Targeting metabolism and membrane: Neferine synergizes with gentamicin against Escherichia coli by disrupting FNR/ArcA-mediated homeostasis

ABSTRACT

The escalating prevalence of antimicrobial-resistant bacteria, particularly in food and environmental settings, poses a severe global health threat. E. coli with rising resistance to gentamicin exemplifies this crisis, necessitating novel strategies to restore antibiotic efficacy. This study identified Neferine, a bisbenzylisoquinoline alkaloid from Nelumbo nucifera, as a potent synergistic adjuvant for gentamicin against E. coli. Integrating high-throughput screening with bioinformatic and molecular biological approaches, we systematically investigated its synergistic efficacy and underlying mechanism. Our results demonstrated that Neferine, while lacking intrinsic bactericidal activity, dramatically enhanced gentamicin’s potency, reducing its MIC by ≥16-fold, accelerating bactericidal kinetics, and delaying resistance development in vitro. Mechanistically, Neferine specifically targeted the global transcriptional regulators FNR and ArcA with high affinity (K_D = 1.3 μM and 658.6 nM, respectively), inducing conformational changes and disrupting their regulatory networks. This interaction hyper-activated the TCA cycle, leading to NAD⁺/NADH imbalance, severe ATP depletion, and ROS burst, ultimately causing metabolic catastrophe. Concurrently, Neferine synergized with gentamicin to severely disrupt bacterial membrane integrity, increasing fluidity and provoking rapid depolarization. In vivo validation demonstrated that the combination significantly enhanced host survival, reduced bacterial loads in organs, and attenuated excessive inflammatory responses.In conclusion, Neferine synergizes with gentamicin through a dual mechanism: targeting FNR/ArcA to induce metabolic collapse and co-disrupting membrane integrity, offering a promising therapeutic strategy to combat multidrug-resistant E. coli infections.

GRAPHICAL ABSTRACT

Introduction

The persistent spread of antibiotic resistance has become a major global challenge threatening public health, food safety, and ecological sustainability [1–3]. The World Health Organization (WHO) has listed antimicrobial resistance (AMR) as one of the top ten public health crises worldwide [4,5]. Particularly concerning is the deep penetration of the resistance transmission chain into the integrated “human-animal-environment” One Health system [6–8]. It is projected that by 2050, drug-resistant infections could cause approximately 40 million deaths annually [9]. Among Gram-negative bacteria, Escherichia coli (E. coli) is a conditionally pathogenic organism widely present in the human intestinal tract, food supply chains, farming environments, and natural water bodies [10,11]. When host immunity declines or strains acquire specific virulence factors, E. coli can breach intestinal barriers, leading to various infectious diseases such as urinary tract infections, sepsis, and gastroenteritis [12,13]. In developing countries, toxigenic E. coli is the primary pathogen responsible for acute watery or bloody diarrhea in infants, young children, and travelers, causing hundreds of millions of cases and hundreds of thousands of deaths each year [14]. Even more alarming is the continuous rise in resistance rates among clinical and livestock-derived E. coli to commonly used antibiotics, including aminoglycosides, fluoroquinolones, and sulfonamides, due to their extensive and prolonged use. Drug-resistant E. coli circulates among humans and animals through the food chain, water bodies, and the fecal-oral route, not only leading to treatment failures and infection recurrences but also exacerbating the horizontal dissemination of resistance genes, forming a persistent resistance cycle that is difficult to break from a One Health perspective [15–17].

Since its introduction in the 1960s, the aminoglycoside antibiotic gentamicin (GM) has been widely used to treat acute intestinal infections, including those caused by E. coli, due to its rapid bactericidal activity against Gram-negative bacteria and low cost [18,19]. However, due to the rapid dissemination of genes encoding aminoglycoside‑modifying enzymes (AMEs), such as aac(3)-II, aac(6′)-Ib and ant(3″)-Ia, on plasmids and transposons, the resistance rate of clinical isolates to GM has been increasing annually [20,21]. This rise in resistance has significantly diminished the clinical efficacy of GM monotherapy, often necessitating higher doses which in turn increases the risks of adverse effects such as ototoxicity and nephrotoxicity [22–24]. Therefore, there is an urgent need to discover novel adjuvants capable of restoring or enhancing the antibacterial activity of GM and extending its clinical lifespan to alleviate the pressure of E. coli infections and delay the evolution of resistance [25].

Traditional Chinese medicine (TCM), with its long-standing clinical practice, offers a unique repository of medicinal resources and therapeutic experiences, providing a natural library of TCM-derived monomers for modern research on antimicrobial potentiation [26,27]. The characteristic “multi-component, multi-target, holistic regulation” approach of TCM-derived monomers offers unique advantages for the study of antibiotic synergy [27]. Modern phytochemical research has shown that TCM-derived monomers, such as alkaloids, flavonoids, and saponins, can significantly enhance the bactericidal efficacy of antibiotics through various mechanisms, including the inhibition of resistance enzymes, blockade of efflux pumps, interference with quorum sensing, and modulation of host immunity [28–30]. Nelumbo nucifera Gaertn is a traditional Chinese medicinal and edible plant. Its mature seeds, known as “lotus seeds” and green embryos, known as “lotus plumules,” possess significant medicinal value. The primary benzylisoquinoline alkaloid found in these parts, neferine (NEF), has been demonstrated to possess various pharmacological activities, including anti-inflammatory, antioxidant, anti-fibrotic, and cardiovascular protective effects [31–34]. However, whether NEF can serve as a GM adjuvant against E. coli infections, as well as its synergistic effects and underlying mechanisms, remains unexplored.

Therefore, we systematically evaluated the in vitro and in vivo antibacterial activity of the combination of NEF and GM against clinically resistant E. coli. Using transcriptomics and molecular biology techniques, we elucidated the synergistic mechanisms and explored the immunomodulatory effects of NEF from the perspective of host-pathogen interactions. This study is expected to provide novel strategies for extending the clinical lifespan of GM.

Materials and methods

Chemicals, bacterial strains, and growth conditions

The TCM-monomer library (L6810) and NEF (standard, purity ≥ 98%) used in this study were purchased from Taoshu Biotechnology Co., Ltd. (Shanghai, China). GM (analytical standard, purity ≥ 590 IU/mg) was obtained from Yuanye Biotechnology Co., Ltd.(Shanghai, China). The standard E. coli strain ATCC 25,922 was purchased from Shanghai Preservation of Microorganisms Co., Ltd., while the E. coli strains A25, B2, and C3 were preserved at −80°C in our laboratory. All chemicals were prepared as high-concentration stock solutions using dimethyl sulfoxide (DMSO), aliquoted, and stored protected from light at −20°C. Before use, they were diluted to the required working concentrations with sterile physiological saline or culture medium. Luria-Bertani broth (LB), Luria-Bertani agar (LBA), and Mueller-Hinton broth (MHB) media used in the experiments were all purchased from Rishui Biotechnology Co., Ltd. (Qingdao, China). Strain revival was performed by streaking on LBA plates, followed by activation at 37°C in a constant temperature incubator for 12–18 hours. Unless otherwise specified, all strains were cultured under standard conditions at 37°C with shaking at 170 rpm in an orbital shaker.

Antimicrobial susceptibility testing

The minimum inhibitory concentration (MIC) determination and combination susceptibility testing were performed using the microbroth dilution method recommended by the Clinical and Laboratory Standards Institute (CLSI). Briefly, E. coli was activated on LBA plates. A single colony was then inoculated into LB broth and incubated with shaking at 37°C until the logarithmic growth phase was reached. The bacterial suspension was adjusted to approximately 1 × 10⁸ CFU/mL using sterile MHB and further diluted 1000- fold to a final concentration of approximately 5 × 10⁵ CFU/mL, which served as the working inoculum.

