidative stress, the liver can synthesize antioxidant enzymes and activate antioxidant pathways to maintain redox balance⁵³. One of the major protective mechanisms is the Nrf2-Keap1 signaling pathway⁵⁴. This pathway is activated when Nrf2 dissociates from Keap1 and translocates into the nucleus, where it promotes the transcription of genes encoding antioxidant enzymes and detoxifying proteins. These downstream targets include antioxidant enzymes, such as SOD, CAT, and GSH-Px, and detoxifying enzymes, such as glutathione S-transferase and NQO1^51,53,55,56. Our results showed that probiotic treatment was associated with activation of the hepatic Nrf2-Keap1 signaling pathway and upregulation of downstream genes in lambs. Our findings are consistent with previous studies describing the protective effects of probiotics against oxidative damage. For example, L. reuteri has been reported to protect mice from hepatic oxidative stress by activating the Nrf2 pathway²⁸.

Weaning can increase harmful factors such as LPS and ROS, which may downregulate ZO-1, Claudin-1, and Occludin expression and increase intestinal permeability^57,58. A compromised intestinal barrier allows LPS to translocate from the gut into the bloodstream and reach the liver, where it may promote oxidative stress, as reflected by increased MDA levels and decreased antioxidant enzyme activities^59,60. In the present study, probiotic treatment significantly reduced plasma DAO and LPS levels and increased the expression of intestinal tight junction proteins, including ZO-1 and Occludin. These barrier-related improvements coincided with increased intestinal SCFA levels. Probiotic supplementation increased acetate and butyrate levels. SCFAs are key microbial metabolites that help maintain epithelial integrity and barrier function^38,61–63. In addition to this barrier-related effect, SCFAs may also contribute to hepatic protection. Butyrate and propionate have been reported to activate the hepatic Nrf2-Keap1 signaling pathway after entering the systemic circulation^64,65. Accordingly, the observed enhancement of hepatic Nrf2-Keap1 pathway activity suggests that probiotics may protect the liver by improving intestinal barrier function, limiting LPS leakage, and enhancing hepatic antioxidant defenses, potentially through SCFA-associated Nrf2 signaling^27,66.

The mucosal barrier serves as the first line of defense against pathogenic microorganisms and consists of epithelial cells and a mucus layer containing commensal microorganisms^67–69. Mucus and mucosa-associated microorganisms help limit pathogen invasion of epithelial cells; therefore, disruption of this community can compromise intestinal barrier integrity⁷⁰. Although most studies have focused on the microbiome of intestinal contents, the mucosal microbiome may serve as a useful biomarker of intestinal inflammatory status^69,71. Weaning-induced dietary changes can disrupt the mucosa-associated microbial community^72,73. Such disturbances are often reflected by shifts in microbial composition, including decreases in lactose-utilizing Firmicutes and increases in polysaccharide-utilizing Bacteroidetes^74,75. In the present study, probiotic treatment significantly reduced the abundance of Prevotella. Although Prevotella is a conserved and functionally important genus in ruminants, its ecological significance may vary under conditions of intestinal stress. Here, the weaning-associated increase in Prevotella was negatively correlated with tight junction markers and positively correlated with DAO and LPS. Together with previous reports linking specific Prevotella taxa to disrupted mucosal homeostasis in young ruminants, this finding suggests that Prevotella enrichment may represent a dysbiosis-associated signal under early-weaning stress rather than evidence that this genus is universally harmful^76,77. Similarly, probiotic treatment decreased the abundances of Streptococcus and Anaerostipes. Streptococcus may activate macrophages to produce ROS, whereas Anaerostipes has been associated with impaired butyrate production under enteritis conditions^78,79. In contrast, probiotic supplementation promoted the enrichment of potentially beneficial taxa. Specifically, probiotic treatment increased Faecalibacterium, a butyrate-producing genus that can generate microbial anti-inflammatory molecules and help maintain mucosal immune and redox homeostasis^33,80. In addition, probiotic supplementation increased the abundances of Acetatifactor and UBA7173, which have been negatively associated with microbial invasion and oxidative stress in previous studies^69,81. Together, these findings suggest that probiotic supplementation was associated with a more favorable mucosa-associated microbiota profile. These microbial changes were accompanied by increased tight junction gene expression and reduced DAO and LPS levels, suggesting improved barrier integrity and attenuated mucosal injury. Because 16S rRNA sequencing mainly reflects community-level taxonomic variation, these changes should be interpreted as probiotic-associated remodeling of the mucosa-associated microbial ecosystem. Further strain-specific and quantitative analyses are needed to determine the persistence and functional contributions of L. plantarum M1 and L. reuteri K4 in the intestinal mucosa.

