The mentioned strain of this study was separated from an autochthonous clinical trial, and the antibiotic susceptibility of the strain was shown in previous publication (16). Genomic information of E. coli 1587 was shown in Supplementary Table 1. E. coli K12 cells were preserved in our lab. Except for special instructions, all isolates were incubated using MacConkey agar plates (Luqiao company, Beijing, China) or Luria-Bertani broth (LB, Qingdao Hope Bio-technology) at 37°C. Antibiotics were obtained from the China Institute of Veterinary Drug Control. All E. coli isolates were screened for the phenotypic identification of ESBL producers on MacConkey agar containing cefotaxime (2 mg/L) prior further confirmation using double-disc synergy testing in accordance with Clinical and Laboratory Standards Institute (CLSI) recommendations. Bacterial isolates were determined to be positive when the clear zone inhibition of ceftazidime plus clavulanic acid or cefotaxime plus clavulanic acid was at least 5 mm larger than their respective single disks (17). Then, part of culture medium was taken out to dilute in fresh LB until it reaches a starting optical density at 600 nm (OD₆₀₀) of 0.2 at a 10 ml conical tube. Filter-sterilized GA (Gallic acid, Sigma-Aldrich, CID 370, 0.08 g/L and 0.8 g/L dissolved in distilled water) were separately appended to groups and further cultured at 37°C with shaking at 200 rpm until harvest. Specifically, cultures were mixed to acquire the optical density data at the indicated time-points post treatment (4, 8, 12, and 24 h).

The animal experiment was conducted in a conventional cow pasture in Heilongjiang Province, China, following the approved protocols of the experiment by Beijing Association for Science and Technology (ID no. SYXK, 2016-0008). The animal experiment was conducted in early spring, with temperatures of 7 ± 2°C, humidity of 60 ± 5%, and light intensity of 180 ± 20 lux for 18–20 h. The neonatal Holstein female calves were separated from dams immediately after birth and transferred to individual pens, following navel disinfection with 7% iodine. No animals received medical treatment in this study and health conditions, i.e., presence or absence of disease, injury, and dehydration, were appraised daily during our experiment. The pasteurized colostrum (60°C for 60 min) was thawed at 37°C using a water bath and calves were fed 4 L of colostrum within 2 h after birth using bottle feeding in the post-natal care service. Animal rectal temperature, fecal score, and respiratory score were recorded as in previously published methods (18). Then, they were bucket fed 4 L of milk replacer daily at 2–7 days after birth (phase I) and 5 L at days 8–14 (phase II) after birth. The milk replacer contained 83.3% of skim milk powder with 260 g/kg crude protein, 160 g/kg crude fat, 10 g/kg crude ash, and 19.2 MJ/kg metabolizable energy on a dry matter basis (Nutrifeed, IN, Netherlands). Calves with diarrhea were monitored and were classified as “diarrhea” when their fecal score was ≥3 for at least 2 days and positive of diarrheagenic E. coli (19). On the other hand, calves of the same age were identified as “healthy” when their fecal score was ≤2 for at least 2 days and without pathogen infections (Supplementary Table 2). In this trial, no calves were treated with antibiotics. Fecal samples (~10 g) were collected directly from the rectum by rectal palpation using sterile gloves and dedicated equipment to prevent cross-contamination. Samples were then subjected to the detection of E. coli 1587, Rotavirus, Coronavirus, and Cryptosporidium antigens using PCR method or commercial ELISA kits (IDEXX, Maine, USA). Eventually, 40 samples were enrolled with phase I (10 healthy calves-H_1, 11 diarrheal calves-D_1) and phase II (10 healthy calves-H_2, 9 diarrheal calves-D_2) over this entire study. All collected fecal samples were stored at −80°C immediately after collection.

Fecal samples were collected, and bacterial genomic DNA extraction was performed using a QIAamp DNA Isolation Kit (Qiagen, Hilden, Germany). Amplicons of the V3-V4 region within the 16S rRNA gene were conducted by PCR using optimized primers (338F: 5′-ACTCCTACGGGAGGCAGCAG-3′, 806R: 5′-GGACTACHVGGGTWTCTAAT-3′). PCR product library was prepared using the TruSeq Nano DNA LT Library Prep Kit (Illumina, SanDiego, USA), following sequenced on the Illumina Miseq platform (2 × 300, pair end). Sequencing data were processed using QIIME2 version 2020.02 (20). Then, taxonomy was assigned to filtered amplicon sequence variants (ASVs) using a pretrained QIIME2-compatible SILVA version 132 database, with 99% identity for the bacteria and representative sequences (21). Species diversity was determined using q2-diversity (http://www.r-project.org/).