For compound MIC determination, two-fold serial dilutions of the compound were prepared in broth within a 96-well plate. An equal volume of the working inoculum was added to each well, resulting in a final inoculum of approximately 5 × 10⁵ CFU/mL per well. For the combination susceptibility test: a two-fold serial dilution series of GM was similarly prepared in a 96-well plate. Subsequently, a final concentration of 200 µM of the TCM monomer compound was added to all wells except the growth control wells, followed by the addition of the working inoculum. All assays included bacterial growth control wells and medium-negative control wells. The 96-well plates were incubated statically at 37°C for 16–20 hours. The MIC results were visually observed and recorded, with the MIC defined as the lowest drug concentration at which no visible bacterial growth was observed. The outcome of the combination susceptibility test was determined by comparing the MIC value of GM in the presence of the fixed 200 µM TCM monomer compound with the MIC value of GM alone.

Growth kinetics assay

Bacterial growth trends were evaluated by monitoring the optical density (OD₆₀₀) of bacterial cultures using a NanoDrop 2000 spectrophotometer. Briefly, an overnight culture of E. coli A25 was subcultured at a 1:1000 ratio into fresh 5 mL LB tubes and incubated with shaking at 37°C until the logarithmic growth phase (OD_600 nm ≈ 0.5) was reached. This culture was then diluted 1000-fold to serve as the starting inoculum. Aliquots of this inoculum were added to 5 mL LB tubes containing different concentrations of NEF and to a drug-free medium control tube. The cultures were continuously incubated with shaking at 37°C, and the OD₆₀₀ was measured every 30 minutes over a 6-hour period. Each experimental group was performed in triplicate.

Time-kill kinetics

An E. coli A25 culture in the logarithmic growth phase was adjusted to a starting concentration of approximately 5 × 10⁵ CFU/mL using fresh Mueller-Hinton broth (MHB) and then incubated with NEF (200 µM), GM (1/2 MIC), or a combination of both. All treatment groups were incubated with shaking at 170 rpm and 37°C. Over a 12-hour period, samples from each group were serially diluted, plated on LBA plates, and incubated inverted at 37°C for 16–20 hours for colony counting. Synergy was defined as a ≥ 2 log₁₀ CFU/mL increase in killing by the drug combination compared to the most effective single drug. Antagonism was defined as a ≥ 2 log₁₀ CFU/mL reduction in killing by the combination compared to the most effective single drug. An indifferent effect was defined when the difference in killing between the combination and the single drugs was within 2 log₁₀ CFU/mL.

Resistance development

The influence of NEF on the development of acquired resistance was assessed through a continuous passage assay. A starting culture of E. coli A25 in the logarithmic growth phase was inoculated into either a sub‑inhibitory concentration of GM alone or a combination of sub‑inhibitory GM and NEF. All groups were incubated with shaking at 37°C. Beginning at the initial passage and daily thereafter, the MIC of GM against the strains in each experimental group was determined by the microbroth dilution method. The dynamic changes in the GM MIC over the 21‑day passage period were plotted and compared between the combination treatment group and the GM monotherapy group to evaluate the effect of NEF on inhibiting the development of GM resistance.

Scanning electron microscopy imaging

An E. coli A25 culture in the logarithmic growth phase was separately exposed to GM (1/2 MIC), NEF (200 µM), or a combination of both. A natural growth group served as the control. After co‑incubation for 4 h at 37°C, bacterial cells were collected by low‑speed centrifugation and washed three times with sterile PBS. The cell pellets were then fixed overnight at 4°C with 2.5% glutaraldehyde, followed by dehydration through a graded ethanol series. The prepared samples were dried, sputter‑coated with gold, and bacterial morphology was observed using an Extreme Resolution Analytical Field Emission SEM (Tescan, Czech Republic).

Molecular docking

The three‑dimensional structures of the transcription factors FNR (P0A9E5), ArcA (P0A9Q1), PhoB (P0AFJ5), NtrC (P0AFB8), PhoP (P23836), FlhD (P0A8S9), LysR (P03030), Fur (P0A9A9), FlhC (P0ABY7), IhfA (P0A6×7), Lrp (P0ACJ0), Nac (Q47005), DksA (P0ABS1), ArgP (P0A8S1), ArgR (P0A6D0), and IhfB (P0A6Y1) were obtained from the AlphaFold Protein Structure Database (https://alphafold.ebi.ac.uk/). Water molecules and original ligands were removed using PyMOL (https://pymol.org/2/), and hydrogen atoms were added to optimize the structures. The two‑dimensional structure of NEF (PubChem CID: 159,654) was retrieved from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/) and converted into a three‑dimensional model using Chem3D. Docking calculations were performed with the AutoDock Vina program (https://discover.3ds.com/discovery‑studio‑visualizer‑download), and the resulting complexes were visualized and analyzed using PyMOL.

Surface plasmon resonance assay

Recombinantly expressed ArcA and FNR proteins were immobilized separately on CM5 sensor chip channels. The NEF analyte was serially diluted in PBS‑T buffer to generate a concentration gradient and injected sequentially into the protein‑immobilized channels at a flow rate of 20 μL/min. All binding experiments were conducted at 22°C with a contact time of 100 s and a dissociation time of 180 s, repeated over eight concentration cycles.

RT‑qPCR

A single colony of E. coli A25 was inoculated into a 5 mL LB tube and incubated at 37°C with shaking at 170 rpm until the OD_600 nm reached approximately 0.3. GM (1×MIC), NEF (200 µM), or a combination of both was then added, and the cultures were co‑incubated for 6 h. Bacterial cells were harvested by centrifugation at 4°C and 8,000 × g for 5 min. Total RNA was extracted using an enhanced RNA extraction kit (TransGen Biotech, China). RNA quality and concentration were assessed with a NanoPhotometer N60 spectrophotometer. cDNA synthesis and subsequent qPCR were performed using the PerfectStart® Uni RT & qPCR Kit (TransGen Biotech, China). The 16S ribosomal RNA gene served as the internal reference gene. Relative expression levels were calculated using the 2^−ΔΔCT method. The experiment was independently repeated three times. Primer sequences are listed in Table S1.

Circular dichroism (CD) spectra analysis

Changes in the secondary structures of ArcA and FNR were investigated using a JASCO circular dichroism spectrophotometer (Jasco, Japan). Measurements were performed at room temperature (23°C). A 2 mL aliquot of PBS buffer was placed in a cuvette with a 0.1 cm pathlength as a blank control, and the wavelength range was set from 210 to 300 nm. Background interference was eliminated by subtracting the blank spectrum before acquiring data for the experimental samples. The concentrations of ArcA and FNR proteins were both 0.1 mg/mL, and the final concentration of NEF was 50 μM. For the binding groups, the protein‑NEF mixtures were incubated at room temperature for 10 min before measurement. CD spectra were recorded sequentially for ArcA alone, FNR alone, the ArcA‑NEF complex, and the FNR‑NEF complex. The acquired spectral scans were averaged and smoothed to reduce noise.

NAD⁺/NADH level assay

E. coli cultures in the logarithmic growth phase were subjected to the following treatments: a blank control, NEF alone (400 μM), GM alone (1/2 MIC and MIC), and a combination of NEF and GM. After incubation for 2 h at 37°C, bacterial cells were collected, washed with ice‑cold PBS, and then processed using an enhanced NAD⁺/NADH assay kit (Beyotime, China) to determine NAD⁺/NADH levels. Absorbance was measured at 450 nm using a microplate reader. The experiment was independently repeated three times.