The liver plays a crucial role in maintaining health and regulating disease processes in both humans and animals⁸². Through the gut–liver axis, weaning can contribute to hepatic inflammation⁸³, oxidative stress^84,85, and lipid metabolism disorders^56,86. Our study showed that weaning mainly altered hepatic glycerophospholipid metabolism. Weaning may promote hepatic PC and LPC synthesis through activation of the CDP-diglyceride pathway, accelerate lipid turnover in response to energy demand, and reduce hepatic lipid accumulation⁸⁷. Increased PE levels may represent a cellular adaptive response to oxidative damage and may also participate in chronic pathological transformation^88,89. Our metabolomic analysis suggested that probiotic treatment was associated with enrichment of the pentose phosphate pathway and glutathione metabolism, which may support hepatic antioxidant defense. The PPP generates NADPH, an important reducing equivalent for maintaining cellular redox balance. NADPH supports the reduction of GSSG to GSH through glutathione reductase, thereby contributing to ROS detoxification and oxidative stress alleviation⁹⁰. Although GSSG is generally considered a marker of oxidative status, Nishimura et al. reported that GSSG may protect the heart from hypoxic stress by maintaining mitochondrial quality control⁹¹. Cys-Gly and 5-L-glutamyl-L-alanine are degradation products of GSH. The sulfhydryl group of Cys-Gly can neutralize hydroxyl radicals, and decreases in these metabolites may serve as indirect biomarkers of oxidative stress⁹². These metabolites can also serve as substrates for de novo GSH synthesis^93,94. Similar to our findings after probiotic treatment, previous work showed that increased GSSG and 5-L-glutamyl-L-alanine levels in ovarian venous serum metabolites of sheep were associated with enhanced antioxidant capacity in follicular fluid⁹⁵. Probiotic treatment reduced cysteine-glutathione disulfide levels. This compound is produced when cysteine residues on proteins combine with glutathione under oxidative stress⁹⁶.

In addition to enrichment of the pentose phosphate pathway and glutathione metabolism, the increased hepatic spermidine abundance suggests that polyamine-related metabolism may also be involved in probiotic-associated antioxidant regulation. Spermidine is a bioactive polyamine involved in redox homeostasis and antioxidant defense^97–99. Recent evidence from a mouse model of metabolic syndrome showed that L. reuteri ZJ617 promoted microbiota-derived spermidine production by providing substrates to spermidine-producing commensal bacteria. This finding highlights the potential importance of probiotic–microbiota interactions in host polyamine metabolism¹⁰⁰. In the present study, hepatic spermidine abundance increased after probiotic supplementation, and cell-based experiments further showed that spermidine attenuated H₂O₂-induced oxidative stress. This protective effect was weakened by Nrf2 silencing but enhanced by Keap1 silencing. These findings suggest that spermidine may be one of the metabolites associated with the improved hepatic antioxidant status observed after probiotic intervention. However, the present in vivo data do not demonstrate whether the increased hepatic spermidine was directly produced by the gut microbiota or transported to the liver through the portal vein. Therefore, we cannot exclude the possibility that hepatic spermidine increased as a consequence of restored host polyamine metabolism after oxidative stress was alleviated. Future studies using targeted polyamine quantification, portal vein sampling, isotope tracing, and microbiota-depletion or gnotobiotic models are needed to determine the origin of hepatic spermidine and its causal role in probiotic-mediated hepatoprotection. Although the present data do not establish a direct causal link between specific microbiota-derived metabolites and hepatic signaling activation, they provide useful evidence for probiotic-associated metabolic and antioxidant regulation. Taken together, this study identified L. plantarum M1 and L. reuteri K4 as promising sheep-derived probiotic candidates and showed that their combined supplementation alleviated weaning-induced liver injury in lambs. These protective effects were associated with coordinated changes along the gut–liver axis, including remodeling of the colonic mucosa-associated microbiota, increased colonic SCFA levels, improved intestinal barrier integrity, hepatic metabolic reprogramming, enhanced antioxidant capacity, and modulation of Nrf2-Keap1 pathway-related factors (Fig. 12). Nevertheless, this study was conducted at a single post-weaning time point. Therefore, future studies incorporating multiple time points are needed to better characterize the temporal dynamics of microbial, metabolic, and host responses during weaning stress.

