Microbiota-derived indole-3-acetic acid alleviates rumen epithelial barrier dysfunction during the peripartum period through AhR signaling

Abstract

Peripartum dairy cows are highly susceptible to metabolic disorders, with ketosis being the most prevalent postpartum disease associated with rumen microbial dysbiosis and systemic inflammation. However, the mechanisms by which microbial alterations compromise rumen epithelial integrity remain poorly understood. Using peripartum cows with ketosis as a model, we demonstrated that perturbations of rumen microbiota disrupt tryptophan metabolism, resulting in pronounced depletion of indole-3-acetic acid (IAA). The loss of IAA-producing taxa (Lactobacillus and Bifidobacterium) contributed to reduced IAA levels and epithelial barrier dysfunction, whereas enrichment of proinflammatory taxa (Candidatus Saccharimonas and Mycoplasma) was associated with exacerbated epithelial inflammation. In vitro, IAA supplementation activated the AhR/IL-22 signaling pathway, promoting bovine rumen epithelial cells (BRECs) regeneration and restoring barrier integrity. These findings identify the microbiota–IAA–AhR/IL-22 axis as a key regulator of rumen epithelial homeostasis and suggest that targeting this pathway represents a promising strategy to prevent metabolic disorders in dairy cows.

Introduction

The ruminal epithelium acts both as a protective barrier for host defense and as a tissue with metabolic importance for energy metabolism in ruminants^1,2. As a representative squamous stratified epithelium, the epithelium of the rumen comprises four cellular layers: the stratum basale, stratum spinosum, stratum granulosum, and stratum corneum^3,4. As a keratinized cellular layer, the stratum corneum immediately contacts the rumen milieu and functions as a physical barrier, preventing microbes and harmful substances in the rumen fluid, such as endotoxins and biogenic amines, from penetrating the ruminal epithelium⁵. Adjacent to the stratum corneum is the stratum granulosum, which contains abundant junctional complexes that allow the selective absorption of nutrients and serve as the permeability barrier of the epithelium^3,6. The cells in the stratum granulosum are tightly connected via tight junction (TJ) proteins, including adapter proteins such as zonula occludens (ZO) and transmembrane proteins such as occludin and claudins⁷. TJs are crucial intercellular interactions for the formation of the epithelial barrier, which strengthens the mechanical defense of the rumen epithelium and plays an important role in sustaining the integrity of the mucosal barrier^6,8,9. TJ proteins can be detected in the stratum basale and stratum spinosum, and the intensities decrease as they approach the basal membrane of the rumen epithelium⁷. For ruminants, the unique and complex structure of ruminal epithelial layers greatly influences optimal rumen digestion, absorption, and defense mechanisms. Therefore, maintaining the integrity of the epithelial structure and function of the rumen is important for effective and normal ruminant production.

Dairy cows are livestock animals that are economically important for milk supply to humans. The rumen serves as a complicated and symbiotic ecosystem involving different microbial species and is important for physiological and metabolic adaptations as well as host energy homeostasis¹⁰. The ruminal epithelial barrier can protect against exogenous pathogen invasion or exposure to harmful substances in rumen fluid⁶. Many studies have associated the gut microbiota with intestinal epithelial barrier function¹¹. Disruptions in the gut microbiota may result in an impaired intestinal barrier, increasing permeability and inflammation¹². Several studies have indicated that changes in the composition and corresponding metabolites of the rumen microbiota facilitate the pathogenesis of metabolic disorders in dairy cows^13–15. During modern dairy production, ketosis can be a frequent metabolic disorder that causes significant economic losses in the dairy farming industry, leading to displaced abomasum, metritis, lameness, and culling, as well as reducing milk yield and reproductive performance^16–18. Studies have identified disturbances in the ruminal microbiota in ketotic cows, suggesting that perturbations in the rumen microbiota are associated with the health status of the host^19,20. Integrated meta-omics analyses revealed that ruminal microbes may contribute to the progression of ketosis by influencing β-hydroxybutyrate (BHBA) accumulation and short-chain fatty acid metabolism²¹. Although microbiota dysbiosis and its impact on epithelial barrier function have been extensively studied in mice and humans, the specific relationship between rumen microbiota alterations and epithelial barrier integrity in dairy cows suffering from metabolic disorders remains poorly understood.

Gut microbial metabolites, which also regulate host-microbe interactions, have important effects on gut health^22,23. Tryptophan metabolism has attracted interest because it can regulate intestinal barrier function and integrity^24,25. Gut microbes such as Clostridium, Lactobacillus, Bifidobacterium, and Bacteroides catabolize tryptophan into various metabolites²⁶. Specifically, indoles, tryptophan metabolites, and associated derivatives such as indoleacrylic acid (IA), indolelactic acid (ILA), indole-3-acetic acid (IAA), and indolealdehyde (IAld) are ligands for aryl hydrocarbon receptor (AhR), causing AhR activation²⁵. AhR serves as a transcription factor and modulates intestinal barrier homeostasis²⁷. AhR protects intestinal barrier integrity by promoting the production of interleukin-22 (IL-22)²⁸. AhR signaling can exert barrier-protective effects by upregulating IL-22, further promoting TJ and antimicrobial peptide expression and cell proliferation^29–31. Although many studies have confirmed that microbial tryptophan catabolites (indole and its associated derivatives) are important for mediating AhR activation to alleviate impaired intestinal barrier integrity in humans and mice, it remains unclear whether the rumen microbiota can modulate tryptophan metabolism and further influence rumen epithelial barrier function in dairy cows with ketosis.

Although the association between the microbiota and altered epithelial barrier function has been established, the mechanisms by which microbiota-driven impairment of ruminal barrier integrity occurs, particularly those involving the host–metabolite–microbe signaling axis at this critical interface, remain poorly understood in peripartum dairy cows with metabolic disorders. In this study, we utilized a model of ketosis in dairy cows to investigate alterations in the compositional and taxonomic dynamics of the ruminal bacterial community, the enrichment of microbial tryptophan catabolism in ruminal fluid, and the resulting compromise of epithelial barrier function during the peripartum period. Additionally, we validated the role of the AhR/IL-22 axis in vitro using immortalized bovine ruminal epithelial cells (BRECs) and pharmacological inhibition of key target molecules. Our results demonstrated that IAA mediates the crosstalk between rumen microbiota and host epithelium, revealed the regulatory effect of AhR on rumen epithelial barrier function, and provided novel directions and therapeutic targets for addressing rumen homeostasis dysbiosis in dairy cows with metabolic disorders.

