Dipiperazine‐Phenyl Derivatives Based on Convergent Molecular Platforms Can Reverse Multidrug Resistance in Gram‐Negative Bacteria by Inhibiting Efflux and Permeabilizing Cell Membranes

Abstract

With the global surge of infections caused by multidrug‐resistant (MDR) Gram‐negative bacteria, there is an urgent need for breakthrough therapeutic approaches. To overcome the intrinsic resistance mechanisms of bacteria, End‐alkyl‐modified dipiperazine‐phenyl derivatives are designed via convergent molecular platforms (CMPs)‐guided multi‐target directed ligand (MTDL) strategy. These dual‐functional compounds not only inhibit the AcrB‐TolC efflux pump system but also enhance bacterial membrane permeability and display a distinctive activity profile across a broad concentration range. Through integrated evaluation combining in vitro activity screening and computational ADMET (absorption, distribution, metabolism, excretion, toxicity) profiling, compound C5 is identified as a promising lead candidate. This compound achieved three notable breakthroughs. First, it reduces biofilm formation by 80% at 1/64 minimum inhibitory concentration (MIC) when combined with antibiotics. Second, unlike conventional antibiotic adjuvants that typically display potentiation within a narrow concentration window (1/4 MIC), C5 maintained robust and consistent synergistic activity across a broad range from 1/64 MIC to 1/4 MIC. Third, C5 markedly enhanced the therapeutic efficacy of antibiotics such as minocycline by over 1000‐fold in in vivo infection models, without causing detectable acute toxicity or cytotoxicity. The established MTDL‐CMPs integrated platform pioneers a novel “pump‐membrane dual blockade” therapeutic paradigm against MDR Enterobacteriaceae infections.

By integrating a convergent molecular platform strategy, this study designed a novel dual‐target C5 to combat multidrug‐resistant Gram‐negative bacteria. C5 synergistically enhances antibiotic efficacy by inhibiting efflux pumps and increasing bacterial membrane permeability. This innovative “efflux pump‐membrane dual blockade” mechanism offers a new therapeutic paradigm for addressing the growing problem of bacterial resistance.

1. Introduction

Antibiotic resistance represents a major challenge in modern medicine, with the rise of MDR bacteria posing a severe threat to public health.^{[ 1} , ^{2 ]} By 2050, over 39 million people are projected to die from antibiotic‐resistant infections; however, improved access to effective antibiotics and optimized infection management from 2025 to 2050 could save ≈92 million lives.^{[ 3} , ^{4 ]} The latest World Health Organization report highlights antibiotic resistance as a global health priority, underscoring the urgent need for innovative therapeutic strategies to address the declining efficacy of existing antibiotics.^{[ 5 ]}

At present, antibiotic research and development is deeply trapped in multiple intertwined challenges, and the translation of new antibiotics from the laboratory to clinical application remains difficult. On the one hand, the evolution of bacterial resistance is much faster than the development cycle of antibiotics. Many traditional antibiotics rapidly lose efficacy in clinical use, forcing researchers to develop new compounds to counter emerging resistance mechanisms, which greatly increases development difficulty and cost.^{[ 6 ]} On the other hand, the investment‐return ratio of antibiotic research and development is severely imbalanced. From early compound screening to late‐stage clinical trials, the process requires over a decade of time and enormous financial resources, while facing a high risk of failure. Consequently, many pharmaceutical companies adopt a cautious attitude toward investment in this field, which further slows the pace of new antibiotic development.^{[ 7 ]}

Therefore, developing effective strategies to expand available treatment options is crucial.^{[ 8 ]} Inspired by the synergistic mechanisms of certain antibiotics and non‐antibiotics when used in combination, researchers have begun to develop antibiotic adjuvants to enhance antibiotic efficacy and delay resistance development.^{[ 9} , ¹⁰ , ¹¹ , ^{12 ]} Specifically, combining polymyxins with other antibiotics enhances bactericidal activity against Gram‐negative bacteria by disrupting their outer membranes.^{[ 13 ]} Additionally, compounds with appended lipophilic chains increase membrane permeability, facilitating antibiotic entry into bacterial cells.^{[ 14 ]} Moreover, compounds like PAβN interfere with bacterial efflux pumps, thus enhancing the antibacterial activity of antibiotics like tetracycline and chloramphenicol.^{[ 15 ]} These strategies have shown promising results, particularly in treating MDR bacterial infections. By using these combination strategies, the lifecycle of existing antibiotics can be prolonged, addressing gaps in antibiotic development. Thus, the development of broad‐spectrum antibiotic adjuvants and their combination with existing antibiotics holds significant potential to improve the treatment of bacterial infections.

To guide the development of effective antibiotic adjuvants, a thorough understanding of bacterial resistance mechanisms is essential. Among these, restricted drug entry and active efflux pumps are the most critical for Gram‐negative bacterial resistance, while others, such as target alterations, enzyme‐mediated drug degradation, and metabolic pathway changes, play a secondary role.^{[ 16} , ¹⁷ , ^{18 ]} Recent studies have extensively investigated influx and efflux mechanisms^{[ 16} , ¹⁹ , ^{20 ]} and proposed diverse strategies for developing inhibitors targeting these pathways.

Although progress has been made in the research of AcrB efflux pump and the cell membrane barrier, current therapeutic approaches predominantly remain single‐targeted.^{[ 21} , ²² , ^{23 ]} Dual‐target strategy that simultaneously inhibit efflux pump and regulate membrane permeability is still underdeveloped. Existing dual‐target inhibitors, such as PaβN, can maintain concentration gradients sufficient for antibacterial activity, but their clinical translation has been hampered by high toxicity and low in vivo efficacy (see Extended Data Figure S1, Supporting Information). Conversely, compounds with lower toxicity often fail to reach therapeutically effective concentrations in vivo, thereby producing suboptimal synergistic effects. For example, although metformin shows promise, it is only effective at milligram‐per‐milliliter (mg·mL⁻¹) levels,^{[ 24 ]} and its activity significantly decreases when the concentration falls below 1/8 minimum inhibitory concentrations (MIC). Consequently, developing low‐toxicity compounds possessing both mechanisms simultaneously is crucial for overcoming bacterial resistance. These compounds should have minimal or no membrane‐damaging effects and significantly enhance antibiotic activity at lower concentrations to meet clinical requirements (see Figure 1a).

Figure 1.

Mechanism of action and structure‐activity relationship analysis of dual‐targeted compounds. a) This figure illustrates the multiple mechanisms by which SDUKJ‐C5 enhances the antibacterial activity of tetracycline. The left section shows that SDUKJ‐C5 can increase bacterial membrane permeability, thereby promoting the entry of tetracycline into bacterial cells. The right section demonstrates that SDUKJ‐C5 inhibits bacterial efflux pump, and effectively prevents the extrusion of tetracycline and certain toxic metabolites (including specific efflux pump substrates), resulting in their significant accumulation within bacterial cells. This intracellular accumulation not only enhances the bactericidal efficacy of the antibiotic but also suppresses biofilm formation, thereby contributing to the reversal of bacterial multidrug resistance. EP:efflux pump. b) The molecular structure of partially synthesized dual‐targeted compounds. c) The interaction mechanism between C20 and the binding pocket of the efflux pump.

2. Results

2.1. Design and Synthesis of Conjugated Compounds CMPs‐001 and CMPs‐002 Using Piperazine as a Linker

To fill the gap in the research on dual‐target antibiotic adjuvants, we reviewed a wide range of methodologies across various fields and proposed the CMPs for Drugs,^{[ 25} , ²⁶ , ^{27 ]} an improved strategy based on the existing Multi‐functional Drug Design (MfDD). MfDD integrates multiple functional modules into a single chemical framework to achieve multi‐target effects. For example, dual‐functional inhibitors used for cancer or antibacterial therapy can simultaneously inhibit protease activity and block signal transduction,^{[ 28 ]} or the combination of antibiotics with cationic peptides can enhance membrane permeability.^{[ 29 ]} However, this approach often relies on the physical combination of existing compounds, lacking in‐depth design of complementary interactions between target binding pockets, which may lead to insufficient activity or increased toxicity. In contrast, CMPs can optimize the complementarity between modules, avoiding the pitfalls of simple stacking that can result in “1 + 1 < 1” or “1 + 1 = 0” effects. Not only that, CMPs can employ stable‐linker designs to ensure synergistic effects, achieving “1 + 1 > 2” outcomes. To standardize this approach, each module is labeled as CMPs‐001, CMPs‐002, etc., to track and name different compound combinations. This strategy aims to create more active and multifunctional molecular platforms to drive the development of multi‐target drugs.

To screen the efflux pump‐targeting antibiotic adjuvant CMPs‐001, we used MDR wild‐type E. coli BW25113 and its efflux pump knockout mutant E. coli BW25113 (ΔacrB) as model strains. wild‐type E. coli BW25113 shows resistance to all major antibiotic classes (see Table S1, Supporting Information). By combining our in‐house compound library with a commercial library (SDUKJ) and using virtual screening, we identified hundreds of candidate compounds from tens of thousands. They were further filtered based on optimization potential, binding affinity, and bioactivity. In the initial screening, candidate compounds combined with minocycline were tested for potentiation effects on wild‐type E. coli BW25113 and E. coli BW25113 (ΔacrB). Molecular docking identified compounds with high binding efficiency and further pharmacophore modeling and ADMET analysis led to the selection of the most bioactive precursor compound (see Extended Data Figure S2a, Supporting Information). Surprisingly, SDUKJ‐4584 specifically enhanced antibiotic activity against wild‐type E. coli BW25113 but had no effect on E. coli BW25113 (ΔacrB) (see Tables S1 and S2, Supporting Information), showing significant efflux pump inhibition in preliminary tests.

Preliminary structure‐activity relationship (SAR) analysis revealed that the activity of SDUKJ‐4584 is driven by three key pharmacophores (see Extended Data Figure S3a, Supporting Information). For example, removal of the piperazine ring in the A‐series compounds resulted in a complete loss of activity, highlighting its structural importance. Further modifications to the benzene ring showed that the increase of heterocyclic nitrogen atoms slightly reduced activity, likely due to the electronic effects constraining the piperazine ring's spatial conformation. Additionally, substitutions at the para‐position of the piperazine ring significantly affected the compound's activity (see Table S1, Supporting Information). Based on these findings, we retained the core benzene and piperazine structures and optimized the substituents using computer‐aided design, ultimately determining the structure of CMPs‐001.

