Duodenal resident Limosilactobacillus improves feed efficiency through ornithine-mediated optimization of gut microbiota and nutrient absorption via the Nrf2 signaling

Abstract

Background

The small intestinal microbiota directly influences host intestinal digestive and absorptive responses to dietary nutrients and plays a crucial role in optimizing feed efficiency in food-producing animals. However, the microbial functions of small intestine in regulating feed efficiency in broiler chickens remain to be elucidated.

Methods

A total of 150 healthy broilers were individually housed under identical feeding conditions to accurately calculate their feed efficiency. The gut microbiota in different intestinal segments of high and low feed efficiency chickens were compared using 16S rRNA sequencing. Gut bacterial candidates associated with feed efficiency were identified through a two-part model, LEfSe, and the Wilcoxon rank-sum test. Another 1725 1-day-old male broiler chicks were fed either a basal diet (BD) or BD supplemented with four different bacterial candidates isolated from the chicken gut to investigate their roles in regulating gut microbiota and nutrient absorption. The underlying molecular mechanisms by which key Limosilactobacillus strains and their metabolite ornithine improve intestinal health were also examined using an intestinal epithelial cell line.

Results

This study found that chickens with high feed efficiency exhibited greater microbial community stability and stronger cooperative interactions compared to low feed efficiency chickens, particularly within the duodenal microbiota. Meanwhile, duodenal resident Limosilactobacillus were significantly positively correlated with feed efficiency. Further validation trials revealed that specific Limosilactobacillus strains (L. vaginalis LD11 and L. ingluviei CC32) significantly improved feed efficiency, concurrently enhancing antioxidant capacity, barrier function, nutrient absorption, as well as increasing Limosilactobacillus abundance in the duodenum. These two bacterial strains could produce high concentrations of ornithine in the duodenum, which effectively alleviated LPS-induced intestinal cell damage by enhancing antioxidant capacity and upregulating the protein expression of nutrient transporters. Mechanistically, both bacterial strains and ornithine enhanced antioxidant capacity and nutrient uptake by activating Nrf2 signaling.

Conclusions

Dietary intervention using L. vaginalis LD11 and L. ingluviei CC32 contributes to high feed efficiency by producing ornithine, which modulates the duodenal microbiota and enhances the intestinal physiological functions for nutrient absorption

Download video file ^{(43.6MB, mp4)}

Video Abstract

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s40168-026-02461-2.

Background

The gastrointestinal tract harbors a complex microbial ecosystem integral to host physiology, metabolism, and nutrient acquisition [1]. Optimizing the functional output of gut microbiome is pivotal for enhancing resource efficiency and sustainability in animal agriculture [2, 3]. Previous studies have demonstrated that gut microbiome profoundly influences animal feed efficiency, with specific taxonomic, functional, and metabolic features closely correlated with feed efficiency in food-producing animals, such as cows, swine, and poultry [4–8]. The poultry gut microbiome, an important factor affecting bird health, plays a pivotal role in processing nutrients and harvesting energy from ingested feed [8–10]. The cecal microbiota have been the most studied in poultry feed efficiency due to their high microbial richness and diversity, which enhance microbial fermentation of undigested carbohydrates to produce bioactive metabolites [11–14]. In contrast, the proximal small intestinal microbiota community has relatively lower diversity and abundance than the cecum, but it is more susceptible to dietary changes and directly influences host intestinal functions involved in regulating digestive and absorptive responses to dietary nutrients [15]. However, the microbial ecology of the proximal small intestine, particularly the duodenum, in poultry remains underexplored, despite its role as a critical and distinct niche where host digestion is initiated and most nutrient absorption occurs.

The chicken duodenum is characterized by rapid transit and high enzymatic activity, creating a distinct ecological niche [16]. These conditions impose strong selective pressures, shaping a microbial community that, although lower in diversity, is metabolically specialized and positioned at the host-diet interface [17]. Unlike the distal fermentation function of the cecum, the duodenal microbiota participates in the earlier, rate-limiting steps of nutrient absorption [18]. Potential mechanisms include the modulation of luminal digestibility, interaction with host digestive secretions, and direct influence on intestinal epithelial function and nutrient transporter activity [19, 20]. Consequently, variations in the duodenal microbial community structure could have direct and profound effects on feed efficiency. However, the specific taxonomic and functional features of the duodenal microbiota associated with high feed efficiency in broiler chickens, as well as the mechanisms by which it contributes to nutrient utilization, are not well understood.

Herein, this study hypothesizes that the duodenal microbiota harbors efficiency-associated taxa that directly influence host digestive physiology. By selecting broiler flocks with high and low feed efficiency under natural feeding conditions, we systematically compared the differences in duodenal microbiota profiles between chickens with varying feed efficiencies. Subsequently, we identified and isolated key candidate bacteria associated with feed efficiency from the chicken gut to investigate their roles in improving intestinal health and nutrient absorption in broilers. This study aims to promote the understanding of the functions of duodenal microbiota in feed efficiency and to provide novel insights to improve productivity and sustainability in animal agriculture.

Materials and methods

Broiler chicken feeding trial and sample collection

The animal experiment (No. HZAUCH-2024–0022) was approved by the Institutional Animal Care and Use Committee of the Huazhong Agricultural University. A total of 200 1-day-old Cobb chicks were housed together during the brooding stage (days 1–14) to satisfy their social needs, avoid solitary stress, and ensure microbiota exchange through typical bird behaviors, including coprophagy [21]. After removing chicks with the highest and lowest body weights, a total of 150 chicks with similar body weights (425.50 ± 20.23 g) were randomly transferred into 150 individual cages and reared separately starting from day 14 [12]. All cages were designed the same size (35 × 35 × 45 cm), placed at a uniform height above the ground, and arranged in five rows with 30 cages per row and a row spacing of 0.6 m to facilitate adequate air circulation. During growth stage (days 15–42), feed intake and body weight of individual chickens were recorded weekly to assess their growth performance, including average daily gain (ADG), average daily feed intake (ADFI), and feed-to-gain ratio (F/G). Birds were housed in an environmentally-controlled room, the temperature started at 35 °C and decreased by 2–3 °C every 7 days until it reached 22 °C; the relative humidity was targeted to approximately 65–70% during the first week and gradually declined to 50–55% in the later growing phase through adjustments in minimum ventilation; and the lighting program was 0 h of dark on day 0, 1 h of dark for days 1 to 3, and 6 h of single continuous dark for days 4 to 42 to ensure adequate rest. All chickens were fed a basal diet (BD, Table S1). Individual housing and measurements were implemented to eliminate competition for feed and reduce behavioral issues affecting feed intake [21], allowing for a more precise correlation between the gut microbiota and feed efficiency. At the end of feeding trial (day 42), broilers with high feed efficiency (HFE, n = 15) and low feed efficiency (LFE, n = 15) were selected based on the F/G rankings of each broiler. After 12 h of feed deprivation, chickens were slaughtered for sampling. The digesta samples from the duodenum, jejunum, ileum, and cecum were immediately collected and stored at − 80 °C for subsequent gut microbial analysis and sections of the small intestine were perfused and fixed in 4% paraformaldehyde for subsequent histological analysis [22].

