Novel nitroxoline derivative combating resistant bacterial infections through outer membrane disruption and competitive NDM-1 inhibition

ABSTRACT

New Delhi metallo-β-lactamase-1 (NDM-1) has rapidly disseminated worldwide, leading to multidrug resistance and worse clinical prognosis. Designing and developing effective NDM-1 inhibitors is a critical and urgent challenge. In this study, we constructed a library of long-lasting nitroxoline derivatives and identified ASN-1733 as a promising dual-functional antibiotic. ASN-1733 can effectively compete for Ca²⁺ on the bacterial surface, causing the detachment of lipopolysaccharides (LPS), thereby compromising the outer membrane integrity and permeability and exhibiting broad-spectrum bactericidal activity. Moreover, ASN-1733 demonstrated wider therapeutic applications than nitroxoline in mouse sepsis, thigh and mild abdominal infections. Furthermore, ASN-1733 can effectively inhibit the hydrolytic capability of NDM-1 and exhibits synergistic killing effects in combination with meropenem against NDM-1 positive bacteria. Mechanistic studies using enzymatic experiments and computer simulations revealed that ASN-1733 can bind to key residues on Loop10 of NDM-1, hindering substrate entry into the enzyme's active site and achieving potent inhibitory activity (K_i= 0.22 µM), even in the presence of excessive Zn²⁺. These findings elucidate the antibacterial mechanism of nitroxoline and its derivatives, expand their potential application in the field of antibacterial agents and provide new insights into the development of novel NDM-1 inhibitors.

GRAPHICAL ABSTRACT

KEYWORDS:

• Nitroxoline
• ASN-1733
• NDM-1
• inhibitor
• antibiotic resistance

Introduction

NDM-1, a prominent member of the metallo-β-lactamase (MBL) family, is characterized by the two Zn²⁺ ions in its active site and the hydrolysis ability of nearly all β-lactam antibiotics including carbapenems [1]. The hydrolysis activity of NDM-1 can be inhibited by EDTA, while it remains unaffected by currently available β-lactamase inhibitors in clinical use, such as clavulanic acid, sulbactam, and avibactam [2]. NDM-1 is predominantly detected within Enterobacteriaceae, including Escherichia coli and Klebsiella pneumoniae [3]. Enterobacteriaceae are capable of causing various community-acquired and nosocomial infections, such as sepsis, intra-abdominal infections, pneumonia, and urinary tract infections (UTI) [4]. Acinetobacter are also a prominent host for NDM-1, with Acinetobacter baumannii being a notorious opportunistic pathogen closely associated with hospital-acquired infections [3,5]. Bacteria harbouring NDM-1 exhibit extensive resistance against virtually all β-lactam antibiotics. Moreover, the presence of the bla_NDM-1 gene on plasmids allows for the co-existence of up to 14 additional resistance genes, facilitating the occurrence of cross-resistance [6,7]. As a result, NDM-1-carrying bacteria are commonly multidrug-resistant or even extensively drug-resistant, thus earning the designation of “superbugs” [8]. Currently, colistin and tigecycline are the only antibiotics effective against NDM-1-producing bacteria [9,10]. However, resistance to these two drugs has also been reported and the clinical use of colistin is limited by its adverse effects, such as nephrotoxicity and neurotoxicity [11–15]. Therefore, NDM-1 poses a formidable challenge in medical settings. The discovery of a potent inhibitor that can effectively rescue the activity of existing β-lactam antibiotics against pathogens expressing NDM-1 holds profound implications for human health, the sustainable utilization of pharmaceutical resources, and the associated economic impacts.

Screening potential new antimicrobial agents from natural products or approved drugs has emerged as an efficient strategy in drug discovery [16,17]. Nitroxoline (5-nitro-8-hydroxyquinoline), an ancient antibiotic, has been approved in certain countries for the treatment of uncomplicated urinary tract infections [18]. Nitroxoline demonstrates potent antibacterial activity against various bacteria, especially Escherichia coli, and its efficacy is influenced by environmental pH and divalent cations [19]. Studies have indicated that nitroxoline can reduce bacterial adhesion to uroepithelial cells or affect biofilm formation, suggesting a potential mechanism for its effectiveness in treating UTI [20,21]. Nitroxoline exhibits mild, broad-spectrum antibacterial activity, and good safety profiles. However, its short plasma half-life and rapid urinary excretion following oral administration limit its usage [22]. Recent research has suggested that nitroxoline can effectively inhibit NDM-1 activity in vitro by chelating Zn²⁺, but there is a lack of evidence for its efficacy in animals other than insects, Galleria mellonella larvae [23,24].

In this study, we constructed a library of long-lasting derivatives based on the nitroxoline skeleton. From this library, we identified a compound, ASN-1733, which exhibits antibacterial activity by disrupting the integrity and permeability of the bacterial outer membrane. In infection models beyond UTI, ASN-1733 demonstrated superior therapeutic efficacy compared to nitroxoline. Furthermore, ASN-1733 exhibits potent inhibitory activity against NDM-1 via a novel competitive mechanism, distinct from EDTA and nitroxoline. These findings expand the potential applications of nitroxoline analogues in the field of antibacterial infections and provide new insights for the development of novel NDM-1 inhibitors.

Methods and materials

Bacterial strains

Standard reference strains including E. coli ATCC25922, ATCC35218, A. baumanii ATCC19606, P. aeruginosa ATCC27853, S. pneumoniae ATCC49619, and S. aureus ATCC29213, ATCC43300 were used as control. All other clinical isolates were obtained from the Antibiotic Research Institute at Huashan Hospital, affiliated with Fudan University in Shanghai, China.

Minimum inhibitory concentration (MIC)

The compounds’ MIC values against the bacterial strains were assessed by employing the broth microdilution method in a sterile 96-well plate, following the guidelines established by the Clinical and Laboratory Standards Institute (CLSI) [25]. Mueller-Hinton broth (MHB) was utilized as the growth medium. The initial bacterial inoculum was prepared to a concentration of 1 × 10⁶ CFU/mL and incubated at 37°C for 18 h. MIC referred to the minimum concentration of the compound that showed no visible bacterial growth in the wells.

Time-killing curve

The bacterial inoculum was adjusted to 1 × 10⁶ CFU/mL in the culture medium containing the corresponding compound concentration and then underwent incubation at 37°C with constant agitation at 160 rpm. During the incubation, aliquots were taken at regular intervals and serially diluted. The diluted samples were then spread onto LB agar plates and subjected to a 24-hour incubation to perform colony counting.

Pharmacokinetics

Pharmacokinetic studies were conducted following oral administration of nitroxoline, ASN-1213, and ASN-1733 at a single dose of 10 mg/kg in ICR mice. Blood samples were collected at various time points after dosing, including 0.25, 0.5, 1, 2, 4, 8, 12, and 24 h, with 3 mice sampled at each time point. The collected samples were centrifuged at 4°C and 4000 g for 10 min to obtain plasma for concentration analysis. A liquid chromatography-tandem mass spectrometry (LC-MS/MS) method was adopted to measure the compound concentrations in the plasma. The lower detection limit was 5 ng/mL.

