Multidrug-Resistant Escherichia coli Antagonized by Luteolin: A Mechanistic Insight into Virulence Suppression and Gut Microbiota Restoration

Abstract

Multidrug-resistant Escherichia coli (MDR-E. coli) poses a serious threat in foodborne infections, highlighting an urgent need for novel antimicrobial strategies. Natural plant-derived compounds, particularly flavonoids, have gained attention for their potential as alternative antimicrobial agents. This study aimed to evaluate the antibacterial efficacy and underlying mechanisms of luteolin (LUT), a dietary flavonoid, against MDR-E. coli, and to assess its immunomodulatory and microbiota-regulatory effects in vivo. (1) Minimum inhibitory concentration (MIC) and minimum bactericidal concentration (MBC) assays were performed. (2) Biofilm formation, ATP synthesis, and alkaline phosphatase (AKP) leakage were measured. (3) Gene expression of resistance (tolC, ant(3″)-Ia) and virulence (fliC, K99, stx1) factors was analyzed via RT-PCR. (4) Network pharmacology and molecular docking identified key targets and pathways. (5) In vivo effects on intestinal pathology, inflammatory cytokines (IL-1β, IL-6, TNF-α), and gut microbiota composition were examined. The results show that (1) LUT exhibited potent antibacterial activity against MDR-E. coli (MIC = 1 mg/mL, MBC = 2 mg/mL). (2) It significantly inhibited biofilm formation, disrupted bacterial cell integrity, and suppressed ATP synthesis. (3) Expression of key resistance and virulence genes was downregulated. (4) In vivo, LUT alleviated intestinal inflammation, reduced pro-inflammatory cytokine levels, and restored gut microbial diversity, notably enriching beneficial bacteria (E. faecalis). (5) Network analysis revealed involvement of interleukin signaling pathways. LUT demonstrates dual antibacterial and immunomodulatory effects against MDR-E. coli through direct microbial inhibition and host immune regulation. It represents a promising food-compatible alternative to conventional antibiotics, with potential applications in controlling multidrug-resistant infections in the food chain. Further clinical studies are warranted to validate its efficacy and safety in humans.

1. Introduction

Diarrhea caused by E. coli infection is one of the most widespread foodborne illnesses in the world, especially in economically underdeveloped areas. Consumption of E. coli-contaminated food and water can lead to severe diarrhea in both humans and animals, which is extremely dangerous for newborns, and if left untreated, can lead to secondary infections, dehydration and other symptoms that can lead to death [1,2]. Antibiotic use remains the primary treatment method today, but with the widespread use of antibiotics, the emergence of drug-resistant bacteria has become a major threat to global public health [3]. Currently, clinical treatment for E. coli infections primarily relies on antibiotics. Commonly used antibiotics include β-lactams (such as amoxicillin and cephalosporins), quinolones (such as ciprofloxacin and levofloxacin), aminoglycosides (such as gentamicin), and sulfonamides. These drugs exert bactericidal or bacteriostatic effects by inhibiting bacterial cell wall synthesis, interfering with nucleic acid replication, or suppressing protein synthesis [4,5]. However, the widespread emergence of MDR-E. coli has significantly diminished the efficacy of traditional antibiotics, with some strains exhibiting complete resistance. Consequently, clinicians often employ combination therapy or adjust treatment regimens based on antimicrobial susceptibility testing to delay resistance development and enhance therapeutic outcomes [6]. The emergence of multidrug-resistant pathogens has surpassed the development of new antibacterial drugs at an alarming rate. There are many reasons for this widening gap, including the exhaustion of medicinal targets, the physical and chemical limitations of penetrating the envelope of Gram-negative bacteria and evolutionary adaptation, such as efflux pumps and biofilm formation [7,8]. The challenge of pathogen resistance is exacerbated by economic factors in drug development and regulatory complexity, as well as the misuse of antibiotics in the clinic, especially in agriculture. Therefore, there is an urgent need to find new ways to combat the growing crisis of pathogen resistance [9].

Natural plant compounds, which are widely distributed in various vegetables, fruits and spices, have become a hot spot for drug discovery due to their structural diversity, high biological activity and low toxicity [10]. Natural plant compounds also show strong potential in dealing with drug resistance of pathogenic bacteria, which is mainly reflected in their multi-target mechanism of action, biofilm inhibitory ability and synergistic effect with traditional antibiotics; plant-derived antimicrobial components (e.g., alkaloids, phenols, terpenoids, and flavonoids) are not only able to directly disrupt the integrity of bacterial cell membranes, inhibit the activity of key enzymes or interfere with the synthesis of nucleic acids, but also inhibit the group sensing. They also reduce the risk of bacterial resistance development by inhibiting the group-sensing system, reducing bacterial virulence and preventing biofilm formation [11,12]. In addition, these compounds often possess antioxidant and anti-inflammatory properties that mitigate infection-associated tissue damage and enhance the host immune response [13]. More importantly, many plant extracts (e.g., allicin, curcumin, and catechins) are capable of reversing bacterial resistance and restoring susceptibility to conventional antibiotics by inhibiting efflux pumps or disrupting the expression of drug-resistant genes, which provides an important direction in the development of novel antimicrobial strategies or antibiotic adjuvants, especially in the clinical challenges of multidrug-resistant bacteria (e.g., MRSA and carbapenem-resistant enterobacteria) application prospects [14,15].