Unlike the screening strategy for CMPs‐001, the selection of CMPs‐002 focuses more on the hydrophobic cavity characteristics of CMPs‐001 and the shared mechanisms through which antibiotics and non‐antibiotics interact with bacterial membranes. Studies indicate that hydrophobic hydrocarbon chains and their modifications are essential for membrane‐permeabilizing activity (see Extended Data Figure S2b, Supporting Information). For example, the N‐acyl fatty acid chain and hydrophobic groups at positions 6 and 7 of polymyxin molecule insert into the lipid bilayer of bacterial membranes, disrupting the packing of adjacent lipid A acyl chains.^{[ 30 ]} This interaction expands the outer membrane monolayer, leading to swelling. Similarly, the in vitro and in vivo antibacterial activity of daptomycin derivatives improves with the increase of chain length (up to 10C atoms).^{[ 31} , ^{32 ]} This highlights the importance of lipopeptide hydrophobicity, which is crucial for daptomycin's membrane‐permeabilizing effects. These findings underscore the critical role of fatty acid chains in bacterial membrane interactions. Professor Johannes Morstein further suggested that medium‐length chains have the strongest impact on bacterial membranes,^{[ 33 ]} while longer chains show lower toxicity. Building on these findings, CMPs‐002 was developed by selecting and optimizing a variety of fatty acid chains to achieve optimal membrane‐permeabilizing activity.

In small‐molecule drug design, linkers are essential for connecting functional modules like CMPs‐001 and CMPs‐002. Unlike those in antibody‐drug conjugates, a small‐molecule linker must remain stable while functioning intracellularly to ensure the active molecule stays intact until it reaches the target site. Piperazine, due to its cyclic structure, high hydrophilicity, and biocompatibility, is regarded as an ideal linker. When modified at its para‐substitution, this nitrogen atom resists enzymatic degradation and enhances compound‐target interactions through hydrogen‐bond donor properties, thereby optimizing the drug's bioactivity and pharmacokinetics. Using piperazine as a linker to join CMPs‐001 and CMPs‐002 significantly improves the solubility and stability of the resulting compounds, paving the way for a CMPs series with strong therapeutic potential (see Extended Data Figures S3d and S4b,c, Supporting Information).

2.2. Antibacterial Activity and Ability to Reverse Bacterial Resistance

The in vitro antibacterial activity of compounds was evaluated by measuring their minimum inhibitory concentrations (MIC) against MDR wild‐type E. coli BW25113 and its antibiotic‐sensitive mutant strain E. coli BW25113 (ΔacrB). None of compounds showed antibacterial activity against either wild‐type E. coli BW25113 or its mutant strain (see Tables S1 and S2, Supporting Information). These findings suggest that compounds at concentrations ≤ 1/4 MIC used in subsequent synergistic experiments will not affect the synergistic results due to their intrinsic antibacterial activity, which ensures a reliable foundation for studying antibiotic combination therapies.

The compounds' ability to reverse bacterial resistance was initially assessed through high‐concentration primary screening. Standard checkerboard assays were then conducted to evaluate the interactions between the compounds and various antibiotics at different concentrations.^{[ 34} , ^{35 ]} The selected antibiotics included minocycline, oxacillin, linezolid, and rifampin, among others. The first three antibiotics are substrates of AcrB efflux pump and membrane permeabilizers,^{[ 36 ]} while rifampin is not. PAβN, a known dual target inhibitor of efflux pump and membrane permeability, was used as a positive control to demonstrate the enhanced effects of dual‐target inhibitors in combination with antibiotics.

At sub‐MIC levels, the compounds demonstrated synergistic effects with multiple classes of antibiotics, including minocycline, oxacillin, linezolid, and rifampin, against E. coli BW25113 (see Table S1, Supporting Information). The interactions were evaluated by calculating the Fractional Inhibitory Concentration Index (FICI), where FICI ≤ 0.5 indicates synergy and 0.5 < FICI ≤ 1 suggests additive effects. The FICI value is a key metric for evaluating the efficacy of drug combinations,^{[ 37} , ^{38 ]} particularly in analyzing therapeutic regimens for MDR bacterial infections. Notably, the optimal compounds achieved FICI values as low as 0.015, demonstrating exceptional synergistic potency.

In this study, we analyzed five optimized compounds: A4, C3, C5, C15, and C20 (see Extended Data Figure S4, Supporting Information). A4 represents the optimal efflux pump inhibitor identified prior to the introduction of CMPs‐002 following the introduction of the dual‐target mechanism. Although C3 and C5 exhibited similar activities, C3 had relatively strong antibacterial activity with a lower MIC (see Table S1, Supporting Information), which does not align with our goal of developing antibiotic adjuvants. Antibiotic resistance arises from the presence of its antibacterial activity, as bacteria must develop some mechanisms to overcome it. Therefore, compounds without antibacterial activity do not cause bacteria to develop resistance.^{[ 39 ]} As a result, C5, C15, and C20 show promising prospects as antibiotic enhancers.

We performed high‐concentration primary screening of the compounds, with three parallel experiments per group to ensure the accuracy of the results (see Extended Data Figure S5a,b, Supporting Information). The MIC of minocycline (MIN) was determined to be 4–8 µg·mL⁻¹, whereas PAβN reduced its MIC to 0.0625‐0.125 µg·mL⁻¹, confirming its enhancing effect on antibiotic efficacy. In contrast, metformin at a low concentration (128 µg·mL⁻¹) showed minimal antibiotic‐enhancing activity. Among the four compounds, C5 and C20 exhibited efflux pump inhibition and membrane permeabilizing effects similar to PAβN. When combined with minocycline at 1/4 MIC, C5 decreased the MIC of minocycline to 0.0625 µg·mL⁻¹, a 64‐fold increase in antibacterial activity compared to its individual use, with an FICI value of 0.0157. In a checkerboard dilution assay, even at 1/64 MIC, C5 was still able to decrease the MIC of minocycline by 16 times, demonstrating its strong synergistic effect (see Table S4b and Extended Data Figure S5c, Supporting Information). During high‐concentration screening, precipitation of the compound was observed, which may interfere with the accuracy of visual inspection because precipitates can be mistaken for bacterial growth. To avoid this interference and obtain more reliable quantitative data, OD₆₀₀ measurement was introduced. This method provides an objective evaluation of bacterial growth at the population level based on light scattering and complements visual inspection by ensuring the detection of residual survival at low concentrations while accurately reflecting overall growth inhibition trends. The results showed that the fold‐change patterns obtained by visual inspection and OD₆₀₀ measurement were consistent, indicating that the error in fold‐change calculation remained within an acceptable range (see Extended Data Figure S5e, Supporting Information).

Both C5 and C20 exhibited excellent antibiotic‐sensitizing activity in the initial assessment. Due to its moderate chain length and complex structure, C20 was selected for molecular docking with the target AcrB (see Figure 1c). This compound, in addition to interacting with the original binding residues of AcrB (Asp408, Val411, and Ile445), perfectly binds to the designed target amino acids, such as new key residues Asp951 and Lys955, and an unexpected residue Glu917. Its modified long‐chain end fits perfectly into the hydrophobic pocket of AcrB, achieving our desired application of CMPs. It should be further noted that C20 exhibits sensitizing activity only at high concentrations. At lower concentrations, individual or few molecules are unable to efficiently insert into the lipid bilayer or the hydrophobic cavity of efflux pumps, as a relatively high energy barrier must be overcome to penetrate the membrane.

2.3. The structure‐activity relationship (SAR) was concluded based on observations of synergistic antibacterial effects with antibiotics and molecular docking analysis

The results indicate that the superior performance of C5 can be attributed to the synergistic contribution of three key structural determinants: i) the positively charged piperazine headgroup, ii) the piperazine linker and its specific connection with the terminal substituent, and iii) the length and substitution pattern of the terminal hydrophobic chain. Together, these structural elements modulate the overall amphiphilicity, membrane‐interaction geometry, and self‐assembly behavior of the molecule, thereby determining whether the compound acts as a mild membrane‐perturbing efflux pump inhibitor or a non‐selective surfactant (see Extended Data Figure S3b–d and S4a–c, Supporting Information).

At physiological pH, the terminal piperazine ring of C5 is partially protonated, providing a “soft” positive charge that enables electrostatic attraction with the anionic components of the Gram‐negative outer membrane. This interaction promotes the interfacial localization of the molecule and facilitates the insertion of its hydrophobic chain into the lipid bilayer. The structural modification of the piperazine moiety completely abolished the antibacterial sensitizing activity (see Table S1 and Extended Data Figure S4a, Supporting Information), indicating that the piperazine‐Asp408 salt bridge, together with the hydrogen‐bonding interaction between the piperazine N‐H and Asp404, plays an essential role in maintaining activity (see Extended Data Figure S3d, Supporting Information). Furthermore, molecular docking analysis revealed that the piperazine linker can also form salt bridges with Asp951 and Glu947, which further stabilize the binding conformation of C5 in the AcrB efflux pump and thereby enhance the antibiotic‐synergistic effect.

The regulation of chain length exhibits a typical biphasic relationship with activity. Short‐to‐medium chains with 9–14 carbon atoms (e.g., C2) act in a “detergent‐like” manner at experimental concentrations, producing direct bactericidal effects but significant cytotoxicity as well. In contrast, long chains with more than 16 carbon atoms tend to aggregate or precipitate due to low solubility and reduced critical micelle concentration (CMC), thereby weakening membrane interactions. The 15‐carbon atom alkyl chain of C5 lies within the optimal window, allowing it to insert deeply into the lipid bilayer to modulate membrane fluidity and curvature stress, promotes antibiotic penetration without causing membrane disruption or cell lysis, and effectively overcomes permeability‐associated resistance barriers.