The pharmacokinetic parameters were calculated using the mean plasma concentration-time data, employing the non-compartmental model in Phoenix WinNonlin 8.2 software.

Scanning electron microscope (SEM)

Cultivating EC1864 to its logarithmic growth phase and achieving a concentration of 1 × 10⁸ CFU/mL in LB medium with varying compound concentrations. The bacterial suspension was subsequently subjected to incubation at 37°C and 160 rpm for a duration of 4 h. After centrifugation, the pellet was washed with PBS and fixed with 2.5% glutaraldehyde. Images were observed and recorded using a scanning electron microscope (Hitachi Regulus 8100, Japan).

Membranes permeability

The bacterial suspension containing 1 × 10⁸ CFU/mL of bacteria in the logarithmic growth phase was incubated in PBS containing 1 µg/mL different compounds at room temperature for 0, 30, 60, 120, and 240 min. At each point, N-phenyl-1-naphthylamine (GENMED SCIENTIFICS, INC., Shanghai, China) and propidium iodide (Meilun Biotechnology, Dalian, China) were added following the instructions provided in the kit. The samples were further incubated in a light-protected environment for an additional 30 min. The fluorescence intensity was determined through a fluorescence microplate reader (BioTek, Synergy H1). NPN excitation wavelength was set at 350 nm, with an emission wavelength at 420 nm. PI excitation wavelength was set at 535 nm, with an emission wavelength at 617 nm.

LPS and Ca²⁺ detection assays

The bacterial suspension containing 1 × 10⁸ CFU/mL of bacteria in the logarithmic growth phase was incubated in PBS containing 0, 10 and 20 µg/mL different compounds at 37°C for 0, 2, 4, and 6 h. After incubation, the samples were centrifuged and the supernatant was collected and filtered with a 0.22 µm filter to remove excess bacteria. Subsequent measurements following the instructions provided in the LPS assay kits (Xiamen Bioendo Technology, China) and Calcium Quantitation Kit (Beyotime, China).

Isothermal titration calorimetry

ITC was utilized to quantify the binding affinity between the compound and Ca²⁺ ions [26]. The experiments were performed using Nano-ITC (TA instruments, USA). Solvents for all compounds, LPS, and CaCl₂ were prepared as 5% DMSO in ddH₂O. CaCl₂ concentration was set to 5 mM, nitroxoline-like compounds concentration was 0.5 mM, and LPS concentration was 0.375 mM. The titration of the calcium pool was carried out by adding nitroxoline-like compounds and LPS at 25°C, and the resulting microcalorimetric data were collected during the titration process. Following the data collection, the area under each peak was integrated, and the molar ratio of ligand to protein was used as the x-axis for plotting. The obtained isotherms were then fitted to a binding model to determine the dissociation constant (K_d) and thermodynamic parameters involved in the binding process.

Fractional inhibitory concentration (FIC)

FIC index was used to evaluate the synergistic effect in vitro among nitroxoline-like compounds and meropenem. A checkerboard method was employed to prepare a drug susceptibility plate by diluting the nitroxoline-like compound and meropenem in MHB medium. Following the procedure of MIC measurement, FICI was defined as (MIC_{A in combination}/MIC_A + MIC_{B in combination}/MIC_B). A FICI value ≤ 0.5 was considered as synergistic effect.

Quantitative polymerase chain reaction (qPCR)

Bacterial DNA extraction was conducted following the instruction of Bacterial Genomic DNA Extraction Kit (Solarbio, China). The bla_NDM-1 was determined using the NDM-1 Diagnostic Kit (Shanghai XuanYa Biotechnology, China). Following the instruction provided by the manufacturer, a sample was considered NDM-1 positive if the Ct value was ≤35 in the detection channel and displayed a clear exponential growth curve.

Meropenem consumption assay

1 mL culture of bacteria in the logarithmic growth phase was harvested by centrifugation to separate the supernatant from the bacterial pellet. The pellet was washed and resuspended with PBS. Subsequently, the bacterial cells were disrupted using an ultrasonic disruptor (Shanghai Lichen Instrument Technology, China) at 4°C, generating a lysate. PBS was included as a negative control. Each sample was supplemented with meropenem to attain a final concentration of 1 mM. The reaction mixtures were incubated at 37°C for 30 min, followed by measuring the OD value at 300 nm to determine the proportion of meropenem depletion.

Enzyme-linked immunosorbent assay (ELISA)

The expression level of bacterial NDM-1 protein was detected using the bacterial β-lactamase NDM-1 ELISA kit (Shanghai Lanpai Biotechnology Company, China). The ELISA procedure involved coating the microplate wells with NDM-1 antibody, followed by the addition of samples, standard samples, and HRP-labelled detection antibody. The plate was incubated at 37°C for 60 min and thoroughly washed. Subsequently, TMB substrate was added for colour development, and the plate underwent a 15-minute incubation in a dark environment. The reaction was halted by introducing the stop solution, and the OD values of each well were measured at 450 nm to calculate the expression level of NDM-1 protein.

Expression and purification of the NDM-1 protein in Escherichia coli system

After the removal of the signal peptide and the addition of a 6 × His gene sequence at the 5’ end, the sequence of the recombinant NDM-1 protein is as follows:

MHHHHHHMPGEIRPTIGQQMETGDQRFGDLVFRQLAPNVWQHTSYLDMPGFGAVASNGLIVRDGGRVLVVDTAWTDDQTAQILNWIKQEINLPVALAVVTHAHQDKMGGMDALHAAGIATYANALSNQLAPQEGMVAAQHSLTFAANGWVEPATAPNFGPLKVFYPGPGHTSDNITVGIDGTDIAFGGCLIKDSKAKSLGNLGDADTEHYAASARAFGAAFPKASMIVMSHSAPDSRAAITHTARMADKLR

The NDM-1 gene was synthesized entirely and incorporated into the pET30a expression vector at NdeI and HindIII restriction enzyme sites. The correctness of the ultimate expression vector was verified through restriction enzyme digestion and sequencing analysis. The constructed vector was then transformed into expression strain BL21(DE3). IPTG was utilized to induce the expression of the NDM-1 protein. Subsequently, purification of the NDM-1 protein was achieved by using affinity chromatography with Ni-IDA resin.

Microscale thermophoresis (MST)

MST was employed to measure the dissociation constant (K_d) between various inhibitor molecules and recombinant NDM-1 enzyme, thereby assessing their binding affinities. The NDM-1 protein was labelled using the RED-NHS dye at a protein concentration of 10 µM and a dye concentration of 30 µM, followed by a 30-minute incubation at room temperature. The labelled protein mixture was then purified using a gel filtration column with a binding buffer solution (PBST) to remove any unbound dye. The compounds were prepared at a maximum concentration of 2 mM and subjected to 2-fold serial dilutions for 15 times. Each diluted solution had a volume of 10, and 10 µL of the diluted protein solution was incorporated into it and mixed thoroughly. The resulting mixture was then drawn into a capillary and placed in the MST instrument for detection. The following parameters were configured: MST-Power was set to “Medium,” Excitation-Power was set to “40%,” and Excitation-type was selected as “Nano-Red.” The K_d values were determined by non-linear fitting of the monomer inhibitor from three independent replicates.