Bioinformatics, as an emerging discipline, provides a new research paradigm for drug discovery, disease mechanism research and personalized medicine by integrating multi-dimensional data and constructing complex network models. Bioinformatics has shown great potential for application in the fields of multi-target drug design, drug repositioning, and modernization of traditional Chinese medicine [16,17]. Molecular docking technology has the unique advantages of high efficiency, precision and low cost in drug design, which can rapidly predict the binding mode and affinity between small molecule ligands and target proteins through computational simulation, thus significantly shortening the drug discovery cycle and reducing the cost of experimental screening [18,19]. Therefore, in this study, we used bioinformatics screening combined with molecular docking methods to explore potential natural plant compounds for the treatment of MDR-E. coli, and integrated in vitro and in vivo assays to investigate their therapeutic efficacy, therapeutic mechanisms, and effects on immunity and gut microbes. The findings may inform the treatment of MDR-E. coli infections and support the unique advantages of natural compounds in tackling drug resistance.

This study aims to systematically evaluate LUT as a potential food-derived antimicrobial agent or antibiotic adjuvant for controlling foodborne infections caused by MDR-E. coli. We specifically focus on its potential applications in food chain safety and animal health, such as serving as a feed additive or veterinary clinical adjunct therapy to reduce antibiotic overuse and curb the spread of resistance. To this end, this study employs in vitro antibacterial assays, mouse infection models, and network pharmacology with molecular docking analysis to multidimensionally investigate LUT’s antibacterial mechanisms, immunomodulatory effects, and gut microbiota restoration capabilities. All experimental designs and result interpretations are centered on their application scenarios in food and veterinary medicine, emphasizing the feasibility of LUT as a natural, safe, and non-resistance-inducing alternative or adjunct strategy. This study aims to provide theoretical foundations and practical references for developing novel antimicrobial agents.

2. Materials and Methods

2.1. Test Material

Test material included MDR-E. coli, Shigella toxin-producing E. coli (STEC) resistant to six major classes of antibiotics and twelve specific drugs; (BA1, Specific details such as drug resistance and detailed gene expression can be found in Supplementary Materials and our published literature [20,21]); specific-pathogen-free (SPF)-grade mice, 8 weeks old and weighing 19–22 g; various types of microbiological media; luteolin (>98% purity, HPLC, CAS No. 491-70-3; MACKLIN, Shanghai, China); and various kinds of molecular test kits, antibodies, and detection instruments, etc.

2.2. Screening of Drugs Against MDR-E. coli Infections

Using “E. coli infection” as keywords, GeneCards (https://www.genecards.org, accessed on 12 May 2025), OMIM (https://www.omim.org, accessed on 12 May 2025), TTD (http://db.idrblab.net/ttd/, accessed on 12 May 2025) and DisGeNET (https://www.disgenet.org/, accessed on 12 May 2025) to search for disease-related gene targets. Disease target data was obtained by summarizing the data and removing the lower-ranked targets. Disease targets were entered into the Enrichr (https://maayanlab.cloud/Enrichr/, accessed on 12 May 2025) database, and the drug prediction module was utilized to find potential therapeutic agents.

2.3. Screening Drugs for Inhibition of MDR-E. coli

2.3.1. Bacteriostatic Effect Test

After a comprehensive evaluation of items, such as the number of integrated genes, p-value prediction and gene ratio for screening drugs, LUT was selected as the target drug for the test. MDR-E. coli bacterial fluids were added to nutrient broth medium and grown on a shaker at 37 °C for 6 h to the logarithmic phase. Briefly, LUT was diluted to 4000 µg/mL, 2000 µg/mL, 1000 µg/mL, 500 µg/mL, 250 µg/mL, 125 µg/mL, 62.5 µg/mL, 31.25 µg/mL, totaling 8 concentration gradients in a 96-well plate, and then bacterial solution was added to the wells, and finally, the concentration of the bacterial solution was adjusted to 1 × 10⁶ CFU/mL by adding MHB liquid medium and mixing the wells. The ninth well without bacterial solution was used as the negative control group (NC), and the tenth well without LUT was used as the positive control group (BC). The results were observed after incubation at 37 °C for 24 h. The wells in which no significant bacterial growth was seen were taken as the MIC. An amount of 100 μL of the liquid from the above wells was added to nutrient broth plates and incubated at 37 °C for 24 h. The lowest concentration in which no bacterial growth was seen was taken as the minimum bactericidal concentration (MBC). The bacterial viability was examined using 0.1% (w/v) resazurin; the holes with bacterial growth after the addition of resazurin were pink in color [22].

2.3.2. Biofilm Elimination

The effects of different concentrations of LUT on the formation of MDR-E. coli biofilm was explored based on the results of the bacterial inhibition test [23]. The different concentrations of drugs and bacteria were mixed and incubated for 24 h, then the culture was discarded, and PBS was added to the wells to wash three times; after air-drying, 200 μL of crystal violet solution (1%) was added, and the wells were incubated at 37 °C for 20 min, and then washed using PBS three times. After drying, ethanol (95%) was added, and the absorbance value (OD600) was measured after incubation at 37 °C for 30 min. Bacteria were treated with different concentrations of LUT, and after fixation, dehydration, and gold spraying, the bacterial growth was observed using a scanning electron microscope.

2.3.3. Disruption of the Pathogen Barrier

Alkaline phosphatase (AKP) is a protease that exists between the cell membrane and the cell wall and only leaks out when the cell wall is damaged [24]. Therefore, AKP activity was used to assess the extent of cell wall damage. MDR-E. coli (10⁶ CFU/mL) were treated with different concentrations of LUT, and a blank growth group was set up to explore the extent of bacterial cell wall damage. The bacterial fluids at 0, 4, 8 and 12 h were centrifuged, and the supernatants were taken for assay. The leakage of bacterial AKP after LUT treatment was detected using an AKP kit and an enzyme marker at 520 nm.