Fig. 12. Schematic illustration of how L. plantarum and L. reuteri mitigate weaning-induced hepatic oxidative injury in lambs via the gut-liver axis.

Dashed arrows indicate hypothesized regulatory links that require further causal validation.

The beneficial effects observed in this study may have potential relevance beyond Hu sheep. Because the gut–liver axis and Nrf2-Keap1 antioxidant pathway are broadly conserved across mammals, L. plantarum M1 and L. reuteri K4 may have similar protective potential in other sheep breeds and young ruminants under comparable weaning-stress conditions. This possibility is supported by Fan et al., who applied sheep-derived Clostridium beijerinckii R8 to goat kids and reported reduced diarrhea and improved intestinal health¹⁰¹. Evidence from monogastric animals further supports this possibility. Dell’Anno et al. reported that supplementation with L. plantarum and L. reuteri reduced diarrhea incidence in weaned piglets¹⁰². Tang et al. showed that L. plantarum CGMCC 1258 enhanced antioxidant function and intestinal immunity, whereas L. reuteri LR1 improved growth performance and reduced diarrhea in pigs. These findings suggest that some probiotic-related functions of these species may be conserved across hosts¹⁰³. The unique genes identified in L. plantarum M1 and L. reuteri K4—including stress adaptation genes (ClpS, glyoxalase, and arsC) and genes related to metabolic diversity, such as glycoside hydrolases and ABC transporters—may support environmental adaptability and functional potential. However, because ruminants and monogastric animals differ physiologically, these potential cross-species effects require direct validation in appropriate models.

Methods

Preliminary analysis of gut microbiota in healthy sheep and isolation of bacteria

We managed all sheep following standard animal husbandry protocols. Three healthy Hu sheep weighing about 50 kg were humanely euthanized. Sodium pentobarbital was administered intravenously via the jugular vein (80–100 mg/kg). Lambs typically lost consciousness within 30–60 s and died within 3–5 min. Death was confirmed by the absence of the corneal reflex, no detectable heartbeat on auscultation and fixed dilated pupils. The animals were slaughtered, and tissue collection was performed. Fresh luminal contents were aseptically collected from the duodenum, jejunum, ileum, cecum, colon, and rectum into sterile tubes, with separate sterile tools and tubes used for each animal/segment to minimize cross-contamination. These samples were promptly snap-frozen in liquid nitrogen and subsequently stored at −80 °C until further analysis. We subjected sample aliquots to DNA extraction, PCR amplification, and Illumina MiSeq sequencing, and then performed subsequent bioinformatic and statistical analyses (n = 3). In addition, because the initial intestinal-segment microbiota profiling was performed using a limited number of healthy sheep, these results should be regarded as exploratory and used primarily to guide the isolation of probiotic candidates. Future studies with larger cohorts are needed to confirm segment-specific microbial patterns and better account for inter-individual variation.