16S rRNA gene sequencing and microbiota data analysis

Digesta contents, including chyme and mucosa, were collected from the duodenum, jejunum, ileum, and cecum at the end of the experiment. Total bacterial DNA was then extracted using the DNA stool kit (Tiangen, Beijing, China) according to the manufacturer’s instructions. The V3–V4 hypervariable regions of the 16S rRNA gene were amplified. The sequencing library was established with NEBNext® Ultra™ II DNA Library Prep Kit for Illumina® (New England Biolabs, MA, USA), and all amplicons were sequenced using an Illumina Nova6000 platform (2 × 250 bp paired-end).

Raw sequences were quality filtered, denoised, and de-chimerized to avoid sequencing inaccuracy using a modified Parallel-Meta Suite (PMS) pipeline [23]. Taxonomy was assigned to amplicon sequence variants (ASVs) using the SILVA 138.2 database at 99% sequence identity. Only ASVs with average relative abundance greater than 0.2% and detected in at least 50% of the chicken samples were included in the downstream analysis. Microbial diversity and composition were conducted using the OmicStudio tools [24]. Analysis of similarities (ANOSIM) was conducted in QIIME2 [25]. Spearman’s rho correlation test was used for correlation analysis, and correlations with coefficients < − 0.4 or > 0.4 and adjusted P-values < 0.05 were considered significant [26]. Microbial co-occurrence networks were constructed using the cnsknowall tools based on pearson correlations between genera, retaining robust associations with |R|> 0.6 and P < 0.001. PICRUSt2 analysis was used to predict functions of microbial communities, and the differentially enriched metabolic pathways based on PICRUSt were analyzed using STAMP (v. 2.1.3) [27]. Significant differences between groups were determined using Wilcoxon rank-sum test, and false discovery rate (FDR) adjusted P values were corrected using the Benjamini–Hochberg method [28].

Identification of specific microbial taxa associated with chicken feed efficiency

The distribution of the taxon abundance did not follow a normal distribution, as some microbial taxa were undetected in many samples. Thus, a two-part model was used to analyze the associations between bacterial features and F/G values as previously described [2, 29]. Briefly, this model is described as: y = β1b + e and y = β2q + e for binary and quantitative models, respectively. Where y is the F/G value, b is a binary feature which coded as 1 for detected or 0 for undetected, q is the abundance of a microbial taxa, e is the residuals, and β1 and β2 are regression coefficients. The adjusted P value < 0.05 from the binary model indicated that the presence or absence of features could affect feed efficiency, while the adjusted P value < 0.05 from the quantitative model represented the relative abundance of features associated with feed efficiency. Meanwhile, the Wilcoxon rank-sum test and LEfSe analysis were performed to determine significantly differential features between high and low F/G chickens. If the adjusted P values from the two-part model analysis, LEfSe analysis, and Wilcoxon rank-sum test were all less than 0.05, this indicated that the bacterial features could significantly influence chicken feed efficiency.

Biochemical parameters analysis

The activity of digestive enzymes (Amylase, Lipase, and Trypsin), total antioxidant capacity (T-AOC), superoxide dismutase (SOD) activity, and concentrations of malondialdehyde (MDA), reduced glutathione (GSH), ammonia, and blood urea nitrogen (BUN) were determined using their specific assay kits. All kits were purchased from the Nanjing Jiancheng Bioengineering Institute (Nanjing, China). The serum diamine oxidase (DAO) and lipopolysaccharides (LPS) levels were measured using Enzyme-Linked immunosorbent assay (ELISA) kits (MM-60017O2 and MM-33277O2, MEIMIAN) according to the manufacturer’s instructions.

Histological analysis

The intestinal samples of chickens were fixed, embedded in paraffin, then sectioned into 6 μm slides and stained with hematoxylin–eosin. After morphological evaluation, 3–5 microscopic fields per section were randomly selected for measuring the villus height (VH) and crypt depth (CD) of small intestine using Image J software [30]. Additionally, the formalin-fixed duodenal sections were subjected to TUNEL staining using assay kit (C1088; Beyotime Biotechnology) according to the manufacturer’s instructions. Before imaging, sections were counterstained with DAPI to stain nuclei and then examined under a Zeiss inverted fluorescence microscope.

Bacterial strains isolation and culture

The bacterial strains used in this study, including L. reuteri (LR), L. vaginalis (LV), L. johnsonii (LJ), or L. ingluviei (LI), were isolated from chicken gut as previously described [31]. Briefly, cecal digesta suspension was serially diluted with PBS and seeded onto de Man-Rogosa-Sharpe (MRS) plates at 37 °C under anaerobic condition. Following incubation, individual colonies were isolated and identified by PCR amplification using the primers (27F 5′-AGRGTTTGATYMTGGCTCAG-3′; 1492R 5′-GGYTACCTTGTTACGACTT-3′). The sequences were then blasted using the NCBI genome database. Individual isolates were assigned species names based on > 99.0% sequence homology and stored in MRS broth with 30% glycerol at – 80 °C.

Bacterial genome sequencing analysis

Each sample of 0.5 µg genome DNA was prepared following Annoroad®

Universal DNA Fragmentase kit V2.0 and Annoroad® Universal DNA Library Prep Kit V2.0 protocols. The DNA libraries were sequenced using a 150 PE run from Illumina NovaSeq 6000 platform. Reads were quality controlled and decontaminated from sequencing artefacts and adapters and merged using fastp. Assembling was performed using SPAdes (v3.15.0) and further gene structure and function annotation was done with Prokka (v1.14.5) [30].

Resistance to simulated gastric and intestinal conditions

The tolerance of Lactobacillus isolated from chicken gut in the simulated gastric and intestinal conditions was assessed in vitro according to previous method [32]. In short, simulated gastric juice (SGJ) was made by dissolving 3.0 g/L pepsin in sterile saline solution, and the pH was adjusted to 3.0. Simulated intestinal juice (SIJ) was formulated by the addition of 0.3 g/L bile salts and 1.0 g/L trypsin in sterile saline solution, and the pH was adjusted to 3.0. Lactobacillus isolates were continuously activated for 18 h in MRS broth at 37 °C. A total of 1 mL bacterial cell suspension was harvested by centrifugation at 3000 × g for 10 min at 4 °C and dissolved in MRS broth (500 μL) and SGJ (500 μL) miscible liquids, and incubated at 37 °C for 3 h and then the bacteria were resuspended in SIJ and incubated in the same condition for 4 h. The cell suspensions were then cultivated on the MRS agar for calculating the number of viable cells.

Administration of Lactobacillus in broiler chicken

A total of 1725 1-day-old male broiler chicks with similar body weight were used to investigate the causal relationship between the gut microbe candidates and feed efficiency. Chicks were randomly divided into 5 treatments with 15 replicates of 23 broilers each. Broilers were fed either a basal diet (BD) or a basal diet supplemented with 1 × 10⁹ colony-forming units (CFU)/kg LR, LV, LJ, or LI for the first 21 days, respectively [33]. These Lactobacillus strains with the strongest tolerance to SGJ and SIJ were isolated from chicken gut in the first trial of this study. Subsequently, all broilers were fed a BD diet from days 21 to 42. Feed and water were given ad libitum for a full commercial broiler production cycle. Feed intake and body weight of broilers were recorded at day 0, 21, and 42 for the calculation of ADG, ADFI, and F/G. On the last day of the trial, 15 broilers from each group were randomly selected and slaughtered for sampling after 12 h of feed deprivation.