2.3.4. Determination of ATP Concentration

The diluted bacterial suspension (10⁷ CFU/mL) was centrifuged, and the bacterial precipitate was collected, treated with different concentrations of LUT for 3 h, centrifuged and collected again. A blank growth control group was also set up. The absorbance of OD at 636 nm was measured by the Na⁺K⁺-ATPase activity assay kit to calculate the change of bacterial ATP content [25].

2.3.5. Suppression of Resistance Genes and Virulence Genes

The purified pathogen was cultured to the logarithmic growth phase and added to nutrient broth medium with a final concentration of 500 µg/mL LUT, incubated at a constant temperature of 37 °C for 24 h, and then centrifuged to obtain the bacterial precipitate, and the bacterial RNA was extracted using the bacterial RNA extraction kit, and the RNA concentration was balanced using the micro-nucleic acid assay instrument for reverse transcription, and the RNA was reverse-transcribed into cDNA using the reverse-transcription kit. The test primers were designed by NCBI. cDNA was used as a template for RT-PCR to detect the effect of LUT on the resistance genes and virulence genes in MDR-E. coli. Cultured bacteria without added drugs were used as a blank control. The experiment employed three independent biological replicates, each with three technical replicates. RT-PCR data were relatively quantified using the 2^−ΔΔCT method, with the 16S rRNA gene serving as the internal control. Statistical comparisons were performed using a two-tailed t-test, with multiple testing corrected via the Benjamini–Hochberg method (FDR < 0.05). Selected target genes included the efflux pump gene tolC, the aminoglycoside modifier enzyme gene ant(3″)-Ia, the flagellar protein gene fliC, the adhesin gene K99, and the Shiga toxin gene stx1. These genes represent key resistance mechanisms and virulence factors of MDR-E. coli.

2.4. Model Building

Refer to our previous experiments [21]. Mice were divided into six groups (n = 6 each): (1) Control group (NC) received intraperitoneal injection of saline and gavage; (2) Pathology group (PG) received intraperitoneal injection of pathogenic bacteria (8 × 10⁷ CFU/mL, 0.005 mL/g) and saline gavage; (3) Low-dose treatment group (LT) received 80 mg/kg drug gavage after intraperitoneal injection of pathogenic bacteria; (4) Medium-dose treatment group (MT) received 160 mg/kg drug gavage after intraperitoneal injection of pathogenic bacteria; (5) High-dose treatment group (HT) received 320 mg/kg drug gavage after intraperitoneal injection of pathogenic bacteria; and (6) Antibiotic treatment group (AG) received 25 mg/kg Ciprofloxacin gavage after intraperitoneal injection of pathogenic bacteria. The medicine was instilled once a day for seven days.

2.5. Pathological Examination

Observe and record the physiological state of mice, anesthetize the mice after the completion of treatment, and observe the gross lesions of mice by dissection. The organs of mice were fixed with 4% formaldehyde, embedded in paraffin, sectioned and stained with H&E, and examined microscopically to observe the histopathological changes in the organs of mice.

2.6. Gut Microbial Metagenome Sequencing and Scanning Electron Microscopy of Intestinal Tissues

Total DNA was collected from mice feces and sequenced, and the sequencing results were analyzed by bioinformatics to investigate the effects of drugs on mice’s intestinal microorganisms. Library construction and quality assessment were carried out using the PacBio and Illumina platforms. Sequencing was performed on the PacBio Sequel and Illumina NovaSeq PE150 systems once the library quality was confirmed. The genome assembly was executed using SOAPdenovo2 software (https://sourceforge.net/projects/soapdenovo2/support, Version 2.0), and the assembly results were refined using GapCloser software (https://sourceforge.net/, Version 1.1) to achieve a high-quality bacterial genome sequence. Microbial community richness was explored by analyzing the α&β diversity index (Shannon index, Chao index, and PCoA analysis) of mice intestinal contents. Mice intestinal tissues were collected, and the following steps were performed sequentially: glutaraldehyde (2.5%) fixation, ethanol gradient dehydration, drying and gold plating. The samples were processed, and the intestinal villus changes were observed using field emission scanning electron microscopy.

2.7. Network Pharmacology

2.7.1. Intersection Target Acquisition

Targets corresponding to drugs were obtained from the TCM Systematic Pharmacology Database and Analysis Platform (TCMSP, https://www.tcmsp-e.com/index.php), DrugBank (https://go.drugbank.com), and Swiss Target Prediction (http://www.swisstargetprediction.ch/). Key targets at the intersection of drug and disease were obtained using the Venny website (https://bioinfogp.cnb.csic.es/tools/venny/index.html, accessed on 28 May 2025). The cross-targets obtained in the above steps were entered into the “Multiprotein” module of the STRING database (https://cn.string-db.org/) to retrieve the protein–protein interaction (PPI) networks of the target genes. The PPI networks were then imported into Cytoscape software (version 3.9.1) and analyzed using the cytoHubba module to construct PPI networks of the cross-targets.

2.7.2. Gene Function and Pathway Enrichment Analysis

The cross-targets were entered into the Enrichr database (https://maayanlab.cloud/Enrichr/, accessed on 5 June 2025) and DAVID database (https://davidbioinformatics.nih.gov/tools.jsp, accessed on 5 June 2025) for gene ontology (GO) biological enrichment analysis, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis and Reactome enrichment analysis. Among them, GO biological enrichment analysis included biological process (BP), cellular component (CC), and molecular function (MF). p < 0.05 was used as the screening criterion, and the top-ranked items of the analysis results were used to construct visualized data plots.