For culture-based isolation, aliquots of the collected samples were incubated overnight in De Man, Rogosa and Sharpe (MRS) broth, serially diluted with sterile saline and then spread onto MRS, Lactobacillus Selective Agar (LBS) and bromocresol green (BCG) milk agar plates. Following anaerobic incubation at 37 °C for 48–72 h, we picked colonies with distinct morphologies and purified them by subculturing in MRS broth. Finally, we cryopreserved the pure cultures in 50% (w/v) sterile glycerol at −80 °C for future analysis.

Molecular identification of the isolated bacteria

We amplified and sequenced the 16S rRNA gene from purified cultures in order to identify the strains (Sangon Biotech Co., Ltd., Shanghai, China). We then confirmed sequence homology via BLASTN against the NCBI database and subsequently constructed a phylogenetic tree in MEGA 10 software (www.megasoftware.net) by aligning the sequences with the neighbor-joining method.

Antibacterial assay

Antimicrobial activity was evaluated against indicator pathogenic strains Escherichia coli K99 (O38), Staphylococcus aureus ATCC6538, and Salmonella Typhimurium ATCC14028 via agar diffusion. LAB isolates were grown overnight in MRS broth at 37 °C, while pathogens were incubated in shaking LB broth for 12 h. We centrifuged the LAB cultures (6000 × g, 10 min) and filter-sterilized the resulting cell-free supernatant through a 0.22 µm membrane. Spread 100 µL of the indicator bacteria suspension (10⁶ CFU/mL) on the surface of the LB agar plates and punch wells in the agar using an Oxford cup. Each well was filled with 200 µL of LAB supernatant, and plates were incubated for 12 h at 37 °C. Every strain was examined three times; MRS broth served as the negative control. We measured the diameter of inhibition zones using a vernier caliper, recording the results in millimeters.

Antioxidant assay

Cell-free supernatants (CFS) and intact cells (IC) were prepared from overnight LAB cultures using a modified protocol by Ragul et al.¹⁰⁴. Briefly, cultures were centrifuged to obtain supernatants (CFS) and washed cell pellets (IC) for subsequent antioxidant assays¹⁰⁴. The 2-2-diphenyl-1-picrylhydrazyl (DPPH) radical scavenging capacity was assayed in 96-well plates as outlined by Xu et al.¹⁰⁵, hydroxyl radical scavenging activity was assessed using the Fenton reaction methodology¹⁰⁶, and superoxide anion scavenging ability was measured following Wu et al.¹⁰⁷.

Lamb feeding trial and sample collection

The feeding trial was conducted at Shangpin Meiyang Technology Co., Ltd. All animals were managed under standard farm husbandry conditions with ad libitum access to clean water and designated diets, and were monitored daily for health status. Housing was kept clean, well-ventilated, and maintained under appropriate temperature and lighting conditions. The protocol was approved by the Bioethics Committee of Shihezi University (A2024-428). Sixty healthy neonatal Hu sheep lambs (average birth weight 4.53 ± 0.54 kg) were enrolled in a 35-day trial (n = 20 per group). Lambs were randomly allocated to three groups: a control group reared with ewes (physiological saline, CON), and two early-weaned groups, with and without probiotic intervention (EWP and EW). Ewes of the weaned groups were removed on day 30 (physiological saline, EW). The EWP group received oral supplementation of Lactiplantibacillus plantarum M1 and Limosilactobacillus reuteri K4 (1 × 10⁹ CFU/g, 2 g/day) daily from day 1 until weaning; the EW group received no intervention. The solution of physiological saline or probiotics was infused into each lamb’s mouth using an injection syringe. All lambs had access to starter feed from day 7 and were provided water ad libitum. Feed composition and nutrient content are detailed in Supplementary Table 1. Probiotic candidates were lyophilized according to Wu et al.¹⁰⁸, and viable counts were determined by plating.