Intracellular ATP determination

E. coli cultures in the logarithmic growth phase were co‑incubated with NEF (400 μM), GM (1/2 × MIC), or their combination for 2 h at 37°C. After washing with ice‑cold PBS, the bacterial cells were immediately placed in a boiling water bath to terminate metabolism and achieve complete cell lysis. Intracellular ATP levels were determined using an enhanced ATP assay kit (Beyotime, China). The experiment was independently repeated three times.

ROS level determination

E. coli cultures in the logarithmic growth phase were co‑incubated with NEF, GM alone, or their combination. After 2 h of incubation in the dark at 37°C, bacterial cells were collected and resuspended in PBS buffer containing 10 μM DCFH‑DA probe, followed by incubation in the dark at 37°C for 30 min to allow sufficient probe uptake. The cells were then washed twice with PBS to thoroughly remove extracellular probe, resuspended, and transferred to a black 96‑well plate. Fluorescence intensity was measured using a fluorescence microplate reader with excitation at 488 nm and emission at 525 nm. A sample treated with the probe but without drugs served as the baseline control. The relative fluorescence intensity of each experimental group compared to the control was used to quantitatively evaluate intracellular ROS accumulation under different treatment conditions. The experiment was independently repeated three times.

Membrane fluidity

The effect of NEF on membrane fluidity was evaluated using the fluorescent dye Laurdan. Briefly, an E. coli A25 culture with an OD₆₀₀ of 0.5 was incubated with 10 μM Laurdan in the dark at 37°C for 30 min. After washing, the cells were treated with GM (1/2 MIC or 1 MIC), NEF (200 μM or 400 μM), or their combination. Fluorescence intensities were measured at excitation 350 nm and emission wavelengths of 440 nm and 490 nm. Membrane fluidity was assessed by calculating the generalized polarization (GP) index according to the formula: Laurdan GP = (I₄₄₀ − I₄₉₀) / (I₄₄₀ + I₄₉₀).

Membrane depolarization analysis

Bacteria in the logarithmic growth phase (OD_600 nm = 0.5) were collected, washed with buffer, and then incubated in the dark at a constant 37°C with 10 µM of the membrane potential‑sensitive fluorescent probe DiSC3(5) (3,3′‑dipropylthiadicarbocyanine iodide; Sigma, Germany). The baseline fluorescence intensity was continuously monitored for 6 min using a fluorescence spectrophotometer set at excitation 622 nm and emission 670 nm. Subsequently, NEF (final concentration: 0, 200, or 400 µM), GM (final concentration: 1/2 MIC or 1 MIC), or a combination of both was added to the system. Fluorescence changes were further monitored for 24 min, with data recorded every 3 min.

Measurement of bacterial membrane permeability

Outer membrane permeability assay: E. coli cultures in the logarithmic growth phase were harvested by centrifugation at 8,000 × g for 5 min. The cell pellets were washed three times with PBS and resuspended to an OD600 of 0.3. The cultures were then treated with NEF (0, 200, or 400 μM), GM (0, 1/2 MIC, or 1 MIC), or their combination for 3 h. Subsequently, the cultures were incubated with N‑phenyl‑1‑naphthylamine (NPN) at a final concentration of 10 μM for 1 h. Fluorescence intensity was measured using a multifunctional microplate reader with excitation at 350 nm and emission at 420 nm. The experiment was independently repeated three times.

Inner membrane permeability assay: E. coli cultures in the logarithmic growth phase were harvested by centrifugation at 8,000 × g for 5 min. The cell pellets were washed three times with PBS and resuspended to an OD600 of 0.3. The cultures were then treated with NEF (0, 200, or 400 μM), GM (0, 1/2 MIC, or 1 MIC), or their combination for 3 h. Thereafter, the cultures were incubated with Propidium iodide (PI) at a final concentration of 3 μM for 1 h. Fluorescence intensity was measured using a multifunctional microplate reader with excitation at 535 nm and emission at 615 nm. The experiment was independently repeated three times.

Direct damage of purified genomic DNA by drugs

To evaluate whether NEF and GM directly damage E. coli genomic DNA, purified genomic DNA of E. coli A25 was first extracted using a bacterial genomic DNA extraction kit. Equal amounts of the purified genomic DNA were then incubated at 37°C for 3 h with the following treatments: control (no drug), GM (1/2 MIC), NEF (200 μM), NEF (400 μM), and the combination groups. After incubation, 20 μL of each reaction mixture was subjected to electrophoresis on a 0.8% agarose gel at 100 V for 30 min. DNA bands were visualized and photographed under a UV gel documentation system. The brightness and integrity of the DNA bands were used to assess the direct DNA‑damaging effect of the drugs.

DNA leakage assay

To evaluate whether the combined treatment of NEF and GM induces intracellular DNA leakage in bacteria, E. coli A25 cultures in the logarithmic growth phase were incubated with the following treatments: control, GM (1/2 MIC), NEF (200 μM), NEF (400 μM), and the combination groups. All treatment groups were incubated at 37°C with shaking for 3 h. After incubation, the bacterial cultures were centrifuged at 12,000 × g for 10 min at 4°C, and the supernatants were collected. Using the culture supernatant of the control group as a blank, the absorbance of the supernatants from each treatment group was measured at 260 nm using a spectrophotometer.

Determination of intracellular gentamicin accumulation

To evaluate the effect of NEF on the intracellular accumulation of GM, E. coli A25 cultures in the logarithmic growth phase were incubated with the following treatments: GM (1/2 MIC), GM (1 MIC), and the combination groups of NEF (200 μM or 400 μM) with GM (1/2 MIC or 1 MIC). All treatment groups were incubated at 37°C with shaking for 3 h. After incubation, bacterial cells were harvested and washed three times with ice‑cold PBS to remove extracellular residual drugs. The washed cell pellets were resuspended in PBS and lysed by three cycles of freeze‑thawing in liquid nitrogen. The lysates were centrifuged at 12,000 × g for 10 min at 4°C, and the supernatants were collected. The gentamicin concentration in the lysates was determined using an ELISA kit (Mlbio, Shanghai) according to the manufacturer’s instructions. The absorbance of each sample was measured at 450 nm using a microplate reader, and the gentamicin concentration was calculated accordingly.

Animals

Ethics statement and animal husbandry

All experimental procedures in this study were reviewed and approved by the Animal Welfare and Ethics Committee of Jilin Agricultural University (Approval No.: 20,230,314,003) and were strictly conducted in accordance with the committee’s guidelines. Female Kunming mice (KM), approximately eight weeks old and weighing 18–22 g, were purchased from Changsheng Biotechnology Co., Ltd. (Liaoning, China). Animal anesthesia was induced using a multifunctional gas anesthesia system (Bolutent Biotech Co., Ltd., China) with 3.5% isoflurane for approximately 3 min. Subsequent experimental steps were initiated only after complete loss of consciousness was confirmed. Mouse cages, water dispensers, and the experimental animal facility were pre‑sterilized with formaldehyde, and temperature and humidity in the animal room were maintained relatively stable throughout the study. For animal inclusion/exclusion: No specific inclusion or exclusion criteria were established a priori. All mice that met the general health and weight criteria were enrolled and randomly assigned to treatment groups. No animals were excluded from the study after enrollment. For data inclusion/exclusion: No exclusion criteria were applied to data points. All collected data were included in the final analysis. Randomization. Animals were randomly assigned to treatment groups using simple random allocation by drawing numbered tags from a container before the experiment.