The linkage mode also exerts a pronounced influence on molecular conformation and interaction behavior. The direct N‐alkyl linkage in C5 preserves the high flexibility of the hydrophobic chain, allowing it to dynamically embed into the membrane and interacts effectively with lipids through hydrophobic interactions. In contrast, the carbonyl linkage in C15 introduces additional polarity and rigidity, altering charge distribution and chain orientation, which prevents the formation of auxiliary salt bridges with Asp951/Glu947, leading to enhanced membrane disruption and increased toxicity. The terminal double bond in C20 increases the rigidity and insertion depth of the chain, resulting in significant sensitization only at high concentrations. Moreover, introducing nitrogen atoms into the intermediate aromatic ring slightly decreases sensitizing activity, suggesting that the electronic properties of the aromatic ring also influence the membrane interaction mode.

Collectively, the flexible cationic piperazine headgroup, the medium length hydrophobic chain with 15 carbon atoms, the non‐polar N‐alkyl linkage, and the non‐functionalized terminal group cooperatively define the optimal pharmacophoric configuration of C5, which enhance antibiotic permeability while preserving membrane integrity and suppressing efflux function, thereby exerting synergistic antibacterial activity without elevating host toxicity.

2.4. Compounds Enhance the Accumulation of Antibiotics in E. coli BW25113 by Inhibiting AcrB Efflux Pump

AcrB is the core component of the RND‐type AcrAB–TolC efflux system, which is organized into a homotrimer spanning the bacterial inner membrane. During the substrate transport cycle, each protomer sequentially adopts three conformational states: loose (L), tight (T), and open (O). These functional conformational transitions are driven by the proton motive force generated by the transmembrane proton gradients, which are transmitted through a conserved proton relay network formed by key residues such as Asp407, Asp408, and Asp951.^{[ 21 ]}

To further elucidate the mechanism by which the compounds inhibit efflux pumps, we separately evaluated their effects on efflux pump inhibition and membrane permeability. We first confirmed the AcrB‐dependent activity of the compounds through gene knockout experiments: when the acrB gene was deleted in E. coli, the compounds showed a markedly reduced ability to enhance antibiotic accumulation, indicating that AcrB is the primary target responsible for their activity (see Table S2, Supporting Information). To further validate this finding, a Nile red efflux assay was performed. Nile red, a well‐characterized AcrB substrate, enables direct quantitative assessment of AcrB‐mediated transport by monitoring its intracellular accumulation.^{[ 40} , ⁴¹ , ^{42 ]} The results showed that, at high concentrations, most compounds exhibited Nile red efflux inhibition comparable to or greater than that of PAβN. Notably, C5 and C15 almost completely inhibited Nile red efflux, showing effects equivalent to those observed in the ΔacrB group (see Figure 2a–c), suggesting that both compounds can block the full transport cycle of AcrB.To characterize the inhibitory kinetics of the lead compound C5, an ethidium bromide efflux assay was conducted using a concentration gradient ranging from 128 to 1 µg·mL⁻¹, with untreated cells and cells treated with a subinhibitory concentration of PAβN serving as control. Fitted curve analysis revealed a clear concentration‐dependent inhibition of efflux activity by C5 (see Figure 2d), consistent with its specific binding behavior toward AcrB. In addition, compound C20, known for its strong membrane‐permeabilizing activity, was selected for a tetracycline accumulation assay. At 128 µg·mL⁻¹, C20 promoted rapid intracellular accumulation of tetracycline, blocked its efflux, and increased the intracellular antibiotic concentration to a level comparable to that of the PAβN‐treated group (see Figure 2g), demonstrating a synergistic effect between AcrB inhibition and enhanced membrane permeability.

Figure 2.

The target compound potently inhibits bacterial efflux pump function and increases cell membrane permeability, while maintaining minimal or no damage to the bacterial membrane architecture. a–c) Nile red efflux inhibition assays for A4, C5, and C20. d) Tetracycline accumulation assay for C20. e,f) Proton gradient disruption validation assays for C5 and C20. g) ethidium bromide efflux assay for C5. h,i) The effects of C5, and C20 on the integrity of the bacterial outer membranes. j,k) The impacts of C5, and C20 on the bacterial inner membranes. l–o) Scanning electron microscopy (SEM) images showing bacteria treated with sub‐inhibitory concentrations of PAβN (positive control) and the tested compounds (l. DMSO group, m. PAβN group, n. C5 group, and o. C20 group). p) Protein leakage from bacteria following treatment with PAβN and the tested compounds. Statistical analysis was performed using One‐way ANOVA. “***” indicates p < 0.001.

Molecular dynamics simulations further revealed the inhibitory mechanism of C5. C5 was found to bind to an allosteric site located at the junction of the TM4, TM5, and TM10 helices within the transmembrane domain of AcrB, adjacent to the conserved proton relay network. The nitrogen atoms on both piperazine rings of C5 form salt bridges with Asp408, Asp951, and Glu947, which stabilize the local conformation and induce regional dynamic changes. Among the above interactions, the exposed piperazine terminus maintains stable binding through persistent hydrogen bonding with Asp404, together with hydrophobic interactions mediated by the long alkyl chain. These interactions collectively cause local compression within the transmembrane region (see Extended Data Figures S3b–d and S10a–d, Supporting Information), locking AcrB in the L conformation and preventing the critical L → T transition required for substrate entry into the deep binding pocket (DBP). These results indicate that C5 acts as a typical allosteric inhibitor of AcrB. Given that the function of AcrB is highly dependent on the proton motive force, we further evaluated the effect of C5 on the bacterial proton gradient. The results showed that both C5 and C20 significantly disrupted the inner membrane proton gradient (see Figure 2e,f), whereas C3 and C15 had relatively weaker effects. Among them, A4 exhibited proton gradient dissipation comparable to that of the classical inhibitor PAβN (see Extended Data Figure S6a–c, Supporting Information). The collapse of the proton gradient depleted the energy source of AcrB and blocked the proton‐driven conformational rotation among its protomers, consequently further impairing the overall efflux function.

In summary, these findings demonstrate that the optimized compounds inhibit AcrB through a dual mechanism. They not only lock AcrB in an inactive conformational state, but also disrupt the proton gradient that provides energy for substrate transport. This multitarget mode of action effectively blocks the AcrB efflux cycle and markedly promotes intracellular antibiotic accumulation, confirming that these compounds are potent and promising antibiotic adjuvants.

2.5. The Compounds Can Enhance the Permeability of Bacterial Membranes without Compromising Their Integrity

The efflux pump validation experiments showed high reproducibility, establishing a strong basis for further exploring its role in enhancing membrane permeability in a dual‐targeting mechanism. This study aims to enhance membrane permeability while minimizing membrane structural damage to reduce the risk of resistance development.^{[ 14} , ^{16 ]} To further clarify the underlying mode of action, checkerboard assays with exogenous LPS, PE, PG, and cardiolipin were performed. These membrane components had negligible effects on antibacterial activity, indicating that the compounds do not rely on specific binding to individual lipid species but instead act through non‐specific modulation of membrane packing and permeability (Extended Data Figure S11, Supporting Information). To ensure rigor and reliability, we performed direct experiments on inner and outer membranes of bacteria,^{[ 43} , ⁴⁴ , ^{45 ]} along with reverse validation methods such as protein leakage assay, complemented by scanning electron microscope (SEM) observation.

The dual‐membrane structure of Gram‐negative bacteria acts as an effective barrier, greatly enhancing their antibiotic resistance.^{[ 13 ]} Our compounds increase the permeability of both inner and outer membranes, promotes penetration of antibiotics and synergizes with their efflux inhibitory activity. Experimental results showed that C5 and C20 markedly increased bacterial outer membrane permeability, with the relative fluorescence intensity of NPN staining in the C5‐ and C20‐treated groups being 1.66‐ and 1.53‐fold that of the untreated control group, respectively, while C3 and C15 had weaker effects (see Figure 2h,i; Extended Data Figure S6d, Supporting Information), aligning with synergistic antibacterial findings. Additional experiments confirmed that C5 and C20 enhanced bacterial inner membrane permeability, with the relative fluorescence intensity of PI staining in the C5‐ and C20‐treated groups being 1.33‐ and 1.55‐fold that of the untreated control group, respectively (see Figure 2j,k), possibly through mechanisms that increase selective permeability. Protein leakage assay showed that neither C5 nor C20 caused protein leakage (see Figure 2p). This indicates that the overall integrity of the bacterial membrane remained largely unaffected. Furthermore, SEM observations consistently showed that the treated bacteria retained an intact membrane morphology, without noticeable shrinkage, pore formation, or rupture‐features typically associated with membrane disruption (see Figure 2l–o). These results provide strong evidence from a reverse perspective that the action of C5 and C20 does not rely on crude membrane disruption, but rather enhances membrane permeability through more refined and controlled mechanisms. This “non‐destructive membrane permeabilization” effect ensures the structural integrity of the membrane while markedly improving the transmembrane entry of exogenous molecules (including antibiotics), thereby further supporting the effectiveness of the dual‐target antibacterial strategy.

2.6. The combination of compounds with antibiotics enhances bactericidal effects through multiple mechanisms, including reactive oxygen species (ROS) generation, biofilm inhibition, and prolonged post‐antibiotic effect (PAE). Furthermore, it exhibits significant potential in reversing antibiotic resistance and maintaining this reversing effect over the long term

This approach strengthens antibacterial efficacy and shows potential in preventing the development of resistance. ROS damage bacterial DNA, proteins, and cell membranes, ultimately leading to bacterial death.^{[ 46 ]} For example, levofloacin and minocycline significantly increase ROS production when treating wild‐type E. coli BW25113 at their MIC.^{[ 47} , ^{48 ]} Compared to untreated controls, the addition of NAC (N‐acetylcysteine) reduced ROS levels and diminished antibacterial efficacy, highlighting the critical role of ROS in the antimicrobial mechanism.^{[ 49 ]} Further studies showed that C5 and C20, when combined with antibiotics at 1/4 MIC, significantly increased ROS production and enhanced the bactericidal activity of the antibiotics at concentrations of 1/4, 1/8, 1/16, and 1/32 MIC. Notably, C5 could restore the ROS‐generating capacity of 1/4 MIC antibiotic to its MIC level at a concentration as low as 1/32 MIC (see Figure 3a). Biofilms, which are microbial communities attached to surfaces, are closely associated with antibiotic resistance. Efflux pumps, directly or indirectly, affect biofilm formation, with many RND transporters playing a role in biofilm development.^{[ 50} , ^{51 ]} Therefore, we assessed the effects of C5 and C20 on inhibiting biofilm formation and eradicating mature biofilms. C5 and C20 were found to inhibit biofilm formation, with C5 showing a 60% inhibition rate at 1/16 MIC. However, neither of both compounds were effective in eradicating mature biofilms, which is consistent with our experimental findings. At high concentrations, C20 induced partial biofilm detachment, likely due to its strong permeabilizing activity, but its overall effect was limited (see Figure 3b,c,n–s; Extended Data Figure S6f,g, Supporting Information). When combined with minocycline at MIC, both C5 and C20, even at concentrations as low as 1/32 and 1/64 MIC, significantly enhanced the antibiotic's ability to reduce the formation of bacterial biofilm by over 80% (see Figure 3d,e).