To assess growth performance, we measured feed intake and body weights on days 1 and 35 and used these data to calculate the average daily gain (ADG) and average daily feed intake (ADFI) (n = 20). On day 35, blood (10 mL) was collected from six lambs per group via jugular venipuncture into sodium heparin tubes and centrifuged (3000 rpm, 4 °C, 10 min); plasma samples were stored at −80 °C. According to the above-mentioned euthanasia method, we euthanized the lambs and harvested the liver, spleen, kidneys, and heart for weighing and examination. Colon contents were collected into cryovials. After the luminal contents were expelled, the remaining colonic tissues were cleaned with PBS, the intestinal segment was incised, and the mucosa was scraped with a coverslip into cryovials. These mucosal scrapings were immediately flash-frozen in liquid nitrogen and stored at −80 °C.

Biochemical analysis of liver function, oxidative stress, and intestinal permeability

Liver function was evaluated by measuring aspartate transaminase (AST), alanine aminotransferase (ALT), total protein (TP), albumin (ALB), globulin (GLO), alkaline phosphatase (ALP), and lactate dehydrogenase (LDH) using a HITACHI 7180 biochemical analyzer (n = 6). We homogenized the liver samples in ice-cold PBS and centrifuged the resulting homogenate for 10 min at 12,000 rpm and 4 °C. Following the manufacturer’s instructions, the supernatant was used to measure the levels of malondialdehyde (MDA), superoxide dismutase (SOD), glutathione peroxidase (GSH-Px), catalase (CAT), and total antioxidant capacity (T-AOC) with commercial kits (Nanjing Jiancheng Bioengineering Institute) (n = 6). For the cell experiments, cells from each group were collected and lysed, and the oxidative stress-related parameters were determined using commercial assay kits according to the manufacturers’ instructions. The levels or activities of MDA, SOD, GSH-Px, and glutathione (GSH) were measured to evaluate intracellular oxidative damage and antioxidant capacity (n = 3). Plasma lipopolysaccharide (LPS) and diamine oxidase (DAO) were also measured according to kit manuals (n = 6).

Histopathological analysis

Liver tissues were fixed in 4% paraformaldehyde, embedded in paraffin, sectioned, and subjected to histological staining. H&E staining was used to evaluate general morphological changes. In addition, Sirius red staining and Masson’s trichrome staining were performed to further assess collagen deposition and liver injury. Histological images were captured under an Olympus Microscope system (BX53, Tokyo, Japan).

Quantitative real-time PCR (qPCR)

The gene expression analysis pipeline started with total RNA extraction from cultured cells, colonic and liver tissue using Trizol reagent (TransGen Biotech, Technology Co., Ltd., Beijing, China). After extraction, we confirmed RNA purity and concentration with the NanoDrop 2000 (Thermo Scientific, Waltham, MA, USA) spectrophotometer. Then, in accordance with the manufacturer’s instructions, we treated 2 µg of the total RNA with an RNase-free DNase kit before reverse transcribing it into cDNA. Quantitative real-time PCR (qPCR) was then performed on a Roche LightCycler96 system, using 1 µL of a 1:20 diluted cDNA sample as the template for each reaction. Relative gene expression was quantified using the 2^−ΔΔCt method, with β-actin serving as the internal reference gene. Refer to Supplementary Table 2 for a complete list of primer sequences.

Measurement of hepatic Nrf2-Keap1 pathway-related proteins

Liver samples were homogenized in ice-cold PBS and centrifuged at 12,000 rpm for 10 min at 4 °C. The resulting supernatants were collected for protein analysis. Total protein concentration was determined using a bicinchoninic acid (BCA) protein assay kit. Subsequently, the protein levels of Nrf2, Keap1, HO-1, and NQO1 in liver tissues were quantified using commercially available enzyme-linked immunosorbent assay (ELISA) kits according to the manufacturers’ protocols. The optical density was measured with a microplate reader, and the concentrations of the target proteins were calculated based on the corresponding standard curves (n = 6).