Results

Rumen microbiota composition and identification of key microbial taxa in healthy and ketotic dairy cows

To assess local inflammation in the rumen, we first measured the LPS content in the ruminal tissues. The LPS contents of the KET group were considerably higher than those in the CON group, indicating localized inflammation of the rumen epithelium (Fig. 1A). To assess the microbial contributions to this type of rumen inflammation, 16S rDNA high-throughput sequencing of the rumen fluid was performed. A saturation plateau was reached on rarefaction curves, suggesting sufficient sample coverage and sequencing depth (Fig. 1B). The results of the alpha diversity analyses showed that the Chao1 and Shannon indices in the KET group increased significantly, indicating an increase in microbial richness and diversity (Fig. 1C). The results of the Bray-Curtis distance-based principal coordinate analysis (PCoA) revealed apparent clustering between the two groups, reflecting the different compositions of the microbial communities (Fig. 1D). An altered Firmicutes-to-Bacteroidota (F/B) ratio is a marker of microbial dysbiosis. The KET group presented a considerably high F/B ratio, suggesting disrupted microbial homeostasis (Fig. 1E). Taxonomic analysis revealed that while the abundance of Firmicutes did not differ significantly at the phylum level, several genera exhibited prominent changes. Specifically, the inflammation-associated genera Mycoplasma and Candidatus Saccharimonas were significantly enriched in the KET group, whereas tryptophan-metabolizing genera, including Limosilactobacillus, Lactobacillus, Ligilactobacillus, and Bifidobacterium, decreased considerably. These reductions were particularly evident at the species level, including Lactobacillus mucosae LM1, Lactobacillus acetotolerans, and Lactobacillus sp. KLDS 10716 (Fig. 1F). Significant depletion of the anti-inflammatory genus Akkermansia, including its representative species Akkermansia muciniphila, was found in the KET group (Fig. 1F). Through linear discriminant analysis effect size (LEfSe) analysis, 62 differential genera were identified as two groups (LDA > 2.5, P < 0.05), with 24 and 38 enriched genera in the CON and KET groups, respectively (Supplementary Table S1). To identify potential microbial biomarkers associated with ruminal epithelial barrier dysfunction, we performed random forest analysis followed by LEfSe and receiver operating characteristic (ROC) analysis. Several genera, including Lactobacillus, Ligilactobacillus, Limosilactobacillus, Bifidobacterium, and Akkermansia (enriched in CON) and Mycoplasma and Candidatus Saccharimonas (enriched in KET), showed strong discriminatory power and may serve as potential biomarkers (Fig. 1G, Supplementary Table S2). The results of the indicator species analysis further supported the biomarker potential of these key genera (Fig. 1H). To determine the relationships among these taxa, we conducted a Spearman correlation network analysis. Lactobacillus exhibited widespread positive correlations with other genera, forming a highly connected microbial network. In contrast, Mycoplasma and Candidatus Saccharimonas were negatively correlated with most taxa, suggesting that they may promote inflammation in the rumen through microbial antagonism (Fig. 1I).

Fig. 1. Rumen microbiota composition and identification of key microbial taxa in healthy and ketotic dairy cows.

A Concentration of LPS in ruminal fluid. B Chao1 curves. C Alpha diversity of rumen microbial richness was determined by measuring the Chao1 and Shannon diversity indices. D Beta diversity was determined by conducting principal coordinate analysis (PCoA) using the Bray-Curtis distance matrix (n = 10). E The relative abundances of Bacteroidetes and Firmicutes and the F/B ratio are shown. F Comparison of abundance between CON and KET at the phylum, genus, and species levels. G The key genera (top 25 at the genus level) were identified by random forest, receiver operating characteristic (ROC), and linear discriminant analysis effect size (LEfSe) analyses. H Differential biomarkers in the CON and KET groups were identified by indicator species analysis. I Correlation network of differential genera. Red, positive correlations; blue, negative correlations. Line thickness represents the strength of the correlation.

Suppression of tryptophan metabolism leads to decreased IAA production in the rumen of ketotic cows

We conducted rumen fluid metabolomics to investigate the mechanisms related to ruminal barrier injury in dairy cows with ketosis. A total of 2412 metabolites were detected, which were primarily classified into the “Organic acids and derivatives”, “Lipids and lipid-like molecules”, “Organoheterocyclic compounds”, “Organic oxygen compounds”, and “Benzenoids” categories (Fig. 2A). We found total metabolic differences through orthogonal partial least squares discriminant analysis (OPLS-DA), revealing a distinct metabolic profile in the KET and CON groups (Fig. 2B). Further differential analysis (fold change ≥2 or ≤0.5, P ≤ 0.05, and VIP ≥ 1) revealed 271 significantly altered metabolites, including 214 and 57 upregulated metabolites in the CON and KET groups, respectively (Fig. 2C). The differential metabolites were enriched in KEGG pathways, which revealed enrichment in tryptophan metabolism (Fig. 2D, Supplementary Table S3). This finding was confirmed through gene set enrichment analysis (GSEA), which also revealed that tryptophan metabolism was distinctly enriched in the CON group (Fig. 2E). Metabolite set enrichment analysis (MSEA) and KEGG pathway enrichment identified seven key metabolites (Fig. 2F, Supplementary Table S4). The results of the redundancy analysis (RDA) suggested that the above key metabolites were strongly correlated with the composition of the ruminal microbiota (Fig. 2G). Variation partition analysis (VPA) further quantified their contributions and indicated that IAA contributed the most to the variation in microbial community structure and function (Fig. 2H). The relative abundance of IAA was significantly greater in the CON group than in the KET group (Fig. 2I). The results of the ROC analysis revealed that IAA exhibited good predictive power in distinguishing between healthy and ketotic states, suggesting its potential as a metabolic biomarker involved in ketosis (Fig. 2J). Further analysis of the indole derivative branch in tryptophan metabolism revealed a decrease in the relative abundance of several intermediate metabolites in the KET group, indicating that impaired indole biosynthesis may be a key cause of a reduction in IAA levels (Fig. 2K). The results of the linear regression analysis showed that the abundances of Lactobacillus, Limosilactobacillus, and Ligilactobacillus were positively related to IAA levels, whereas the decrease in the abundance of these potential IAA-producing taxa in the KET group may have contributed to a decrease in IAA production (Fig. 2L).