Untargeted quantification for Lactobacillus metabolites

The metabolites of Lactobacillus were generated as previously described [34]. Briefly, Lactobacillus strains were inoculated into anaerobic MRS medium at 37 °C, and monitored the growth phase of the strains by determining the OD values. The bacterial cultures were immediately collected when the OD values indicated late log phase. The bacteria-free supernatants were extracted, and the metabolic extracts of Lactobacillus were analyzed by high-resolution liquid chromatographic mass spectrometry (LC–MS) as previously described [35]. Dynamic exclusion was implemented to remove irrelevant data in the MS/MS spectra, and the metabolic features and MS/MS spectra were matched according to their accurate masses (± 10 ppm). A heatmap of different metabolites of Lactobacillus was generated by TBtools-II (v 2.040).

Determination of ornithine content in duodenum

The ornithine content in the supernatant of duodenum was detected by ninhydrin chromogenic spectrophotometry. The duodenal samples were added with normal saline at a mass-to-volume ratio of 1:9, and then homogenized by a grinder. The supernatant was taken to determine the ornithine content. Accurately absorb 2.0 mL of ornithine monohydrochloride standard solution in a 10 mL plug colorimetric tube, add 6 mol/L H₃PO₄-glacial acetic acid (1/3, V/V) to prepare a 25 mg/mL ninhydrin mixed acid solution as a chromogenic solution, react in a water bath at 100 °C for 60 min, and determine ornithine concentration at a wavelength of 510 nm. The blank control group was replaced with distilled water and mixed acid solution for the same analysis and comparison.

High-performance liquid chromatography

The polyamine content in duodenal tissues and IPEC-1 cells was determined as previously described [36]. Briefly, samples were homogenized in 1.0 mL 5% HClO₄, and the supernatant was collected after centrifugation. Then, 2 mL of 2.5 mol/L NaOH and 4.0 μL of benzoyl chloride were added to the supernatant. After vortex oscillation, the supernatant was derivatized in a water bath at 40 °C for 30 min. After passing through a C18 solid-phase extraction column, the samples were eluted with 0.5 mL of chromatographic methanol, and the eluent was filtered using a 0.22-μm needle filter for detection. The detection conditions were as follows: mobile phase composition: methanol:water = 66:34 (V:V), the flow rate was 0.9 mL/min, the detection wavelength of the ultraviolet detector was 229 nm, and the column temperature was 25 °C.

Cell culture and treatments

Porcine intestinal epithelial cells (IPEC-1) were cultured in DMEM/F12 containing 10% FBS, 0.1% ITS, 1% penicillin–streptomycin, and 5 ng/mL epidermal growth factor in 5% CO2 at 37 ℃ [37]. To investigate the effect of ORN and SPD on intestinal health and nutrient absorption, IPEC-1 cells were pretreated with 50 μM ORN or 10 μM SPD in serum-free medium for 12 h, and then cultured with 5 μg/mL LPS for 12 h. To evaluate the effect of the Nrf2 pathway on ORN and SPD-mediated optimization of intestinal health and nutrient absorption, IPEC-1 cells were pretreated with 100 nM Brusatol (Nrf2 inhibitor; MedChemExpress, HY-19543) along with 50 μM ORN or 10 μM SPD in the serum-free medium for 12 h and then cultured in 5 μg/mL LPS for 12 h.

Intracellular reactive oxygen species analysis

Intracellular reactive oxygen species (ROS) production was monitored using ROS assay kit (S0033S, Beyotime, Shanghai). Briefly, IPEC-1 cells were seeded in 12-well plates, stained with 10 µM 2′,7′-dichlorohydro-fluorescein diacetate (DCFH-DA) for 20 min at 37 °C, and then washed with serum-free medium. The fluorescence signals of ROS were examined under a Zeiss inverted fluorescence microscope. After imaging, cells were scraped and resuspended in PBS to determine the intracellular ROS production with excitation at 488 nm and emission at 525 nm. The fluorescence signals were normalized to the control group and used to express intracellular ROS contents.

Molecular docking analysis

The 3D structures of chicken Nrf2 were obtained from the UniProt database (https://www.uniprot.org/). The 2D molecular structure of the ornithine ligand was obtained by PubChem database (http://pubchem.ncbi.nm.nih.gov/), and the 2D structure was input into Chem Office software to make its 3D structure. Docking was performed with AutoDock Vina software (TheScripps Research Institute, USA). The conformation with the lowest binding energy was selected and visualized by Discovery Studio 2019 and Pymol2.6 software, and the 3D diagrams of the interaction between ornithine and key residues were drawn.

Quantitative PCR

Total RNA was extracted from the duodenum and IPEC1 cells for gene expression analysis as described previously[38]. Briefly, the mRNA levels of genes were determined by qPCR (Bio-Rad, USA) using SYBR qPCR Mix (ABclonal, China) and specific primers (Table S2). The genes involved in glucose, fatty acid, and amino acid transporters, as well as polyamine metabolism were determined. Housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used to normalize the relative expression.

Western blot analysis

Total proteins from tissues and cell samples were extracted for western blot analysis as previously described methods [39]. The primary antibodies used were as follows: GLUT2 antibody (Proteintech, 20,436–1-AP), FABP1 antibody (Proteintech, 12,326–1-AP), FATP4 antibody (Proteintech, 11,013–1-AP), CAT1 antibody (Proteintech, 14,195–1-AP), HO-1 antibody (Proteintech, 10,701–1-AP), and Nrf2 antibody (Servicebio, GB113808). All whole-cell lysate protein expression levels were normalized to the housekeeping protein β-actin (ABclonal, AC026), and nuclear protein expression levels were normalized to the protein Lamin B1 (ABclonal, A1910). Densitometric quantification of the western blot bands was performed using ImageJ software.

Immunofluorescence analysis

IPEC-1 cells were seeded in 24-well plates and then exposed to different treatments. After treatment, cell plates were incubated in 4% paraformaldehyde for 20 min, permeated with 0.5% Triton X-100 for 10 min, and then blocked with 5% bovine serum albumin for 30 min. Cell plates were incubated with primary Nrf2 antibody (Servicebio, GB113808) at 4 °C overnight, then incubated with the Cy3 goat anti-rabbit secondary antibody (ABclonal, AS007) for 2 h. Before imaging, sections were counterstained with DAPI to stain nuclei and examined under a Zeiss inverted fluorescence microscope. The mean fluorescence intensity of the Nrf2 protein was measured by ImageJ software.

Statistical analysis

In addition to the above-mentioned specific statistical methods of sequencing data, all data analyses were conducted by GraphPad Prism 8.0. Two-tailed unpaired Student’s t-test or one-way analysis of variance (ANOVA) with Tukey’s multiple comparison tests were used to determine statistical significance. Detailed statistical methods are described in each figure legend. Data are presented as mean ± standard error (SE) with a significance level of P < 0.05.