The animal infection model was established using the clinically isolated pathogenic E. coli strain A25 [35]. In the Galleria mellonella and mouse infection models, the bacterial inoculum for survival analysis was set at the minimum lethal dose, whereas the inoculum for bacterial burden analysis was set at a sublethal dose. The drug doses were determined based on the pharmacokinetic profile of NEF and preliminary safety assessments.

Galleria mellonella infection model

Galleria mellonella larvae were purchased from Keyun Biotechnology Co., Ltd. (Henan, China). Healthy, active Galleria mellonella larvae weighing between 0.8 and 1 g were selected and randomly assigned to different treatment groups. After surface disinfection, each larva was injected with 10 μL of a PBS suspension containing 1 × 10⁹ CFU of E. coli A25 into the last pair of prolegs to establish the infection model. Two hours post‑infection, the larvae were administered via the same injection route with PBS buffer, GM monotherapy (10 mg/kg), NEF monotherapy (5 mg/kg), or combination therapy (10 mg/kg GM +5 mg/kg NEF). After treatment, the larvae were placed in sterile Petri dishes and incubated at 28°C in the dark. Survival status was observed and recorded daily for 5 days; larvae were considered dead when they showed no response to touch and displayed pronounced melanization.

For bacterial colonization analysis, larvae were infected with 10 μL of a suspension containing 1 × 10⁷ CFU of E. coli A25. At 0 and 12 h post‑infection, they received GM (10 mg/kg), NEF (5 mg/kg), or combination treatment. After 24 h of infection, the larvae were homogenized, serially diluted, and plated for bacterial enumeration.

Mouse infection model

For survival analysis, female KM mice were infected via intraperitoneal injection of 3 × 10¹⁰ CFU of E. coli A25 to establish a systemic infection model. Post‑infection, the mice were randomly divided into four groups (n = 10 per group): an infected model group, a GM monotherapy group (60 mg/kg), an NEF monotherapy group (10 mg/kg), and a combination therapy group (60 mg/kg GM +10 mg/kg NEF). This sample size is consistent with previous studies on antibiotic synergy in murine infection models. The respective treatments were administered by subcutaneous injection at 0, 12, and 24 h post‑infection. Survival was monitored continuously for 96 hours post-infection, and the survival status of each group was recorded.

In addition, a bacterial burden assay was performed to further evaluate the antibacterial efficacy of the treatments in infected mice. Female KM mice were infected intraperitoneally with 3 × 10⁹ CFU of E. coli A25 to establish a systemic infection, and then randomly assigned to four groups (n = 6 per group): an infected control group, a GM monotherapy group (60 mg/kg), an NEF monotherapy group (10 mg/kg), and a combination therapy group. Treatments were repeated every 12 h via subcutaneous injection. Forty‑eight hours post‑infection, blood samples were collected under deep anesthesia, after which the mice were euthanized by cervical dislocation. Liver, spleen, and kidney tissues were immediately harvested; part of each tissue was homogenized for bacterial colony counting, while the remaining portion was fixed in 4% paraformaldehyde for hematoxylin‑eosin (H&E) staining to assess pathological changes. Meanwhile, serum levels of the inflammatory cytokines TNF‑α, IL‑2, IL‑1β, and IL‑4 were measured using commercial ELISA kits (mlbio, China) according to the manufacturer’s instructions. A total of 70 mice were used in this study. Randomization. Animals were randomly assigned to treatment groups using simple random allocation by drawing numbered tags from a container before the experiment. Minimization of confounding factors. To minimize potential confounders, animals were randomly assigned to treatment groups using simple random allocation prior to the experiment. Cage positions were randomized on the rack and rotated daily. All experimental procedures were performed by the same investigator to ensure consistency. The primary outcome measure was mouse survival at 96 h post‑infection. Secondary outcomes included bacterial burden in the liver, spleen, and kidney tissues, serum levels of cytokines (TNF‑α, IL‑1β, IL‑2, and IL‑4), and histopathological changes in liver sections. Survival data were analyzed using the Kaplan‑Meier method. 95% confidence interval was calculated.

Hemolytic activity assay

Fresh rabbit blood was collected into centrifuge tubes containing 3.8% sodium citrate as an anticoagulant and mixed gently. After centrifugation at 1,000 × g for 10 min, the plasma was removed, and the pelleted red blood cells were collected and washed three times with sterile PBS buffer. The washed red blood cells were then diluted with PBS to prepare a 2% (v/v) red blood cell suspension. Different concentrations of NEF were added to the reaction mixture, and the suspensions were incubated statically at 37°C for 2 h. Normal saline was used as the negative control, and 0.1% Triton X-100 was used as the positive control. Following incubation, the samples were centrifuged at 1,800 × g for 10 min. The supernatants were transferred to a 96‑well plate, and the absorbance was measured at 540 nm using a microplate reader. The hemolytic index was calculated based on the absorbance values of each group.

Statistical analysis

All experimental data in this study are expressed as the mean ± standard deviation (SD) from three independent replicates. Statistical analysis was performed using GraphPad Prism software (version 10.1.2). Comparisons between two groups were conducted using the Student’s t‑test, and comparisons among multiple groups were performed by one‑way analysis of variance (ANOVA). A p-value < 0.05 was considered statistically significant.

Results

NEF potentiates the activity of GM against E. coli in vitro

Based on high-throughput screening of a Traditional Chinese Medicine monomer library, the synergistic antibacterial activity of each monomer with GM against E. coli strain A25 was evaluated at a fixed concentration of 200 μM (Figure 1(A)). The results indicated that NEF significantly enhanced the antibacterial effect of GM. Further determination of the MIC of GM in the presence of varying concentrations of NEF revealed that the MIC of GM against E. coli A25 decreased markedly (by ≥ 16‑fold) compared to GM alone, demonstrating a clear concentration-dependent synergistic effect of NEF (Figure 1(B)). Similar results were observed in clinically isolated multidrug-resistant E. coli strains B2 and C3, where NEF increased the susceptibility of these strains to GM.

Figure 1.

In vitro synergistic bactericidal activity of NEF with GM. (A) schematic of the screening workflow for traditional Chinese medicine monomers that potentiate GM againstE. coliin this study, created usinghttps://app.biorender.com. The MIC of GM againstE. coliA25 was 1024 μg/mL. (B) Fold change in the MIC of GM in the presence of different concentrations of NEF. (C) Growth kinetics ofE. coliA25 cultured in LB broth in the absence of NEF or in the presence of 200 μM or 400 μM NEF. (D) Time‑kill curves ofE. coliA25 showing the synergistic enhancement of antibacterial activity by the combination of GM (1/2 MIC) and NEF (200 μM). (E) Continuous passage ofE. coliA25 under sub‑inhibitory concentrations of GM alone or GM combined with NEF; MIC tests were performed each passage to monitor resistance development. (F) Scanning electron microscopy images ofE. coliA25 treated with GM (1/2 MIC), NEF (200 μM), or the combination (GM 1/2 MIC + NEF 200 μM).

To clarify whether the synergistic effect of NEF originated from its direct influence on bacterial growth, growth kinetics of E. coli treated with different concentrations of NEF alone were examined. The growth curves of NEF-treated groups showed no significant difference from the control (CTRL) group, with no notable inhibition in either growth rate or final cell density (Figure 1(C)). It is particularly noteworthy that in the MIC assay of NEF alone against E. coli, after excluding the effect of the maximum soluble concentration of the solvent DMSO, no inhibitory activity of NEF was detected. These findings confirm that NEF itself does not affect the normal growth of E. coli. Therefore, the observed enhanced antibacterial efficacy can be unequivocally attributed to the synergistic interaction between NEF and GM, rather than any intrinsic antibacterial activity of NEF.