Fig. 2. Suppression of tryptophan metabolism leads to decreased IAA production in the rumen of ketotic cows.

A Percentages of the main lipid subclasses. B Orthogonal partial least squares discriminant analysis (OPLS-DA) score plot for discriminating the rumen metabolome from the CON and KET groups (n = 10). C Volcano plot showing differential metabolites. D KEGG analysis of differentially metabolic pathway. E GSEA results of tryptophan metabolism. F Venn diagram of the KEGG analysis and Metabolite set enrichment analysis (MSEA). G Redundancy analysis (RDA) of key metabolites and the composition of the ruminal microbiota. H Variation partition analysis (VPA). I Comparison of the relative abundance of IAA. J ROC analysis of IAA. K Summary diagram of disrupted metabolites in the rumen tryptophan metabolism pathway. L Association of IAA concentration with the abundance of Limosilactobacillus, Lactobacillus, and Ligilactobacillus.

Histopathological alterations and the inflammatory response in the rumen epithelium of ketotic cows

Cows with ketosis presented abnormal rumen epithelial structures characterized by mild epithelial thinning and a loose overall organization (Fig. 3A). The intercellular spaces in the granular and spinous layers appeared slightly enlarged in the KET group relative to those in the CON group (Fig. 3A). TUNEL staining revealed a greater number of apoptotic nuclei in the rumen epithelium in ketotic cows (Fig. 3B). The inflammatory response is a key characteristic of cows with ketosis. To further investigate this, we analyzed the expression of the proinflammatory cytokines IL-1β, IL-6, and TNF-α in the rumen epithelium via immunofluorescence, ELISA, qPCR, and Western blotting. Compared to the CON group, the KET group presented substantially upregulated IL-1β, IL-6, and TNF-α mRNA and protein levels, indicating severe inflammatory damage to the rumen epithelium in ketotic cows (Fig. 3C–G). The subsequent correlation analysis between IAA and the above inflammatory factors in the rumen epithelium of cows with ketosis revealed a strong inverse relationship between IAA and these inflammatory factors (Fig. 3H). Consistently, we observed significant negative correlations between ruminal LPS concentrations and the expression of tight junction proteins (ZO-1, Occludin, and Claudin), indicating that elevated LPS levels were closely associated with reduced tight junction integrity (Fig. 3I). Together, these findings suggest that both decreased IAA and increased LPS contribute to inflammation and epithelial barrier dysfunction in ketotic cows.

Fig. 3. Histopathological alterations and the inflammatory response in the rumen epithelium of cows with ketosis.

A H&E staining of the rumen epithelium tissue. B Representative TUNEL images. C Immunofluorescence images of proinflammatory cytokines (IL-1β, IL-6, and TNF-α). D Average fold change in the fluorescence intensities of proinflammatory cytokines. E qRT-PCR analysis of proinflammatory cytokines at the mRNA level. F Concentrations of IL-1β, IL-6, and TNF-α in the rumen epithelium. G Western blotting analysis of the protein levels of IL-1β, IL-6, and TNF-α. H Correlation analysis between IAA and proinflammatory cytokines. I Correlation analysis between LPS and tight junction proteins. The results are presented as the mean ± SEM (n = 3). Unpaired Student’s t-tests (two-sided) were conducted to determine differences between the groups; ^*P < 0.05, ^**P < 0.01, and ^***P < 0.001 indicated significant differences between the CON group and the KET group.

Role of the AhR/IL-22 signaling pathway in regulating rumen epithelial function in ketotic cows

We evaluated rumen epithelial barrier function by assessing the levels of TJ proteins, cell proliferation markers, and antimicrobial peptides. Immunofluorescence analysis revealed significantly lower levels of TJ proteins (ZO-1, Occludin, and Claudin-3) in the rumen epithelial cells of cows with ketosis (Fig. 4A, C). The results of Western blotting and qPCR analyses confirmed a reduction in these protein levels in the rumen epithelium of cows with ketosis (Fig. 4D, E). PCNA and Ki67, which serve as important markers of cell proliferation, were also significantly downregulated in the rumen epithelium, indicating that the proliferative capacity of rumen epithelial cells in cows with ketosis was impaired (Fig. 4B–D). Similarly, the level of antimicrobial peptides decreased considerably in the rumen epithelium of ketotic cows (Fig. 4I, J). AhR is a nuclear receptor that functions as a transcription factor following its nuclear translocation. Therefore, the expression of AhR was analyzed by immunofluorescence staining. The results indicated impaired and suppressed expression of AhR in cows with ketosis (Fig. 4F). AhR signaling can exert barrier-protective effects by upregulating IL-22. Similar to AhR expression, the expression of IL-22 in the KET group decreased considerably relative to that in the CON group and was significantly positively correlated with the expression of AhR (Fig. 4G, H). AhR activation by ligand binding induces the transcription of CYP1A1 and CYP1B1, whose expression levels are commonly used as markers of AhR activation³². The expression of AhR, ARNT, and their downstream targets (CYP1A1 and CYP1B1) was confirmed by conducting Western blotting and qRT-PCR assays. The protein and mRNA levels of AhR, ARNT, CYP1A1, and CYP1B1 were lower in the KET group than in the CON group (Fig. 4I, J). Then, the expression of STAT3 and phosphorylated STAT3 in the rumen epithelial tissue was examined. The p-STAT3-to-total STAT3 ratio decreased substantially in the KET group (Fig. 4I), suggesting that STAT3 phosphorylation was inhibited, probably because the AhR/IL-22 pathway was suppressed. These findings also indicated a close association of the AhR/IL-22 pathway with rumen epithelial barrier function, according to the TJ protein, cell proliferation marker, and antimicrobial peptide levels (Fig. 4K).

Fig. 4. Role of the AhR/IL-22 signaling pathway in regulating rumen epithelial function in ketotic cows.