Results

Selection of broiler flocks with high and low feed efficiency based on feed-to-gain ratio performance

To select high and low feed efficiency chicken flocks, we conducted a feeding trial for selecting the broilers with high feed efficiency (HFE) and low feed efficiency (LFE) from 150 chicken flocks (Fig. 1a). No significant difference in body weight among groups when they were transferred into individual cages at day 14 (Additional file 1: Fig. S1a). The HFE broilers had significantly higher body weight at day 42, as well as ADFI and ADG during days 15 to 42 compared to the LFE broilers (Additional file 1: Fig. S1). As shown in Fig. 1b and Table S3, the F/G ratios of all 150 broilers exhibit a significant normal distribution (1.875 ± 0.131, P < 0.001). Excluding outliers, broilers with the highest 10% F/G were defined as the LFE group, while those with the lowest 10% F/G were assigned to the HFE group (Fig. 1b, c). Based on this, the HFE broilers had significantly lower F/G during days 15 to 42 compared to the LFE and overall broilers (Fig. 1d).

Fig. 1.

Differences in profiling the small intestinal microbiota of chickens with varying feed efficiency. a Schematic diagram of the experimental design. b Scatter distribution graph of feed-to-gain ratio (F/G) of 150 chickens during days 15–42. Each point represents an independent chicken. Excluding outliers, broilers with the highest 10% F/G were defined as the low feed efficiency (LFE) group, while those with the lowest 10% F/G were assigned to the high feed efficiency (HFE) group. c Histogram frequency distribution plot of F/G of 150 chickens during days 15–42. d The F/G ratio in each group. Overall indicates the whole chicken flock (n = 150). e The numbers of observed ASVs. f Changes in Shannon index. g Changes in β-diversity in microbial composition based on PCoA analysis. h Pairwise comparison of Bray–Curtis dissimilarity between HFE and LFE groups across different gut segments. i Changes in the microbial network of duodenum. j Modularity of microbial networks. It refers to the number and structure of modules composed of nodes in the network. Highly modular networks have obvious substructures, while lowly modular networks do not have this structure. k Robustness analysis of microbial networks. Data are presented as mean ± SE, and differences between HFE and LFE groups were conducted by unpaired Student’s t-test (d, k). n = 15 per group. Labeled means with different superscript letters are significantly different (P < 0.05) by Kruskal–Wallis test (h). ***, P < 0.001

Differences in profiling the microbiota in different intestinal segments of chickens with varying feed efficiency

To investigate the of the small intestinal microbial community in boilers, we examined luminal content samples from the duodenum, jejunum, and ileum using the 16S rRNA gene sequencing. The alpha diversity exhibited an increasing trend along the small intestine from front to back (Additional file 1: Fig. S2a), and the beta diversity based on Bray–Curtis dissimilarity also suggested obvious differences in bacterial communities among duodenum, jejunum, and ileum (Additional file 1: Fig. S2b). Meanwhile, a total of 884 (HFE vs. LFE, 772 vs 520), 1001 (HFE vs. LFE, 842 vs. 712), and 1217 (HFE vs. LFE, 1102 vs. 787) ASVs were identified in duodenum, jejunum, and ileum, respectively (Fig. 1e), which showed that HFE broilers had more bacterial taxa in each gut segment compared to LFE broilers. The Shannon index between HFE and LFE broilers showed no significant differences in this three gut segments (Fig. 1f). Notably, significant separation between HFE and LFE groups in duodenal microbiota profiles, not jejunal or ileal microbiota, was observed based on the beta diversity analysis (Fig. 1g, and Additional file 1: Fig. S2c). The pairwise Bray–Curtis dissimilarity between samples from HFE and LFE broilers further indicated that the most substantial changes in bacterial communities occurred in the duodenum and ileum (Fig. 1h). Taxonomic composition analysis showed that Lactobacillus was more abundant in the duodenum than in the jejunum and ileum (Additional file 1: Fig. S2e). To examine differences in microbial interactions within the gut of chickens exhibiting varying feed efficiencies, we constructed co-occurrence networks at the genus level across different gut segments. More nodes (117 vs. 83 in the duodenum, 120 vs. 94 in the jejunum, and 112 vs. 95 in the ileum) and more edges (2,463 vs. 1334 in the duodenum, 2,322 vs. 1451 in the jejunum, and 3,117 vs. 949 in the ileum) were observed in HFE broilers than in LFE broilers (Fig. 1i and Additional file 1: Fig. S3). Additionally, compared to LFE broilers, HFE broilers exhibited the highest modularity of network in the duodenum relative to the jejunum and ileum (Fig. 1j). Based on the random removal of module hubs, HFE broilers demonstrated greater robustness than LFE broilers in both the duodenum and ileum (Fig. 1k). These results suggest that HFE broilers exhibit stronger collaborative interactions within microbial communities and higher community stability in the small intestine, particularly in the duodenum.

We also investigated the differences in the microbial community of cecum. A total of 1435 (HFE vs. LFE, 1392 vs 1325) ASVs were identified, while the Shannon index and beta diversity between HFE and LFE broilers showed no significant differences (Additional file 1: Fig. S4a–e). Taxonomic composition analysis showed that Firmicutes and Bacteroidota were two dominant phyla, and Bacteroides was more abundant genus in the cecum (Additional file 1: Fig. S4d, e). Further co-occurrence network analysis showed that more nodes (130 vs. 127) and more edges (1036 vs. 698) were observed in HFE broilers than in LFE broilers (Additional file 1: Fig. S4f).

Screening and isolation for microbial candidates associated with feed efficiency in broilers

To accurately screen potential gut microbial taxa associated with feed efficiency, we performed four methods (quantitative, binary, Wilcoxon test, and LEfSe analysis) to assess differential bacterial features in each gut segment. Only bacterial features identified by at least three methods were considered differential. A total of 4, 1, and 0 bacterial features were identified in the duodenum, jejunum, and ileum, respectively (Figs. 2a and Additional file 1: Figs. S4g and S5a). In the duodenum, all four features (F14, F1568, F1547, and F1609) exhibited higher relative abundance in HFE broilers compared to LFE broilers (Fig. 2b). Ternary analysis further revealed that the duodenum had a higher abundance of these four features than the jejunum and ileum (Additional file 1: Fig. S5b). In the cecum, a total of 11 bacterial features were identified, and all these features, expect for F1585 exhibited higher relative abundance in HFE broilers compared to LFE broilers (Additional file 1: Fig. S4g, h). Furthermore, histological analysis showed that only the duodenum exhibited markedly increased VH and VH to CD ratio in HFE broilers, whereas the jejunum and ileum did not (Figs. 2c, d and Additional file 1: Figs. S5c, d). Analysis of digestive enzymes revealed significantly elevated activities of amylase and lipase in the duodenum of HFE broilers, while trypsin activity remained unchanged (Fig. 2e). These outcomes suggest that these four bacterial features in the duodenum could be key candidate bacteria contributing to feed efficiency by enhancing duodenal nutrient digestion and absorption.

Fig. 2.