To dynamically validate the synergistic bactericidal effect of NEF and GM, a time‑kill kinetic assay was performed. As shown in Figure 1(D), during the 12‑h monitoring period, the colony counts of the control group, the GM monotherapy group (1/2 MIC), and the NEF monotherapy group (200 μM) all increased over time with similar trends, indicating that neither drug alone effectively inhibited the growth of E. coli at these concentrations. In contrast, when the two drugs were combined, a pronounced synergistic bactericidal effect was observed: the viable bacterial count in the combination group fell below the detection limit within just 2 h of treatment, resulting in a sterile state. These results confirm that NEF enhances the killing rate of GM against E. coli and exhibits rapid synergistic bactericidal action.

NEF inhibits the development of GM resistance

To investigate whether NEF suppresses resistance development, a 21‑day in vitro continuous passage assay was conducted. In the group treated with a sub‑inhibitory concentration of GM alone, the MIC of E. coli against GM began to increase from day 9 and rose by 8‑fold by day 21 (Figure 1(E)). In contrast, in the group treated with the combination of GM and NEF, the MIC of E. coli against GM remained relatively stable throughout the 21‑day passage period, with no significant change. These results demonstrate that NEF not only enhances the bactericidal activity of GM but also effectively delays the acquisition of GM resistance under prolonged drug pressure.

Combination treatment induces synergistic structural damage

Morphological differences among E. coli cells under different treatments were observed by scanning electron microscopy (Figure 1(F)). The control group exhibited smooth, plump, and intact bacterial surfaces. Cells treated with either 1/2 MIC GM or 200 μM NEF alone showed only minor irregularities in morphology while largely maintaining relatively intact cellular contours. In contrast, treatment with the combination of GM and NEF induced severe and extensive structural damage: bacterial surfaces displayed pronounced shrinkage, collapse, and lysis; cell membrane integrity was substantially compromised; and cytoplasmic leakage was evident, collectively leading to accelerated bacterial death. These observations indicate that NEF significantly enhances the antibacterial activity of GM against E. coli in vitro.

The molecular network underlying NEF-mediated synergistic enhancement of GM antibacterial activity

To elucidate the molecular mechanisms by which NEF enhances the antibacterial effect of GM against E. coli, transcriptomic sequencing was performed following treatment with NEF. The raw sequencing data have been deposited in the Genome Sequence Archive (https://ngdc.cncb.ac.cn/gsa) under accession number GSA: CRA035022. Principal component analysis (PCA) revealed good clustering of biological replicates within each group and clear separation among different treatment groups, confirming the high reliability of the data and its suitability for subsequent differential expression analysis (Figure. S1).

Initially, we compared gene expression changes between the GM monotherapy group and the control group (Figure. S2A). The volcano plot identified 414 significantly upregulated and 273 significantly downregulated genes (p < 0.05). To preliminarily explore the functional and pathway distribution of these differentially expressed genes (DEGs), Gene Ontology (GO) enrichment analysis (Figure. S2B) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis (Figure. S2C) were conducted [36,37]. GO analysis, which describes the attributes and functions of genes and gene products, revealed that these DEGs were primarily enriched in biological processes such as “aerobic respiration,” “cytosolic large ribosomal subunit,” and “structural constituent of ribosome.” KEGG pathway analysis indicated that the genetic alterations induced by GM treatment were predominantly enriched in the ribosomal pathway, consistent with the known primary mechanism of action of GM [38,39].

Subsequently, we assessed the gene expression differences between the combination treatment group (GM + NEF) and the GM monotherapy group (Figure. S2D). Compared to GM alone, the combination treatment significantly upregulated 510 genes and downregulated 456 genes. These DEGs were mainly associated with transmembrane function and transporter-related pathways (Figure. S2E, S2F). Given the absence of intrinsic bactericidal activity of NEF, we hypothesized that NEF amplifies the gene regulatory effects of GM. Therefore, we performed integrated statistical analysis of the two comparative groups to identify key pathway genes. By screening for DEGs common to both comparisons and conducting KEGG pathway enrichment analysis, we retained pathways with a False Discovery Rate (FDR) <1 and identified eight candidate pathways (Figure 2(A)).

Figure 2.

NEF influences multiple biological pathways. (A) Groups: D, control group; Q, GM monotherapy group; JQ, GM + NEF combination group. The heatmap illustrates the expression patterns of the shared differentially expressed genes (DEGs) identified through comparative analysis between the combination group and the GM monotherapy group, and between the GM monotherapy group and the control group. (B) Transcription factor enrichment analysis of the remaining 20 common DEGs after filtration using cytoscape. (C) Molecular docking scores of NEF and GM with 16 transcription factor protein structures. (D) Interaction diagram showing hydrogen bonds formed between NEF and amino acid residues at the active site of FNR protein. (E) Interaction diagram showing hydrogen bonds formed between NEF and amino acid residues at the active site of ArcA protein. (F) Surface plasmon resonance (SPR) sensorgrams obtained when different concentrations of NEF (15.60, 31.30, 62.5, 125, 250, 500, 1000 nM) were flowed over the immobilized FNR protein chip surface. The increase in binding response (RU) with rising NEF concentration indicates concentration‑dependent specific binding. (G) SPR sensorgrams obtained when different concentrations of NEF (7.81, 15.60, 31.30, 62.5, 125, 250, 500 nM) were flowed over the immobilized ArcA protein chip surface. The rise in binding response (RU) with increasing NEF concentration demonstrates concentration‑dependent specific binding. (H – J) Relative proportions of secondary‑structure components of FNR protein, along with the circular dichroism (CD) spectra and relative compositional changes of FNR upon NEF binding. (K – M) relative proportions of secondary‑structure components of ArcA protein, along with the CD spectra and relative compositional changes of ArcA upon NEF binding.

The shared DEGs enriched in these eight pathways were subjected to protein-protein interaction (PPI) network analysis using the STRING database, with a minimum required interaction score of 0.700, to eliminate unrelated independent genes and further refine the screening (Figure. S3) [40]. The resulting PPI network was imported into Cytoscape, and three algorithms – CytoNCA, CytoHubba, and MCODE – were employed to perform key node analysis, identify densely connected modules within the network, and systematically evaluate the topological importance of nodes, respectively [41–43]. The intersection of genes identified by these three algorithms yielded 43 common key protein-coding genes (Figure. S4A). Transcription factor enrichment analysis of these 43 key genes identified 18 candidate transcription factors (Figure 2(B)).

NEF targets FNR/ArcA, induces conformational changes, and enhances GM susceptibility

We successfully obtained the three‑dimensional structures of 16 transcription factors for molecular docking simulations. The results demonstrated that NEF exhibited favorable binding potential to most of the tested transcription factors. Notably, NEF showed binding affinities of −9.7 kcal/mol with the global regulator FNR and −8.6 kcal/mol with ArcA, which were not only significantly strong but also higher than its binding affinities with GM (−9.2 kcal/mol for FNR and −8.0 kcal/mol for ArcA) (Figure 2(C)). FNR primarily regulates anaerobic respiratory enzyme systems, while ArcA senses redox status and modulates global metabolic pathways such as the TCA cycle [44,45]. Molecular docking visualization and surface plasmon resonance (SPR) further supported our hypothesis. NEF interacts with both FNR and ArcA proteins via hydrogen bonds, and the SPR response increased with rising NEF concentrations, yielding affinity constants (K_D) of 1.3 μM for FNR and 658.6 nM for ArcA (Figure 2(D–G); Figure. S4B – C).