A, B Immunofluorescence images of tight junction (TJ; ZO-1, Occludin, and Claudin-3) and cell proliferation marker (Ki67 and PCNA) expression. C Average fold change heatmap of the fluorescence intensities of TJ and cell proliferation markers. D The heatmap shows the mRNA expression of TJ and cell proliferation markers. E Western blotting analysis of the protein levels of ZO-1, Occludin, and Claudin-3. F Representative IF images of AhR. G Representative IF images and fluorescence intensity analysis of IL-22. H Correlation analysis between AhR and IL-22. I Western blotting analysis of the AhR/IL-22 pathway-related protein levels. J Heatmap of relative AhR/IL-22 pathway-associated gene expression levels. K Chord diagrams. The results are presented as the mean ± SEM (n = 3). Unpaired Student’s t-tests (two-sided) were performed to determine differences between the groups; ^*P < 0.05, ^**P < 0.01, and ^***P < 0.001 indicate significant differences between the CON group and the KET group.

IAA alleviates LPS-induced BRECs injury via AhR/IL-22 signaling pathway

To analyze the ability of IAA to protect against LPS-mediated inflammatory injury, an in vitro evaluation was performed using BRECs (Fig. 5A). In our previous study, 10 µg/mL LPS was identified as the most effective concentration for establishing optimal inflammatory conditions in BRECs³³. After treating BRECs with IAA at doses > 500 µM for 2 h, 6 h, or 12 h, cell proliferation decreased continuously and significantly (P < 0.05) (Fig. 5B). However, at 6 h, the reduction in cell proliferation was moderate, indicating a less dramatic change. Based on the above results, we selected 100, 250, and 500 µM IAA for 6 h for subsequent analyses. By performing CCK-8 viability assays, we established an LPS-induced BREC injury model and an IAA antagonism model. To assess the protective effect of IAA on BRECs exposed to LPS, we evaluated the AhR/IL-22 pathway, proinflammatory cytokine, and TJ protein levels. Compared to those in the CON group, the AhR, IL-22, P-STAT3/STAT3, TJ (ZO-1, Occludin, Claudin-3), and Reg3g levels were lower, whereas the levels of proinflammatory cytokines (IL-1β, IL-6, and TNF-α) were significantly greater in the LPS group. (Fig. 5D–F). In the IAA group, the levels of AhR, IL-22, P-STAT3/STAT3, TJ proteins (ZO-1, Occludin, and Claudin-3), and Reg3g increased, whereas the protein levels of proinflammatory cytokines (IL-1β, IL-6, and TNF-α) decreased significantly in a dose-dependent manner as IAA concentrations increased (Fig. 5D–F). These findings indicated that IAA activated the AhR/IL-22 pathway, thereby alleviating LPS-induced damage in BRECs. The expression patterns detected by immunofluorescence for AhR, IL-22, STAT3, ZO-1, Occludin, Claudin-3, and Reg3g generally matched the results identified through Western blotting and qPCR assays (Fig. 5G, H). Additionally, the LPS group presented a lower number of Ki67-positive cells (Fig. 9D). In contrast, the IAA groups presented a larger number of Ki67-positive cells (Fig. 5G, H), suggesting that IAA promotes epithelial cell proliferation and injury recovery.

Fig. 5. IAA alleviates LPS-induced BRECs injury via AhR/IL-22 signaling pathway.

A LPS and IAA at different doses were added to the BRECs. B Identification of BRECs by Cytokeratin 18 (CK18). C A CCK-8 assay was performed to evaluate cell viability following IAA treatment at different doses for 2, 6, and 12 h. D Western blotting analysis was performed to analyze the protein levels of the AhR/IL-22 pathway, proinflammatory cytokines, TJs, and antimicrobial peptides. E Protein abundance analysis. F mRNA expression analysis. G, H Immunofluorescence images and fluorescence intensity analysis of AhR, IL-22, STAT3, ZO-1, Claudin-3, Occludin, Reg3g, and Ki67. The results are presented as the mean ± SEM (n = 3). One-way ANOVA followed by Fisher’s LSD post hoc test was performed to determine differences between the groups. ^a–d Values without identical letters are statistically significant (P < 0.05).

Inhibiting AhR activation abolishes the protective effects of IAA

To determine whether IAA protects against BREC injury via the AhR pathway, CH223191, an AhR antagonist, was added to LPS-treated BRECs. The molecular docking results revealed the binding of IAA to the ligand-binding domain of AhR, and the binding energy was –6.469 kcal/mol, indicating that IAA can strongly bind to AhR (Fig. 6A). As shown in Fig. 6B–D, CH223191 administration markedly decreased the protein and mRNA levels of AhR, IL-22, STAT3, TJ proteins (ZO-1, Occludin, and Claudin-3), and Reg3g compared to their respective levels in the LPS + IAA group. In contrast, the protein and mRNA levels of proinflammatory cytokines (IL-1β, IL-6, and TNF-α) were greater in the LPS + IAA + CH group than in the LPS + IAA group (Fig. 6B–D). The results of the immunofluorescence analysis (Fig. 6E, F) supported these findings, showing that CH223191 administration markedly reduced the levels of AhR, IL-22, STAT3, TJ proteins, Reg3g, and the proliferation marker Ki67 compared to their levels in the LPS + IAA group Fig. 7. To summarize, IAA protects against LPS-mediated inflammatory injury by activating the AhR/IL-22 pathway.

Fig. 6. Inhibition of AhR activation abolished the protective effects of IAA.

A Molecular docking of IAA was performed with the ligand-binding domain of the AhR protein, and its binding energy was –6.469 kcal/mol. B Western blotting analysis of the expression of the AhR/IL-22 pathway, proinflammatory cytokines, TJ proteins, and antimicrobial peptide proteins. C Protein abundance analysis was performed. D mRNA expression analysis was performed. E, F Immunofluorescence images were captured, and fluorescence intensity analysis of AhR, IL-22, STAT3, ZO-1, Claudin-3, Occludin, Reg3g, and Ki67 was performed. The results are presented as the mean ± SEM (n = 3). One-way ANOVA followed by Fisher’s LSD post hoc test was conducted to determine the differences between the groups. ^a–c Values without identical letters are statistically significant (P < 0.05).