Figure 3.

a) When the combination of C5 or C20 with antibiotics was used, the production of ROS in bacteria was significantly enhanced. b,c) Bacterial biofilm formation inhibition. d,e) The effect of C5 and C20 at different concentrations in combination with antibiotics on biofilm formation. f,g) Growth inhibition curves against E. coli of C5 and C20 at sub‐inhibitory concentrations and minocycline at varying concentration gradients. h. Growth inhibition fitting curves of C5, C20, minocycline, as well as C5 and C20 at 1/4 MIC each in combination with minocycline at 1/4 MIC. i,j) Time‐kill curves under combination treatment. k‐m. Post‐antibiotic effect (PAE) curves. n∼s. SEM of bacterial biofilm formation inhibition by C5 and C20 under different concentrations. n. untreated control. o. minocycline at 1 µg·mL⁻¹. P. C5 at 64 µg·mL⁻¹. q. C5 at 32 µg·mL⁻¹. r. C20 at 64 µg·mL⁻¹. s. C20 at 32 µg·mL⁻¹. t. Long‐term drug‐induced resistance experiment. Statistical analysis was performed using One‐way ANOVA. “***” indicates P < 0.001. “**” indicates P < 0.005.

To further evaluate the potential of antibiotic adjuvants in enhancing antimicrobial efficacy, we analyzed the synergistic effects of the compounds and antibiotics on bacterial growth and survival inhibition. Additionally, we also assessed the compounds' ability to sustain bacterial growth inhibition after antibiotic clearance, under simulated in vivo conditions. The bactericidal curve experiment showed that minocycline alone had a weak inhibitory effect on wild‐type E. coli BW25113.^{[ 52 ]} However, when combined with either C5 or C20, its complete bacterial eradication was achieved within 24 h, demonstrating exceptional bactericidal activity (see Figure 3i,j). Subsequently, we conducted growth inhibition curve experiment under high inoculum conditions to simulate clinical scenarios involving high bacterial loads and the antimicrobial efficacy of the antibiotic‐adjuvant combinations. This experiment also aimed to evaluate whether the adjuvants could effectively inhibit the emergence of resistant bacteria under high bacterial density.^{[ 53 ]}

In the above experiments, the compound concentration was maintained at sub‐minimum inhibitory concentrations (sub‐MIC), while minocycline was diluted in a concentration gradient (from 1/16 MIC to 1/256 MIC), with minocycline at 1/16 MIC and the untreated group as control. The results indicated that even at a minocycline concentration as low as 1/256 MIC, C5, C15 and C20 still exhibited significant inhibitory activity (see Figure 3f,g; Extended Data Figure S7d, Supporting Information). Further comparison of the growth inhibition effects showed that all compounds alone exhibited inhibition similar to or slightly stronger than the untreated group, while minocycline alone demonstrated moderate inhibition, but stronger than either the individual compounds or the untreated group. Notably, when combined with the compounds, minocycline consistently and completely inhibited wild‐type E. coli BW25113 growth, regardless of bacterial load, demonstrating superior growth suppression capability (see Figure 3h; Extended Data Figure S7a–c, Supporting Information).

In the PAE experiment, minocycline alone induced 4 h PAE against wild‐type E. coli BW25113. When combined with C5, C20, minocycline's PAE was extended to 5 h (100% increase), 5 h (100% increase), or 8 h (400% increase), respectively (see Figure 3k–m). The significant extension of PAE is crucial for antibiotic efficacy, as it reduces dosing frequency, prolongs the drug's effective duration in vivo, and more effectively suppresses bacterial growth and reproduction.^{[ 54 ]} To further verify the long‐term resistance control potential of the adjuvant‐antibiotic combination regimen, we conducted a 30‐day continuous induction experiment on wild‐type E. coli BW25113. In the minocycline monotherapy group, the bacteria developed obvious acquired resistance to minocycline, while no significant resistance was observed in the minocycline + C5 combination group (see Figure 3t).

Comprehensive activity evaluations revealed that although both C5 and C20 exhibited significant antibiotic‐potentiating activity, C5 maintained superior potentiation effects at an extremely low concentration of 1/64 MIC, whereas C20 exhibited markedly diminished activity at 1/16 MIC. Notably, C5 demonstrated exceptional biofilm inhibition efficacy, which not only broadens its effective dosage window, but also mitigates bacterial virulence by reducing biofilm formation. Based on these critical advantages, C5 was selected as the lead candidate for further evaluation.

2.7. Based on comprehensive analysis of computer‐aided ADMET prediction models and in vitro/in vivo biological experiments, this compound demonstrates excellent pharmacokinetic properties, low toxicity, and significant biological activities, indicating high drug‐like potential

Computer‐based ADMET prediction demonstrated that C5 exhibited favorable drug‐like properties, with its molecular structure strictly adhering to the physicochemical criteria of Lipinski's Rule of Five. Specifically, its Caco‐2 permeability and HIA (Human Intestinal Absorption) classification met the requirements for intestinal absorption of oral drugs, while its blood‐brain barrier penetration probability fell within the borderline range but did not reach the high‐risk threshold. Additionally, the low plasma clearance rate (<5 mL·(min·kg)⁻¹) of C5 indicated excellent metabolic stability, facilitating the maintenance of its effective in vivo concentrations (see Tables S5 and S6 and Data Figure S9, Supporting Information). Given the developable drug potential of this molecule, we subsequently conducted its extensive bioactivity evaluations, primarily focusing on in vivo safety assessment and pharmacodynamic validation.Computer‐based ADMET prediction demonstrated that C5 exhibited favorable drug‐like properties, with its molecular structure strictly adhering to the physicochemical criteria of Lipinski's Rule of Five. Specifically, its Caco‐2 permeability and HIA (Human Intestinal Absorption) classification met the requirements for intestinal absorption of oral drugs, while its blood‐brain barrier penetration probability fell within the borderline range but did not reach the high‐risk threshold. Additionally, the low plasma clearance rate (<5 mL·(min·kg)⁻¹) of C5 indicated excellent metabolic stability, facilitating the maintenance of its effective in vivo concentrations (see Tables S5 and S6 and Data Figure S9, Supporting Information). Given the developable drug potential of this molecule, we subsequently conducted its extensive bioactivity evaluations, primarily focusing on in vivo safety assessment and pharmacodynamic validation.

Our study systematically evaluated the safety of C5. At a concentration of 256 µg·mL⁻¹, no hemolytic effect was observed against murine or ovine erythrocytes, indicating the absence of significant membrane‐disruptive properties in this compound (see Figure 6d; Extended Data Figure S8c, Supporting Information). At 128 µg·mL⁻¹, C5 induced no cytotoxicity in human embryonic kidney (HEK293) cells, whereas C20 exhibited substantial cytotoxicity (see Extended Data Figure S8d, Supporting Information). To further assess its in vivo toxicity, we employed the Galleria mellonella larval model and monitored the body weight changes of mice following intraperitoneal administration of the compound (see Figure 4a). The results demonstrated minimal lethality of C5 toward G. mellonella even at a high dose of 2000 mg·kg⁻¹, with all treatment groups exhibiting high survival rates and robust viability (see Figure 4a,b; Extended Data Figure S8a, Supporting Information) and the body weight of C5‐treated mice showed no reduction but a significant increase after administration (see Figure 5c). 3D fluorescence whole‐slide imaging (3D‐WSI) analysis revealed preserved tissue architecture across all organs: alveolar integrity in the lung, orderly myocardial fiber alignment in the heart, normal splenic pulp distribution, regular hepatic trabeculae, and structurally sound renal glomeruli and tubules. Quantitative assessment of characteristic regions demonstrated no significant histopathological differences between C5‐treated and control groups, indicating no overt organ damage (see Figure 4h). Topical application of C5 (200 mg·kg⁻¹) on the depilated dorsal skin of mice caused no erythema or swelling after 10 days, and subsequent Hematoxylin and eosin (H&E) staining revealed intact skin tissue structure (see Figure 5 h). Subchronic toxicity of C5 was further evaluated through serum biochemical analysis (see Figure 4i). At 3 and 14 days after administration, hepatic and renal functions assessed via corresponding biomarkers such as alkaline phosphatase (ALP), blood urea nitrogen (BUN), and creatinine levels, and electrolyte balance remained comparable to those of the untreated group. Collectively, these findings indicate that C5 maintains a favorable safety profile at elevated doses, supporting its further evaluation for in vivo pharmacological applications.

Figure 6.

The compound possesses both the characteristic of complete non‐toxicity and significant biological activities in vivo. a) Daily inspection photos of mouse skin wound experiments, skin wound colony images, and tissue staining. W represents the sterile wound group, (6) represents the bacterial wound group, (5) represents the minocycline treatment group (10 mg·kg⁻¹), (4) represents the C5 compound treatment group (20 mg·kg⁻¹), and (3, 2, and 1) represent the combination groups of C5 (20, 10, and 5 mg·kg⁻¹) with minocycline (10 mg·kg⁻¹). b) Wound healing progression diagram in mice. c) Wound healing curve in mice. d) Hemolysis chart. e) Scatter plot of colony counts in skin wounds. f) Scatter plot of colony counts in bacteremia. g) Thermal imaging of bacterial colony counts in organs (A‐E: Heart, Liver, Spleen, Lungs, and Kidneys, W represents the trauma group). Statistical analysis was performed using One‐way ANOVA. “***” indicates p < 0.001. “**” indicates P < 0.005. “ns” indicates no significance.