Fecal samples (5 mg) from healthy or diarrheal calves were homogenized with zirconium oxide beads for 3 min and mixed with 145 μl extraction solution (containing 25 μl of water and 120 μl of methanol) to extract the metabolites. The samples were then homogenized for another 3 min using a high-throughput tissue disruptor and centrifuged at 1,800 g for 20 min. Acquired supernatant was transferred to a 96-well plate and mixed with 20 μl derivative reagents at 30°C for 60 min, following procedures of Eppendorf epMotion Workstation (Eppendorf Inc., Humburg, Germany). Samples were further diluted with 330 μl of ice-cold 50% methanol, stored at −20°C for 20 min, and followed by centrifuged (4,000 g for 30 min at 4°C). The supernatants were obtained for liquid chromatography-mass spectrometry (LC-MS) analysis.

The metabolite extracts from calf fecal samples were analyzed using an ultra-high performance liquid chromatography coupled to tandem mass spectrometry (UPLC-MS/MS) system (ACQUITY UPLC-Xevo TQ-S, Waters Corp., Milford, MA, USA). Chromatographic separation was performed using a BEH C18 column (2.1 mm × 100 mm, 1.7 μm, Waters). The desolvation and source temperatures were set at 500°C and 150°C, respectively. Mobile phases containing acetonitrile/isopropanol (1:1, 0.1% formic acid) and 0.1% formic acid were used as carried liquid at a constant flow rate of 0.4 ml/min.

The raw data were processed using the MassLynx software (version 4.1, Waters, Milford, MA, USA). Each sample was analyzed by ultra-performance liquid chromatography coupled to tandem mass spectrometry (UPLC-MS/MS) in both positive and negative ionization modes to acquire metabolite profiles. Analysis order of all test samples was randomized. The quality-control (QC) samples were pooled samples in which both the metabolite composition of the samples and sample matrix were mixed, and then analyzed using the same methods to evaluate the quality and variance of the acquired data. Self-developed platform iMAP (version 1.0, Metabo-Profile, Shanghai, China) was used for further statistical analyses including PCA, PLS-DA, univariate analysis, and pathway analysis.

Human colon adenocarcinoma cell line Caco-2 (ATCC-HTB-37, Manassas, USA) was maintained in Modified Eagle medium (MEM, Gibco, NY, USA) supplemented with 10% fetal bovine serum (FBS, Gibco) and 1 × penicillin-streptomycin-L-glutamine (Gibco, 10378-016) at 37°C in 5% CO₂ incubator. As for the adhesion assay, cells were grown in 24-well tissue culture plates until reaching 80% confluence, and then incubated with E. coli strains at a multiplicity of infection (MOI) of 10 post washing in MEM. Cell-associated bacteria was quantified via removing non-adhering bacteria, which was performed following washing and cell lysis. Then, the number of Colony Forming Unit (CFU) was determined using serial dilution method and adherence percent was calculated via dividing the final CFU number per ml by the initial CFU number per ml for each replicate.

Then, LPS component of E. coli 1587 was extracted using a Lipopolysaccharide Isolation Kit (Sigma, MAK339, St.Louis, USA) and diluted to a storage concentration of 1 mg/ml in MEM. The effect of GA on LPS-treated cells was further detected via adding GA at a final concentration of 0.01 g/L to normal culture medium, and the cells were then cultured for 9 h followed by LPS (10 μg/ml of final concentration) treatment for 4 h. The same volume of MEM was used as negative controls.

Littermate pregnant specific pathogen-free (SPF) CD-1 mice (Sipeifu Biotechnology, Beijing, China) were adapted to standardized environmental conditions (temperature = 25 ± 2°C; humidity = 55 ± 10%), with a 12-h cycle of light and darkness, food, and sterile water ad libitum. Mice maintenance was strictly observed in accordance with the rules for the Administration of Affairs Concerning Experimental Animals approved by the State Council of People's Republic of China (14-11-1988). As for neonatal mice, the day of birth was considered day 0. Female mice were screened at day 2 and verified to have ingested breast milk (abdominal milk spot). After 1 week of acclimatization, murine infection test was done by intraperitoneally injected or oral gavage with a sublethal dose of E. coli 1587 persisters suspension (3.0 × 10⁵ CFU per mouse). Mice were weaned at day 21. The murine experiments were approved by the institutional ethics committees of China Agricultural University (XXMBB-2012-03-15-1).