Fig. 7. Schematic diagram of the microbial metabolite indole-3-acetic acid alleviates rumen epithelial barrier dysfunction by modulating AhR activation in ketotic dairy cows during the peripartum period.

Rumen microbiota dysbiosis drives tryptophan metabolism disruption, leading to epithelial barrier dysfunction, and supplementation with IAA in vitro alleviates LPS-induced injury to BRECs by activating AhR.

Discussion

The rumen epithelium is involved in host defense responses and acts as the immune interface with the ruminal environment to prevent the invasion of pathogenic bacteria and harmful metabolites. Although many studies have assessed the human gut microbiota and intestinal barrier, only a few studies have investigated the relationship between the rumen microbiota and epithelial barrier functions. To identify the key microbial species and microbial metabolites in the rumen, we integrated 16S rDNA gene sequencing and nontargeted metabolome analyses of the rumen fluid and elucidated the rumen epithelial host-microbe interactions in dairy cows with ketosis. Our findings showed that dairy cows with ketosis exhibit perturbations in the rumen microbiota, which consequently increases LPS production and disrupts microbiota-tryptophan metabolism, thereby impairing the rumen epithelial barrier. Moreover, IAA, an important tryptophan catabolite, alleviated the damage to ruminal epithelial cells exposed to LPS by activating the AhR/IL-22 pathway, thus decreasing the inflammatory response while increasing the repair of the epithelial barrier function.

The rumen microbiota plays an important role in host energy metabolism and immune regulation. Studies have shown that the rumen microbiota dynamically modulates epithelial inflammatory responses and metabolic disorders in dairy cattle^34,35. High-grain feeding causes structural and compositional alterations in rumen bacteria, altering rumen metabolism and fermentation, increasing epithelial permeability, and triggering inflammation^13,36. Additionally, rumen epithelial barrier function and structural integrity are impaired in dairy cattle with diet-induced metabolic disorders, probably affecting host metabolism and activating systemic and local inflammation^37–39. Ketosis is the most prevalent peripartal disease in prolific dairy cows. Ketotic dairy cows exhibit typical features of an overt systemic inflammatory response and dysbiosis of the rumen microbiota^19,40. However, studies on the interplay between the microbiota and the ruminal epithelium remain unclear. Our results showed that ketotic cows experience rumen microbiota dysbiosis and damage to their epithelial structure. Further assessment of epithelial cell proliferation, inflammation, and TJs and antimicrobial peptide expression revealed compromised barrier function characterized by inhibited cell proliferation, reduced antimicrobial peptide expression, heightened inflammation, and disrupted TJs in dairy cows with ketosis. Thus, we speculated that rumen microbiota dysbiosis may drive epithelial inflammation, impairing barrier integrity and disrupting rumen homeostasis, which adversely affects cow health. It should be noted that species-level taxonomic assignments based on 16S rDNA sequencing carry inherent uncertainty due to limitations in the amplified regions, reference databases, and classification algorithms. Future studies using metagenomic sequencing or isolation-based approaches could further validate and explore these findings.

Rumen microbiota dysbiosis is a key etiological factor that influences volatile fatty acid metabolism in dairy cows with ketosis²¹. The postpartum rumen bacterial composition undergoes considerable changes and takes longer to recover in ketotic cows⁴¹. The microbiota analysis in this study revealed a significant difference in the structure and composition of the rumen microbiota of ketotic dairy cows. Microbiota dysbiosis is related to reduced probiotic levels and increased enteric pathogen and LPS production. The genera Lactobacillus and Akkermansia are safe and effective probiotics that regulate host metabolism, particularly for maintaining the intestinal barrier functional integrity, by producing beneficial metabolites^42,43. We found a substantial decrease in commensal symbionts (Lactobacillus and Akkermansia) and a significant increase in LPS, which improved the impaired rumen epithelial barrier function of dairy cows with ketosis. Additionally, Mycoplasma and Candidatus Saccharimonas expansion is related to inflammatory disorders, such as colitis and endometrial inflammation^44,45. Our findings revealed enrichment of the proinflammatory genera Mycoplasma and Candidatus Saccharimonas. The pathogenic expansion of inflammation-associated taxa may exacerbate barrier dysfunction by promoting inflammation. Such microbial alterations occurred with an increase in LPS generation in the rumen, thereby negatively affecting the rumen epithelium. Overall, these results suggested that ketotic dairy cows presented significant shifts in microbial composition characterized by depletion of probiotics and enrichment of the proinflammatory microbiota. The dysbiosis of the rumen microbiota increased intestinal permeability and endotoxin LPS levels, further exacerbating rumen epithelial inflammation and disrupting barrier function; these adversities are associated with detrimental effects on host health.

Gut microbiota-generated metabolites play key roles in mediating host-microbial crosstalk. Rumen bacterial community structural and compositional alterations lead to corresponding changes in rumen fluid metabolites. In this study, our metabolomic analysis revealed significant enrichment of the tryptophan metabolism pathway. MSEA and KEGG analysis revealed that IAA, a central node metabolite, was considerably depleted in the perturbed tryptophan metabolism network. Additionally, ROC analysis revealed that IAA is a biomarker in healthy and ketotic dairy cows. Based on the above findings, the microbiota may modulate ruminal epithelial barrier function through tryptophan metabolism. Tryptophan, an essential amino acid in mammals, can be metabolized by the gut microbiota into indole and its derivatives, which maintain the integrity of intestinal barrier function and epithelial homeostasis⁴⁶. Lactobacillus, Limosilactobacillus, Ligilactobacillus, and Bifidobacterium generate indole derivatives and positively affect host metabolism regulation^47–49. We also found that Lactobacillus is an important differential microbial genus in healthy dairy cows by conducting ROC and random forest analyses. The abundance of Lactobacillus was positively related to IAA. Based on the above findings, the marked depletion of Lactobacillus in ketotic cows probably mediates the reduction in IAA production, which in turn compromises the integrity of the ruminal epithelial barrier.