Figure 4.

The compound possesses both the characteristic of complete non‐toxicity and significant biological activity in vivo. a) In vivo experimental flowchart. b) Acute toxicity of C5 and C20. c) Impact of bacterial infection on G. mellonella larvae survival at 48 h. Control: Larvae injected with bacterial suspension. PBS: Larvae injected with sterile PBS as negative control. d) H&E staining analysis of randomly selected G. mellonella larvae. e) Six G. mellonella larvae were randomly selected from each experimental group, homogenized, and subjected to bacterial colony counting. Representative petri dishes from the first experimental group were labeled and photographed for documentation. f) Bacterial load in different experimental groups of G. mellonella. g) Pharmacokinetic profiles of the compound in mice following i.p. and i.v. administration. The four panels represent elimination half‐life (t_1/2), systemic exposure (AUC_0‐∞), peak plasma concentration (C_max), and mean residence time (MRT_0‐∞), respectively. The labels 4c, 2c, and c denote concentration gradients, with “c” corresponding to 5 µg·mL⁻¹ for i.p. administration and 1 µg·mL⁻¹ for i.v. administration. h. Representative 3D fluorescence whole‐slide images (3D‐WSI) of major organs (heart, liver, spleen, lung, and kidney) from mice treated with compound C5. i) Liver function (alkaline phosphatase), kidney function (blood urea nitrogen), and electrolyte balance were evaluated following intraperitoneal administration of C5 (n = 6).

Figure 5.

Comprehensive in vivo pharmacodynamic evaluations were performed across multiple murine infection models via intraperitoneal administration of C5 and minocycline. a) In vivo experimental flowchart. b) Daily observations of wound healing progression and observations of Hematoxylin and Eosin (H&E) stained sections in mice. (A) represents the bacterial wound group, (B) represents the minocycline treatment group (10 mg·kg⁻¹), (C) represents the C5 treatment group (20 mg·kg⁻¹), (D–F) represent the combination groups of C5 (20, 10, and 5 mg·kg⁻¹) with minocycline (10 mg·kg⁻¹), and (G) represents the sterile wound group. c) Daily weight changes of mice. d) Wound healing curve in mice. e) Scatter plot of colony counts in skin wounds. f) Scatter plot of colony counts in bacteremia. g) Quantification of bacterial load in homogenized thigh muscle tissues from the neutropenic thigh infection model. h) Dermal toxicity of C5 assessed by visual observation and H&E staining. i) Plate culture images showing CFUs from blood samples in the systemic infection model. j) Quantification of CFUs in blood samples from the systemic infection model. k) Quantitative analysis of CFUs in spleen tissues. l) Quantitative analysis of CFUs in lung tissues. m) Quantitative analysis of CFUs in liver tissues. n) Quantitative analysis of CFUs in kidney tissues. o) Selected organ tissues were stained with Hematoxylin and Eosin (H&E) for the systemic infection model. The control group consists of normal mouse organ sections, the saline group represents the bacterial infection group injected with physiological saline, and the subsequent groups are treated with minocycline alone and a combination therapy of minocycline and C5. p) Expression profiles of relevant inflammatory factors across different groups in the systemic infection model. Statistical analysis was performed using One‐way ANOVA. “***” indicates P < 0.001. “ns” indicates no significance.

To investigate the therapeutic effect of C5 combined with antibiotics in the treatment of drug‐resistant bacterial infections, we established both G. mellonella larval infection and murine multi‐organ infection models (see Figures 4a and 5a). In G. mellonella infection experiment, the combination treatment group (C5 + antibiotic) and PBS control group maintained a 100% survival rate during the 48 h observation period. In contrast, single‐agent control groups (minocycline or C5 alone) exhibited significant mortality such as partial larval death and prevalent body rigidity in surviving individuals. Notably, under treatment with low‐dose of minocycline or C5, larval mortality markedly increased, with surviving individuals frequently showing loss of mobility (see Figure 4c). Homogenate analysis following 100‐fold dilution revealed that when the C5 dose was 100 mg·kg⁻¹, minocycline demonstrated effective in vivo bactericidal activity across a dose range of 100–0.1 mg·kg⁻¹. However, when the C5 dose decreased to 0.1 mg·kg⁻¹, the enhancement of antibacterial activity became substantially diminished even at a dose of 100 or 0.1 mg·kg⁻¹ of minocycline (see Figure 4e,f).

To further validate the mechanism of C5, we performed H&E staining on G. mellonella larvae (see Figure 4d). Results showed significant melanin deposition in both the E. coli‐infected group and minocycline monotherapy group (100 mg·kg⁻¹), whereas the tissue morphology of the C5‐antibiotic combination treatment group closely resembled that of the PBS control group. This findings strongly demonstrate that C5, as an antibiotic adjuvant, exerts synergistic therapeutic effect by inhibiting host hyperimmune responses.

We first performed pharmacokinetic (PK) studies to guide dose selection for subsequent in vivo experiments. The results showed that the compound exhibited a moderate half‐life of ≈11 h following intravenous administration, with comparable systemic exposures across different doses and a moderate bioavailability (≈32%). These findings indicate that plasma uptake remains relatively consistent among the tested doses (see Figure 4g; and Table S9, Supporting Information).

Based on the above results, we subsequently conducted in vivo experiments in a murine model, in which the maximum dose of C5 was set at 20 mg·kg⁻¹. The dose of minocycline was estimated based on available information to be within the minimum inhibitory concentration (MIC) range of 4–8 µg·mL⁻¹ in vitro. Therefore, we selected an in vitro reference concentration of 1–2 µg·mL⁻¹ for in vivo extrapolation, corresponding to an approximate in vivo dose of 10 mg·kg⁻¹.

In addition, to further validate the therapeutic potential of C5 against systemic infections, multiple murine infection models were established, including intraperitoneal E. coli infection to induce systemic bacterial dissemination, neutropenic thigh infection (modeling and treatment of neutropenic thigh infections), and cutaneous wound infection. In these models, in vivo studies were primarily conducted via intraperitoneal administration (see Figure 5a). Upon intraperitoneal administration, the combination of C5 and minocycline markedly eradicated E. coli colonization in both the bloodstream and major organs in the systemic infection model, whereas monotherapy with either agent resulted in bacterial loads comparable to the saline control (see Figure 5i–n). The histological examination of selected organs revealed that, compared with the PBS or monotherapy groups, the combination therapy group exhibited markedly reduced inflammatory cell infiltration and better protection of tissue integrity (see Figure 5o). To further validate the therapeutic efficacy of C5, we supplemented the expression profiles of relevant inflammatory factors across different treatment groups. The results demonstrated that the combination treatment significantly reduced the production of inflammatory cytokines (see Figure 5p). To further evaluate the compound's tissue penetration and antibacterial performance at infected site, a neutropenic thigh infection model was established. The combination treatment achieved pronounced bacterial clearance within 24 h (see Figure 5g). In the cutaneous wound infection model, dynamic monitoring over seven days revealed that the wound‐healing rate in the combination groups was comparable to that observed in the uninfected control (see Figure 5b,d). The histological examination of wound tissues collected on day 7 demonstrated markedly enhanced epithelial regeneration and reduced inflammatory infiltration in the combination group compared with PBS or monotherapy groups (see Figure 5b). Moreover, the bacterial loads in blood and skin tissues were largely comparable among the different combination‐dose groups (see Figure 5e,f). This consistency may be attributed to the pharmacokinetic properties of C5, as intraperitoneal administration at 5 mg·kg⁻¹ was sufficient to reach plasma saturation levels, thereby minimizing dose‐dependent differences in tissue bacterial clearance and resulting in bacterial levels similar to those observed in uninfected control.

In a mouse wound infection model by wild‐type E. coli, Groups 6 (control), 5 (C5 monotherapy), and 4 (minocycline monotherapy) exhibited significantly delayed wound healing. During days 2–3 post‐infection, purulent inflammation with yellow pus accumulation was observed in subcutaneous tissues of these groups. In contrast, the combination therapy group (C5 + minocycline, hereafter “combination group”) demonstrated accelerated wound healing with effective suppression of purulent exudate. Dynamic observation revealed gradual transition of wound coloration from deep red at initial infection to pink in the combination group, achieving near‐complete epithelialization by day 7 (see Figure 6a). Quantitative analysis of wound area (see Figure 6b,c) indicated that the healing rates in the combination groups matched or exceeded natural skin regeneration levels. The therapeutic mechanism may involve two aspects: 1) The combination regimen enhances bactericidal efficacy through rapidly eliminating high concentrations of wild‐type E. coli at wound sites; 2) Environmental microbiota inhibition simultaneously suppresses opportunistic pathogen colonization, which reduces secondary infection interference. Subsequent bacterial culture of wound homogenates confirmed near‐complete bacterial clearance in high‐dose C5 combination groups. A dose‐dependent increase in bacterial counts was observed as the concentration of C5 decreased. Nonetheless, all combination groups maintained significantly lower colony‐forming units compared to controls (see Figure 6a,e).

To validate the bacteremia model, we collected blood from the infected mice via eye puncture. The bacterial load in the blood mirrored that in the skin infection site (see Figure 6f). We then triturated the organs from sacrificed mice to assess the effects of combined treatment on bacterial infections in the heart, liver, spleen, lungs, and kidneys. Clinical and pathological analysis showed that E. coli primarily infected the spleen, lungs and kidneys.^{[ 55 ]} In the absence of a functional immune system, bacteria can infect multiple organs. In the three control groups, all organs except the heart exhibited significant bacterial infections. In contrast, the C5 combined treatment group showed almost no bacterial infections in the heart, liver, spleen, lungs, or kidneys (see Figure 6g; Extended Data Figure S8b, Supporting Information).

H&E staining revealed more inflammatory cells and necrotic tissue in the wounds of the three control groups. However, after the C5‐antibiotic combination treatment, the wound site almost completely restored the structure of the dermis and epidermis, while promoting the formation of new blood vessels and hair follicles (see Figure 6a). In conclusion, these results suggest that C5 enhances the bactericidal activity of antibiotics, overcomes bacterial resistance, and holds promise as a therapeutic strategy for treating drug‐resistant infections.