Screening, identification, isolation and pure culture of key microbial candidates related to feed efficiency in duodenum. a Number of microbial features associated with F/G detected by a two-part model analysis (Binary and Quantitative), LEfSe analysis, and Wilcoxon test and their overlaps in the duodenum. b Comparison analysis of the relative abundances of gut microbial features identified by a two-part model analysis (binary and quantitative), LEfSe analysis, and Wilcoxon test. c, d Representative H&E staining images of duodenum (scale bar, 500 µm), and statistical analysis of villus height, crypt depth, and the ratios of villus height to crypt depth. e The activities of digestive enzymes, including amylase, lipase, trypsin, in duodenum. f The relative abundance of different Lactobacillus species were determined by quantitative PCR in the duodenum of broilers between HFE and LFE groups. g Spearman’s correlation analysis of four differential Lactobacillus species with growth performance and biochemistry parameters. Correlation coefficients and P-values were determined via Spearman’s rho correlation test. h Schematic diagram of the pure culture of bacterial isolation from chicken gut. i Isolation of L. reuteri, L. vaginalis; L. ingluviei, and L. Johnsonii, and their resistance to simulated gastric and intestinal conditions. Data are presented as mean ± SE, and differences between HFE and LFE groups were conducted by unpaired Student’s t-test (D, E). *, P < 0.05; **, P < 0.01. SGJ, simulated gastric juice; SIJ, simulated intestinal juice

As these four bacterial features belong to the genera Lactobacillus and Limosilactobacillus, we performed quantitative PCR analysis to determine which Lactobacillus species were significantly altered in the duodenum. The results showed that four species, including L. reuteri, L. vaginalis, L. johnsonii, and L. ingluviei, exhibited a markedly higher abundance in HFE broilers (Fig. 2f). Spearman’s correlation analysis further indicated that all four species had a strong negative correlation (R < − 0.72, P < 0.01) with F/G values and a positive correlation (R > 0.57, P < 0.05) with ADFI, ADG, and VH (Fig. 2g). To screen for highly active strains, we then isolated and cultured these bacterial species from chickens to investigate their role in regulating feed efficiency (Fig. 2h). A total of 25 strains were isolated and evaluated for tolerance to SGJ and SIJ, and the results showed that L. reuteri LC7, L. vaginalis LD11, L. johnsonii LC4, and L. ingluviei CC32 demonstrated the strongest tolerance within their respective species (Fig. 2i).

Dietary supplementation of L. vaginalis and L. ingluviei improved feed efficiency by enhancing antioxidant and nutrient uptake in duodenum

For the validation trial, broiler chicks were fed a basal diet supplemented with four candidates, including L. reuteri LC7, L. vaginalis LD11, L. johnsonii LC4, and L. ingluviei CC32, during the first three weeks to facilitate bacterial colonization in the small intestine (Fig. 3a). No significant differences were observed in ADFI from 1 to 6 weeks among the five groups (Fig. 3b). Compared to the BD-treated broilers, LR, LI, and LV-treated broilers exhibited higher body weight at day 42 and greater ADG during the 1–6 week period (Fig. 3c and Additional file 1: Fig. S6a). Notably, only LI and LV-treated broilers showed lower F/G values during each period throughout the trial compared to BD-treated broilers (Fig. 3d and Additional file 1: Fig. S6b, c), suggesting that dietary supplementation with L. vaginalis and L. ingluviei improved feed efficiency in broilers. Further histological analysis of the duodenum revealed that LI treatment increased VH, but did not affect CD or the VH to CD ratio compared to the other groups (Fig. 3e, f). Digestive enzymes break down dietary macromolecules into smaller molecules, which are then absorbed by intestinal nutrient transporters to support animal growth, thereby contributing to improved feed efficiency (Additional file 1: Fig. S6d). As expected, both LI and LV-treated broilers exhibited elevated amylase activity in the duodenum and pancreas, as well as increased lipase activity in the pancreas (Fig. 3g and Additional file 1: Fig. S6e). Meanwhile, the expression of amylase and lipase genes was upregulated by dietary LI and LV treatments (Additional file 1: Fig. S6f). The expression of SGLT1, GLUT2, FATP4, and CAT1 genes, along with GLUT2, FATP4, and CAT1 proteins, which are important nutrient transporters involved in glucose, fatty acids, and amino acids absorption, was significantly upregulated in the duodenum by dietary LI and/or LV treatment (Fig. 3h, i).

Fig. 3.

Dietary supplementation of L. vaginalis and L. ingluviei improved feed efficiency by enhancing nutrient absorption and intestinal health in broilers. a Schematic diagram of a validation trial on the effect of candidate Lactobacillus species on broiler feed efficiency. b–d Growth performance including ADFI, ADG, and F/G of broilers throughout the validation trial. e, f Representative H&E staining images of duodenum (scale bar, 500 µm), and statistical analysis of villus height, crypt depth, and the ratios of villus height to crypt depth among three groups. g The activities of digestive enzymes (amylase, lipase, and trypsin) in duodenum. h Relative gene expressions of the genes involved in glucose, fatty acid, and amino acid transporters in duodenum. i Relative protein expressions of GLUT2, FATP4, FABP1, and CAT1 proteins in duodenum. j Duodenal antioxidant capacity, including T-AOC level, SOD activity, and concentrations of GSH and MDA. k Representative fluorescent images of TUNEL staining of duodenal sections. Scale bar, 200 μm. Statistical comparison performed by one-way ANOVA. Data are presented as mean ± SE. Graph bars with different letters indicate significant difference (P < 0.05)

Intestinal health serves as the foundation for optimal nutrient digestion and absorption, as well as overall host health, contributing to better feed efficiency, we therefore assessed whether LI and LV are beneficial for improving duodenal functions. Analysis of oxidative stress-related parameters in the duodenum revealed that the levels of T-AOC, SOD, and GSH increased significantly, whereas MDA concentrations decreased in both LI and LV-treated broilers (Fig. 3j). Further histological analysis of the duodenum revealed that both LI and LV-treated broilers had more mucus-producing goblet cells compared to BD broilers (Additional file 1: Fig. S6g), while had fewer TUNEL positive signals (Fig. 3k). Moreover, serum LPS levels were significantly decreased in both LI and LV-treated broilers, whereas no differences were detected in DAO levels across the three groups (Additional file 1: Fig. S6h, i). These results indicated that dietary supplementation of L. vaginalis and L. ingluviei improved duodenal antioxidant and barrier functions.

L. vaginalis and L. ingluviei reshaped microbial community and functions in the duodenum

The effects of L. vaginalis and L. ingluviei on the microbial composition and functions in the duodenum were further investigated. LV treatment led to a significant reduction in the observed OTUs and shannon index of the duodenal microbiota (Fig. 4a). Meanwhile, beta diversity based on Bray–Curtis dissimilarity also indicated clear differences in bacterial communities among the three groups (Fig. 4b). Taxonomic composition analysis showed that the abundance of the genera Lactobacillus, Rothia, Limosilactobacillus, and Streptococcus increased, while the abundance of Brevibacterium, Paracoccus, Cellulomonas, and Nocardioides decreased in LI or LV-treated broilers compared to BD broilers (Fig. 4c). Further co-occurrence networks constructed at the genus level revealed more edges in LV and LI-treated broilers than in BD broilers (BD vs. LI vs. LV: 2911 vs. 3887 vs. 7347) (Fig. 4d). Additionally, both LV and LI-treated broilers exhibited higher average degree and network robustness than BD broilers (Fig. 4e, f).