AhR serves as a key regulatory factor for maintaining intestinal barrier homeostasis. Gut microbiota-produced tryptophan metabolites, especially IAA, are mostly endogenous ligands for AhR⁴⁶. AhR activation ameliorates colitis by reducing inflammation and improving epithelial barrier function^50–52. Moreover, highly expressed AhR stimulates IL-22 production and downstream STAT3 phosphorylation, which directly enhances the expression of tight junction proteins such as ZO-1, Occludin, and Claudin3, as well as antimicrobial peptides, thereby improving epithelial barrier integrity^53,54. Following intestinal mucosal injury, IL-22 can restore the epithelial barrier by promoting regeneration and proliferation of the epithelium^30,31. In ketotic cows, rumen microbiota dysbiosis leads to decreased IAA levels, resulting in suppressed AhR/IL-22 signaling, reduced tight junction protein expression, and compromised barrier function. These findings highlight a mechanistic link between microbial metabolites, AhR/IL-22 signaling, and the maintenance of rumen epithelial integrity.

To elucidate the protective effects of IAA against inflammatory damage to the rumen epithelium in ketotic cows and the key regulatory role of AhR, we conducted an in vitro study in which BRECs were stimulated with LPS. The in vitro data suggested that IAA alleviated the LPS-induced inhibition of the proliferation of BRECs and decreased TJ expression and antimicrobial peptide production. We also determined whether the protective effects of IAA against LPS-mediated injury in ruminal epithelial cells were orchestrated through the AhR signaling pathway. CH223191, an AhR inhibitor, was combined with LPS and IAA to investigate the involvement of AhR signaling. Compared to those in the LPS + IAA group, the levels of intestinal inflammatory factors, as well as cell proliferation, increased considerably, and the mRNA and protein expression levels of TJ proteins and antimicrobial peptides decreased significantly in the CH223191 treatment group. These results indicated that the protective effects of IAA on rumen epithelial cell injury were weakened by CH223191 treatment. Therefore, IAA activated AhR/IL-22 signaling to improve LPS-mediated rumen epithelial cell injury.

In summary, our findings provide mechanistic insight into how IAA mediates rumen microbiota–host epithelial crosstalk to maintain rumen barrier integrity in peripartum metabolic disorder dairy cows. Specifically, perturbations in the rumen microbiota disrupted tryptophan metabolism and led to increased LPS production in ketotic cows. Moreover, microbial dysbiosis contributed to a decrease in the production of tryptophan metabolite IAA due to the depletion of tryptophan-producing bacteria, especially Lactobacillus. The depletion of IAA and elevated LPS trigger TJ protein dysregulation and inflammation in the rumen epithelium, resulting in the dysfunction of the epithelial barrier. Disruption of the AhR/IL-22 signaling axis compromises ruminal barrier function, whereas in vitro IAA supplementation alleviates LPS-induced damage to BRECs via AhR activation (Fig. 7). These findings suggest that targeting the microbiota–IAA–AhR/IL-22 axis may offer novel strategies to enhance the productivity and health of dairy herds.

Methods

Ethics statement

All animal experiments were approved by the Animal Welfare Committee of China Agricultural University (permit no. AW12305202-2-02).

Animal experiment design and sample collection

The experiment was conducted at a commercial dairy in Dingzhou City (Hebei Province), and all cows were raised in an intensive system. Before the start of the experiment, dairy cows (n = 89) with similar numbers of lactations (median = 3, range = 2–4) and days in milk (DIM) (median = 7 d, range = 5–10) were enrolled in this study. To exclude cows with potential comorbidities, all animals underwent routine physical examinations by a licensed veterinarian for 3 consecutive days. A total of 10 cows were excluded due to clinical signs of conditions such as displaced abomasum, mastitis, endometritis, or laminitis. Blood samples were collected from the tail vein using 22-gauge blood collection needles and non-anticoagulant evacuated tubes (Hebei Kangweishi Medical Technology Co., Ltd., China). After allowing the blood to clot at room temperature for 2 h, the samples were centrifuged at 3500 × g for 15 min at 4 °C, and the serum was collected for analysis. Blood BHBA concentrations were then measured using a hand-held electronic BHBA meter (Precision Xtra, Abbott, USA). Cows with blood BHBA concentrations ≤0.6 mmol/L were classified as healthy, whereas those with blood BHBA concentrations ≥3.0 mmol/L were diagnosed with clinical ketosis. Based on clinical symptoms and BHBA levels, 20 cows were ultimately selected and assigned to two groups: the healthy control group (CON; n = 10; BHBA ≤ 0.6 mmol/L) and the clinical ketosis group (KET; n = 10; BHBA ≥ 3.0 mmol/L). All cows were fed a total mixed ration (TMR) and had ad libitum access to clean drinking water. The detailed composition of the basal diet is provided in Supplementary Table S5, and the basic physiological parameters of the selected cows are summarized in Table 1.

Table 1.

Basic characteristics of normal and ketosis cows^a

Itemsb	CON (n = 10)	KET (n = 10)	P value
Parity	3.300 ± 0.260	2.800 ± 0.249	0.1825
DIM	10.700 ± 0.700	10.900 ± 0.690	0.8411
BW (kg)	615.100 ± 13.171	633.600 ± 16.912	0.3995
BCS	3.000 ± 0.105	3.350 ± 0.113	0.0361
DMI (kg/d)	21.280 ± 0.950	19.652 ± 0.858	0.3945
Milk yield (kg/d)	34.616 ± 0.859	27.734 ± 1.060	<0.001
BHBA (mmol/L)	0.450 ± 0.040	3.320 ± 0.099	<0.001

^aData are expressed as the mean ± SEM.

^bDIM days in milk, BW body weight, BCS body condition score, DMI dry matter intake.

Before feeding, blood was collected via coccygeal venipuncture from 0730 to 0830 h for 3 consecutive days, and the samples were centrifuged at 4 °C for 15 min at 3500 × g to obtain the serum. Ruminal fluid was collected from oral stomach tubes as described previously⁵⁵ and was stored at –80 °C. Ten cows per group were used for microbiota sequencing and metabolomic profiling analysis. For rumen epithelial tissue sampling, three cows (n = 3) per group were selected. The ruminal papillae were obtained through biopsy using the rumenotomy procedure, as reported in another study⁵⁶. The ruminal tissue samples were obtained through a standard rumenotomy approach via the left paralumbar fossa under local infiltration anesthesia with 2% lidocaine hydrochloride (L7780, Solarbio, China) and maintained in a standing position. Ruminal papillae (approximately 1 cm pieces) were collected and rinsed with 0.9% NaCl solution. The procedures used in this study followed the guidelines of the Animal Welfare Committee of China Agricultural University. All freshly collected samples were classified into two parts, with one being immersed in 4% paraformaldehyde during hematoxylin and eosin (H&E) and immunofluorescence staining, and the other being stored at –80 °C before use.