2.8. Total RNA Extraction and RT-PCR

Molecular test validation based on network pharmacology analysis. Mice’s intestinal tissues were collected, RNA was extracted using a genome extraction kit, and RNA was reverse transcribed into cDNA by a reverse transcription kit after the RNA concentration was balanced by a micro-nucleic acid analyzer. Then, the mouse TNF-α, IL-6 and IL-1β primers were designed on the NCBI website, and GAPDH was used as a control to conduct RT-PCR (Primer information: Table 1), to explore the effect of LUT on the transcription of inflammatory factors in mice. Perform relative quantification using the 2DDCT method.

Table 1. Primer information.

Cytokine		Primer Sequence	Product Length (bp)
IL-1β	NM_008361.4	Forward primer: 5′-ATCTCGCAGCAGCACATCAA-3′	209
		Reverse primer: 5′-ATGGGAACGTCACACACCAG-3′
IL-6	NM_001314054.1	Forward primer: 5′-TCTTGGGACTGATGCTGGTG-3′	213
		Reverse primer: 5′-CTGCAAGTGCATCATCGTTGT-3′
TNF-α	NM_001278601.1	Forward primer: 5′-CAGTTCTATGGCCCAGACCC-3′	227
		Reverse primer: 5′-TAGCAAATCGGCTGACGGTG-3′
GAPDH	NM_001289726.2	Forward primer: 5′-GAGAGTGTTTCCTCGTCCCG-3′	201
		Reverse primer: 5′-ACTGTGCCGTTGAATTTGCC-3′

2.9. Immunohistochemical Staining

Mice intestinal tissues were embedded and sectioned, and incubated with antibodies, stained with hematoxylin and examined at 100× magnification. TNF-α and IL-6 were localized in the cell membrane or cytoplasm. The proportion of positive cells in each high magnification field (×100) and the intensity of cell coloration were scored using the semi-quantitative integration method [26]. Color intensity score: no coloration was scored as 0, light yellow coloration as 1, brown coloration as 2, and tan coloration as 3. Proportion of positive cells scored: <1% as 0, 1–25% as 1, 26–50% as 2, and >50% as 3. Immunohistochemical score obtained by adding color intensity and percentage of positive cells. Score sum 0–1 expression negative, 1–2 weakly positive (+), 3–4 positive (++), 5–6 strongly positive (+++) [27]. All observations and ratings of the samples were conducted independently by two experienced researchers in a blinded manner.

2.10. Molecular Docking and Molecular Dynamics Simulation of Key Targets

Important targets of the PPI network and key proteins of the pathway, TNF-α (2AZ5), IL-6 (1ALU), IL-1β (1IOB), NF-κB (1IKN) and TP53 (1UOL) were selected as receptors for molecular docking with the drug. AutoDock 4.2 software was used to simulate the possible combination of the drug and Protein ligand. The 3D structure of the drug was obtained from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/), and the structure was optimized by using Sybyl X 2.0 software, Tripos force field. The crystal structure of the protein ligand was obtained from the protein database (https://www.rcsb.org/, accessed on 16 June 2025), and the dehydration, hydrogenation and Gasteiger charge treatment were carried out by using PyMOL 2.5 software. The grid of protein ligand active center was set to 60 Å × 60 Å × 60 Å, the grid spacing was 0.375 Å, and the number of runs was 100. The Lamarckian genetic algorithm (LGA) algorithm was selected, the other parameters defaulted, the treated protein ligand docked with LUT, and the conformation with the lowest binding energy and the most docking times was selected for further analysis.

The protein with low binding energy in the molecular docking test was targeted, and an important PPI network was selected as a ligand to carry out a molecular dynamics simulation test with LUT. Molecular dynamics simulations (MD) were performed using the Gromacs2022 program with the GAFF force field for small molecules, AMBER14SB force field for proteins and TIP3P water model, merging the files of proteins and small-molecule ligands to construct the simulation system of the complex. During MD simulations, all involved hydrogen bonds were constrained using the LINCS algorithm with an integration step of 2 fs. Electrostatic interactions were calculated using the (Particle-mesh Ewald) PME method with the cutoff value set to 1.2 nm. The cutoff value for non-bonded interactions was set to 10 Å and updated every 10 steps. The simulation temperature was controlled to be 298 K using the V-rescale temperature coupling method, and the pressure was controlled to be 1 bar using the Berendsen method. 100 ps of NVT and NPT equilibrium simulations were performed at 298 K. A 100 ns MD simulation was performed for the complex system, and the conformation was saved every 10 ps. After the simulations were completed, the simulation trajectories were analyzed using VMD and PyMOL, and MMPBSA binding free energy analysis between the protein and the small molecule ligand was performed using the g_mmpbsa program.

2.11. Statistics and Analysis

The above trials were conducted at least three times. Statistical analyses were performed using SPSS version 22.0 (IBM Inc., Chicago, IL, USA). Data were analyzed by a t-test (between two groups). Statistical graphs were generated using Prism version 5.0 (GraphPad software Inc., La Jolla, CA, USA). p < 0.05 indicated statistically significant differences.