Fig. 4.

Dietary supplementation of L. vaginalis and L. ingluviei reshapes duodenal microbiota composition. a Changes in shannon index and chao1 index. b Changes in β-diversity in microbial composition based on PCoA analysis. c Changes in microbial compositions at genus levels. Arrow labeling indicates a significant increase or decrease in LI and/or LV broilers compared to the BD broilers. d Changes in the microbial network. e Average degree of microbial networks. f Robustness analysis of microbial networks. g Differential genera in the duodenum were determined by LEfSe analysis. h Relative abundance of Limosilactobacillus genus among three groups. i Correlation analysis between the abundance of Limosilactobacillus genera and the F/G values of broilers. j Ternary analysis of differential genera. k Functional pathways analysis of four crucial genera predicted using PICRUSt2. l Changes in L-ornithine biosynthesis pathways. Graph bars with different superscript letters are significantly different (P < 0.05) by Kruskal–Wallis test

LEfSe analysis identified 24, 8, and 4 differential genera in BD, LI, and LV broilers, respectively (Fig. 4g). Among these genera, the abundance of Lactobacillus, Rothia, Limosilactobacillus, and Streptococcus was elevated in response to LI and/or LV treatment (Fig. 4h and Additional file 1: Fig. S7a). Linear regression analysis indicated that Limosilactobacillus had a strong negative correlation with F/G values (R = − 0.512, P = 0.013), whereas the other three genera showed no significant correlation with F/G values (Fig. 4i and Additional file 1: Fig. S7b). Ternary analysis further confirmed that LI and LV treatment had higher abundances of Lactobacillus, Rothia, Limosilactobacillus, and Streptococcus (Fig. 4j), suggesting that these four genera may play crucial roles in the duodenum contributing to feed efficiency. To infer their functional roles, PICRUSt2 prediction analysis was applied. The results showed that 24 microbial pathways involved in various amino acid biosynthesis processes were significantly enriched in LV and LI-treated broilers, including L-ornithine biosynthesis (Fig. 4k, l and Additional file 1: Fig. S8a, b). Similarly, eight different amino acid biosynthesis pathways, including L-ornithine biosynthesis, were also enriched in HFE broilers based on PICRUSt2 analysis of the duodenal microbiota (Additional file 1: Fig. S8c, d).

Ornithine produced by Lactobacillus enhanced intestinal cell antioxidant and nutrient uptake

To explore which Lactobacillus metabolites play a crucial role in duodenal antioxidant and nutrient uptake, culture supernatants from MRS and four Lactobacillus species were determined by untargeted metabolomics. Among the 32 Lactobacillus metabolites, ornithine, stearidonic acid, and linoleic acid were significantly increased in both LI and LV bacteria (fold change > 10, Fig. 5a). Notably, whole-genome sequencing analysis showed that the bacteria of LR, LI, and LV had contigs of citrulline and ornithine biosynthesis gene clusters, whereas LJ bacteria did not (Fig. 5b). Increased ornithine concentrations were further confirmed in the duodenum of HFE broilers, as well as in LI and LV-treated broilers (Fig. 5c, d). Correlation analysis further indicated that duodenal ornithine concentrations had a positive correlation (R > 0.48, P < 0.05) with T-AOC, SOD, GLUT2, SGLT1, and VH values in duodenum (Fig. 5e). Meanwhile, duodenal ornithine concentrations had a positive correlation (R > 0.65, P < 0.01) with ADG values and a negative correlation (R = − 0.505, P = 0.023) with F/G values (Fig. 5e, f), suggesting that ornithine produced by Lactobacillus may be a key metabolite for improving broiler feed efficiency. Thus, we next explored whether ornithine could regulate the redox status and nutrient uptake in IPEC1 cells (Fig. 5g). LPS treatment decreased the levels of T-AOC, SOD, and GSH, and increased MDA concentrations. However, ornithine effectively inhibited LPS-induced oxidative stress by increasing T-AOC level and SOD activity (Fig. 5h–j). We then used ROS analysis to assess the cellular lipid peroxidation level. The results showed that intracellular ROS was elevated by LPS treatment, which was largely ameliorated by ornithine supplementation (Fig. 5k). Further ornithine treatment upregulated the expression of GLUT2 and CAT1 proteins compared to CON or LPS groups (Fig. 5l).

Fig. 5.

Ornithine produced by L. vaginalis and L. ingluviei contributes to intestinal health and nutrient uptake. a Heatmap of untargeted metabolomics analysis for culture supernatant of MRS and four Lactobacillus species. Relative concentrations of Lactobacillus metabolites normalized to MRS concentrations and the fold change > 5 are shown. b Whole genome analysis of five Lactobacillus species showing the contigs of citrulline and ornithine biosynthesis gene clusters. c The concentrations of ornithine in the duodenum between LFE and HFE broilers. d The concentrations of ornithine in the duodenum of broilers with four Lactobacillus species treatment. e Lollipop chart represents correlation analysis between ornithine and growth performance, as well as other biochemical parameters. f Correlation between the duodenal ornithine contents and the F/G values of broilers. g Schematic diagram of the cell experimental design. h–j Antioxidant capacity, including T-AOC level, SOD activity, and MDA concentrations of IPEC1 cells with ornithine (ORN) treatment. k Representative fluorescent images (scale bar, 100 μm) of ROS production detected by DCFH-DA fluorescence and ROS content. l Relative protein expressions of key nutrient transporters (GLUT2, FATP4, and CAT1) with ORN treatment. Statistical comparison performed by one-way ANOVA. Data are presented as mean ± SE. Graph bars with different letters indicate significant difference (P < 0.05)

Ornithine primarily plays a vital role in urea cycle and polyamine catabolism in the body (Additional file 1: Fig. S9a). However, no differences were observed in serum ammonia and BUN, the initial substance and final product of the urea cycle, among three groups (Additional file 1: Fig. S9b, c), suggesting that ornithine might affect host polyamine catabolism. As expected, higher spermidine and spermine concentrations were found in the duodenum of both LI and LV-treated broilers (Additional file 1: Fig. S9d). Further, the expression of ODC1, SRM, SMS, SMOX, and PAOX genes in duodenum was upregulated in LI and/or LV-treated broilers (Additional file 1: Fig. S9e), suggesting that ornithine enhanced host intestinal polyamine metabolism. Thus, we next investigated whether spermidine could improve cell antioxidant and nutrient transportation in IPEC1 cells (Additional file 1: Fig. S10a). Similar to ornithine treatment experiment, spermidine also alleviated LPS-induced oxidative stress by increasing the T-AOC level and SOD activity, as well as decreasing MDA concentrations and ROS production (Additional file 1: Fig. S10b–e). Moreover, spermidine treatment upregulated the expression of GLUT2 and CAT1 proteins compared to CON or LPS groups (Additional file 1: Fig. S10f).