To further validate the docking predictions and examine whether NEF binding induces conformational changes in the target proteins, circular dichroism (CD) spectroscopy was performed on FNR and ArcA. As shown in Figure 2(H–J), the presence of NEF altered the CD spectrum of FNR, causing a spectral shift and a marked reduction in β‑sheet content. Similarly, NEF binding modified the CD profile of ArcA, increasing the proportion of antiparallel structure in the complex (Figure 2(K–M)). These data confirm that NEF binding induces substantial changes in the structures of both FNR and ArcA.

To further validate that these two targets indeed participate in the synergistic effect, we overexpressed FNR and ArcA in the E. coli BL21 strain. The MIC of GM alone showed no change compared to the wild‑type strain. However, when 200 μM NEF was added to the culture medium of the overexpression strains, a clear synergistic effect was observed both before and after induction, with the MIC of GM significantly decreasing by 2‑ to 4‑fold (Figure 3(A)). This indicates that FNR and ArcA are not resistance targets for GM, but rather the potentiating targets of NEF. NEF exerts its synergistic effect by directly binding to the pre‑existing FNR/ArcA proteins and interfering with their regulatory functions, a finding consistent with the results presented in Figure 2(D–M).

Figure 3.

NEF acts through the ArcA and FNR transcription factors to modulate downstream gene transcription. (A) Effect of arcA and fnr gene overexpression on GM MIC under non‑induced and induced overexpression conditions. (B) Schematic illustration of the downstream regulatory genes corresponding to the FNR and ArcA transcription factors. (C‑F) Relative expression levels of each downstream transcriptional target gene in different treatment groups compared to the control group.

To verify at the transcriptional level whether FNR and ArcA are involved in the synergistic effect of NEF, we designed specific primers targeting the downstream genes of FNR and ArcA identified by transcriptomic analysis (Figure 2(B), 3(B), Table S1) and examined their expression changes under different treatments using RT‑qPCR. The results showed that, compared with the control group, combined treatment with NEF and GM significantly altered the transcript levels of the FNR- and ArcA‑regulated downstream genes (Figure 3(C–F)). These findings confirm our hypothesis that neferine binds to FNR and ArcA, thereby modulating the transcriptional levels of their downstream genes, which is highly consistent with the transcriptomic data.

NEF‑mediated metabolic collapse via FNR/ArcA synergizes with GM for bacterial killing

To elucidate the molecular pathway through which NEF affects bacterial energy metabolism via FNR and ArcA, key metabolic indicators were monitored. Measurement of the NAD⁺/NADH ratio revealed that NEF treatment alone lowered this ratio, whereas combination with GM significantly elevated it, indicating that the combined treatment hyper‑activates the TCA cycle and shifts the intracellular redox balance toward an over‑oxidized state (Figure 4(A,B)). ATP content measurement further revealed that NEF exacerbated the ATP depletion induced by GM. Compared with the single-agent treatment groups, the combination group exhibited the lowest ATP level, indicating that the synergistic action significantly disrupted bacterial energy homeostasis (Figure 4(C,D)). Different detection indicators exhibit varying sensitivity to drug treatment. The production of reactive oxygen species (ROS) often occurs in a “burst-like” manner. Consistent with our hypothesis, ROS detection showed that the combination of NEF and GM provoked a dramatic accumulation of ROS, significantly exceeding that in either monotherapy group (Figure 4(E,F)). These findings confirm that NEF, by targeting the FNR/ArcA regulatory nodes, hyper‑activates the TCA cycle, leading to excessive ATP consumption and massive ROS generation, ultimately triggering oxidative‑stress collapse and energy exhaustion in bacteria, thereby synergistically enhancing the bactericidal efficacy of GM.

Figure 4.

NEF affects bacterial oxidative stress processes and membrane fluidity. (A – B) Effect of the control, NEF (400 μM), GM (1/2 MIC), and their combination on bacterial NAD⁺/NADH levels. (C – D) Effect of the control, NEF (400 μM), GM (1/2 MIC), and their combination on bacterial ATP levels. (E – F) Changes in ROS following treatment with NEF alone or in combination with GM. (G – H) Changes in bacterial cell membrane fluidity following treatment with NEF alone or in combination with GM. (I) Schematic for monitoring bacterial transmembrane potential using DiSC3(5) under different treatment conditions. (J – K) Membrane depolarization induced by NEF, shown as fluorescence intensity values of DiSC3(5)‑loadedE. coliA25 cells before and after the addition of NEF, GM, or their combination.

NEF increases membrane fluidity and exacerbates depolarization

Given that scanning electron microscopy revealed pronounced morphological damage in bacteria treated with the drug combination, we hypothesized that this phenomenon might be associated with alterations in the physical properties of the cell membrane. To test this hypothesis, membrane fluidity was quantitatively assessed using fluorescence polarization by measuring the generalized polarization (GP) value of the Laurdan dye. The results showed that compared with the control group, NEF treatment alone caused a decrease in the GP value, which nevertheless remained above 0.25 (Figure 4(G)). Notably, after combined treatment with GM and NEF, the GP value decreased further and significantly (below 0.02), which was substantially lower than that of any monotherapy group (Figure 4(H)). This finding clearly demonstrates that the two drugs act synergistically to disrupt the integrity of the bacterial cell membrane, leading to abnormally increased fluidity and consequent severe dysfunction of membrane structure and function.

A close functional relationship exists between membrane fluidity status and membrane potential [46–48]. We further investigated the effect of NEF on membrane depolarization using the membrane potential‑sensitive probe DiSC3 (5) (Figure 4(I)). The results showed that after the addition of 200 μM or 400 μM NEF at 9 min, the fluorescence intensity (RFU) increased rapidly and significantly, indicating that NEF effectively induces bacterial membrane depolarization (Figure 4(J)). Interestingly, treatment with a sub‑inhibitory concentration (1/2 MIC) of GM alone also elicited a certain degree of membrane depolarization (RFU ≤ 110), suggesting that GM itself possesses some membrane‑perturbing activity [49]. However, when the two drugs were combined, the depolarization effect was markedly enhanced: the RFU value of the combination group rose rapidly to above 130, far exceeding that of the GM monotherapy group, indicating potent membrane‑disrupting capability (Figure 4(K)).

We further quantitatively assessed outer and inner membrane damage using NPN and PI fluorescent probes (Figure S5A-D) [50]. The results showed that NEF treatment alone significantly increased outer membrane permeability, with fluorescence intensity approximately 2–4 fold higher than that of the control group, while no obvious effect on the inner membrane was observed. Compared with the GM monotherapy group, the combination of NEF and GM led to a marked increase in both outer and inner membrane permeability, indicating severe disruption of the bacterial membrane structure, which is highly consistent with the morphological changes observed in our previous scanning electron microscopy analysis (Figure 1(F)).

NEF promotes GM accumulation via membrane disruption, not direct DNA damage

To investigate whether NEF directly damages bacterial genomic DNA, purified E. coli genomic DNA was incubated with NEF alone or in combination with GM [51]. Agarose gel electrophoresis showed that, compared with the control group, none of the treatment groups exhibited DNA smearing or reduced band intensity, indicating that neither NEF alone nor its combination with GM directly damages genomic DNA (Figure S5E-F). To further explore whether NEF treatment induces intracellular DNA leakage, E. coli cultures were treated with NEF, GM, or their combination, and genomic DNA was then extracted from the harvested bacterial cells. The results showed that no genomic DNA band was detected in the combination group, suggesting that extensive intracellular DNA leakage occurred following bacterial lysis, which prevented the recovery of intact genomic DNA (Figure S5G).