Histopathological analysis

The rumen epithelial papillae were fixed in 4% paraformaldehyde, dehydrated through a graded ethanol series, cleared in xylene, embedded in paraffin, and sectioned at 3 μm for H&E staining. Stained sections were observed under a light microscope (BX51, Olympus, Japan). Apoptosis of rumen epithelial papillae was evaluated by TdT‑mediated dUTP nick end labeling (TUNEL) staining. Paraffin sections were processed according to the manufacturer’s instructions using a commercial TUNEL kit (G1501, Servicebio, China) and observed under a fluorescence microscope (Nikon DS-U3, Japan).

Immunofluorescence staining

The paraffin section preparation for immunofluorescence staining was conducted using the methods described in the Histopathological analysis section. The tissue was sectioned (3 μm thick), followed by xylene deparaffinization and gradient ethanol dehydration. Next, EDTA buffer (pH 8.0) (G1206, Servicebio, China) was added for antigen retrieval. The slices were permeabilized through 10 min of incubation with 0.1% Triton X-100 (GC204003, Servicebio, China), blocked with 3% bovine serum albumin (BSA) (GC305010, Servicebio, China) and subsequently probed with primary and secondary antibodies, including IL-1β (Servicebio, GB11113), IL-6 (Servicebio, GB11117), TNF-α (Servicebio, GB11188), ZO-1 (Servicebio, GB115686), Occludin (Servicebio, GB111401), Claudin-3 (ABClone, A2946), Ki67 (Servicebio, GB121141-100), PCNA (Servicebio, GB12010-100), AhR (Bioss, BS-21600R), and IL-22 (Servicebio, GB11259).

For the immunofluorescence staining of BRECs, 4% paraformaldehyde was added for 30 min to fix cells at 4 °C. The permeabilization, blocking, and immunostaining procedures were the same as those used in the IF analysis of the tissue sections. The primary antibodies used included AhR (Bioss, BS-21600R), IL-22 (Servicebio, GB11259), STAT3 (Servicebio, GB150001), ZO-1 (Proteintech, 21773–1-AP), Occludin (Proteintech, 27260–1-AP), Claudin-3 (ABClone, A2946), Ki67 (Servicebio, GB121141-100), and Reg3g (Affinity, DF6869). A fluorescence microscope (Leica, Germany) was used for image acquisition. The positive cell ratio and fluorescence density were analyzed using the ImageJ software (NIH, Bethesda, MD, USA).

Measurement of proinflammatory cytokines

Interleukin (IL)–6, IL-1β, and tumor necrosis factor (TNF)-α concentrations in rumen epithelial tissues and cell supernatants were determined by enzyme-linked immunosorbent assay (ELISA) kits (USCN, China) following the corresponding protocols. Sample absorbance values were measured at 450 nm using a spectrophotometer (Epoch, BioTeK, USA).

Microbiota sequencing and analysis

Using HiPure Stool DNA Extraction kits (Magen, Guangzhou, China), total genomic DNA in the rumen fluid was isolated. Next, the V3–V4 region in the 16S rDNA gene was amplified using the primers 341 F (5′-CCTAYGGGRBGCASCAG-3′) and 806 R (5′-GGACTACNNGGGTATCTAAT-3′). After constructing the libraries, the rumen microbiota underwent high-throughput sequencing with double-ended amplification using the Illumina NovaSeq platform, and subsequently, QIIME2 was applied to process the sequencing data. The DADA2 pipeline was used to process the sequences and obtain high-quality clean reads, which were then clustered into amplicon sequence variants (ASVs). Alpha and beta diversity analyses were performed using QIIME 2. To identify microbial biomarkers, Linear Discriminant Analysis Effect Size (LEfSe) analysis was initially performed to compare the microbial community composition between the CON and KET groups, followed by the selection of significantly different genera. The differentially identified genera were used as input variables to construct a random forest classification model via the randomForest package in R. Next, ROC curves were generated, and the area under the curve (AUC) values were determined using the pROC package to evaluate model performance. To validate the specificity and indicative value of the candidate biomarkers across different groups, indicator species analysis (ISA) was performed using the labdsv package, and the IndVal index, along with the corresponding P-values, was computed. Finally, Spearman’s rank correlation coefficients between key differential genera were calculated using the psych package to examine their collinearity relationships.

Metabolomic profiling and analysis

The extraction of rumen fluid metabolites and LC-MS (AB Triple TOF 6600, AB Sciex Pte. Ltd., USA) analysis were performed as described in another study⁵. The SIMCA-14.1 software (MKS Data Analytics Solutions, Umea, Sweden) was first used to process and convert the original data, such as sample name, peak number, and normalized peak area, followed by OPLS-DA. Variable importance in the projection (VIP) was used to evaluate metabolites in the first principal component of each OPLS-DA model. Univariate regression (t-test) was also performed to assess the fold change (FC) and statistical significance. Metabolites satisfying VIP > 1, P < 0.05, and FC ≥ 2 or FC ≤ 0.5 were considered to be differentially expressed and subjected to GO and KEGG pathway enrichment. To further validate the enrichment trends, overrepresentation analysis (ORA) was performed using the R package MSEAp. Metabolic pathways and their core metabolites were identified through the integration of KEGG-based and MSEA-based analytical frameworks. To investigate the associations between key metabolites and the rumen microbiota, RDA was conducted for ordination and visualization with the vegan package in R. Additionally, variation partitioning analysis (VPA) was performed to quantify the extent to which key metabolites explained the observed variation in the microbial community structure. The pROC package was used to generate ROC curves and compute the AUC to assess the diagnostic potential of key metabolites as candidate biomarkers. Linear regression analysis was conducted with the lm function from the R stats package to evaluate the relationship between the abundance of putative IAA-producing taxa and the IAA concentration, thereby elucidating the contribution of microorganisms to metabolite dynamics.