Kinetics of enzyme-catalyzed reactions

The enzymatic function of the recombinant NDM-1 protein was determined by measuring the kinetic parameters, K_cat and K_m, using the substrate meropenem. The reaction was carried out in a 96-well plate format. The reaction mixture consisted of 5 nM NDM-1, meropenem concentrations ranging from 10 to 300 µM, and 10 mM HEPES (pH 7.5). Three replicates were set for each substrate concentration, and a control group without the enzyme was included. The OD values of the system were measured at 300 nm, with sequential readings taken every 30 s for a total of 30 readings. By plotting OD₃₀₀ against the substrate concentration, the reaction velocity corresponding to each substrate concentration was determined within the linear range of change using the equation: V = Δ[S] / Δt, where V was the initial reaction velocity, [S] represents the substrate concentration, t stands for the time. The data was modelled according to a Michaelis–Menten curve. K_m and V_max was determined by GraphPad Prism 9.5.1. K_cat was calculated as: K_cat = V_max / [E], where [E] was the concentration of the enzyme NDM-1.

Inhibitory effects assessment

EDTA served as a reference. The reaction system consisted of NDM-1 enzyme at a concentration of 5 nM, meropenem at 1 mM, and inhibitor concentrations ranging from 0.002 to 100 µM. Five points were tested for each logarithmic concentration. The reaction was conducted in 10 mM HEPES buffer at pH 7.5. To initiate the reaction, NDM-1 and the inhibitors were incubated together for 5 min before the addition of the substrate, meropenem. The measurement of the initial reaction velocity was performed as described previously, and the relative velocity was calculated as follows: Relative velocity (%) = V_inhibitor / V_negative × 100%.

For the inhibitory selectivity assays, the substrate nitrocefin was used at a concentration of 1 mM, while the concentration of all β-lactamases was kept constantly at 0.5 nM. The inhibitors were tested at two different concentrations: 1 µM as the low concentration and 10 µM as the high concentration. The absorbance at 490 nm was measured considering the different substrates. The inhibitory effect was calculated using the formula: Inhibition effect (%) = (V_negative-V_inhibitor) / V_negative × 100%.

For the activity recovery experiment, the concentrations of EDTA, nitroxoline, ASN-1213, and ASN-1733 were set to 0.5, 8, 4, and 1 µM, respectively, based on their IC₅₀ values. The inhibitors were incubated with the enzyme at room temperature for 5 min, followed by the addition of 1-fold or 10-fold concentration of Zn²⁺ (in the form of ZnCl₂) and further incubation for 10 min. The meropenem was added to initiate the reaction. The relative activity restoration (%) of NDM-1 is defined as: Restoration (%)  = V_recovery / V_negative× 100%.

In all equations, V represents the initial reaction velocity.

Molecular docking

Docking analyses were carried out utilizing AutoDock Vina 1.1.2 software. Prior to docking, receptor protein underwent hydrogenation using PyMol academic open-source version. The processed receptor protein and ligands were converted into the required PDBQT format for docking using ADFR suite 1.0. Docking was performed using Vina, with a thoroughness setting of 32 for global search and keeping the remaining settings maintained at their default values. In the end, the binding conformations were determined as the one with the highest scores. The docking results were then visually inspected and analyzed using PyMol.

Molecular dynamics simulation

Molecular dynamics simulations were performed using Gromacs2022.3 software [27]. The ligand was prepared using AmberTools22 and parameterized with the GAFF force field and subjected to hydrogenation and RESP potential calculations using Gaussian 16W. The generated potential data were incorporated into the topology file. The simulations were conducted under constant temperature (300 K) and pressure (1 Bar) conditions, utilizing the Amber99sb-ildn force field. Water molecules were employed as the solvent, and an appropriate number of Na⁺ ions were added to neutralize the total charge. The simulation system was energy-minimized using the steepest descent method. Following that, equilibration was performed separately for an NVT (constant number of particles, volume, and temperature) ensemble and an NPT (constant number of particles, pressure, and temperature) ensemble, with a coupling constant of 0.1 ps and a duration of 100 ps. Finally, the system underwent a production run of free molecular dynamics. Upon completion of the simulations, trajectory analysis was conducted, including computations of RMSD, RMSF, and protein radius of gyration, in conjunction with free energy calculations and free energy profiles.

Cell viability

Cell viability was evaluated using the CCK-8 assay. A cell density of 5 × 10⁴ cells/mL in 100 µL of culture medium was used to seed cells in 96-well plates. Different concentrations of compounds were introduced into the wells and the plates were placed in an incubator set at 37°C. Cell viability was assessed after 24 and 48 h following the protocol of the CCK-8 assay kit (Meilun Biotechnolog, Dalian, China).

Haemolytic activity

A 5% red blood cell suspension was prepared using fresh sterile sheep blood (Lezhen Biotechnolog, Nanjing, China) and inoculated into a 96-well plate. The wells were treated with varying concentrations of compounds and incubated for a duration of 1 h. The OD values were measured at 576 nm, with 1% NP40 serving as the positive control. Haemolysis rate was defined as: Haemolysis (%) = (A – A _negative)/ (A _positive – A _negative), where A was OD_{576 nm}.

Resistance acquisition

Continuous exposure of bacteria to sub-MIC concentrations of a compound can induce resistance. In a 96-well plate, the highest concentration that allows bacterial growth in the MIC test is selected, and the suspension is diluted 100-fold with fresh MHB medium. The next MIC test is repeated using the diluted suspension. Repeat this cycle 20 times. Meropenem and colistin are used as controls, and the increase in drug MIC is evaluated to assess the resistance acquisition.

Animal studies

Animal experiments were conducted using 6-week-old ICR mice weighing 18–22 g. Female mice were used for the UTI model, while male mice were used for all others. The experimental procedures and animal care protocols were approved by the Animal Ethics Committee of Fudan University (2019-08-WY-FMQ-01).

Urinary tract infection model

ICR mice were subjected to water deprivation for 12 h prior to the experiment. Anaesthesia was induced by intraperitoneal injection of 200 µL of 5% chloral hydrate solution. Once fully anaesthetized, a suspension of 1 × 10⁶ CFU of Escherichia coli EC1864 in 50 µL LB medium was transurethrally injected into the bladder of the mice. Two hours after infection, treatment was initiated. All compounds were orally administered at a dose of 10 mg/kg, and a total of two doses were given. Twelve hours after the second dose, urine samples and bladder and kidney homogenates were collected and cultured on LB agar plates. The plates were then incubated at 37°C for 18 h, followed by colony counting to determine bacterial load.