Immunofluorescence staining of tight junction proteins

To further evaluate intestinal barrier integrity at the protein level, immunofluorescence staining was performed for tight junction proteins. Intestinal tissue sections were fixed, permeabilized, and blocked, followed by incubation with primary antibodies against ZO-1, Claudin-1, and Occludin at 4 °C overnight. After washing with PBS, the sections were incubated with fluorescence-labeled secondary antibodies in the dark. Nuclei were counterstained with DAPI. Images were captured under a fluorescence microscope, and the relative fluorescence intensity was quantified using ImageJ software.

Analysis of SCFAs in colonic contents

SCFAs in colonic contents were measured by gas chromatography. We homogenized 1 g samples of colon content in 5 mL of ultrapure water and centrifuged the mixture (4000 rpm, 10 min, 4 °C). We then took 1 mL of the resulting supernatant and treated it with 0.2 mL of 25% (w/v) metaphosphoric acid before vortexing and re-centrifuging it at a higher speed (15,000 rpm, 15 min, 25 °C). We filtered the final supernatant through a 0.45 µm membrane, and the filtrate was then injected into an Agilent 7890B gas chromatograph. Standard curves for acetic, propionic, butyric, isobutyric, valeric, and isovaleric acids were established for quantification (n = 6).

Whole-genome sequencing and functional genomic analysis of L. plantarum M1 and L. reuteri K4

Genomic DNA from L. plantarum M1 and L. reuteri K4 was extracted and subjected to whole-genome sequencing. After quality control, clean reads were assembled into draft genomes, followed by gene prediction and functional annotation. Functional genes were annotated against multiple databases to characterize the metabolic potential of each strain.

Comparative genomic analysis was subsequently conducted between L. plantarum M1 and L. reuteri K4 to explore genomic features that may underlie their functional roles in the present study. Special attention was paid to genes associated with metabolic pathways, stress resistance, antioxidant function, and other biological processes potentially relevant to gut-liver axis modulation. When interpreting the combined administration of L. plantarum M1 and L. reuteri K4, the annotated genomes were further examined for possible complementary functional traits and potential metabolite-mediated interactions between the two strains.

Gut microbiota analysis

We selected colonic mucosal samples for microbiota analysis because mucosa-associated microbiota is more closely linked to the intestinal epithelium, barrier integrity, and host–microbe interactions than luminal microbiota in colonic contents. We extracted genomic DNA from both intestinal contents and colonic mucosa using an Omega Bio-Tek kit (Norcross, GA, USA). DNA integrity was assessed via 1% agarose gel electrophoresis. Universal primers (341 F:5′-CCTAYGGGRBGCASCAG-3′, 806 R:5′-GGACTACNNGGGTATCTAAT-3′) targeting the bacterial 16S rRNA gene V3–V4 region were used for amplification. Amplicons were quantified using the QuantiFluor™-ST system and processed for library construction with the NovaSeq 6000 Sp Reagent Kit, followed by sequencing on the Illumina NovaSeq platform. Subsequent data analysis was conducted in QIIME2 and R (v3.2.0) to analyze the microbial community structure, including assessments of microbial community α-diversity and β-diversity (n = 6). We employed Linear Discriminant Analysis Effect Size (LEfSe) to identify differentially abundant taxa and used random forest models to distinguish between experimental groups.

Untargeted metabolomics of the liver

To profile the liver metabolome, we prepared samples beginning with the pulverization of 50 mg of tissue in liquid nitrogen. We then homogenized this powder in a 2:2:1 mixture of methanol, acetonitrile, and deionized water. After extraction and centrifugation, supernatants were dried and reconstituted in acetonitrile/methanol (1:1, v/v) and filtered through a 0.22 µm nylon membrane for subsequent liquid chromatography–mass spectrometry (LC-MS) analysis. Both positive and negative ESI modes were used with TOF-MS. Peak extraction and alignment from the resulting spectra were accomplished using Progenesis QI software. Metabolite levels are reported as relative abundance, expressed as normalized peak area in arbitrary units, rather than absolute concentration. We defined significant differential metabolites as those with a variable importance projection (VIP) score >1 and P < 0.05 (Student’s t-test) (n = 6). Finally, we used the KEGG database to perform a pathway analysis on these metabolites to assess their functional enrichment (http://www.genome.jp/kegg).