3. Results

3.1. Results of Drug Screening Against MDR-E. coli

Through bioinformatics screening, 12 drugs with a high comprehensive ranking of potential inhibitory effect on MDR-E. coli infection were obtained, including four natural plant compounds and eight antibiotic and other small molecule compounds. LUT had the highest comprehensive ranking with gene ratio = 137.19, p-value = 1.60 × 10⁻⁴⁵ and number of genes = 51, so LUT was selected as a potential therapeutic drug to be tested (Figure 1).

Figure 1. Results of drug screening against MDR-E. coli. The picture shows the visual display of bioinformatics screening results and the chemical structure of drugs.

3.2. Screening Drug In Vitro Test Results

3.2.1. LUT Inhibits the Growth of MDR-E. coli and Biofilm Formation

The results of the minimum inhibitory concentration assay showed that the minimum inhibitory concentration of LUT against MDR-E. coli was 1 mg/mL, and the minimum bactericidal concentration was 2 mg/mL (Figure 2C), and the formation of the biofilm was significantly inhibited by 0.25 mg/mL of LUT (Figure 2D). Scanning electron microscopy results showed that MDR-E. coli underwent significant crumpling accompanied by efflux of cellular contents at 1 mg/mL LUT concentration (Figure 2B).

Figure 2. Comprehensive exploration results of LUT antibacterial effect. (A) LUT, a natural active substance derived from various plants. (B) Electron microscope results of the antibacterial effect of LUT (1: LUT concentration = 0 mg/mL, 2: LUT concentration = 0.5 mg/mL, 3: LUT concentration = 1 mg/mL). (C) MIC test results. (D) Results of biofilm inhibition assay (**: compared with NC, ##: compared with BC). (E) Results of bacterial AKP activity assay (same letter: p > 0.05, different letters: p < 0.05). (F) Results of ATPase activity assay. (G) Results of resistance and virulence gene inhibition. *: p < 0.05, **: p < 0.01, #: p < 0.05, ##: p < 0.01.

3.2.2. LUT Causes Leakage of MDR-E. coli Cell Contents

The extracellular AKP of the control group was maintained at about 0.05 Kings/100 mL, and after treatment with different concentrations of LUT, the bacterial extracellular AKP was significantly increased (p < 0.05) and increased with the increase of treatment time and LUT concentration (Figure 2E).

3.2.3. LUT Inhibits MDR-E. coli ATP Synthesis

The ATPase activity of the control bacteria was between 0.8 and 1.0 U/mg, and after treatment with different concentrations of LUT for 3 h, the bacterial ATP production was significantly reduced (p < 0.05), and the inhibition effect was more obvious with the increase of drug concentration (Figure 2F).

3.2.4. LUT Inhibits the Expression of Drug Resistance Genes and Virulence Genes

Treatment with LUT at a concentration of 0.5 mg/mL showed a significant reduction (p < 0.05) in the expression of pathogen resistance genes tolC, ant(3″)-Ia and virulence genes fliC, K99 and stx1 (Figure 2G).

3.3. LUT Reduces Inflammation Caused by MDR-E. coli Infection

Mice in the model group showed significant pathologic responses, including decreased feeding, weight loss, diarrhea, and other clinical signs. Gross autopsy results showed different layers of lesions in the viscera of the mice, including thinning of the intestinal wall, enlargement of the spleen, and an increase in the viscera/body weight ratio (Figure 3B–D). Histopathological results showed significant epithelial cell detachment, hemorrhage, and inflammatory cell exudation in the small intestinal tissues of the mice in the model group. The treatment of LUT significantly improved the pathological changes in the mice, and the effect got better with the increase of the therapeutic dose (p < 0.05) (Figure 3E).

Figure 3. Animal model test results. (A) Experimental procedure. (B) Results of body weight changes in mice. (C) Results of gross anatomy of the mouse intestine and spleen. (D) Visceral coefficients of liver, spleen and lungs. (E) Results of H&E staining of mouse intestines (1: NC, 2: PG, 3: LT, 4: MT, 5: HT, 6: AG). Indicated by the arrow: Obvious changes in tissue structure, pathological manifestations such as hemorrhage, etc. All bar graphs are PG compared to NC and individual treatment groups compared to PG. *: p < 0.05, **: p < 0.01, #: p < 0.05, ##: p < 0.01, ns: p > 0.05.

3.4. LUT Improves Gut Microbes and Promotes Healing of Intestinal Villi in Mice

The results of mouse fecal macro-genome sequencing showed that the Shannon, chao and PCoA indices of gut microorganisms were significantly lower (p < 0.05) in the model group compared with the control group, and the species diversity of gut microorganisms in mice was significantly higher (p < 0.05) after LUT treatment, but still lower than that in the control group (Figure 4(A1–A3)). Compared to the control group, the model group had a decrease in gut microbes such as E. faecalis, which are currently considered beneficial, and an increase in E. coli and P. aeruginosa. LUT treatment reversed this process (Figure 4B). The results of scanning electron microscopy showed that the microvilli of the intestinal tract of control mice were arranged densely and neatly; the microvilli of the small intestinal tissues of the model group of mice were detached, fragmented, and disorganized; the pathological symptoms of the treatment group were significantly improved, but the phenomenon of enlarged microvilli gaps still existed (Figure 4C–E).

Figure 4. Macrogenomics of intestinal microorganisms and results of scanning electron microscopy of intestinal structures. (A,B) Visualization map of intestinal microbial species diversity index (1: Shannon index, 2: Chao index, 3: PCoA index), *: p < 0.05. (C) Observation results of the intestinal scanning electron microscope in NC. (D) Observation results of the intestinal scanning electron microscope in PG. (E) Observation results of the intestinal scanning electron microscope in mice in MT. Arrows: abnormal intestinal villus morphology.