Gallic acid was dissolved in the sterile water to a concentration of 40 mg/kg body weight. Two-day-old female mice were randomly divided into three treatments with 12 mice and one dam per group. These mice received daily oral gavage of 50 μl GA or the same volume of sterile water for 14 days.

Bacterial infection route was responsible for the virulence and invasion of clinical isolates. Hence, they were firstly challenged with E. coli 1587 via intraperitoneal injection to induce peritonitis at day 0. Mice were randomly divided into three groups (n = 12 per group) as follows: CON group, mice were access to regular sterile water for 2 weeks without infection; Placebo group, free access to regular sterile water for 2 weeks with E. coli infection; GA group, daily gavage of GA with E. coli infection. Body weight was recorded daily, and disease activity index (DAI) was rated to appraise colitis severity from day 0 to day 7 as previously described (22). Murine eyeball blood was collected, and serum was generated by centrifugation (2,500 g for 15 min at 4°C) post euthanasia on day 3 and day 7. The colon was quickly removed for length measurement, and a 5-mm segment of hollow mid-colon was collected for further assessment. Colonic tissues were homogenized in sterile PBS for bacterial loads using CFU calculation method.

Then, another test was conducted to detect the effect of GA in oral infection model of neonatal mice. Two-day-old female mice were casually divided into four groups with six mice and one dam per group. These mice also received daily oral gavage of 50 μl GA for 14 days. Mice were then infected with E. coli via oral gavage to imitate the natural infection at day 0 to induce colitis. The treatment groups are as follows: CON group, daily gavage of regular sterile water normally for 2 weeks; GA group, oral GA for 2 weeks without infection; 1,587 group, daily gavage of regular sterile water for 2 weeks with E. coli 1,587 infection; 1,587 + GA group, oral GA for 2 weeks with E. coli 1,587 infection. Body weight and DAI were monitored daily from day 0 to 7, followed by anesthesia on day 7. Colon length was removed for measurement, and colonic tissues were collected for subsequent assessment. Colonic tissues were homogenized in sterile PBS for bacterial loads using CFU calculation method. Mice serum was generated by centrifugation of blood samples (2,500 g for 15 min at 4°C) which were frozen at −80°C following assessment of serum cytokines.

For the fecal transplantations, 2-day-old female mice received daily oral gavage of 50 μl GA or the same volume of sterile water alone for 14 days. Meanwhile, mice were challenged with E. coli 1587 via oral gavage at day 0. They were randomized to three groups (n = 4–6 per group) as follows: CON group, free access to regular sterile water for 2 weeks without infection; Placebo group, oral gavage of regular sterile water for 2 weeks, with E. coli 1587 infection; and GA group, oral gavage of GA daily for 2 weeks with E. coli 1587 infection. Fresh feces of each group were pooled and homogenized in sterile saline to a liquid solution (final concentration of 100 mg feces/ml). The fecal supernatant was obtained by centrifugation and filtration and was then orally inoculated into recipient mice. Two-day-old female mice were administered with FMT via oral gavage two times a day (50 μl per mouse) during 1-week period. Mice were followed with E. coli 1587 infection via oral gavage at day 8. Then, body weight and DAI were measured daily post infection. Mice fresh feces were collected sterilely during day 14 and day 15 for future assessment of microbiota profiles. All mice were sacrificed by anesthesia at day 15, and the colon was quickly collected for length measurement and future detection of colonic contents.

Total RNA from Caco-2 cell or colon tissue was extracted using the EASYspin Plus kit (Aidlab, Beijing, China) following the instructions of manufacture. Total RNA was treated with Dnase?, and 500 ng of total RNA was used for cDNA synthesis by reverse transcription using the PrimeScript RT reagent Kit with gDNA Eraser (Takara, Beijing, China). The cDNA was amplified with specific primers for the indicated genes, including IL-1β, IL-6, TNF-α, IL-10, and TGF-β genes, as previously described (23, 24). The real-time PCR was performed, and products were detected using QuantStudio 6 Flex Real-Time PCR System (Applied Biosystem, CA, USA). The PCR was performed in a 20-μl volume containing 1 μl of cDNA, 10 μl of 2 × SYBR green Premix Ex Taq (TaKaRa Bio Inc., Japan), and a 0.4 μM concentration of each gene-specific primer. Thermal cycling parameters were as follows: 95°C for 2 min, followed by 42 cycles of 95°C for 15 s, 60°C for 15 s, and 72°C for 15 s, and 1 cycle of 95°C for 30 s, 60°C for 30 s, and 95°C for 30 s. The final step was to obtain a melting curve for the PCR product to determine the specificity of the amplification. Each sample was run in triplicates on the 96-well sample plate, and the expression level of each gene was calculated relative to the expression of the glyceraldehyd-3-phosphate dehydrogenase (GAPDH) gene.