Nrf2 signaling is essential for the ornithine mediated enhancement on the intestinal cell antioxidant and nutrient uptake

We hypothesized that Nrf2 signaling is activated in duodenal epithelial cells to improve intestinal homeostasis and nutrient absorption efficiency in broilers. To investigate this, computational docking analysis of ornithine with Nrf2 was conducted, revealing that this compound can bind to the Nrf2 protein (Fig. 6a). Nrf2 gene expression was upregulated in response to ornithine supplementation in IPEC-1 cells (Fig. 6b). Further immunofluorescence analysis confirmed that Nrf2 signals significantly decreased under LPS challenge, but this decrease was markedly counteracted following ornithine treatment (Fig. 6c). Moreover, spermidine can also bind to the Nrf2 protein based on docking analysis (Additional file 1: Fig. S11a), and both ornithine and spermidine effectively ameliorated LPS-induced downregulation of Nrf2 and HO1 (a downstream protein in the Nrf2 signaling pathway) (Additional file 1: Fig. S11b). In the duodenum of broilers, the expression of Nrf2 and HO1 was also significantly upregulated at both mRNA and protein levels in response to LI and LV treatment (Fig. 6d, e).

Fig. 6.

Nrf2 signaling is essential for the ornithine-mediated enhancement on the intestinal health and nutrient uptake. a Molecular docking analysis of ornithine (ORN) on Gallus gallus Nrf2 protein. b Relative mRNA expression of Nrf2 genes in IPEC-1 cells with ORN treatment. c Representative images (scale bar = 100 μm) of the IPEC-1 cells stained with anti-Nrf2 antibody (red) and DAPI (blue), and mean fluorescence intensity of the Nrf2 protein was measured by ImageJ software. d Relative mRNA expression of Nrf2 and HO1 genes in broiler duodenum with LI and LV treatment. e Relative Nrf2 and HO1 protein expression in broiler duodenum with LI and LV treatment. f Schematic diagram of cell experimental design. g Representative immunofluorescence staining images of Nrf2 (scale bar = 100 μm), and mean fluorescence intensity was measured by ImageJ software. h–j Antioxidant capacity, including T-AOC level, SOD activity, and MDA concentration in IPEC-1 cells. k The ROS production was detected by DCFH-DA fluorescence. Scale bar, 50 μm. l Relative protein expressions of key nutrient transporters (GLUT2, FATP4, and CAT1). Statistical comparison performed by one-way ANOVA. Data are presented as mean ± SE. Graph bars with different letters indicate significant difference (P < 0.05)

We then evaluated whether Nrf2 signaling is essential for the ornithine and spermidine mediated improvements in intestinal epithelial cell health and nutrient uptake (Fig. 6f). The results showed that supplementation with brusatol, an Nrf2 inhibitor, blocked the effects of ornithine and spermidine on Nrf2 signaling activation (Fig. 6g). Additionally, brusatol treatment inhibited the regulatory effects of ornithine and spermidine on T-AOC levels, SOD activity, and MDA concentration in IPEC-1 cells (Fig. 6h–j). Further ROS analysis revealed that brusatol suppressed the regulatory roles of ornithine and spermidine on intracellular ROS levels (Fig. 6k). Moreover, ornithine and spermidine treatment under LPS challenge increased the expression of nutrient transporter proteins GLUT2 and CAT1, but this increase was significantly reduced following brusatol treatment (Fig. 6l). Altogether, these findings demonstrate that ornithine and spermidine enhance intestinal epithelial cell health and nutrient uptake through activation of Nrf2 signaling.

Discussion

The small intestinal microbiome in animals is known to play a pivotal role in host nutrient absorption and the maintenance of homeostasis [40], thereby underscoring its significance in regulating feed efficiency. Notably, the microbial community within the proximal small intestine is more vulnerable to dietary alterations [20]. In the present study, it was observed that the small intestinal microbiota, particularly the duodenal microbiota, in HFE broilers demonstrated more robust cooperative interactions within microbial communities and exhibited greater community stability. Concurrently, the duodenum of HFE broilers showed significantly enhanced digestive enzyme activities, specifically amylase and lipase, alongside increased VH and an elevated VH to CD ratio. These findings indicate that a stable duodenal microbiota community plays a crucial role in the physiological regulation of the host digestive and absorptive functions in response to dietary nutrients. Previous research has established that cecal microbiota facilitate microbial fermentation of undigested carbohydrates, producing bioactive metabolites that contribute to feed efficiency in chickens [16–19]. In this study, we identified 11 bacterial features significantly associated with feed efficiency in the cecum. Notably, all these features, except for F1544 Limosilactobacillus, are well-documented carbohydrate fermenters for producing short-chain fatty acids (SCFAs). Cecal SCFAs serve as an important energy source for animal growth and also possess immunomodulatory properties that positively impact animal health and performance [2]. Consistent with previous studies [8–14], our cecal microbiota results also revealed that the cecum of high feed efficiency chickens harbors a higher abundance of SCFA-producing taxa. Taken together, these findings refine our understanding of spatial gut microbial function: the duodenal microbiota acts as a regulatory hub for nutrient absorption efficiency via specific microbial interactions, whereas the cecal microbiota serves as a metabolic engine, providing energy through microbial fermentation.

Precise manipulation of probiotics has been recognized as an important alternative for livestock production following the ban on antibiotics as growth promoters [41]. Identifying gut probiotic species that can enhance intestinal health and improve feed efficiency is an urgent research priority. In this study, we screened four bacterial species (L. reuteri, L. vaginalis, L. johnsonii, and L. ingluviei) as key candidate bacteria contributing to feed efficiency in broiler duodenum, using comprehensive analytical methods. The Lactobacillus genus is widely used as probiotics because it effectively resists pathogenic bacterial colonization in the gut, thereby maintaining microbial stability and improving feed efficiency in poultry production [42–44]. Based on these findings, we isolated and cultured these bacterial species from chickens to investigate their role in regulating feed efficiency. As expected, dietary supplementation with L. vaginalis and L. ingluviei significantly improved feed efficiency in chickens. However, L. reuteri and L. johnsonii did not demonstrate improvements in feed efficiency, which is inconsistent with previous chicken feeding studies reporting that several strains of L. reuteri and L. johnsonii can effectively enhance production performance [45–47]. Notably, different strains of the same species can exhibit very different effects because of substantial genomic variations [48, 49]. For example, L. reuteri L6798 was linked to weight gain, whereas L. reuteri ATCC PTA 4659 showed the opposite effect on body weight [50]. These findings may explain why dietary supplementation with L. reuteri and L. johnsonii strains isolated in this study did not significantly improve broiler growth performance and suggest that further research should focus on strain-level relationships between microorganisms and feed efficiency. Most bacterial species in the genus Limosilactobacillus play an important role in inhibiting the proliferation of pathogenic bacteria and reshaping the intestinal microbiota [51, 52]. Furthermore, we found that supplementation with L. vaginalis and/or L. ingluviei, two bacterial species in the genus Limosilactobacillus, reduced duodenal microbial diversity, enhanced interactions within microbial networks, and increased the abundance of Limosilactobacillus, which showed a strong positive correlation with feed efficiency. Taken together, our findings highlight the potential application of microbial strategies to improve chicken feed efficiency by enriching Limosilactobacillus to reshape the microbiota in the duodenum.