To validate these findings, we directly measured the DNA content in the extracellular culture supernatant. Normalized to the control group, the combination group exhibited the highest level of extracellular DNA leakage, reaching up to 60 ng/μL, which was significantly higher than that in the single-agent or control groups, further confirming that the combination treatment causes substantial leakage of bacterial intracellular contents (Figure S5H). Additionally, we measured the intracellular accumulation of GM. The results showed that the presence of NEF significantly increased the intracellular accumulation of GM. Compared with the GM monotherapy group, the combination group exhibited a markedly higher intracellular GM level, indicating that NEF promotes GM uptake by disrupting the bacterial membrane barrier, thereby enhancing its bactericidal efficacy (Figure S5I).

Based on the integrated experimental findings, we propose a mechanistic model for the synergistic action of NEF in enhancing GM against E. coli (Figure 5). NEF targets the global transcriptional regulators FNR and ArcA, binding with high affinity and altering their protein conformations, thereby disrupting their normal regulatory functions. This disturbance leads to hyper‑activation of the TCA cycle and uncoupling of oxidative phosphorylation, resulting in an imbalance of the NAD⁺/NADH ratio, depletion of ATP, and a burst of ROS, ultimately causing metabolic collapse in the bacteria. Simultaneously, NEF increases bacterial membrane fluidity and, in synergy with GM, induces rapid depolarization of the membrane potential, compromising membrane integrity and facilitating increased intracellular influx of GM. This establishes a positive‑feedback cycle of “membrane damage – enhanced drug accumulation – potentiated bacterial killing.”

Figure 5.

NEF enhances the efficacy of GM againstE. colithrough a dual‑hit strategy that simultaneously perturbs bacterial metabolism and disrupts membrane integrity. By impairing energy and redox homeostasis on the one hand, and by accelerating drug uptake through membrane damage on the other, this combined action leads to rapid bacterial killing.

NEF enhances the in vivo activity of GM

To evaluate the synergistic therapeutic effect in vivo, we first established a Galleria mellonella larval infection model (Figure 6(A)). In the survival assay, after infection with 1 × 10⁹ CFU of E. coli, all larvae in the control group and the 5 mg/kg NEF monotherapy group died within 60 and 72 h, respectively (Figure 6(B)). The 10 mg/kg GM monotherapy group provided complete protection during the first 60 h, but mortality gradually increased thereafter, resulting in a 40% survival rate at 120 h. In contrast, the combination of 10 mg/kg GM and 5 mg/kg NEF significantly prolonged larval survival; deaths first occurred only after 96 h, and the survival rate reached 70% at 120 h, indicating that the combination effectively improves infection outcome in vivo. To further quantify the antibacterial effect, bacterial burden in larvae was examined after sub‑lethal infection (1 × 10⁷ CFU). The results showed that the colony counts in homogenates from the combination group were significantly lower than those in the monotherapy groups, decreasing from approximately 10⁷ CFU/g to 10⁵ CFU/g (mean ± SD, n = 6)(Figure 6(C)).

Figure 6.

NEF effectively enhances GM activity in a Galleria mellonella infection model. (A) Experimental scheme for Galleria mellonella survival and bacterial colonization assays. (B) Survival of Galleria mellonella larvae infected withE. coliA25 (1 × 10⁹ CFU) (n = 10 per group). In the combination group, larval survival was improved after 120 h of continuous observation. (C) Bacterial colonization in Galleria mellonella larvae infected with E. coli A25 (1 × 10⁷ CFU) (n = 10 per group) and subsequently treated with PBS, GM (10 mg/kg), NEF (5 mg/kg), or a combination of GM + NEF (8 + 5 mg/kg).

To verify the in vivo potentiation effect of NEF in a mammalian model, a systemic mouse infection model was established (Figure 7(A)). Survival analysis revealed that after infection with 3 × 10¹⁰ CFU of E. coli, all mice in the control and 10 mg/kg NEF monotherapy groups died within 4 days. Although 60 mg/kg GM monotherapy extended survival time, the 7‑day survival rate was only 20%. In contrast, the combination therapy (60 mg/kg GM +10 mg/kg NEF) significantly improved survival outcomes, raising the 7‑day survival rate to 50% (Figure 7(B)). In the sub‑lethal infection model (3 × 10⁹ CFU), the combination group exhibited a marked reduction (by two orders of magnitude) in bacterial loads in the liver, spleen, and kidneys compared to the monotherapy groups, demonstrating efficient clearance of pathogens from organs (Figure 7(C–E)).

Figure 7.

NEF effectively improves the in vivo activity of GM in a mouse infection model. (A) Experimental scheme for mouse survival and bacterial colonization assays. (B) Survival of mice infected withE. coliA25 (3 × 10¹⁰ CFU) (n = 10 per group) after treatment with PBS, NEF (10 mg/kg), GM (60 mg/kg), or their combination. (C – E) Bacterial colonization in the liver, spleen, and kidneys of mice infected withE. coliA25 (3 × 10⁹ CFU) (n = 6 per group) and subsequently treated with PBS, GM (60 mg/kg), NEF (10 mg/kg), or the combination of GM + NEF (60 + 11 mg/kg). (F – I) Expression levels of pro‑inflammatory cytokines in mouse serum measured by enzyme‑linked immunosorbent assay(ELISA). (J) H&E staining of liver, spleen, and kidney tissues from mice in the control group and each treatment group.

To further assess the regulatory effect of the combination treatment on infection‑related inflammatory responses, serum levels of key cytokines were measured. The results showed that the expression of pro‑inflammatory cytokines (TNF‑α, IL‑1β, IL‑2) in the combination group was significantly lower than in the monotherapy groups, with IL‑2 levels returning to nearly those of the uninfected control. Conversely, the anti‑inflammatory cytokine IL‑4 was relatively up‑regulated Figure 7(F–I). This altered cytokine profile suggests that the combination therapy helps restore immune balance during infection. Furthermore, H&E staining pathology analysis revealed that hepatocytes in the untreated, NEF monotherapy, and GM monotherapy groups exhibited extensive vacuolization, erythrocyte congestion in splenic tissue, and renal tubular epithelial cell nuclear fragmentation in kidney cells. In contrast, tissue lesions in all organs from the combination treatment group showed significant improvement, with reduced inflammatory cell infiltration and pathological changes comparable to those in the PBS‑treated group (Figure 7(J)).

To evaluate the biosafety of NEF, the hemolytic activity of different concentrations of NEF against red blood cells was assessed. The results showed that at 50 μM, NEF induced a hemolysis rate of less than 5%, indicating minimal impact on erythrocyte membrane integrity and favorable safety at this concentration (Figure S5J). When the NEF concentration was increased to 200 μM, the hemolysis rate reached approximately 15%, suggesting a certain degree of cellular damage at higher concentrations. Nevertheless, based on its pharmacokinetic profile, the peak plasma concentration of NEF at the therapeutic doses used in this study is far below 50 μM, supporting good hemocompatibility of NEF at in vivo therapeutic concentrations [52].