3.5. Network Pharmacology Results

3.5.1. Presentation of Results for Important Targets

A total of 150 LUT targets and 2696 E. coli diarrhea targets were obtained online, with 105 intersecting targets identified after cross-analysis (Figure 5A). The protein–protein interaction (PPI) network showed that the important intersecting proteins were TNF-α, IL-6, AKT1, IL-1β, and TP53 (Figure 5B).

Figure 5. Network Pharmacology Results. (A) Garget intersection Venn plot, (B) Intersection target interactions plot. (C) GO biology enrichment plot (Bars represent p-values and gene counts). (D) Reactome enrichment analysis results. (E) KEGG enrichment analysis results. (Bubble size represents the number of enrichment targets, and bubble color represents the p-value).

3.5.2. Biology Enrichment Analysis and Signaling Pathway Enrichment Analysis Results

GO enrichment analysis showed that key targets were mainly enriched in 490 Biological process (BP), 69 Cell component (CC), and 130 Molecular function (MF). BP mainly included response to xenobiotic stimulus, inflammatory response, response to lipopolysaccharide, positive regulation of gene expression, negative regulation of apoptotic process, etc; CC included extracellular space, cytosol, mitochondrion, endoplasmic reticulum lumen, etc; MF included enzyme binding, identical protein binding, nuclear receptor activity, etc (Figure 5C). The top-ranked signaling pathways in KEGG enrichment analysis were the TNF signaling pathway, IL-17 signaling pathway, Pathways in cancer, PI3K-AKT signaling pathway, etc (Figure 5D). The top-ranked pathways in the Reactome enrichment analysis were Signaling by Interleukins, Cytokine Signaling in Immune System and PI3K-AKT Signaling in Cancer, etc (Figure 5E).

3.6. Molecular Test Validation Results

3.6.1. Differential Expression of mRNA for Inflammatory Factors

Compared with the control group, the mRNA expressions of IL-1β, IL-6 and TNF-α in the model group increased significantly (p < 0.05). Compared with the model group, the mRNA expressions of IL-1β, IL-6 and TNF-α in the treatment group were significantly decreased (p < 0.05), but the expression of TNF-α in the low-dose LUT group was not significantly different from that in the model group (p > 0.05). Among all treatment groups, the inhibition of inflammatory factor mRNA expression was best in the medium-dose LUT treatment group (Figure 6A).

Figure 6. Molecular experiments to verify the results. (A) RT-PCR results. (B) Immunohistochemical staining results (TNF-α, 1–5: NC, PG, LT, MT, HT; IL-6, 6–10: NC, PG, LT, MT, HT). (C) Visualization of immunohistochemical results. Compared to NC, **: p < 0.01, Compared to PG, ##: p < 0.01. ns: p > 0.05.

3.6.2. Altered Protein Expression in Mice Tissues

Immunohistochemistry results showed that there were large brownish-yellow deposits in the liver of the model group, and the brownish-yellow deposits were significantly reduced in the treatment group, and the positive areas were smaller with the increase of the treatment dose (Figure 6B). The mean scores of TNF-α and IL-6 were C-group (2.2, 1.64), P-group (8.39, 6.47), L-AE (4.16, 4.14), M-AE (3.23, 2.79), and H-AE (2.9, 2.86), respectively (Figure 6C).

3.7. Molecular Docking and Molecular Dynamics Results of LUT with Important Target Proteins

The binding energies of LUT to each target were TNF-α (−8.1 kcal/mol), IL-6 (−7.2 kcal/mol), IL-1β (−7.2 kcal/mol), NF-κB (−7.6 kcal/mol) and TP53 (−7.8 kcal/mol), respectively (Figure 7A–F). The results show that LUT has good binding ability to each target, in which van der Waals forces, hydrophobic forces and electrostatic interactions play important linking roles. LUT and TNF-α molecular dynamics results showed that the Root Mean Square Deviation (RMSD) of the complex structure and the RMSD of the protein remained basically stable as the simulation proceeded, which indicated that the structure of the complex remained basically stable. The trajectories of the complexes in the steady state were selected by combining RMSD, Rg, Distance, Buried SASA, and interaction energies under consideration of solvation energies and were calculated using the Molecular Mechanics-Poisson Boltzmann Surface Area (MM-PBSA) method to obtain the binding energy-related energy terms. The results show that the binding energy and affinity of the two are high. The number of hydrogen bonds between small molecules and proteins basically kept fluctuating between 0–1 (Figure 7G).

Figure 7. Molecular docking and molecular dynamics results. (A) TNF-α, (B) IL-6, (C) IL-1β, (D) NF-κB, (E) TP53, (F) binding energy, (G) Results of molecular dynamics simulations of LUT with TNF-α. 1: Drug-target protein binding space conformation. 2: RMSD of the complex, protein and small molecule ligand; 3: Rg of the complex; 4: RMSF of the protein in the complex; 5: Spacing of the protein–small molecule binding site (Dock site-ligand); 6: Encapsulated area of the small molecule-protein relationship (Buried SASA).