Thigh infection model

Following anaesthesia using the previously described method, a suspension of 1 × 10⁶ CFU of EC1864 in 50 µL LB medium was injected into the left thigh muscle of each mouse. Two hours after infection, treatment with the compounds was initiated. All compounds were orally administered at a dose of 10 mg/kg, and a total of two doses were given. Twelve hours after the second dose, the mice were euthanized. The left and right thighs of the mice were then weighed, homogenized, and the homogenates were plated onto LB agar plates for bacterial colony counting.

Sepsis model

Each mouse was intraperitoneally injected with a bacterial suspension containing 6 × 10⁷ CFU in 200 µL of LB medium. Two hours after infection, treatment was initiated. All compounds were orally administered at a dose of 10 mg/kg, twice daily for a total of three days. Following the infection period, the mice were observed for a period of 7 days to monitor their survival rate.

In the combination therapy study, nitroxoline-like compounds were administered orally via gavage, whereas meropenem was administered via intravenous injection through the tail vein. All drugs were given at a dose of 10 mg/kg, with a reduced dosing frequency of once daily.

Mild abdominal infection model

Each mouse was intraperitoneally injected with a bacterial suspension containing 3 × 10⁶ CFU in 200 µL of LB medium. Two hours after infection, treatment was initiated. Each compound was given orally at a dose of 10 mg/kg, and a total of two doses were given. Twelve hours after the second dose, mice were euthanized. The main organs were weighed, and blood and homogenates from the main organs were plated onto LB agar plates for bacterial colony counting.

In the combination therapy study, ASN-1733 was administered orally via gavage, whereas meropenem was administered via intravenous injection through the tail vein. Both were given at a dose of 10 mg/kg, with a reduced dosing frequency of once daily. At the final point, the major organs of the mice were collected additionally for H&E staining and histological analysis.

Toxicity in vivo

Following a dose-tolerance experimental design, ICR mice were orally administered ASN-1733 once daily, and their body weight was recorded daily. After completion of four dosing cycles, the mice were euthanized, and major organs were collected for H&E staining and histopathological examination.

Statistical analysis

The data are expressed as Mean ± SD (standard deviation). P-values were calculated using Student’s t-test for comparisons between two groups, one-way ANOVA for comparisons among multiple groups and two-way ANOVA for certain analyses involving multiple factors. The significance levels were denoted as follows: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. ns, not significant. Unless specified otherwise, the comparisons were made against the control group. The statistical analysis was performed using GraphPad Prism 9.5.1 software.

Results

Novel nitroxoline derivatives displayed enhanced antibacterial effects

A library containing nitroxoline and 49 long-lasting derivatives was constructed. Their antibacterial activity was preliminarily assessed by measuring the MIC against Gram-negative Escherichia coli ATCC25922 and Gram-positive Staphylococcus aureus ATCC29213, in which ASN-1213 and ASN-1733 caught our attention due to their promising performance (Figure 1(A, B)). Therefore, further investigation was launched to explore their antibacterial spectrum against a total of 19 bacteria strains belonging to 5 different species. The results showed that ASN-1733 exhibited more stable MIC values (2-8 µg/mL), and no cross-resistance was observed with meropenem (Table S1). For bactericidal kinetics, both ASN-1213 and ASN-1733 exhibited similar killing curve to nitroxoline. Within 24 h, the compounds at 1 MIC can effectively inhibit bacterial growth. The effect was enhanced with increasing concentration and a compound at 10 MIC almost eradicates all bacteria (Figure 1(C), S1A-K). In terms of pharmacokinetics, ASN-1213 and ASN-1733 illustrated a longer half-life in mouse plasma compared to nitroxoline, accompanied with improved C_max and AUC_0-24 (Figure 1D, S2A, B).

Figure 1. The long-lasting nitroxoline derivatives ASN-1213 and ASN-1733 demonstrated antibacterial activity both in vitro and in vivo. (A, B) Minimum inhibitory concentration of nitroxoline and its derivatives against ATCC25922 and ATCC29213. Among them, ASN-1213 and ASN-1733 were identified through screening. NDs represent nitroxoline and its derivatives. (C) Time-killing curve of ASN-1733 against EC1864. (D) Pharmacokinetics of ASN-1733 in ICR mouse plasma. (E) Four different ICR-mouse infection models used to evaluate the in vivo antimicrobial efficacy. All infection models were induced by EC1864, n = 6 per group. (F) Bacterial loads in urine of the urinary tract infected mice. (G) Bacterial loads in thigh of the thigh infected mice. (H) Survival curve of mice with sepsis. (I) Bacterial loads in the blood, lung, liver, and spleen of mice with mild abdominal infection. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Nitroxoline is predominantly used for the treatment of UTI considering its metabolic features. Hence, here we established 4 different infection models to evaluate whether the improved pharmacokinetics can support a wider range of application (Figure 1(E)). In the UTI model, the two derivatives demonstrated similar efficacy to nitroxoline. They were able to reduce the bacterial load in the urine and bladder, but not in the kidney (Figure 1(F), S3A, B). However, in other localized infection or systemic infection models, the derivatives exhibited better therapeutic effects. In the thigh infection model, both derivatives reduced the bacterial load in the infected thigh (Figure 1(G)) and also alleviated the extent of thigh swelling (Figure S3C). In the sepsis model, ASN-1733 improved the survival rate of infected mice from 0% to 66.7% (Figure 1(H)). In the mild abdominal infection model, both ASN-1213 and ASN-1733 reduced the bacterial load in the blood and major organs (Figure 1(I)), although efficacy did not reflect in organ indices (Figure S3D-F).

These results indicate that as long-lasting derivatives, ASN-1213 and ASN-1733 retain the antibacterial activity of nitroxoline in vitro. Nevertheless, due to the improved pharmacokinetic properties, they presented a broader application in vivo beyond UTI.

Nitroxoline and its derivatives disrupt bacterial outer membrane by binding to metal ions

During our investigation, we found that bacteria sequentially displayed surface wrinkling, indenting and rupturing, after incubated with increased concentration of nitroxoline and its derivatives (Figure 2(A), S4A, B). It suggested that these compounds may disrupt the bacteria surface structure thereby exerting their antibacterial effect. 1-N-phenylnaphthylamine (NPN) and propidium iodide (PI) are commonly served as markers of bacterial outer and inner membrane damage [28]. Our study revealed that the accumulated fluorescence intensity of NPN and PI increased in the culture supernatant at 60 and 120 min respectively (Figure 2(B, C)), after incubated with sub-MIC nitroxoline and its derivatives. These findings further confirm the morphological changes.