Mice serum sample was generated by centrifugation of eye-blood sample at 2,500 g for 15 min. The inflammatory state of serum was assessed by measurement of inflammatory cytokines including IL-1β and IL-10 (ABclonal Technology, Wuhan, China) according to the instructions of manufacturer.

Colonic tissues were fixed in 10% formalin for 48 h, dehydrated, and embedded into paraffin following standard steps. About 4 μm thick paraffin sections were subjected to hematoxylin and eosin (H&E) staining. Visual images were further analyzed using the Image J software (version 1.8.0_172, National Institutes of Health, Bethesda, MD, USA). The extent of colitis (including ulcer, ulceration, crypt structure, crypt loss, and presence or absence of edema) was assessed, and the histopathology scoring was determined in a blind manner according to the previously described protocols (25).

Colonic contents or fecal samples were collected for SCFA (such as acetate, propionate, and butyrate) measurement via gas chromatography as previously described (26). In detail, 100 mg of samples from mice were dissolved, homogenized in ultrapure water, and suspension pH was adjusted to 2–3. The clear supernatant was collected post centrifugation at 5,400 rpm for 20 min. Zero-point-two milliliters deproteinized solution of 2-ethylbutyric acid was mixed with supernatant in an ice water bath for 30 min, followed by centrifugation at 10,000 rpm for 10 min. Acquired supernatant was transferred to a 2 ml screw-cap vial, and then subjected for SCFA analysis with Gas Chromatography System (Agilent Technologies, Wilmington, Delaware, USA).

Following treatments, harvested cells or colonic tissue samples used for western blotting were prepared using a lysis buffer (50 mM Tris-HCl, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 5 mM MgCl₂, 10% glycerol, and 1 mM PMSF). Protein contents were obtained post centrifugation (12,000 g for 10 min at 4°C), and their respective concentrations were measured using bicinchoninic acid assay (BCA; Beyotime Biotechnology, Beijing, China) following the instructions of manufacturer. Protein samples were separated using 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gel and transferred onto polyvinylidene difluoride membranes. Non-specific bindings were blocked using 5% nonfat milk at 37°C for about 1 h. Then, the membranes were incubated with the primary antibodies, followed by incubation with peroxidase-conjugated secondary antibody at 37°C for 1 h. Finally, the bands were visualized using enhanced chemiluminescence kit (Beyotime Biotechnology, Beijing, China). The antibodies used in the western blotting assays were shown as follows: phospho-IκBα (Ser36) antibody (Abcam, ab133462), IκBα antibody (Abcam, ab32518), β-actin antibody (ABclonal Technology, AC004), occludin antibody (ABclonal Technolog, A2601), and Horseradish peroxidase (HRP)-conjugated goat anti-mouse or anti-rabbit IgG antibodies (DingGuo, Beijing, China).

Statistical analysis was performed using Graphpad prism 5.0 program (Version 5.2.1350.0 for Windows) and SPSS software (version 19.0). All data were presented as Mean ± SEM unless otherwise indicated. P-values were determined by unpaired t-test, nonparametric Kruskal-Wallis test between two groups, two-way ANOVA among various treatment groups, and Mann-Whitney-Wilcoxon (U-test) in multiple comparisons. Spearman correlation coefficient analysis was used to analyze the correlation between gut microbiota and metabolites using the “ggplot2” and “pheatmap” packages of R software (version 3.3.1). The two-sided, Log-rank (Mantel-Cox) test was used for contrast of mortality curves. All experimental data were representative of three independent experiments with similar results. Statistical significance was determined for ^*p < 0.05, ^**p < 0.01, ^***p < 0.001.