3. Discussion

This study introduces a novel therapeutic strategy to address the challenge of multi‐drug resistant E. coli through the combined inhibition of efflux pumps and modulation of the cell membranes. Currently, the spread of antibiotic resistance among Gram‐negative bacteria has become a major global public health threat, with very limited effective treatment options available in the clinic. In this context, exploring novel mechanisms of action is urgently needed to combat resistant bacterial infections. Efflux pumps are critical determinants of antibiotic resistance in E. coli and other Gram‐negative bacteria, as they expel antibiotics from the bacterial cells, thus mitigating antibiotic efficacy. Our research confirms that C5 specifically binds to AcrB and inhibits its normal function, thereby significantly reducing bacterial resistance to antibiotics. This effect is particularly pronounced in highly resistant bacterial strains, further supporting the potential of C5 as an antibiotic adjuvant. In parallel, C5 enhances cell membrane permeability and facilitates greater antibiotic uptake. This mechanism is consistent with current literature on the role of membrane permeability in bacterial resistance. Notably, C5 increases membrane permeability without compromising membrane integrity, which minimizes cytotoxicity and enhances antibiotic effectiveness, and prevents the development of bacterial resistance.

Building on its in vitro activity results, we extended our investigation into the in vivo efficacy of C5 in a murine infection model. The results demonstrated that C5 significantly potentiated the effectiveness of antibiotics, reduced bacterial load, and extended survival time in infected animals. These findings indicate that C5 not only enhances antibiotic efficacy in vitro but also improves therapeutic outcomes in vivo, thereby alleviating the burden of resistant bacterial infections. Furthermore, C5 did not induce significant toxicological effects, which further supports its potential as a safe antibiotic adjuvant. Through a combination of PAE and bacterial resistance induction assays, we also established that C5 could further prolong the PAE of antibiotics against E. coli. This sustained effect is likely due to the prolonged impact of C5 on both the cell membranes and the efflux pump system.

In conclusion, C5 described in this study, with its dual‐targeting mechanism against the efflux pumps and cell membranes, offers a promising strategy for addressing the growing problem of multi‐drug resistant bacteria. Future studies will focus on its potential in combination with other antibiotics, particularly in diverse infection models. In addition, advancing the “pump‐membrane dual blockade” therapy requires exploration of several key directions. First, it is essential to understand the synergistic mechanism of simultaneously targeting efflux pumps and bacterial membranes. For example, studies can investigate whether membrane perturbation alters pump conformational dynamics or substrate‐binding affinity, and vice versa. Second, the balance between efficacy and specificity needs optimization. The membrane‐targeting component should selectively disrupt bacterial membranes without affecting host cells. At the same time, efflux pump inhibitors should be designed to recognize bacterial‐specific pockets, avoiding cross‐reactivity with human transport proteins. Finally, a standardized preclinical evaluation framework should be established. This framework should assess not only acute efficacy but also long‐term resistance evolution and tissue‐specific delivery efficiency. Such evaluations will help ensure that these therapies can be effectively translated into clinical applications.

4. Experimental Section

Medicinal Chemistry

The detailed procedures for chemical synthesis and structural confirmation of compounds are provided in the Extended Data information.

Bacterial Strains, Cells, and Reagents

The bacterial strains used in this study include standard strains as well as clinical or environmental isolates. The HEK293 cell line (RRID: CVCL_0045), purchased from Wuhan Servicebio Technology Co., Ltd., was also utilized. The cell line was confirmed to be free of contamination (including mycoplasma, bacteria, fungi, and yeast) as verified by the supplier's quality control testing prior to use. Compounds and reagents were sourced from Shanghai Titan Scientific Co., Ltd., Biotopped Pharmaceutical Technology Co., Ltd., Beijing Bailingwei Technology Co., Ltd., Tianjin Fuyu Fine Chemical Co., Ltd., and China National Pharmaceutical Group Chemical Reagent Co., Ltd. All new compounds were independently designed and synthesized. Animal experiments were conducted at Tianjin Huiyude Biological Technology Co., Ltd., among others.

Antimicrobial Assay

The intrinsic antibacterial activity of the compound was assessed by measuring its MIC.^{[ 56 ]} The MIC was determined using the standard two‐fold broth dilution method. After revival, bacterial strains were cultured to prepare a 0.5 McFarland bacterial suspension (1 MCF = 3 × 10⁸ CFU mL⁻¹). The compound stock solution was serially diluted two‐fold in a 96‐well plate, followed by the addition of an appropriate bacterial suspension. The plate was incubated at 37 °C for 18 to 20 h. The MIC value was defined as the lowest concentration of the compound that visibly inhibited bacterial growth. The experiment was performed three times to ensure the reproducibility of the results. This method ensures accurate concentration gradients and reproducibility of results.

Evaluation of Antimicrobial Sensitization Assay

To directly evaluate the potentiation fold of bacterial sensitivity when the compound is combined with antibiotics, different types of antibiotics were selected and investigated the compound's antibacterial potentiation potential using high‐concentration primary screening, gradient dilution checkerboard methods and Measurement of IC₉₀ values.^{[ 57} , ⁵⁸ , ^{59 ]}

1. High‐Concentration Primary Screening Method: The antibacterial potentiation of the compound against wild‐type E. coli BW25113 was assessed by comparing MIC of antibiotics alone and in combination with the compound (1/4 MIC). MIC was determined using a high‐concentration primary screening method in 96‐well plates. Antibiotics were two‐fold serially diluted (columns 1–10) with equal volumes of compound solution (1/4 MIC). Columns 11 and 12 served as positive and negative controls. After adding bacterial suspension, plates were incubated at 37 °C for 18–20 h. The lowest concentration inhibiting visible growth was recorded as the MIC, and FICI values were calculated. The experiment was performed three times to ensure the reproducibility of the results.

   (1)

2. Gradient Dilution Checkerboard Method. The checkerboard dilution method was employed to evaluate the synergistic effects of the compound with antibiotics and to explore potential targets. Wild‐type E. coli BW25113 and BW25113 (ΔacrB) were tested following previously described procedures for culture and suspension preparation. In 96‐well plates, antibiotics were serially diluted across columns, while compound concentrations ranging from 1/4 to 1/64 MIC were applied across rows. Columns 11 and 12 served as positive and negative controls. Plates were inoculated and incubated at 37 °C for 18–20 h. MIC was defined as the lowest concentrations inhibiting visible growth. Minocycline, oxacillin, and linezolid were tested as representative antibiotics, with all assays performed under biosafety conditions. The experiment was performed three times to ensure the reproducibility of the results.

3. Measurement of IC₉₀ Values Using OD₆₀₀ Assay: The IC₉₀ values of antibiotics alone and in combination with the compound (1/4 MIC) were determined to evaluate its antibacterial potentiation against wild‐type E. coli BW25113. Antibiotics were serially diluted in 96‐well plates, with compound solution (1/4 MIC) added to columns 1–10, and columns 11 and 12 served as positive and negative controls. Bacterial suspensions, prepared as previously described, were inoculated and incubated at 37 °C for 18–20 h. Growth was monitored at OD₆₀₀, and IC₉₀ was defined as the lowest antibiotic concentration inhibiting ≧90% of growth relative to the drug‐free control. All assays were conducted under biosafety conditions, and FICI values were calculated.

Nile Red Efflux Assay

Ideal AcrB inhibitors block substrate efflux, causing intracellular accumulation.^{[ 40} , ⁴¹ , ^{42 ]} To test specificity, experiments were conducted using wild‐type E. coli BW25113 and its ΔacrB mutant strain. Overnight cultures were centrifuged and resuspended twice in 20 mm phosphate buffer (pH 7.0) with 1 mm MgCl₂, adjusted to OD660 = 1.0. Samples in glass tubes received CCCP, test compound, and 5 µm Nile red, then incubated at 37 °C with shaking for 3 h. Cells were centrifuged, resuspended, and fluorescence measured using a spectrophotometer with an excitation wavelength of 552 nm and an emission wavelength of 636 nm. After 100 s, 1 m glucose (≈50 mm final) was added, and fluorescence was recorded for 200 s to monitor efflux.

EtBr Efflux Assay

Wild‐type E. coli BW25113 was cultured in 5 mL LB medium until the logarithmic growth phase, then centrifuged at 3500 rpm for 15 min. After adding EtBr (5 µm), the suspension was incubated at 25 °C for 60 min to ensure effective dye loading. The bacterial cells were then centrifuged again, resuspended in cold PBS, and transferred to a black 96‐well plate for detection. The compound was tested at different concentrations (4–128 µg·ml⁻¹) and incubated at 37 °C for 8 min. Fluorescence intensity was recorded every minute until the signal reached equilibrium. The experiment was performed under both glucose‐free and glucose conditions. Fluorescence intensity was measured using a Cytation 5 multi‐mode reader, with excitation at 530 nm and emission at 585 nm, for 30 min. Fluorescence values with glucose were subtracted from those without glucose. CCCP and PAβN were used as positive controls.^{[ 60 ]} All experiments were performed in six biological replicates (n = 6).

Tetracycline Accumulation Assay

Tetracycline accumulation was measured using its strong fluorescence in the near‐visible range.^{[ 54} , ^{61 ]} Wild‐type E. coli BW25113 was cultured to OD₆₀₀ = 1 and incubated with 50 mg·l⁻¹ ERV in the presence or absence of compound for 15 min. Bacteria were collected, washed with PBS, and lysed. The lysate was centrifuged at 13,000 g for 2 min to collect the supernatant. Methanol was added to the pellet, and centrifuged again, and the supernatants were combined. Residual debris was removed by centrifuging at 13,000 g for 10 min. ERV fluorescence was detected at 400/520 nm using a fluorometer.

Inner Membrane Proton Gradient Assay

The AcrAB‐TolC efflux system utilizes the proton motive force to expel substrates without ATP hydrolysis. diSC₅, a membrane potential‐sensitive dye, self‐quenches under normal membrane potential, reducing fluorescence. Disruption of the proton gradient expels diSC₅, and significantly increases fluorescence. Wild‐type E. coli BW25113 was prepared and adjusted to 0.33 MCF in a buffer containing HEPES, glucose, and KCl. After incubation with diSC₅ and EDTA at 37 °C for 1 h, fluorescence intensity was measured at 622/670 nm every 2 min for 8 min. Test compound was added, and fluorescence intensity was monitored every 2 min for 12 min. Each concentration was tested in triplicate^{[ 62 ]} and all experiments were performed in six biological replicates (n = 6).