Figure 2. Nitroxoline and its derivatives disrupt bacterial outer membrane by binding to Ca²⁺. (A) Morphological change of EC1864 observed under scanning electron microscopy after incubation with ASN-1733. Scale bar = 5 μm. (B, C) Fluorescence intensity analysis of NPN and PI in culture supernatant of EC1864 after incubation with sub-MIC of compounds. (D, E) LPS levels and Ca²⁺ concentration in the bacterial culture supernatant following incubation with ASN-1733. (F) LPS detachment induced by a 4-hour treatment of ASN-1733 was alleviated by the addition of Ca²⁺ in the culture medium. (G) Effect of divalent metal ions, calcium and zinc, on the MIC value of ASN-1733 against EC1864. (H, I) Isothermal titration calorimetry assays determined the microscopic heat changes during the binding processes of LPS and ASN-1733 with Ca²⁺. All results were obtained from three independent replicates. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Damage to bacteria outer membrane often accompanies the loss of LPS and metal ions (such as calcium and magnesium), which jointly constitute “stabilizers” for the outer membrane through electrostatic interactions [29]. Accordingly, we observed an increase of LPS (Figure 2(D), S5A, B) while a decrease of calcium (Figure 2(E), S5C, D) in the culture supernatant after compound incubation. Furthermore, the releasing of LPS was delayed by using a calcium-enriched culture medium (Figure 2(F), S5E, F). The presence of excessive calcium ions can also weaken the antibacterial effect of nitroxoline and its derivatives (Figure 2(G), Figure S5G, H). We utilized isothermal titration calorimetry (ITC) to characterize the binding affinity of three compounds and LPS for Ca²⁺. The results showed that the equilibrium dissociation constant (K_d) of LPS for Ca²⁺ was 1.56 × 10⁻⁴ M (Figure 2(H)), while the values of nitroxoline and its two derivatives were much lower (Figure 2(I), S5I, J), indicating a stronger binding affinity.

These results demonstrate that nitroxoline and its derivatives can bind to bridging metal ions, destabilize LPS on the surface and cause membrane damage, finally leading to bacteria death.

ASN-1733 combined with meropenem effectively killed NDM-1-producing bacteria

Given that the antibacterial activity of nitroxoline and its derivatives is attributed to divalent metal ions binding, and NDM-1 hydrolytic activity is relied on zinc in its active site, we therefore hypothesized that these compounds could potentially inhibit NDM-1 activity in a similar way.

We firstly investigated the influence of Zn²⁺ on the antibacterial effect. Similar to calcium, excessive zinc also diminishes the antibacterial potency of nitroxoline and its derivatives (Figure 2(G), S5G, H). Checkerboard assay demonstrated that nitroxoline and its derivatives were able to recover the sensitivity of NDM-1 positive bacteria EC1864 to meropenem. Specifically, sub-MIC concentrations of nitroxoline, ASN-1213, and ASN-1733 caused a significant decrease in the MIC value of meropenem from 128 µg/mL to 8, 8, and 4 µg/mL, respectively (Figure 3(A–C)). We expanded our study by including additional strains resistant to meropenem (Table S2), among which EC1864, EC18201 and AB2092 were blaNDM-1 positive, while AB1845 was bla_NDM-1 negative (Figure S6A, B). The synergistic effect was only evident in NDM-1 positive pathogens (Figure 3(D-G), S6C, D).

Figure 3. ASN-1733 restore the activity of meropenem against NDM-1-producing bacteria in vitro and in vivo. (A-C) Checkerboard assay showed the synergistic killing effects of nitroxoline, ASN-1213, and ASN-1733 in combination with meropenem against EC1864. The heat plot showed an average of three technical replicates. (D) The bactericidal effect of sub-MIC concentrations of ASN-1733 and meropenem against EC1864 and AB1845. Sub-MICs for EC1864 are 64 µg/mL for meropenem, and 1 µg/mL for nitroxoline, ASN-1213, and ASN-1733. For AB1845, sub-MICs are 32 µg/mL for meropenem, 2 µg/mL for nitroxoline, and 1 µg/mL for ASN-1213 and ASN-1733. (E, F) Sub-MIC concentrations of nitroxoline and its derivatives (s-ND) reverse meropenem resistance in NDM-1-producing EC1864, but not NDM-1 negative AB1845. (G) Fractional inhibitory concentration index (FICI) of nitroxoline, ASN-1213 and ASN-1733 combined with meropenem against various meropenem-resistant bacteria. FICI below 0.5 demonstrated synergistic activity. (H) Combination therapy effect for the treatment of mouse sepsis. (I, J) Combination therapy effect for the treatment of mice with mild intra-abdominal infection. Bacterial burden in blood, lung, liver, and spleen. Representative histopathological images of major organs examined with H&E staining, Scale bar = 200 μm. All infections were induced by EC1864. For animal models, n = 6 per group. *P < 0.05, ***P < 0.001, ****P < 0.0001.

The in vivo results indicated that the combination therapy of meropenem with ASN-1213 or ASN-1733 significantly improved the survival rate of sepsis mice to 66.7% and 83.3%, respectively. However, the group receiving combination therapy of nitroxoline or monotherapy did not exhibit a protective effect (Figure 3(H)). In chronic infection, ASN-1733 combined meropenem further reduced bacterial load in mice blood and organs compared to the control and monotherapy groups (Figure 3(I)). Additionally, the combination strategy resulted in alleviated organ congestion and inflammatory cell infiltration (Figure 3(J)).

These results indicate that ASN-1733 can recover the susceptibility of NDM-1 positive bacteria to meropenem. The combination therapy of ASN-1733 and meropenem holds promise as a valuable approach for managing infections caused by such strains.

ASN-1733 competitively inhibits the hydrolysis ability of NDM-1

The NDM-1 enzyme produced by resistant bacteria can hydrolyze meropenem, but the hydrolyzation is diminished after incubation with ASN-1733 (Figure 4(A)). We examined the expression of NDM-1 protein and found no difference whether ASN-1733 was present or not (Figure 4(B)). It suggested that ASN-1733 may impact bacterial resistance by inhibiting the protein activity rather than affecting the expression of NDM-1. To further elucidate the mechanism, we employed an E. coli expression system to produce and purify NDM-1 enzyme (Figure 4(C), S7A). Its hydrolytic activity was validated with a K_m value of 54.72 µM and a K_cat value of 11.71 s⁻¹, which were comparable to those reported in previous studies [30] (Figure S7B).

Figure 4. ASN-1733 competitively inhibits the hydrolytic activity of NDM-1. (A) EC1864 lose the capacity of meropenem hydrolysis after 1-hour incubation with sub-MIC of ASN-1733. Lys, lysates; Sup, supernatants. (B) ELISA detection reveals that incubation with sub-MIC of ASN-1733 does not affect the expression of NDM-1 in EC1864. (C) Recombinant expression of NDM-1 protein in vitro. The left panel shows SDS-PAGE analysis, where M1 represents the marker, Lane 1 corresponds to BSA as a control, and Lane 2 represents the NDM-1 protein. The right panel depicts Western Blot analysis, where Lane 2 displays the NDM-1 protein, and M1 represents the marker. (D) Binding affinities of recombinant NDM-1 protein with EDTA, nitroxoline, ASN-1213 and ASN-1733 were determined by MST. (E) Inhibition of NDM-1 hydrolytic activity by different compounds determined through enzymatic assays. All data are relative to the standard condition with 50 µM Meropenem. IC₅₀ values were obtained from Hill equation fitting under this standard condition. (F) Selective inhibition of various β-lactamases by different inhibitors using nitrocefin as a substrate. “L” denotes low concentration, while “H” represents high concentration. (G) Restoration of NDM-1 activity through the addition of excess Zn²⁺. (H) Lineweaver-Burk plots of inhibition of the NDM-1 hydrolysis activity by EDTA, nitroxoline, ASN-1213 and ASN-1733. All results were obtained from three independent replicates.