Colibacillus diarrhea closely correlates with the pathogenesis of ESBL-EAEC and is accompanied by the deterioration of antimicrobial resistance in pasture. With the aim of plumbing more about the self-regulatory effects of neonatal calves on ESBL-EAEC-infection control, pooled fecal samples of calves were subjected to 16S ribosomal RNA (rRNA) sequencing analyses to evaluate the diversity and community of fecal microbiota in different groups (Supplementary Table 1). As a result, Firmicutes, Proteobacteria, Actinobacteria, Bacteroidetes, and Fusobacteria were the major phyla in fecal microbiota (Supplementary Figure 1A), with enriched Coriobacteriaceae, Ruminococcaceae in healthy groups and Streptococcaceae, Lactobacillaceae in diarrheal calves (Supplementary Figure 1B). Among them, considerably abundant Collinsella and Faecalibacterium were found in both the two healthy groups at the genus level (Figures 1A,B). Then, microbial compositions in groups exhibited a decline trend in D_1 group, as induced by chao1 and Simpson indices (Figure 1C). The overall difference in the microbial structure of diarrheal groups was distinct from healthy ones, according to principal co-ordinates analysis (PCoA) based on weighted UniFrac distance (Figure 1D). The linear discriminant analysis (LDA) effect size (LefSe) algorithm ranked Collinsella, Coriobacterium, Faecalibacterium, Butyricicoccus, and Enterococcus as the main distinguished bacterial taxa in the H_1 group, while the diarrheal groups were differentiated with Streptococcus, Peptococcaceae, and Gallibacterium. Similarly, Clostridia, Collinsella, and Coriobacterium was the main differential bacterial in the H_2 group, while Halomonas and Devosia in the D_2 group (Supplementary Figures 1C,D). Differentially expressed genes (DEGs) were shown by Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment pathway analysis and the results implicated some general pathways, including amino acid, carbohydrate, lipid metabolism, biosynthesis of other secondary metabolites, and infectious diseases (Supplementary Figure 2). Briefly, these data demonstrated that ESBL-EAEC-infection was closely related to the alteration of hindgut microbial community structure, especially the temporal changes of microbial biomarker Collinsella and Coriobacterium in the first 2 weeks of life.

Fecal metabolites were further detected using untargeted metabolomics to gain a systematic understanding of the interactions among intestinal epithelium, gut microorganisms, and their associated metabolome. Here, fecal metabolites of healthy (n = 20) or diarrheal calves (n = 20) were analyzed with UPLC-MS/MS system. The comparison of different metabolites revealed a dramatically high relative abundance of benzenoids (2.47 vs.22%), phenols (0.13 vs 0.024%), indoles (0.0090 vs.00084%), SCFA (27.31 vs 13.26%), and amino acids (40.11 vs 23.24%) in healthy calves when compared with diarrheal ones (Supplementary Figure 3A). Similarly, the healthy fecal metabolome separated widely from diarrheal groups based on PLS-DA analysis considering their allocated groups (Component 1, p = 7.86e−07; Component 2, p = 5.55e−05; Figure 2A, Supplementary Figure 3B). Undeniably, dispersed data points on plots of the metabolome were clearly displayed between H_1 and D_1, H_2 and D_2, H_1 and H_2, and D_1 and D_2 groups (Supplementary Figure 4).

Differentially abundant fecal metabolites were further identified using an unweighted paired group method with arithmetic mean (UPGMA) dendrogram and appeared in a heatmap, including 57 most diverse metabolites (Figure 2B). Among them, several benzenoid compounds were listed potential biomarkers by a comparison of the healthy and diarrheal calves, including GA (p = 1.4e−06), 4-hydroxybenzoic acid (p = 1.1e−05), vanillic acid (p = 2.9e−05), and phthalic acid (p = 2.1e−02) (Figures 2C–F). In addition, relative fold change of quantities and variation significances were presented by volcano plots, as indicated in H_1 vs D_1, H_2 vs D_2, H_1 vs H_2, and D_1 vs D_2 groups (Supplementary Figure 5). According to the markedly altered metabolites, enriched KEGG pathway analyses were involved with valine, leucine, and isoleucine biosynthesis, pantothenate and CoA biosynthesis, propanoate metabolism, linoleic acid metabolism, suggested with alanine, aspartate, and glutamate metabolism (Supplementary Figure 3C). Moreover, as was shown by random forest supervised machine learning algorithm, 10 prominent fecal metabolites contributed to the discrimination power of dairy calf health status, including GA and 4-hydroxybenzoic acid. The relative rank of relative abundance of those 10 metabolites biomarkers was plotted against the healthy status represented by the score of Mean Decrease Gini (Supplementary Figure 3D). Then, the correlation of significantly altered metabolites with specific differentiated microbial taxa was directly revealed. Spearman rank correlation analysis indicated a strong positive correlation between abundant GA and Collinsella (R > 0.62, p = 0.0026), Faecalibacterium (R > 0.67, p = 0.00073), Butyricicoccus (R > 0.62, p = 0.0026), Enterococcus (R > 0.47, p = 0.027), and Coriobacterium (R > 0.47, p = 0.029) in the H_1 phase (Figure 3A). Similarly, H_2-enriched GA was closely linked to two hindgut commensals, including Collinsella (R > 0.49, p = 0.033), and Coriobacterium (R > 0.48, p = 0.036) (Figure 3B). Notably, the microbes mentioned above were closely associated with enriched 4-hydroxybenzoic acid, vanillic acid, and SCFA, and reduced hippuric acid, octanoic acid, and α-linolenic acid. However, it should be noted that there existed a strong correlation between Flavobacterium and octanoic acid (R > 0.70, p = 0.00074), Gallibacterium and octanoic acid (R > 0.66, p = 0.0010), and Streptococcus and hippuric acid (R > 0.55, p =0.0096), suggesting that hippuric acid, octanoic acid, and linolenic acid could be identified as metabolite biomarkers in those diarrheal calves. Thus, massive cuts in GA and some other benzenoid materials were probably attributed to the decrease of Collinsella and Coriobacterium abundance observed in healthy calves. These observations indicated that declines of some commensal bacteria reshaped by ESBL-EAEC infection were linked to the impaired biosynthesis of GA or benzenoid pools, further inducing impaired hindgut bacterial structure. Of note, our findings suggested that GA administration could ameliorate intestinal homeostasis by some specific patterns.