Cell-based validation of spermidine-mediated antioxidant signaling

Hepatic cells were maintained in complete medium supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin at 37 °C in a humidified atmosphere containing 5% CO₂. Oxidative stress was induced by H₂O₂, and spermidine (Spd) was administered as a pretreatment or cotreatment as indicated. To investigate the involvement of the Nrf2-Keap1 pathway, cells were transfected with siRNAs targeting Nrf2 or Keap1, with a scrambled siRNA used as a negative control. Three candidate siRNAs were tested for each gene, and the most efficient sequence was selected based on RT-qPCR and immunoblot analysis. For mechanistic validation, cells were assigned to six groups: si-NC + H₂O₂, si-NC + H₂O₂ + Spd, si-Nrf2 + H₂O₂, si-Nrf2 + H₂O₂ + Spd, si-Keap1 + H₂O₂, and si-Keap1 + H₂O₂ + Spd. Total and nuclear proteins were extracted and subjected to immunoblot analysis using antibodies against Nrf2, Keap1, HO-1, and NQO1. H3 and GAPDH were used as the nuclear and total protein loading controls, respectively (Supplementary Table 3). Their densities were quantified with ImageJ v1.8.0 software after normalization to H3 and GAPDH (n = 3).

Intracellular and tissue ROS measurement

Intracellular reactive oxygen species (ROS) levels were determined using the DCFH-DA fluorescent probe. After treatment, cells were incubated with DCFH-DA in the dark at 37° C for the indicated time and then washed with PBS to remove excess probe. The fluorescence intensity was analyzed by flow cytometry, and ROS levels were expressed as the percentage of ROS-positive cells or relative fluorescence intensity. For the tissue experiments, fresh intestinal and liver tissues were prepared as frozen sections and incubated with a ROS fluorescent probe under dark conditions at 37 °C. After washing, the nuclei were stained with DAPI, and the sections were examined under a fluorescence microscope. ROS fluorescence was visualized as red staining, and nuclei were visualized by blue DAPI staining. The fluorescence intensity was used to evaluate ROS accumulation in the intestinal and liver tissues.

Statistical analysis and visualization

All analyses were performed in SPSS software (v19.0; IBM, Chicago, IL, USA). A one-way ANOVA followed by Fisher’s LSD post hoc test was used to compare differences between treatment groups. Results are presented as mean ± SEM. Growth performance was assessed in n = 20 lambs per group, whereas blood and tissue outcomes were evaluated in n = 6 lambs per group at day 35. Statistical significance was defined as *P < 0.05; **P < 0.01; ***P < 0.001; and 0.05 ≤ P < 0.1 indicated a trend. Different lowercase letters in the same figure denote significant differences (P < 0.05). Graphical data were generated with GraphPad Prism 8 (CA, USA).

Author contributions

Q.C.L.: Conceptualization, Methodology, Data curation, Writing—original draft. P.Z.: Formal analysis. Y.J.N.: Software. C.Y.W.: Investigation. Y.Y.Q.: Supervision, Investigation. J.L.N.: Supervision, Formal analysis. C.C.: Supervision, Investigation. W.B.Z.: Supervision, Project administration, Funding acquisition. W.J.Z.: Conceptualization, Project administration, Funding acquisition.

Data availability

The sequencing data produced in this study have been deposited in the Sequence Read Archive (SRA) and are available from the National Center for Biotechnology Information (NCBI) under accession numbers PRJNA1286083 and PRJNA1287334. The raw data supporting the findings of this article are available from the authors upon reasonable request.

Code availability

No custom code was used in this study.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41522-026-01104-6.