To demonstrate the direct interaction between NDM-1 and our compounds, the binding affinities were assessed by MST. We measure a K_d value of 0.32 µM for ASN-1733, which was approximately 1/10 of nitroxoline and comparable to EDTA (Figure 4(D)). The strong affinity enables ASN-1733 to effectively inhibit the catalytic activity of NDM-1, with an IC₅₀ of 1.03 µM (Figure 4(E)). Its specific inhibitory effect was observed exclusively towards metallo-β-lactamase, distinguishing ASN-1733 from clinical inhibitors that show a preference for serine-β-lactamase (Figure 4(F)).

The NDM-1 activity inhibited by zinc depriver can be partially or fully restored by zinc supplementation [31]. It has also been demonstrated with EDTA, nitroxoline, and ASN-1213 in our studies (Figure 4(G)). However, ASN-1733 retained an appreciable portion of its inhibitory effect even in the presence of zinc ions at a 10-fold higher relative concentration. These results indicate that the potent inhibition effect of ASN-1733 is independent of zinc deprivation. Next, we adopted enzyme kinetic analysis at various concentrations of substrate and inhibitor to determine the inhibitory mode of these compounds. The Lineweaver–Burk plot revealed that ASN-1733 demonstrates competitive inhibition against NDM-1 (K_i = 0.22 µM). On the other hand, nitroxoline and ASN-1213 showed non-competitive inhibition features consistent with EDTA (Figure 4(H)). These results indicate that ASN-1733 may exert inhibition on NDM-1 through mechanisms other than zinc deprivation.

ASN-1733 bind to key loops of NDM-1 to effectively occlude its active site

The hydrolysis ability of NDM-1 is dependent on the presence of two Zn²⁺ ions within its active pocket. The pocket is surrounded by flexible amino acid residue loops, among which Loop3 (L65-V73) and Loop10 (C208-L221) contribute to shield the Zn²⁺ ions and stabilize the enzyme–substrate complex [32] (Figure 5(A)). To better understand the inhibitory mechanism of ASN-1733 on NDM-1 and its distinctions from other compounds, we conducted molecular docking and dynamics simulations to depict its binding site and binding process.

Figure 5. Binding mode of ASN-1733 with NDM-1 determined through molecular docking and molecular dynamics simulations. (A) Structure of NDM-1 (PDB ID: 3S0Z). The red spheres represent zinc ions, while the blue and green residues represent Loop 3 and Loop 10, respectively. (B) The interaction between ASN-1733 and the residues in the binding sites of NDM-1 obtained through molecular docking. (C) The stable 3D structure of the ASN-1733/NDM-1 complex obtained through molecular dynamics simulations. (D) The RMSD variation of NDM-1 (Blue) and ASN-1733 (Red) during the 100 ns simulation. (E, F) The RMSF values of each residue in NDM-1 and each atom in ASN-1733 during the simulation process. (G) The Gibbs energy landscape of the NDM-1/ASN-1733 complex during the simulation process. (H) Interactions between ASN-1733 and NDM-1 were categorized. The stacked bar charts are normalized over the course of the trajectory. (I) Timeline summary for the interactions and contacts (H-bonds, Hydrophobic, Ionic, Water bridges). The top panel shows the total number of specific contacts. The bottom panel shows which residues interact with the ligand in each trajectory frame. The molecular docking work was performed using AutoDock Vina 1.1.2 software. The molecular dynamics simulations were performed using Gromacs 2022.3 software. (J) The reduction or elimination of the inhibitory effect of ASN-1733 on NDM-1 in the mutant variants.

Basic information about NDM-1 was obtained from the RCSB database (Figure S8A). The docking results reveal that in the nitroxoline-NDM-1 complex, the nitroxoline molecule binds within the active pocket of NDM-1. The nitrogen atom on the nitro group engages in a Pi-cation interaction with His250. The hydroxyl group generates salt bridges with Zn301 and Zn302 via hydrogen bonding. Additionally, the nitrogen atom on the quinoline backbone bond with Zn301 by forming a metal coordination (Figure S8B). ASN-1213 exhibits a similar binding site as nitroxoline, although its mode of action differs slightly (Figure S8C). The binding mode of nitroxoline and ASN-1213, directly interacting with the Zn²⁺ ions in NDM-1, can provide an explanation for their similar inhibitory behaviour resembling that of EDTA in enzymatic assays. However, during our analysis of the ASN-1733-NDM-1 complex, we discovered that the interaction does not involve any specific zinc ion. The quinoline ring of ASN-1733 engages in Pi-Pi stacking with HIS250, establishing a favourable aromatic interaction. The nitro group of ASN-1733 forms hydrogen bonds with ALA215 and ASP212 and ASP212 participates in a salt bridge interaction with ASN-1733, further facilitating additional stabilization. Moreover, residues PRO68, GLY69, and VAL73 on the Loop3, as well as LYN211, LYS214, and ASN220 on the Loop10, contribute to the overall stability of the complex (Figure 5(B)). In this way, the ASN-1733 molecule forms a tunnel-shaped narrow cleft in the catalytic pocket of NDM-1, thus locking its active cavity and hindering substrate binding to the enzyme, which may be the molecular mode of ASN-1733 inhibiting NDM-1 activity (Figure 5(C)).

During the molecular dynamics simulation, the root-mean-square deviation (RMSD) of the NDM-1 protein remained stable within the range of 2.0-2.6 Å, while the RMSD of the ASN-1733 ligand exhibited fluctuations before eventually stabilizing between 4.8 and 5.6 Å (Figure 5(D)). These observations indicate the formation of a new conformation through their mutual interactions, which is relatively stable and indicative of a strong binding affinity. Root-mean-square fluctuation (RMSF) analysis can help to characterize the local variations in protein chains and small molecule ligands during the simulation process. In our analysis, we observed that the RMSF of amino acid residues interacting with the ligand were generally higher (Figure 5(E)). This indicates that these residues play a critical part in stabilizing the conformation of the complex. Additionally, the RMSF of the ASN-1733 showed higher fluctuations in certain atoms, further supporting our previous observation (Figure 5(F)). Energy analysis reveals a Gibbs free energy minimum in the protein complex within the RMSD range of 0.25-0.3 nm, indicating a more stable conformation at this point (Figure 5(G)). The contacts between the protein and ligand were classified into four categories: Hydrogen bonding, hydrophobic interactions, ionic interactions, and water bridges (Figure 5(H)). The timeline analysis of the interactions showed an increase in total contacts of the complex after 70 ns of simulation. Among these, PHE70, LYS211, SER217, GLY219, and ASN220 exhibited the most prominent contribution (Figure 5(I)). To validate the computational simulation results, we engineered three mutated NDM-1 proteins: D212A-NDM-1, A215G-NDM-1, and H250A-NDM-1, and assessed the inhibitory effect of ASN-1733 on the activity of both wild-type and mutant proteins (Table S3, Figure S9). As anticipated, the mutant proteins exhibited a pronounced reduction in sensitivity to ASN-1733 (Figure 5(J)). These findings indicate that ASN-1733 achieves competitive inhibition of NDM-1 activity by binding to its key loops and blocking its active site.