As reported in the literature (12), GA suppressed E. coli and viable bacteria biofilm formation and activity in vitro. We therefore detected the direct effects of GA on the bacterial growth of clinically relevant ESBL-EAEC strain 1587 in vitro. As shown by Figure 4A, ESBL-EAEC growth was markedly inhibited post GA administration at 8, 12, and 24 h in a dose-dependent manner compared to control group. In addition, the 1,587 isolate adhered significantly better than a laboratory E. coli K12 strain when adding to human colon adenocarcinoma cell line Caco-2 cell culture. Furthermore, GA addition markedly mitigated cell adherence of both strains, indicating the direct effect of GA to effectively cut off the adhesion of bacteria from the surface of intestinal epithelial cells (Figure 4B).

Considering that LPS was an inseparable component of E. coli, the extracted LPS component from ESBL-EAEC was directly added to human colon adenocarcinoma cell line Caco-2 which was pretreated with optimal concentration of GA or sterile water control. As a result, the treatment of GA did not impact cell survival in this specific concentration (data not shown), and cell shrinkage induced by LPS were recovered by GA addition (Figure 4C). To illuminate the underlying mechanism, cells were harvested and were used to quantify the relative mRNA expressions of pro- and anti-inflammatory cytokines. LPS enhanced the expression of IL-1β, IL-6, and TNF-α, and GA pretreatment suppressed the increase of IL-1β (Figures 4D–F), with elevated IL-10 and TGF-β (Figures 4G,H). As TLR4/NF-κB pathway was closely related to bacterial invasion process in intestinal epithelial cells (27), the relative phosphorylation levels of IκBα protein in different treatment groups were further detected. Indeed, GA cotreatment prominently blocked the increased IκBα phosphorylation level induced by LPS (Figures 4I,J). Interestingly, cell integrity analysis showed that GA pretreatment countered the decreased expression level of cell tight-junction related protein occludin (Figures 4I,K). Thus, these results demonstrated that the antibacterial effects of GA administration were mainly mediated by blockage of bacterial growth and cell adherence in vitro. GA exposure significantly mitigated NF-κB signaling activation, thus regulating inflammatory cytokines expressions.

To further study the alleviated effects of GA on ESBL-EAEC infection in vivo, 2-day-old mice received daily oral administration of GA for 2 weeks, accompanied by a challenge via intraperitoneal injection (Figure 5A). Over that period, clinical symptoms of mice were measured daily. Herein, ESBL-EAEC infection resulted in body weight loss (Figure 5B), high morbidity (Figure 5C), colitis (Figure 5D), reduction in the colonic length (Figure 5E), and histological damage including destruction of glandular structure, deep ulceration of mucosal muscularis, and inflammatory infiltrate (Figure 5F). Notably, GA intervention improved the body weight loss significantly in comparison with Placebo group. The remarkably lower DAI, comprehensive evaluation score of body weight loss, rectal bleeding, stool consistency, and relieved colonic growth inhibition were also displayed. Moreover, GA markedly reversed ESBL-EAEC-induced colonic inflammation and epithelial damage in the blind histological assessment, consisting with largely cut of overall clinical score in both 3 days and 7 days post infection (Figures 5G,H).