4. Discussion

LUT is widely distributed in nature and can be isolated from a variety of natural herbs and vegetables [28]. As a natural flavonoid, LUT has a wide range of anti-inflammatory, antioxidant and anti-tumor effects [29]. Pharmacokinetic studies show that LUT has an oral utilization of 26.6%, with the gastrointestinal tract as the main site of absorption, and enters the bloodstream as a prototype component for wide distribution in the body [30]. Despite its low systemic circulation concentration, LUT may accumulate to higher concentrations locally in the gut, particularly when administered in appropriate formulations such as nanomedicines, microencapsulated preparations, or in combination with bioavailability enhancers. In recent years, multiple studies have focused on enhancing the solubility, stability, and intestinal retention time of LUT. For instance, nanocarriers (such as liposomes and polymeric nanoparticles) and self-emulsifying microemulsion systems have been demonstrated to significantly improve the oral bioavailability of LUT while maintaining elevated local concentrations in the gut [31,32]. It has been shown that LUT can help prevent and treat ulcerative colitis by exerting immunomodulatory and antioxidant effects, repairing intestinal mucosal damage, protecting the intestinal mucosal barrier, regulating related signaling pathways (including the MAPK signaling pathway, TLR/NF-κB signaling pathway, JAK/STAT signaling pathway, endoplasmic reticulum stress, and autophagy pathway), and regulating the balance of intestinal flora [33,34,35]. The potential of LUT in antimicrobials has been underestimated compared to the studies of LUT in anti-inflammatory and anti-tumor areas. Modern pharmacological studies have shown that LUT can inhibit the growth of a wide range of pathogens, such as Staphylococcus aureus, Proteus mirabilis, herpesvirus, leishmania, etc. [36,37,38]. As a natural plant extract, LUT has various advantages, such as low toxicity, low cost, and is not easy to develop drug resistance, which makes it an excellent antimicrobial compound in the global environment of reducing or banning the use of antibiotics. However, more in-depth studies on the therapeutic efficacy and pharmacological mechanism of LUT against MDR-E. coli infections are still needed.

The experimental outcomes of this study provide robust evidence supporting the dual antimicrobial and immunomodulatory roles of LUT against MDR-E. coli. Specifically, the in vitro assays demonstrated that LUT not only exerted direct bactericidal effects by disrupting biofilm formation and inhibiting ATP synthesis, but also significantly downregulated key resistance and virulence genes such as tolC, ant(3″)-Ia, fliC, and K99. These findings are directly linked to the proposed mechanisms by which LUT counteracts bacterial resistance, including efflux pump inhibition and virulence attenuation. Furthermore, the in vivo results corroborated the anti-inflammatory efficacy of LUT, as evidenced by reduced expression of pro-inflammatory cytokines (IL-1β, IL-6, TNF-α) and ameliorated intestinal pathology. The restoration of gut microbial diversity and intestinal villi integrity further aligns with the network pharmacology predictions, which highlighted the involvement of interleukin and NF-κB signaling pathways. Collectively, these results form a coherent chain of evidence that bridges the observed phenotypic improvements with the underlying molecular and cellular mechanisms, thereby validating the proposed multi-target action of LUT in combating MDR-E. coli infections.

Currently, there are four main ways in which pathogenic bacterial resistance arises, which are the production of resistance genes (especially the expression of antibiotic efflux genes), the formation of biofilms (the formation of dense protective membrane structures between bacteria that prevent antibiotics from entering the interior of the cell), the mutation of the antibiotic target proteins (a change in specific conformation that results in the inability of the antibiotic to bind to the target proteins), and the production of antibiotic enzymes (e.g., carbapenem-resistant penicillinase) inactivates the antibiotic [39,40]. The results of this study show that LUT can play a role in inhibiting the growth of bacteria by inhibiting the production of ATP in drug-resistant bacteria, disrupting the biological periplasmic structure and cell wall and other processes. Meanwhile, LUT can also exert unique bacterial inhibitory activity by suppressing the expression of antibiotic efflux gene tolC, aminoglycoside gene ant(3″)-Ia and flagellar gene fliC, toxin factor K99 and other resistance and virulence genes of the pathogenic bacteria.

Entry of pathogenic bacteria into mice triggers a strong inflammatory and immune response, accompanied by the release of inflammatory factors (IL-1β, IL-6, TNF-α, etc.), which, if left untreated, may lead to systemic inflammatory response syndrome and multi-organ dysfunction [41]. As an important immune organ, the spleen plays an important role in the immune response of the body. When the body is invaded by pathogens, the spleen, as the largest secondary lymphoid organ, is responsible for filtering pathogens from the blood in its marginal zone, B cells differentiate into plasma cells to produce specific antibodies in the splenic white marrow, and macrophages in the red medulla clears the pathogens that have been conditioned, and the spleen enhances the systemic Inflammatory response [42,43]. As a result, mice in the model group showed significant changes in the internal organs, especially the immune organs. MDR-E. coli has a variety of virulence factors, including lipopolysaccharide, endotoxin, adhesin and exotoxin, etc. The intestinal tract, as an important colonization area for bacterium infection, is invaded by pathogenic E. coli, which secretes a variety of virulence factors, which have a toxic effect on the organism and cause changes in the intestinal microenvironment, further leading to pathological reactions such as imbalance of intestinal microorganisms and destruction of intestinal barriers in the body [44,45]. LUT treatment significantly improved the pathological symptoms, reduced the enlargement of visceral organs, and restored the intestinal microbial homeostasis and intestinal barrier integrity in the mice.