ASN-1733 exhibits low toxicity and low resistance acquisition

The cytotoxicity and haemolytic activity of ASN-1733 were evaluated by in vitro assessment. RAW264.7, HEK-293 T and L-02 cells were subjected to different concentrations of ASN-1733 for a duration of 48 h, and the impact on cell viability was measured. The results indicated that ASN-1733 exhibited an IC₅₀ value of approximately 52.17 µg/mL for RAW264.7, 52.69 µg/mL for HEK-293 T and 49.61 µg/mL for L-02 cells (Figure 6(A–C)), which were notably higher than the MIC₅₀ value of ASN-1733 against bacteria (4 µg/mL). In the haemolysis assay, ASN-1733 exhibited a haemolysis rate below 5% within a concentration range of 64 µg/mL (Figure 6(D)). This indicates that the compound does not pose a risk of cell growth inhibition or haemolysis at bactericidal concentrations.

Figure 6. In vivo and in vitro toxicity of ASN-1733 and its inducible drug resistance. (A, B, C) Assessment of cell viability in RAW264.7, HEK-293 T and L-02 cells using the CCK-8 assay kit following 24-hour and 48-hour incubation with ASN-1733. (D) Haemolysis rate of red blood cells treated with ASN-1733. 1% NP-40 was used as a positive control. (E) The dose tolerance experiment in ICR mice. (F) Changes in mice body weight during the dose tolerance experiment, n = 6 per group. (G) Histopathological examination of major organs in mice using H&E staining following the completion of the dose tolerance experiment. Scale bar = 200 μm. (H, I) Resistance acquisition of pathogens during serial passaging in sub-MIC levels of antimicrobials. All results were obtained from three independent replicates. **P < 0.01, ***P < 0.001, ****P < 0.0001.

In terms of in vivo toxicity, a dose escalation tolerance test of ASN-1733 was conducted in mice (Figure 6(E)). Throughout four treatment cycles, no mouse deaths were observed, and there were no significant reductions in mice body weight (Figure 6(F)). Furthermore, no apparent organ damage was observed during the pathological examination (Figure 6(G)). Regarding the rate of resistance acquisition, a 20-day study was conducted where ASN-1733 was continuously administered at sub-MIC concentrations to induce drug resistance. Compared to meropenem and colistin, the resistance acquisition of ASN-1733 was noticeably reduced in both occurrence and degree (Figure 6(H, I)). These results indicate that ASN-1733 possesses favourable safety profiles both in vitro and in vivo, presenting a promising alternative for further exploration.

Discussion

In recent decades, the escalation of bacterial resistance has posed significant challenges, while the discovery and development of novel antibiotics often demand substantial time and resources [33–36]. In response to this situation, the exploration and repurposing of existing drugs have emerged as practical strategies in the field of new drug development [37,38].

Since the 1960s, several European countries have approved the use of nitroxoline for treating uncomplicated urinary tract infections [39]. Nitroxoline exhibits antibacterial efficacy against a wide range of uropathogens, encompassing both gram-positive and gram-negative bacteria such as Escherichia coli [40]. Besides, there have been few reports of toxicity or resistance associated with its use [39,40]. However, due to its pharmacokinetic properties, characterized by rapid excretion in urine, its application has been mainly limited to UTI [22]. Our findings also confirmed the challenge of observing therapeutic effects of nitroxoline in other localized or systemic infection models (Figure 1).

To address these limitations, we introduced two derivatives, ASN-1213 and ASN-1733, through chemical modifications on nitroxoline. These derivatives retained the in vitro antibacterial properties of the parent compound, including a comparable antibacterial spectrum and time-killing curves (Figure 1(A–C)). Crucially, they exhibited improved pharmacokinetic profiles in vivo, resulting in remarkable therapeutic efficacy in various infection models (Figure 1(D, F–I)). Our findings propose the liberation of compounds within the nitroxoline class from their exclusive UTI-focused usage, potentially expanding their utility to a broader spectrum of infectious diseases and providing more choices for future preclinical studies and even clinical candidate drugs.

The antibacterial mechanism of nitroxoline is commonly attributed to its ability to chelate essential cations required for bacterial survival. These chelation interactions can influence bacterial adhesion, biofilm formation, and the activity of vital enzymes [41–43]. Previous research has suggested that the antibacterial activity of nitroxoline is hindered by interference from ions like Mg²⁺ and Mn²⁺, while K⁺, Na⁺, and Ca²⁺ have a less pronounced effect [19]. However, our study presents a different perspective, indicating that nitroxoline induces bacterial outer membrane damage and bacterial death by chelating Ca²⁺, leading to the detachment of LPS. This disruptive effect can be mitigated by culturing bacteria in a calcium-rich environment, thus reducing nitroxoline’s antibacterial activity (Figure 2). The discrepancy in conclusions may stem from the use of solvents with varying properties. Nitroxoline’s antibacterial activity relies on its chelation capability, which is closely related to the pH value. Therefore, in our experiments, we take care to avoid using alkaline solvents to dissolve nitroxoline to ensure compatibility with the physiological environment it would encounter in vivo. Our findings are further supported by ITC results, demonstrating a significantly higher affinity of natural calcium ions to nitroxoline compared to LPS (Figure 2(H,I)). This new perspective enhances our understanding of the antibacterial mechanism of compounds like nitroxoline. Interestingly, this mechanism bears resemblance to cationic peptides such as colistin. It is widely accepted that colistin are positively charged peptides that directly bind to the negatively charged lipid A portion of bacterial outer membranes, leading to membrane damage and cellular contents leakage [28]. The bacterial outer membrane serves as a formidable obstacle, impeding the penetration of hydrophobic and large-molecule antibiotics. Disruption of the outer membrane facilitates the entry of these antibiotics into the bacterial interior, resulting in a synergistic bactericidal effect [44–46]. Considering the similar antibacterial mechanism to colistin, we have high expectations for the synergistic effects of nitroxoline-like compounds when co-administered with a wider range of other antibiotics. Nevertheless, the subtle differences in mechanisms between these compounds and colistin are also reflected in their acquisition of resistance. Resistance to colistin often arises from modification pathways of their target, LPS [47]. In contrast, the targets of nitroxoline-like compounds are relatively more conserved, which may explain their resistance acquisition difficulty (Figure 6(H,I)).