Molecular docking

As described in previous studies⁵⁷, the computational docking analysis of protein-ligand interactions was performed using AutoDock Vina (version 4.2.6), with subsequent molecular visualization and structural interpretation performed using PyMOL Molecular Graphics System 2.5.5.

Western blotting analysis

Proteins were extracted from rumen epithelial tissues and BRECs using RIPA buffer. The protein concentration was measured using a BCA kit (Bicinchoninic Acid Protein Assay Kit, KTD3001, Abbkine, China) and adjusted to a uniform level. SDS-PAGE was conducted to separate the proteins, after which they were transferred to a polyvinylidene difluoride (PVDF) (IPVH00010, MerckMillipore, USA) membrane. The membranes were blocked with 5% skim milk for 2 h before incubation with primary antibodies. Primary antibodies against IL-1β (ABClone, A22257), IL-6 (Proteintech, 26404-1-AP), ZO-1, (Proteintech, 21773-1-AP), TNF-α (ABClone, A11534), Occludin (Proteintech, 27260-1-AP), Claudin-3 (ABClone, A2946), AhR (ABClone, A1451), ARNT (ABClone, A19532), CYP1A1 (Bioss, bs-1606R), CYP1B1 (ABClone, A1377), IL-22 (ABClone, A23665), STAT3 (Proteintech, 10253-2-AP), p-STAT3 (ABClone, AP0474), and Reg3g (Affinity, DF6869) were used. After washing, the membrane was incubated with secondary antibodies and developed with ECL reagents, followed by visualization with a chemiluminescence system (Tannon, China).

RNA isolation and measurement

Total cellular and tissue RNA was extracted using TRIzol reagent and then reverse-transcribed into cDNA. Quantitative PCR was performed using a SYBR® Premix Ex Taq™ kit (Takara, Japan). Target gene expression was determined using the 2^−ΔΔCT approach, with GAPDH being the endogenous reference. The specific primers used are shown in Supplementary Table S6.

Cell viability assay

Immortalized BRECs were obtained from Dr. Zhan and his research group from Yangzhou University. The isolation, cultivation, and identification of BRECs were handled according to the protocols optimized by Dr. Zhan’s group⁵⁸, Yangzhou University, Yangzhou, China. Briefly, primary BRECs were isolated from rumen tissues of 6- to 7-month-old Holstein calves. Tissues were excised, rinsed with PBS containing antibiotics, and digested with 0.25% trypsin-0.02% EDTA. Cells were cultured in DMEM supplemented with fetal bovine serum, antibiotics, non-essential amino acids, glutamine, insulin-transferrin-selenium, and epidermal growth factor. BRECs were immortalized using SV40 large T antigen and expanded as clonal lines. Identity of immortalized BRECs was confirmed by Western blot for SV40T and RT-PCR detection of SCFA transporter genes⁵⁸. BRECs were inoculated in 96-well plates and subsequently treated with various concentrations of IAA (0, 25, 50, 100, 150, 250, 500, 750, and 1000 µM) for 2, 6, and 12 h at 37 °C. Next, a CCK-8 assay was conducted to measure cell viability following specific instructions. Optical density (OD) was measured using an automatic microplate reader (Molecular Devices, China) at 450 nm.

Cell culture and treatment

BRECs were maintained in DMEM/F12 supplemented with 10% FBS (Gibco, USA) and 1% (v/v) penicillin/streptomycin at 37 °C with 5% CO₂. Later, the cells (5 × 10⁶/well) were inoculated in a 12-well plate, followed by 6 h of treatment with IAA (500 mM, I3750, Sigma), lipopolysaccharide (LPS, 10 ng/mL, L2880, Sigma-Aldrich), or IAA + LPS with or without an AHR inhibitor (10 μM, CH223191, HY-12684, MedChemExpress). To determine whether IAA can mitigate LPS-mediated inflammatory injury to ruminal epithelial cells in a dose-dependent manner, the cells were classified into several groups, including CON (control; non-IAA, non-LPS treatment), LPS (non-IAA, 10 µg/mL LPS treatment), IAA (500 µM IAA, non-LPS induction), IAA + LPS (100 µM IAA, 10 µg/mL LPS induction), IAA + LPS (250 µM IAA, 10 µg/mL LPS induction), and IAA + LPS (500 µM IAA, 10 µg/mL LPS induction). To assess whether IAA exerts its effects through AhR, the AhR inhibitor CH223191 was used. The experimental groups were as follows: (1) the LPS group, in which the cells were exposed to LPS (10 µg/mL) for 6 h; (2) the LPS + IAA group, in which the cells were treated with LPS and IAA for 6 h; (3) the LPS + IAA + CH-223191 group, in which the cells were treated with 10 μM CH-223191 for 2 h before treatment with LPS and IAA; (4) the LPS + CH-223191 group, in which the cells were treated with 10 μM CH-223191 for 2 h before LPS stimulation.

Statistical analysis

All statistical analyses were conducted using the R software (v4.3.1, Posit Software, PBC, Boston, MA, USA) and GraphPad Prism (v10.0, Dotmatics, Boston, MA, USA). The data are presented as the mean ± SEM. One-way ANOVA followed by Student’s t-test (two groups) or Fisher’s LSD post hoc test (across diverse groups) was performed to determine the differences between the groups. All results were considered to be statistically significant at P < 0.05.

Author contributions

C.X. conceived and designed the study. L.M.L. and S.Q.Z. contributed to writing the original draft, methodology development, data visualization, formal analysis, investigation, and data curation. contributed to writing the original draft, data visualization, formal analysis, investigation, and data curation. Y.H.H. and Q.Q.C. were responsible for review and editing of the manuscript, data visualization, formal analysis, methodology development, and data curation. Z.J.D., K.L., Y.L., J.J.L., J.W., and J.Q. contributed to data visualization, methodology development, formal analysis, investigation, and data curation. The authors read and approved the final manuscript.

Data availability

The rumen 16S rDNA sequencing data analyzed in this study have been archived in the OMIX database at the China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences, with the accession number OMIX010888 (https://ngdc.cncb.ac.cn/omix/select-edit/OMIX010888).

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41522-025-00898-1.