While an appropriate inflammatory response contributes to the clearance of pathogens, an uncontrolled inflammatory response may lead to severe damage to the organism. Network pharmacology results showed that tumor necrosis factor and interleukins are important targets for LUT to exert anti-inflammatory effects. GO analysis showed that LUT affects the function of cell organelles such as the nucleus, mitochondria and endoplasmic reticulum through binding to macromolecules and proteases, and then regulates biological processes such as apoptosis and proliferation. KEGG and Reactome signaling pathway prediction showed that the interleukin signaling pathway may be an important pathway in the regulation of LUT. Toll-like receptor 4 (TLR4) is a major sensor in the interleukin pathway recognizing pathogen-associated molecular patterns. Pathogenic bacterial infection-stimulated TLR4 activates nuclear factor-κB (NF-κB), which is involved in inflammatory responses, immunomodulation, stress and apoptotic pathways in vivo [46]. NF-κB plays an important role in regulating cytokine-induced gene expression. Activation of NF-κB has been shown to be associated with a variety of diseases, including arthritis, peritonitis, colon cancer, rheumatoid arthritis, and Alzheimer’s disease [47]. Inhibition of NF-κB expression provides a new approach to treat these diseases. Pathogenic bacterial infections cause increased expression of inflammatory factors in vivo, and inflammatory factors such as IL-1β, IL-6, and TNF-α stimulate phosphorylation of the P65 protein and the release of NF-κB from the dimer-bound state, which activates the NF-κB classical pathway [48,49]. RT-PCR results showed that mRNA expression of IL-1β, IL-6, and TNF-α in mice was significantly reduced after LUT treatment (p < 0.05). Molecular docking results of selected targets in the pathway also showed that LUT was tightly linked to key target proteins in the pathway through hydrogen bonding and hydrophobic interactions, which inhibited the production of inflammatory factors and attenuated the inflammatory response. However, the precise causal relationship requires further validation through subsequent intervention studies using pathway-specific inhibitors and receptor knockout models. The results of the above experiments showed that LUT has good anti-inflammatory activity and has good therapeutic effects on mice infected with pathogenic bacteria. Network pharmacology results indicate that the potential effects of LUT may not be mediated through generalized “cancer pathways” or broad inflammatory pathways. Instead, it likely exerts comprehensive therapeutic effects by synergistically regulating the aforementioned specific pathway networks closely associated with intestinal infections. This action occurs across multiple levels, including inhibiting bacterial virulence, mitigating excessive inflammatory responses, promoting epithelial repair, and maintaining microbial balance. Enriched pathways encompassing pathogen recognition, immune effect, and barrier repair modules support this interpretation.

This comprehensive study suggests that LUT, as a natural plant compound, has great potential for application in the current era of increasing antibiotic resistance (Visualization of antibacterial and anti-inflammatory mechanism of LUT in this study, Figure 8). It is worth noting that although LUT has extremely low toxicity (oral LD50 > 5 g/kg in rats) [50], there is still a lack of data from human clinical trials to validate its specific efficacy, and thus, further studies are needed to investigate the effects of LUT as a drug therapy or food additive in human diseases. Meanwhile, although this study confirms that LUT inhibits MDR-E. coli, modulates immunity, and restores microbiota in animal models; its long-term safety and potential effects on commensal microorganisms require systematic evaluation before translating into practical applications. Future research should follow these key steps: First, conduct long-term toxicity studies across multiple species to evaluate their potential effects on gut microbiota homeostasis, immune balance, and organ function. Second, perform targeted pharmacokinetic studies to determine its effective exposure concentration in the intestine. Third, develop gut-targeted or sustained-release formulations to enhance local efficacy and reduce systemic exposure. Finally, phased randomized controlled clinical trials must rigorously validate their safety, efficacy, and appropriate dosage in target animal models or human populations. Only through such systematic evaluation can the prudent application of LUT as an antibiotic alternative be scientifically advanced.

Figure 8. Schematic diagram of the mechanism by which luteolin regulates MDR-E. coli infection. The figure shows the effects of luteolin in inhibiting MDR-E. coli infection through multiple pathways, including in vitro inhibition of pathogen growth and the formation of drug resistance, as well as an analysis of anti-inflammatory mechanisms in vivo.

5. Conclusions

This study confirms that LUT combats MDR-E. coli infection through the following dual-action mechanism: (1) Direct antibacterial action—disrupting biofilms, inhibiting ATP synthesis, and downregulating resistance genes (tolC, ant(3″)-Ia) and virulence genes (fliC, K99, stx1); (2) Immunomodulatory and anti-inflammatory effects—alleviating intestinal inflammatory responses by regulating interleukin-related signaling pathways (e.g., TNF-α/IL-6/IL-1β). In infected mouse models, LUT not only alleviated clinical symptoms and intestinal pathological damage but also significantly restored gut microbiota diversity, promoted proliferation of beneficial bacteria (e.g., Enterococcus faecalis), and repaired intestinal barrier structure. Network pharmacology and molecular docking analyses further validated its multi-target mechanism involving key biological processes including inflammatory response, immune regulation, and apoptosis. As a natural flavonoid compound, LUT offers advantages including broad availability, high safety profile, and low potential for inducing drug resistance. It demonstrates significant application potential in reducing antibiotic use and controlling the spread of multidrug-resistant bacteria within the food chain, positioning it as a candidate substance for feed additives or veterinary clinical adjunctive therapies. However, this study has certain limitations: Current findings are based on animal models, and their actual efficacy and safety in humans require further validation through clinical trials. Optimization of LUT’s bioavailability and local intestinal concentration necessitates formulation support. Potential long-term effects on gut microbiota homeostasis also warrant systematic evaluation. Future research should focus on pharmacokinetic optimization, formulation improvements, and cross-species clinical translation trials to advance its scientific application as an antibiotic alternative strategy.