The significance of disrupting the bacterial outer membrane extends beyond its bactericidal effects. It also plays a role in inhibiting the NDM-1 enzyme. Unlike other MBLs, NDM-1 is anchored to the bacterial outer membrane, which offers various advantages, such as protection from periplasmic proteases, secretion into outer membrane vesicles, and evasion of extracellular inhibitors [48,49]. We propose that compounds like nitroxoline, which disrupt the integrity and permeability of the bacterial outer membrane, may expose the NDM-1 enzyme, thus facilitating its subsequent inhibition. In terms of inhibition mechanisms, reported NDM-1 inhibitors can be broadly categorized into two groups. The first class directly targets the Zn²⁺ ions in the enzyme’s active centre, leading to its inactivation, as exemplified by compounds like EDTA or bismuth complex [50,51]. In contrast, the second class of inhibitors binds to critical amino acid residues, obstructing substrate binding or hindering the release of hydrolyzed products, as seen in compounds like carnosic acid [11].

In our study, we found that the antibacterial activity of nitroxoline-like compounds can be disrupted by Zn²⁺ (Figure 2(G)), leading us to hypothesize that they may also inhibit NDM-1 activity through chelation of Zn²⁺. Excitingly, recent research has explored this possibility. Anna et al. conducted in vitro experiments that demonstrated the effectiveness of nitroxoline and its derivatives as NDM-1 inhibitors [23]. The prototype compound nitroxoline completely inhibited NDM-1 activity at 30 µM, consistent with our enzymatic research data. Furthermore, the authors confirmed the inhibition mechanism of the synthetic compound 5c, showing its correlation with Zn²⁺ chelation through zinc replenishment experiments. However, it appears that the study might benefit from further exploration in terms of establishing an appropriate animal infection model to evaluate the therapeutic or potential synergistic effects of nitroxoline or its derivatives against NDM-1-producing bacterial infections. In another study by Luigi et al., an insect infection model using Galleria mellonella larvae was employed [24]. The findings suggested that the concurrent utilization of nitroxoline and meropenem enhanced the survival of larvae infected with Escherichia coli, Streptococcus pneumoniae, and Klebsiella pneumoniae in a meningitis sepsis model. Considering the metabolism characteristics of nitroxoline, when assessing its potential as an NDM-1 inhibitor and considering its derivatives, it is essential to evaluate not only their inhibitory activity against NDM-1 but also their pharmacokinetic properties. Only then can we genuinely consider their application in conventional mammalian bacterial infections and even their inclusion in the developmental pipeline for clinical therapies. Our in vitro and in vivo experiments clearly demonstrate that co-administering ASN-1733 with meropenem effectively combats NDM-1 producing bacteria (Figure 3). The structure of ASN-1733 is not previously reported in publicly available studies. We've optimized its pharmacokinetic properties based on nitroxoline, making it more suitable for mammalian infection applications. This represents a novel and innovative finding compared to previous research. ASN-1733 exhibits notably higher inhibition of NDM-1 activity, with approximately a 10 folds greater affinity for binding to NDM-1 than nitroxoline (Figure 4(E)). It exhibits distinct inhibitory characteristics compared to nitroxoline or EDTA. In the zinc replenishment experiment, NDM-1 activity in the ASN-1733 group failed to recover effectively (Figure 4(G)). In enzyme kinetic assays, ASN-1733 displayed a distinctive inhibitory curve (Figure 4(H)). These features suggest that ASN-1733 might possess a mechanism of NDM-1 inhibition beyond chelation, possibly involving binding to crucial amino acid residues that influence substrate binding.

Computational simulations revealed specific interactions between ASN-1733 and amino acid residues in the Loop 3 and Loop 10 regions of NDM-1, with the strongest impact observed on the Loop 10 region (Figure 5(B,C)). Considering the subtle structural difference between ASN-1213 and ASN-1733, there could be various reasons accounting for the distinct inhibitory effects and mechanisms of the two compounds. The bromine substitution at C-6 diminishes the electron cloud density in the metal-binding region formed by nitrogen on the quinoline ring and oxygen on the phenyl ring. Consequently, this results in a reduced capacity of ASN-1733 to bind to the Zn 301 in the catalytic centre of NDM-1. This could be one of the reasons. Additionally, spatial hindrance stemming from bromine substitution may also play a significant role. The catalytic centre of NDM-1 is a deep and narrow cavity, after the substitution of hydrogen with bromine at C-6, due to both electronegativity and spatial hindrance, the compound may encounter difficulties entering the cavity and directly interacting with Zn. Simultaneously, this “blocking effect” also impedes the substrate's entry into the catalytic centre. Besides, the impact of bromine substitution on the polarity and electron cloud distribution of the compound could enhance its interaction with the critical residues on Loop 10. This constitutes the second inhibitory centre, explaining why ASN-1733 doesn't penetrate the active centre of NDM-1 but can still effectively inhibit it in another mode. Several studies have indicated that certain NDM-1 variants have evolved to possess highly efficient mono-zinc enzyme activity [52,53]. These findings explain the evolutionary direction of NDM to overcome the impact of zinc deficiency and maintain enzyme activity. This highlights the significance of developing inhibition strategies that go beyond targeting zinc. Hence, we propose a novel inhibitory mechanism of ASN-1733 against NDM-1, displaying potent inhibition unaffected by environmental Zn²⁺ ions. These findings suggest the future possibility of discovering superior NDM-1 inhibitors targeting specific binding sites on Loop 10.

Our findings provide new insights and potential therapeutic strategies against NDM-1-producing bacterial infections. However, further research is required to a comprehensive understanding. For instance, we've focused solely on ASN-1733s acute toxicity, but assessing it's its long-term toxicity and validating its effectiveness in a broader range of clinical and animal models is essential. Additionally, the dosing regimen and frequency when co-administering ASN-1733 with meropenem are empirical. Modifying this strategy may further enhance its therapeutic efficacy. Lastly, there's promise in developing more potent NDM-1 inhibitors through the novel inhibitory sites. These crucial steps will help ascertain the clinical potential of ASN-1733 as a multifunctional antibiotic.

In summary, our study encompassed the construction of a long-lasting derivative library of nitroxoline and subsequent screening that led to the identification of the active compound ASN-1733. We also elucidated the updated antimicrobial mechanism of this compound class and successfully broadened its application range in the treatment of infections. Notably, we discovered that ASN-1733 exhibits potent competitive inhibition against the NDM-1 enzyme in bacteria, achieving this effect through a novel mechanism. These findings enhance our comprehension of this compound category, providing additional options for clinical or preclinical research, and offering new insights and solutions for the design of NDM-1 inhibitors. Overall, our study presents a significant advancement in the field, with potential implications for combating antimicrobial resistance and facilitating the development of improved therapeutic strategies against NDM-1-mediated infections.

Acknowledgements

The authors would like to extend their sincere gratitude for the invaluable structural information and technical support extended by Jiangsu Asieris Pharmaceutical Co., Ltd. in the design and synthesis of nitroxoline derivatives.

Disclosure statement

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

Additional information

Funding

This work was supported by National Natural Science Foundation of China [grant number 32200745]; Shanghai Sailing Program [grant number 21YF1401900].