Intestinal inflammatory status was closely linked to ESBL-EAEC pathogenesis. We thus detected the pro-inflammatory and anti-inflammatory cytokines production levels in serum and colon tissues. Here, increased serum concentrations of IL-1β were blocked by GA (Supplementary Figures 6A,B). In contrast, reduced IL-10 was recovered to normal status in GA-treated mice (Supplementary Figures 6C,D). Similarly, E. coli 1587 infection dramatically stimulated the upregulation of IL-1β, IL-6, IL-10, and TNF-α mRNA transcription levels in mice colons 3 days post infection (Supplementary Figures 6E,G,I,M), but this phenomenon mostly diminished by 7 days post infection except for IL-6 (Supplementary Figures 6F,H,J,N). IL-1β and TNF-α upregulations were significantly reduced by oral GA (Supplementary Figures 6E,M), accompanied by elevated TGF-β mRNA expression level (Supplementary Figure 6L). In addition, oral GA markedly blocked the raised level of IκBα phosphorylation induced by ESBL-EAEC infection (Figures 5I,J), and the suppression of occludin protein expression level was significantly blocked by GA pretreatment (Figures 5I,K).

Colon physiology can be mostly affected by active SCFA-producing commensal microorganisms (28). We further evaluated the impact of oral GA on SCFA (acetate, propionate, and butyrate) production in colonic contents using a Gas Chromatography System. Compared with CON group, both acetate and butyrate productions were reduced a lot in the Placebo group except for propionate, but dampened acetate and butyrate recovered to baseline by oral GA in 3 days or 7 days post infection (Figures 6A–C). Importantly, oral GA dramatically suppressed E. coli 1587 colonization over the time period (Supplementary Figure 8A). Taken together, mice symptoms and colitis were alleviated by oral GA. Specifically, restored SCFA productions in ESBL-EAEC-infected mice in response to oral GA suggested the reconstruction of hindgut microbial patterns and the beneficial effects were mediated by some commensals in relieving clinic symptoms.

Considering that infection route was responsible for the virulence and invasion of clinical isolates (29), effects of prophylactic GA supplementation were further explored in neonatal mice oral infection test. As before, 2-day-old mice received daily oral delivery of GA, accompanied by challenge via oral gavage of E. coli 1587 (Figure 7A). Over that period, detailed records of body weight and disease symptoms of mice were obtained. As a result, ESBL-EAEC infection induced body weight loss (Figure 7B), colitis (Figure 7C), reduced colonic length (Figure 7D), and histological damage consisting of inflammatory cell infiltration, colonic edema, and mucosa damage (Figure 7E). Notably, GA intervention significantly mitigated the body weight loss post infection in comparison with the other groups. The reduced DAI and relieved colonic growth inhibition were also shown. Moreover, oral GA dramatically reversed ESBL-EAEC-induced colonic inflammation and epithelial damage in comparison with 1,587 group, consisting with largely cut overall clinical scores at 7 days post infection (Figure 7F). Of note, oral GA mediated the downregulation of IL-1β expression level in serum (Supplementary Figure 7A) and the upregulation of IL-10 (Supplementary Figure 7B). Similarly, ESBL-EAEC infection elevated transcription levels of IL-1β (Supplementary Figure 7C), IL-6 (Supplementary Figure 7D) and TNF-α (Supplementary Figure 7E) in colonic tissues, but recovered after GA administration with an obvious decline of IL-10 and TGF-β (Supplementary Figures 7F,G). As before, the raised level of IκBα phosphorylation post ESBL-EAEC infection was markedly blocked by GA intervention (Figures 7G,H), with marked upregulation of occludin protein expression level (Figures 7G,I).

Importantly, GA intervention rescued the SCFA concentrations of colonic contents. Results showed that only the concentration of butyrate decreased in the 1,587 group relative to the CON or GA group, but recovered post pretreatment with GA (Figure 6F). In fact, both oral 1,587 infection and GA addition could not affect fecal acetate and propionate productions (Figures 6D,E). As before, oral GA dramatically suppressed 1,587 colonization in colonic tissues (Supplementary Figure 8B).

In summary, the data demonstrated that oral GA could also effectively ameliorate clinic symptoms, colitis, and bacterial colonization inside hindgut of neonatal mice oral infection model. Notably, the enriched butyrate production of ESBL-EAEC-infected mice in response to oral GA administration suggested the improvement of gut microbiota dysbiosis and upregulatory effects of oral GA on SCFAs productions in the hindgut of neonatal animals.