Membrane Permeability Assay

To verify that the target compounds can strongly affect bacterial membrane permeability and facilitate the penetration of antimicrobial agents into bacterial cells, thereby exhibiting synergistic antibiotic activity, inner and outer membrane permeability assays were conducted:

1. The Method of Outer Membrane Permeability: Wild‐type E. coli BW25113 was subcultured and collected. Bacterial cells were washed with a 1:1 solution of 5 mm glucose and 5 mM HEPES (pH 7.2), centrifuged at 3500 rpm for 5 min, and resuspended in the same solution to 0.33 MCF. A total of 150 µL of cell suspension was added to each well of a 96‐well black plate with a clear bottom, followed by adding 50 µL of 10 µm NPN dye.^{[ 43 ]} After mixing, plates were pre‐incubated at 37 °C. Fluorescence intensity was measured every 2 min for 8 min (excitation: 350 nm, emission: 420 nm). Test compound was then added to each well to the target concentration, and mixed gently. Fluorescence intensity was monitored every 2 min for another 12 min. All experiments were performed in six biological replicates (n = 6).

2. The Method of Inner Membrane Permeability: To assess the effect of compound on the inner membrane permeability of wild‐type E. coli BW25113, logarithmic‐phase bacterial cultures were collected, centrifuged at 3500 rpm, washed with 5 mm glucose and HEPES buffer (pH 7.2), and adjusted to a concentration of 1 × 10⁸ CFU mL⁻¹. Then, 150 µL of the bacterial suspension was mixed with 50 µL of propidium iodide (PI, 10 µm), incubated at room temperature for 30 min, and placed in a black 96‐well plate. Baseline fluorescence intensity was recorded (excitation at 535 nm, emission at 617 nm) every 2 min for 8 min. Afterward, 10 µL of compound solution was added, and fluorescence intensity was recorded for an additional 12 min. All experiments were performed in six biological replicates (n = 6).

Protein Leakage Assay

Wild‐type E. coli BW25113 was cultured to the logarithmic phase, then collected and resuspended in medium containing the compound, and incubated for 2 h. After treatment, the bacteria were removed by centrifugation, and the supernatant was collected. Protein leakage was measured using the Bradford method by mixing the supernatant with Bradford reagent. The absorbance is measured at 595 nm using a spectrophotometer, and the protein concentration in the supernatant is calculated using a bovine serum albumin (BSA) standard curve.^{[ 63} , ^{64 ]} All experiments were performed in six biological replicates (n = 6).

Scanning Electron Microscope (SEM) Analysis

SEM analysis includes SEM observation of Bacterial Envelope Architecture and SEM observation of bacterial biofilm architecture.

1. SEM Observation of Bacterial Envelope Architecture: To investigate the effect of the compound on the inner and outer membranes of wild‐type E. coli BW25113, SEM was used to observe bacterial morphology. The revived strain was cultured in LB liquid medium for 18–20 h, then diluted to OD620 = 0.05 and further cultured to the logarithmic phase. The culture was divided into two groups, one with C5 and the other with DMSO as a control. After 24 h of incubation, bacteria were collected by centrifugation and fixed with glutaraldehyde. After gradient dehydration, the samples were dried at 37 °C and gold‐coated. Finally, the samples were observed using SEM at 4.0 kV.

2. SEM Observation of Bacterial Biofilm Architecture: Wild‐type E. coli BW25113 was streaked onto agar plates. The culture was transferred to LB broth and incubated in a shaking incubator at 37 °C for 20 h. The bacterial suspension was adjusted to OD620 = 0.05, cultured to the logarithmic growth phase, and diluted again to the same OD value. Three milliliters aliquot was added to a sterile EP tube, mixed with test compound or control drug to a final concentration of 128 µg·mL⁻¹, and incubated with positive/negative controls in a shaking incubator at 37 °C for 24 h. After discarding the supernatant, 4 mL of 2.5% glutaraldehyde was added to fix the biofilm for 30 min, and washed three times using PBS. Dehydration was performed using 4 mL absolute methanol for 15 min, and the sample was dried at 30 °C in a oven. The dried biofilm was mounted on conductive adhesive, sputter‐coated with platinum for 30 min, and prepared for SEM analysis.

ROS Measurement

ROS levels in wild‐type E. coli BW25113 were measured as described in an article^{[ 65 ]} with modifications. Bacteria were cultured on C agar plates, washed with PBS, and resuspended in PBS to an OD₆₀₀ of 0.5. The suspension was incubated with 10 µm DCFH‐DA at 37 °C for 1 h. After incubation, the bacteria were treated with levofloxacin, minocycline, and their combinations at sub‐inhibitory concentrations, and transferred to a 96‐well black plate. Antioxidant NAC (6 mm) was used as a control. Fluorescence intensity was measured at excitation/emission wavelengths of 488/535 nm after an additional 2‐h incubation. All experiments were performed in six biological replicates (n = 6).

Inhibition of Biofilm Formation Assay

The crystal violet staining method was used to determine the inhibitory effect of the compound on biofilm formation in wild‐type E. coli BW25113.^{[ 66 ]} Bacteria in the logarithmic phase were prepared into a suspension of 1 × 10⁶ CFU mL⁻¹ and added to a 96‐well plate with 50 µL of bacterial suspension and 50 µL of serially diluted compound solution, with the final concentration chosen based on the compound properties. The 96‐well plate was incubated at 37 °C for 24 h. After incubation, the supernatant was discarded, and the wells were washed 2–3 times with PBS. The biofilm was fixed with 120 µL of absolute methanol for 15 min, and then stained with 120 µL of 0.1% crystal violet for 30 min. After staining, the wells were washed with distilled water until the wash solution was colorless and dried. Finally, the biofilms were dissolved with 120 µL of 1% SDS at 37 °C for 30 min, and the absorbance was measured at 620 nm using a microplate reader. All experiments were performed in six biological replicates (n = 6).

Removal of Mature Biofilm Assay

The crystal violet staining method was used to determine the eradication activity of the compound on pre‐formed biofilms of wild‐type E. coli BW25113.^{[ 67 ]} Bacteria at a concentration of 1 × 10⁶ CFU mL⁻¹ were cultured with broth for 24 h. After treatment with the compound or PBS for 24 h, the biofilms were fixed, stained with crystal violet for 30 min, washed until colorless, dried, and then dissolved with 1% SDS. Absorbance was measured at 620 nm. All experiments were performed in six biological replicates (n = 6).

Time‐Kill Curve Assay

In this experiment, the bactericidal kinetics of wild‐type E. coli BW25113 was assessed by continuous culturing in the presence of the compound. A bacterial suspension of 5 × 10⁵ CFU mL⁻¹ was inoculated into a 96‐well plate, and the compound was added at concentrations of MIC/2, MIC, 2MIC, and 4MIC, with DMSO as a control. Samples were taken every 2 h, diluted tenfold, and plated on sterile agar plates for incubation at 37 °C for 18–24 h. Bacterial concentrations (CFU mL⁻¹) were calculated from colony counts, and bactericidal curves were plotted using Origin 2022 with three parallel experiments. All experiments were performed in three biological replicates (n = 6).

Growth Kinetics Curve Assay

Wild‐type E. coli BW25113 was taken from a −78 °C freezer, streaked onto an agar plate, and incubated at 37 °C for 18–20 h, followed by subculturing in LB broth for an additional 18–20 h. After incubation, the bacterial suspension was diluted to OD620 = 1 and aliquoted into a 96‐well plate with 200 µL per well, with LB broth added to the un‐inoculated wells as a control. The 96‐well plate was incubated at 37 °C for 25 h, and OD620 values were measured at various time points to generate growth curves. All experiments were performed in six biological replicates (n = 6).

Antibiotic PAE Assay

Wild‐type E. coli BW25113 was cultured to mid‐log phase (OD620 = 0.25), revived on nutrient agar, and subcultured into LB liquid medium. The bacterial cultures were treated either with minocycline alone (4 µg·mL⁻¹) or in combination with sub‐minimal inhibitory concentrations (sub‐MIC) of compounds C5, C15, and C20 and the sub‐MIC of minocycline., and incubated at 37 °C with shaking for 2 h. The culture was then diluted 50‐fold, and 50 µL was added to 2450 µL of LB broth to eliminate drug carryover effects. The diluted culture was distributed into a 96‐well plate, and OD₆₀₀ value was measured after 25 h using a microplate reader.^{[ 68 ]} All experiments were performed in six biological replicates (n = 6).

Hemolysis Assay

This experiment evaluates the hemolytic activity of the target compound on red blood cells using fresh blood from male Kunming mice and purchased sheep blood. Mouse blood was extracted from the eye socket, centrifuged to separate red blood cells, and washed with PBS to prepare a 5% red blood cell suspension. The compound solutions at different concentrations were mixed with the red blood cell suspension, incubated for 1 h, and centrifuged. The supernatant's OD value is measured to calculate the hemolysis rate. 2% Triton X‐100 served as a positive control, and PBS as a negative control. All experiments were performed in six biological replicates (n = 6).

$$\mathrm{Hemolysis}\%=\left[\left(\mathrm{O}{\mathrm{D}}_{\mathrm{Compound}}-\mathrm{O}{\mathrm{D}}_{\mathrm{PBS}}\right)/\left(\mathrm{O}{\mathrm{D}}_{\mathrm{PBS}}-\mathrm{O}{\mathrm{D}}_{\mathrm{Contral}}\right)\right]\times 100\%$$ (2)

Cytotoxicity Experiment

Cytotoxicity was assessed using the CCK‐8 kit to evaluate compounds' toxicity against human embryonic kidney 293 cells (HEK293 cells). HEK293 cells in logarithmic growth phase were seeded in 96‐well plates (5 × 10³ cells per well). After 24‐h incubation, the medium was replaced with fresh medium containing either serially diluted compounds (8‐128 µg·mL⁻¹) or 0.1% DMSO (vehicle control), followed by further 24‐h culture. Subsequently, 10 µL CCK‐8 reagent was added to each well and incubated at 37 °C for 2 h. Absorbance was measured at 450 nm using a microplate reader. Experiments were performed with six technical replicates in three independent experiments. Compounds demonstrating >80% viability at 128 µg·mL⁻¹ were considered non‐cytotoxic, and calculated values exceeding 100% were recorded as 100%. All experiments were performed in six biological replicates (n = 6).