Intestinal health serves as the foundation for overall host health and optimal feed efficiency, given that the intestine is the primary site for nutrient digestion and absorption as well as pathogen defense [4, 5]. Our data showed that L. vaginalis and L. ingluviei treatment increased the levels of T-AOC, SOD, and GSH, as well as increased the number of mucus-producing goblet cells. Furthermore, it reduced MDA concentration and number of apoptotic cells in the duodenum of broilers. These results revealed that L. vaginalis and L. ingluviei could provide a favorable intestinal environment for nutrient uptake in chicken. Meanwhile, L. vaginalis and L. ingluviei treatment enhanced intestinal digestion and absorption capacity in chickens, as evidenced by increased activity of digestive enzymes (amylase and lipase), elevated villi height, and upregulated expression of nutrient transporter proteins (GLUT2, FATP4, and CAT1) in duodenum. These results indicated that L. vaginalis and L. ingluviei might facilitate nutrient absorption from digested diets, which could account for the high feed efficiency observed in the treated chickens. Moreover, probiotics can maintain optimal gut pH, inhibit pathogen proliferation, enhance nutrient digestion and absorption, and improve animal growth and development [53, 54]. Consistent with these findings, L. vaginalis and L. ingluviei demonstrated high efficacy in improving broiler feed efficiency, gut health, and nutrient absorption, suggesting their potential role as a promising antibiotic alternative in livestock industry.

Microbial metabolites play a crucial role as signaling molecules that influence host physiological homeostasis, including energy harvest, nutrient metabolism, and immune modulation [55, 56]. This evidence suggests that these metabolites may serve as promising molecular targets for regulating animal feed efficiency. In this regard, we demonstrated that ornithine, a product of L. vaginalis and L. ingluviei, contributed to the regulation of intestinal health and nutrient uptake in IPEC-1 cells. This was evidenced by enhanced antioxidant capacity and upregulated nutrient transporter expression under LPS challenge, which supports that ornithine plays an important role in maintaining healthy gut mucosa [57]. Ornithine exerts beneficial effects through its roles in detoxifying ammonia as an intermediate in the urea cycle [58]. Additionally, ornithine participates in polyamine catabolism within the body, thereby influencing physiological processes such as growth, development, reproduction, and aging [59]. Here, we found that the treatment with L. vaginalis and L. ingluviei increased spermidine and spermine concentrations and upregulated the expression of polyamine metabolism-related genes in the duodenum of broiler, without altering serum ammonia or BUN levels. These findings suggested that ornithine produced by Lactobacillus contributed to improved intestinal health and nutrient uptake potentially through enhanced host intestinal polyamine metabolism rather than via the urea cycle. Notably, polyamine depletion induces ROS accumulation, leading to cell damage and the inhibition of cell growth [60, 61]. Spermidine, an intermediate in polyamine metabolism, has been shown to exert various physiological health effects, including antioxidant, anti-aging, and anti-inflammatory properties [62–64]. As expected, our data showed that spermidine supplementation ameliorated LPS-induced oxidative stress in IPEC-1 cells by increasing T-AOC levels and SOD activity, while decreasing MDA concentrations and intracellular ROS. Meanwhile, spermidine contributed to increased intestinal nutrient uptake, as evidenced by the upregulated nutrient transporter (GLUT2 and CAT1) expression under LPS challenge. Taken together, our findings underscore the critical role of intestinal Lactobacillus in supplying ornithine to enhance host polyamine metabolism, thereby improving gut health and nutrient absorption. This research further deepened our understanding of gut microbial metabolites as intermediates in host-gut microbiota interactions.

Nrf2 signaling pathway activation contributes to a pivotal defense mechanism that enables cells to maintain homeostasis under oxidative stress [65]. Moreover, it also plays an essential role in regulating energy metabolism by prioritizing glucose allocation for energy production in brain cells [66]. Here, our findings showed that both ornithine (the bacterial metabolite) and spermidine (ornithine metabolite) promoted Nrf2 and HO1 expression. Moreover, brusatol (a Nrf2 inhibitor) inhibited the regulatory roles of ornithine and spermidine in improving T-AOC levels, SOD activity, MDA concentrations, intracellular ROS, as well as the expression of GLUT2 and CAT1 proteins. These results highlighted the importance of Nrf2 in regulating intestinal health and nutrient uptake, suggesting important targets for improving feed efficiency in food-producing animals.

Conclusion

Our work revealed that broilers with high feed efficiency exhibit greater stability and stronger cooperative interactions within microbial communities in the small intestine, particularly in the duodenal microbiome. The genus Limosilactobacillus, especially L. vaginalis and L. ingluviei, could reshape the duodenal microbiota and enhance feed efficiency by producing ornithine, which promotes duodenal antioxidant capacity and nutrient transport through activation of the Nrf2 signaling pathway in broilers. Our results highlight that these two bacterial species and ornithine could serve as effective additives in poultry and other food-producing animals to improve intestinal health and feed efficiency.

Abbreviations

Nrf2

Nuclear factor erythroid 2-related factor 2

ADG

Average daily gain

ADFI

Average daily feed intake

F/G

Feed to gain ratio

ASVs

Amplicon sequence variants

ANOSIM

Analysis of similarities

LEfSe

Linear Discriminant Analysis Effect Size

RF

Random forest

T-AOC

Total antioxidant capacity

SOD

Superoxide dismutase

MDA

Malondialdehyde

GSH

Reduced glutathione

BUN

Blood urea nitrogen

DAO

Diamine oxidase

LPS

Lipopolysaccharide

VH

Villus height

CD

Crypt depth

MRS

De Man, Rogosa and Sharpe

ROS

Reactive oxygen species

Authors’ contribution

L.-H. S. and Z.-C. D. designed the study; Z.-C. D., Y.-X. H., K.-X. C., Z. P., A. R., and L. Z. performed the experiments and statistical analyses; Z.-C. D. wrote the original manuscript; L.-H. S., L. L. G., and M. M. K. review and editing the manuscript. All authors have reviewed and approved the final version of the manuscript.

Funding

This research was supported in part by the project of National Natural Science Foundation of China (W2412097), the National Key Research and Development Program of China (2022YFD1300402), the Fundamental Research Funds for the Central Universities (2662026DKPY007), the Academy of Scientific Research and Technology of Egypt (RESPECT-25505), Hubei Provincial Natural Science Foundation of China (2026AFB103), and the Postdoctoral Fellowship Program (Grade C) of China Postdoctoral Science Foundation (GZC20261931).

Data availability

The 16S rRNA gene sequencing data in this study are available in the NCBI SRA repository under accession number PRJNA1253976 and PRJNA1401285. The bacterial genome sequencing data have been deposited in NCBI WGS Batch under accession number PRJNA1419528. The metabolomics data have been deposited in the NGDC OMIX repository under accession number OMIX011423.

Declarations

Ethics approval and consent to participate

Animal experiments were approved by the Institutional Animal Care and Use Committee of Huazhong Agricultural University (approval number: HZAUCH-2024–0022).

Competing interests

The authors declare no competing interests.