It has been well studied that gut microbial community patterns contributed a lot to intestinal normal state (30, 31). FMT assays were thus performed in neonatal mice, which were then modeled to colitis by ESBL-EAEC oral infection. Fresh feces were sterilely collected and used for FMT courses, thus allowing for a quantifiable effect of GA-modulated microbiota on protection against syndromes (Figure 8A). As a result, GA-FMT group significantly increased body weight (Figure 8B), attenuated daily DAI (Figure 8C), and lead to recovery of colonic growth inhibition (Figure 8D). Unlike Placebo-FMT group, GA-FMT inoculation dramatically induced recovery of colonic inflammation and epithelial damage (Figure 8E), and largely cut of overall clinical scores (Figure 8F).

A fecal microbiota community was further detected post FMT using 16S rRNA sequencing analysis. As a result, FMT significantly enhanced fecal bacterial community richness of recipients in contrast to their corresponding donor mice. In addition, the α-diversity of GA-FMT markedly raised, as reflected by chao1 or shannon index (Supplementary Figures 9A,B). Specifically, phylogenetic analysis indicated differentiated microbial communities between donor and recipient mice. Phylum-based distinct microbiota profiles were shown by reduced Firmicutes, Actinobacteria, and richer Bacteroidetes in all FMT-treating groups (Supplementary Figure 9C). Diverse microbial compositions were also detected among groups in the family level. A tendency of lower Lactobacillaceae, enriched Bacteroidaceae, and Muribaculaceae were observed in the GA-FMT group compared with Pacebo-FMT or CON-FMT, while obvious increases of Enterobacteriaceae were observed in both Placebo and Placebo-FMT groups (Figure 8G). PCoA differentiated the microbiota of those groups based on OTU-level Bray-Curtis dissimilarity matrix, revealing a dispersed data points on plots of donor groups (R² = 0.6840, p = 0.001) and among recipient groups (R² = 0.1402, p = 0.296) (Supplementary Figures 9D,E). Notably, the LEfse analysis indicated that oral GA induced enriched Oscillospiraceae, Enterococcaceae, Lachnospiraceae, and Clostridia_UCG-014 and induced abundant Lactobacillus in the Placebo or Placebo-FMT group, which were similar to the microbial profiles of healthy calves (Figure 8H). The dominance of various specific members of Clostridia correlated with SCFA metabolism among commensal intestinal bacteria (32). Thus, fecal SCFA concentrations of donor and recipient mice were investigated. Unlike propionate, the productions of acetate and butyrate were prominently reduced by E. coli 1,587 infection and were then recovered by oral GA (Supplementary Figures 10A–C). Interestingly, GA-FMT markedly upregulated the acetate, propionate, and butyrate productions compared to CON-FMT or Placebo-FMT group (Supplementary Figures 10D–F).

Spearman correlation analysis was further performed to interpret the relevance between differentially enriched bacteria and SCFA productions. g__norank_f__norank_o__Clostridia_UCG-014 exhibited a strong positive correlation with propionic acid (R > 0.68, p = 0.0015), butyric acid (R > 0.63, p = 0.0050), g__Enterococcus correlated with enriched propionic acid (R > 0.51, p = 0.029), abundant acetic acid correlated with g__Aerococcus (R > 0.74, p = 0.00037), and g__Bacteroides (R > 0.61, p = 0.00648) (Supplementary Figure 11). In addition, genus Lachnospiraceae was associated with increased butyric acid. The data proved the LEfSe results of FMT, an indication of the beneficial effect of GA on SCFA productions of commensal microbes. Not be overlooked was the strong negative correlation between g__Lactobacillus and acetic acid (R < −0.59, p = 0.0091), propionic acid (R < −0.54, p = 0.020), and butyric acid (R < −0.55, p = 0.017), which might bring about by enriched lactic acid. This phenomenon was similar to abundant lactic acid and Lactobacillus of diarrheal calves in our previous study.

Taken together, clinical symptoms and colitis remission induced by FMT from GA-treated mice could be partly ascribed to hindgut microbial reconstruction of recipients with more abundant Clostridia_UCG-014, Lachnospiraceae, Oscillospiraceae, and Enterococcaceae, and the recovery of SCFA production, indicating the direct alleviation effects of improved commensal microbiota structure from GA-dosed mice on clinical symptoms and colitis, apart from the direct antibacterial effects on ESBL-EAEC invasion process.