Acute Toxicity Assay

To evaluate the in vivo acute toxicity of the compound, an acute toxicity assay was performed using G.mellonella and mice.

1. Acute Toxicity Test in Galleria mellonella Larvae: Eight healthy larvae (250–320 mg, excluding those with cuticle darkening) were randomly assigned to each group. The compound, dissolved in 10% DMSO or PBS, was injected into the hemolymph through the last pair of forelegs (left abdomen) using a 300 µL insulin syringe, at doses of 2000, 1000, 500, 200, 100, 50, 25, and 12.5 mg·kg⁻¹. Survival was monitored daily for 5 days. According to established criteria, if no mortality occurs at 2000 mg·kg⁻¹ or more than 5 larvae survive per group after 5 days, the compound is considered non‐toxic at that dose; if 3 or more larvae die at any dose, it is classified as highly toxic.^{[ 69} , ⁷⁰ , ^{71 ]}

2. Acute Toxicity Test in Mice: While conducting the in vivo infection experiment, continuous monitoring of body weight changes in the drug‐treated mice can not only reflect the therapeutic effect of the compound on the infection but also evaluate the compound's own impact on the mice's body weight, thereby indirectly reflecting its acute toxicity characteristics.

Organ Toxicity Assay

The in vivo organ toxicity study was conducted by Tianjin Jinke Biological Technology Co., Ltd. Mice (n = 3) were administered the compound through both intraperitoneal and tail vein injection at the same dose levels. Mice were monitored daily for severe distress (e.g., lethargy, >20% weight loss, impaired access to food/water, or severe wound deterioration) and humanely euthanized if criteria were met. Organ slice scanning was performed using WSI analysis.

Subchronic Toxicity Assay

Subchronic toxicity was evaluated on female Balb/c mice (4 weeks old, 18–22 g) according to the FDA guidelines for subchronic toxicity testing in rodents. Mice (n = 6) received intravenous injection of the polymer solution (20 mg·kg⁻¹ in 200 µL sterile PBS, pH 7.4), while control mice only received vehicle. Blood samples were collected from the orbital sinus on days 2 and 14 post‐administration to measure ALP, BUN, and electrolyte levels (Na⁺, K⁺, and Cl⁻). Data were expressed as mean ± SD (n = 6), and statistical significance was determined using Student's t‐test (p < 0.05).^{[ 80 ]}

Skin Irritation Assay

Skin irritation was evaluated in mice following topical application of the compound. Each mouse received 200 µL of the compound solution (200 mg·kg⁻¹), which was evenly applied to the shaved dorsal skin. The treated area was observed daily for 7 days to see if there are any signs of erythema, edema, or lesions.^{[ 80 ]}

G.Mellonella Larval Infection

Healthy final‐instar larvae (250 ± 20 mg) were pre‐adapted at 28 °C for 24 h, followed by injection of 10 µL wild‐type E. coli BW25113 suspension (5 × 10^⁵ CFU/larva) into the hemocoel of the fourth right posterior abdominal segment. Experimental groups included PBS negative control, bacterial positive control, monotherapy (C5 or minocycline), and combination treatment groups (C5 with gradient concentrations of minocycline), The dose screening adopted a concentration gradient design from high to low, with clearly visible bacterial colonies to the naked eye after 1000‐fold dilution of the larval tissue suspension as the basis for setting the initial concentration. Corresponding drugs (10 µL) were administered into the fourth left posterior abdominal segment within 4 h post‐infection. Therapeutic effects were quantified through 48 h survival monitoring (death defined by complete loss of motor function and stimulus response), combined with multiparametric analysis including melanization degree, locomotor velocity, and mechanical response latency. The double‐blind method and biological replicates (10 larvae per group) were implemented throughout the experiment to ensure data reliability.^{[ 72} , ^{73 ]}

Mouse Infection Model

1. Murine Wound Infection Model via Intraperitoneal Administration: Female Kunming mice (6‐8 weeks old, 28.0 ± 2 g) were obtained from a licensed facility and ethically approved by the Shandong Academy of Chinese Medicine. After a week of acclimation with free access to water and standard diet, mice were randomly assigned to seven groups (n = 6): wound, PBS, MIN, C5, C5+MIN‐1, C5+MIN‐2, and C5+MIN‐3 groups. Under ether anesthesia, the dorsal fur was shaved and full‐thickness skin wounds (10 ± 1 mm) were created. All groups except the wound group were inoculated with 50 µL of bacterial suspension (1.0 × 10^{^9} CFU). Three hours after infection, mice received intraperitoneal injections of PBS, MIN (10 mg·kg⁻¹), C5 (20 mg·kg⁻¹), or a combination of C5 and MIN at varying doses. Wound healing was monitored on days 0, 1, 3, and 7 by photography and measurement of wound area. At the endpoint, one mouse per group was used for H&E histological analysis, while remaining wound tissues were homogenized for bacterial enumeration. Blood samples were collected for bacteremia analysis, and major organs were harvested to evaluate bacterial distribution. Mice were observed daily for signs of severe distress (e.g., lethargy, more than 20% body weight loss, inability to access food/water, or wound deterioration), and were humanely euthanized if humane endpoint criteria were met.

2. Murine Wound Infection Model via Topical Administration: Female Kunming mice (6–8 weeks old, 28.0 ± 2 g) were obtained from a licensed facility and ethically approved by the Shandong Academy of Chinese Medicine. After one week of acclimation with free access to water and standard diet, mice were randomly assigned to seven groups (n = 6): wound, PBS, MIN, C5, C5+MIN‐1, C5+MIN‐2, and C5+MIN‐3. Under ether anesthesia, dorsal fur was shaved and full‐thickness skin wounds (10 ± 1 mm) created. All groups except the wound group were inoculated with 50 µL bacterial suspension (1.0 × 10⁹ CFU). Three hours later, wounds were treated with PBS, MIN (10 mg·kg⁻¹), C5 (20 mg·kg⁻¹), or combinations of C5 and MIN at varying doses. Wound healing was monitored on days 0, 1, 2, 3, 4, and 7 by photography and size measurement. At endpoint, one mouse per group was used for H&E histology; remaining tissues were homogenized for bacterial counts. Blood was collected for bacteremia analysis, and major organs were harvested to assess bacterial distribution. Mice were monitored daily for severe distress (e.g., lethargy, >20% weight loss, impaired access to food/water, or severe wound deterioration) and humanely euthanized if criteria were met.^{[ 74} , ⁷⁵ , ^{76 ]}

3. Systemic Bacterial Infection Model Induced by Intraperitoneal Injection of E. coli: All animal experiments were conducted in compliance with institutional ethical guidelines and approved by the Animal Ethics Committee of the Shandong Academy of Chinese Medicine. Sepsis was induced by intraperitoneal injection of bacterial suspension containing 6 × 10⁸ CFU. Two hours post‐infection, mice (n = 6) received intraperitoneal injections of either vehicle control or the compound at 20, 10, or 5 mg·kg⁻¹, following the same dosing regimen as in the skin infection model. Bacteremia levels were determined by collecting blood from the orbital sinus, serially diluting, and plating on agar. Among these steps, the detection of inflammatory factors was performed by Sevencyd Biotechnology Co., Ltd. At the experimental endpoint, major organs were harvested‐part of each was homogenized for bacterial enumeration, while intact organs were fixed for histological analysis.^{[ 80 ]}

4. Modelling and Treatment of Neutropenic Thigh Infection in Mice: Mice were intraperitoneally injected with cyclophosphamide at concentrations of 150 and 100 mg·kg⁻¹ at days 0 and 3 before bacterial administration, respectively, to induce neutropenia. E. coli BW25113 was suspended in sterile PBS and adjusted to a concentration of 10⁸ CFU per infection site. The bacterial suspension was then injected into the left thigh of mice in the corresponding experimental group. Two hours post‐infection, mice (n = 6) received an intraperitoneal injection of either vehicle control or the compound at 20, 10, or 5 mg·kg⁻¹, following the same dosing regimen as in the skin infection model. At 24 h post‐infection, the mice were euthanized. The thigh wound tissues were collected, weighed, homogenized, and serially diluted in sterile PBS. The number of bacterial colonies in each thigh wound tissue was calculated by diluting the thigh wound homogenate.^{[ 79 ]}

Metabolic Stability Assay

The in vivo metabolism study was performed under contract by Tianjin Jinke Biological Technology Co., Ltd. Male Kunming mice (n = 3) received the compound via intraperitoneal injection (20, 10, and 5 mg·kg⁻¹) and tail vein injection (4, 2, and 1 mg·kg⁻¹) to evaluate its pharmacokinetics. Mice were monitored daily for severe distress (e.g., lethargy, >20% weight loss, impaired access to food/water, or severe wound deterioration) and humanely euthanized if criteria were met.

In Silico Prediction of ADMET

ADMETlab 3.0 predicts over 40 ADMET parameters (e.g., solubility, and hepatotoxicity, etc.) by inputting molecular structures (SMILES or files), leveraging machine learning models and physicochemical calculations. Its workflow includes data preprocessing, feature extraction, parallel multi‐model computation, integration of internal databases and experimental validation, generating interactive reports with risk assessments to optimize early‐stage drug development.^{[ 77 ]}

Picture Production

Thanks for the graphic production license provided by the BioGDP and Biorender platform. All uses require permission.^{[ 78 ]}

Conflict of Interest

The authors declare no conflict of interest.

Author Contributions

S.M. conceived and supervised the project. J.D. performed all the experiments. J.D., T. G., Y.H.,Y.M.,W.C., J.X., J.J., E.D.,Y.K., H.D., and W.Z. did the data analysis. J.D. wrote the manuscript. All authors read and approved the manuscript.

Supporting information

Supporting Information

Data Availability Statement

Research data are not shared.