Discussion and conclusion

The global crisis of antibiotic resistance continues to escalate, with the rapid evolution of bacterial resistance posing a severe challenge to public health worldwide [53]. GM, once a key agent for treating E. coli infections due to its potent bactericidal activity against Gram‑negative bacteria, has seen its clinical efficacy decline sharply as a result of the spread of aminoglycoside‑modifying enzyme genes and the upregulation of efflux pump systems following its extensive use in clinical and veterinary settings [54]. The development of resistance not only leads to treatment failure and prolonged disease courses, but also compels clinicians to increase dosages or resort to higher‑tier antibiotics, thereby exacerbating toxicity risks and healthcare burdens. In this context, the search for adjuvants capable of restoring or enhancing the activity of existing antibiotics has become a crucial strategy for addressing the resistance crisis [55,56]. TCM‑derived monomers, owing to their structural diversity, multi‑target activity, and long‑standing clinical experience, represent an important resource for screening novel antibacterial adjuvants [57]. Through systematic screening of a TCM monomer library, this study has for the first time identified NEF as a compound that, despite lacking intrinsic antibacterial activity, significantly potentiates the bactericidal effect of GM against E. coli. This synergy was demonstrated by rapid cooperative killing in time‑kill kinetic assays and by the effective delay in the evolution of GM resistance during long‑term passage experiments.

It is encouraging that our findings indicate the synergistic effect of NEF with GM does not stem from its own direct antibacterial activity, but rather from its function as a “metabolic perturbator.” Transcriptomic analysis and subsequent target validation focused the mechanism of action on two core global transcriptional regulators: FNR and ArcA [58,59]. Molecular docking and surface plasmon resonance experiments confirmed that NEF binds to these proteins with higher affinity than GM, and circular dichroism spectroscopy further revealed that this binding induces substantial changes in the secondary structure of the target proteins. FNR and ArcA serve as central hubs for bacterial sensing of environmental oxygen partial pressure and redox status, co‑regulating the expression of genes involved in pathways ranging from the tricarboxylic acid (TCA) cycle and electron transport chain to anaerobic respiration. Our results suggest that NEF binding may mimic or stabilize a dysregulated conformation, leading to aberrant regulatory function. This abnormal activation of the targets triggered a downstream metabolic cascade. We observed that following co‑treatment with NEF and GM, the intracellular NAD⁺/NADH ratio in bacteria abnormally increased, ATP levels were sharply depleted, and there was a burst‑like accumulation of reactive oxygen species (ROS). This series of phenotypes points to hyper‑activation and uncoupling of the TCA cycle and oxidative phosphorylation. Under normal conditions, bacteria finely tune metabolic flux via FNR/ArcA to adapt to environmental stress; however, NEF intervention causes this regulatory system to become “uncontrolled,” driving metabolic processes into an unsustainable, self‑destructive state. The over‑active TCA cycle consumes large amounts of reducing equivalents and generates excess electrons, leading to overload of the electron transport chain and massive ROS production; meanwhile, ATP depletion directly dismantles the cell’s energy currency. This dual crisis of energy and redox imbalance, triggered by an overloaded “metabolic engine,” fundamentally weakens the bacterial survival foundation, rendering it hypersensitive to the killing effect of GM. The observation that the synergistic effect persisted in FNR/ArcA‑overexpressing strains indicates that the mechanism of NEF is not achieved by suppressing the expression or reducing the activity of these proteins, but likely depends on its interaction with FNR/ArcA proteins to modulate their downstream pathways.

In addition to metabolic perturbation, we identified synergistic disruption of the cell membrane structure. Scanning electron microscopy visually demonstrated severe shrinkage, rupture, and cytoplasmic leakage of bacterial membranes after combined treatment. Fluorescence polarization and membrane potential assays provided quantitative data: NEF treatment alone increased membrane fluidity, while co‑treatment with GM led to an abnormal surge in membrane fluidity and rapid, intense depolarization of the membrane potential. As an amphipathic alkaloid, NEF may further insert into and disrupt the arrangement of the phospholipid bilayer. The two agents act in concert, greatly accelerating the loss of membrane integrity. The collapse of the membrane potential not only directly affects the proton motive force and energy generation but may also promote the intracellular influx of drug molecules such as GM, forming a positive feedback loop of “membrane disruption – increased drug influx – enhanced intracellular killing.” Thus, the synergy between NEF and GM results from the combined action of metabolic perturbation and membrane disruption mechanisms.

Quantitative assessment of membrane damage further elucidated the underlying mechanism of the synergistic antibacterial activity between NEF and GM. NEF treatment alone significantly increased the outer membrane permeability of E. coli without affecting the inner membrane, whereas the combination treatment led to severe disruption of both membranes. This observation suggests that NEF may compromise the barrier function of the outer membrane, thereby creating a favorable condition for GM entry into the periplasmic space. Subsequently, NEF and GM act together to induce irreversible inner membrane damage. Notably, NEF did not directly cleave or degrade genomic DNA, and the DNA leakage observed upon combination treatment should be attributed to the release of cytoplasmic contents resulting from the loss of membrane integrity rather than a direct DNA-targeting effect. Furthermore, NEF significantly enhanced the intracellular accumulation of GM, further confirming that impairment of the membrane barrier function is a key determinant of the synergistic efficacy.

Currently reported mechanisms of antibiotic potentiators mainly include inhibition of blockade of efflux pumps and suppression of biofilm formation [60,61]. The uniqueness of NEF’s mechanism lies in its direct targeting of metabolic transcription factors. Compared to strategies targeting a single resistance enzyme, targeting global regulators such as FNR/ArcA may make it more difficult for bacteria to develop resistance through simple mutations, because altering the core regulatory proteins themselves or their key regulatory networks likely imposes a substantial fitness cost. This aligns with our observation that NEF significantly delays the development of GM resistance.

Furthermore, this study confirmed the in vivo potentiation effect of NEF in both invertebrate and mammalian infection models. The combined treatment not only significantly improved host survival and reduced bacterial loads in various organs but also demonstrated remarkable immunomodulatory functions. We observed that serum levels of pro‑inflammatory cytokines (TNF‑α, IL‑1β, IL‑2) were significantly reduced in the combination group, while the anti‑inflammatory cytokine IL‑4 was relatively upregulated, accompanied by a clear alleviation of pathological damage in liver tissue. It is known that NEF has activity in inhibiting inflammatory pathways such as NF‑κB in various disease models [62–64]. In the context of infection, NEF may, on one hand, rapidly reduce the pathogen burden through enhanced bactericidal action of GM, thereby diminishing the source of inflammatory stimulation; on the other hand, it directly modulates the excessive response of host immune cells, thereby mitigating cytokine storms and tissue damage. This dual regulatory action targeting both the “pathogen” and the “host response” holds promise for achieving superior clinical outcomes in infection treatment.

This study also has certain limitations. First, although molecular docking, SPR, and conformational analyses strongly suggest direct interaction of NEF with FNR/ArcA, obtaining co‑crystal structures would provide the most direct evidence. Second, the regulatory network of FNR/ArcA is extremely complex; how NEF binding precisely affects their interactions with other regulatory proteins requires further investigation. On the other hand, we were unable to directly verify whether neferine blocks the binding of FNR/ArcA to target DNA, which awaits further investigation. In summary, this study discovered and confirmed that neferine is a GM potentiator that acts through an innovative mechanism. By uniquely targeting the FNR/ArcA regulatory nodes, it induces collapse of bacterial metabolic homeostasis and synergistically disrupts the cell membrane with GM, thereby generating powerful synergistic bactericidal effects both in vitro and in vivo, while delaying resistance development and modulating immunity.

Funding Statement

This work was financially supported by the National Natural Science Foundation of China [U23A20242].

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

All data supporting the findings of this study are publicly available. The relevant data have been deposited in the public repository Figshare under DOI: 10.6084/m9.figshare.31839745.

The link is: https://doi.org/10.6084/m9.figshare.31839745.

Ethics and consent to participate

In this study, informed consent was obtained from the owners to allow the use of these animals in the research.

Supplementary Information

Supplemental data for this article can be accessed online at https://doi.org/10.1080/21505594.2026.2690833