3‑Methylthiopropionic Acid Kills Carbapenem-Resistant Klebsiella pneumoniae by Disrupting Membrane Integrity and Bioenergetics Authors: Lirui Ge, Liang Zhao, Saiwa Liu, Jingjing Du, Jian Xu, Kexin Cui, Yu Ga, Jingyu Liu, Dongmei Zhao, Shuwen Wang, Xiaowei Li, Xi Xia Published: 2026 ABSTRACT The rapid dissemination of New Delhi metallo-β-lactamase (NDM) genes, such as bla NDM‑1 and bla NDM‑5, has rendered carbapenems ineffective against Klebsiella pneumoniae, posing a severe public health threat. Here, we employ comparative metabolomics to reveal that NDM-mediated resistance is underpinned by profound metabolic reprogramming. From this altered metabolic landscape, we identify 3-methylthiopropionic acid (MTPA) as a potent endogenous metabolite capable of directly killing NDM-producing strains. MTPA exhibits rapid bactericidal activity in vitro and significantly improves survival in murine infection models. Mechanistically, MTPA disrupts membrane integrity, dissipates proton motive force, depletes intracellular ATP, and induces ROS accumulation via respiratory chain complex I inhibition. Integrated transcriptomics and metabolomics revealed a catastrophic collapse of central carbon metabolism, CoA biosynthesis, and redox homeostasis. This work establishes a chemical strategy to combat carbapenem resistance by targeting pathogen metabolic reprogramming. Introduction Antibiotic resistance, mediated by resistance genes, has become a major global health threat due to high mortality rates and limited therapeutic options. Carbapenems are frequently reserved as the last-line defense against multidrug-resistant bacterial infections. However, the emergence and global dissemination of carbapenemase genes, particularly bla NDM‑1 and its prevalent variant bla NDM‑5, have severely compromised their clinical utility. , Bacteria expressing these genes can hydrolyze a wide range of β-lactam antibiotics, most critically carbapenems. − The rising prevalence of Klebsiella pneumoniae strains carrying bla NDM‑1 and bla NDM‑5, isolated from diverse clinical specimens (e.g., blood, sputum, pus, and urine), has been increasingly documented. These isolates exhibit high-level carbapenem resistance, posing a challenge to existing therapeutic frameworks. Consequently, there is an urgent need to develop innovative strategies to combat bla NDM-mediated resistance in Gram-negative pathogens, particularly in high-risk clones of K. pneumoniae. Previous studies have established that the bacterial metabolic state is a critical determinant of antibiotic efficacy. , Antibiotics exert their bactericidal or bacteriostatic effects partly by disrupting key metabolic processes, whereas bacteria can adapt their metabolism to evade these effects. − This interplay is evident in the oxygen-dependent killing of Mycobacterium tuberculosis, where hypoxia significantly diminishes antibiotic efficacy. Furthermore, exogenous metabolite supplementation has been shown to resensitize bacteria to antibiotics. For instance, replenishing glucose and amino acids can reverse fluoroquinolone resistance in nutrient-starved Staphylococcus aureus by reactivating metabolic pathways and inducing lethal oxidative stress. Similarly, the addition of mannitol potentiates nitrofurantoin (NIT) activity against Escherichia coli, reducing bacterial survival by 6-fold compared to NIT treatment alone; cotreatment with lactic acid also shows a 2-fold enhancement. Collectively, these findings underscore the potential of modulating bacterial metabolism to potentiate existing antibiotics, providing a strong rationale for metabolomics-guided strategies for developing novel antibacterial adjuvants. In this study, we employed untargeted metabolomics to interrogate the metabolic profiles of antibiotic-susceptible and -resistant strains (carrying bla NDM‑1 or bla NDM‑5). This revealed a critical resistance-associated metabolic reprogramming that we then leveraged therapeutically. Selected metabolites were evaluated for their ability to resensitize resistant bacteria to carbapenem antibiotics using both in vitro and in vivo models. Methods Bacterial Strains To investigate NDM-1 and NDM-5-induced metabolic perturbations, we employed three recombinant K. pneumoniae ATCC 13882 strains with well-defined genetic backgrounds: the empty vector control (pACYC184) and its derivatives harboring bla NDM‑1 or bla NDM‑5. The bactericidal efficacy of the selected metabolites was further evaluated using two clinical isolates: K. pneumoniae Z49 (bla NDM‑1-positive) and K. pneumoniae Z220 (bla NDM‑5-positive). All strains were cultured in Luria–Bertani (LB) medium at 37 °C. The primers used for plasmid construction are listed in Table S1. Metabolic Profiling Sample preparation was performed according to a published method with the following modifications. Briefly, bacterial cultures were harvested during the logarithmic growth phase at an OD600 of 0.7, diluted to an OD600 of approximately 0.3, and immediately quenched in a cold methanol/ethylene glycol solution (45:55, v/v). After quenching, the cells were washed with ice-cold physiological saline. Metabolites were extracted by resuspending the cell pellets in hot ethanol/water (75:25, v/v) with constant agitation at 500 rpm for 5 min, followed by precipitation at −20 °C for 1 h. The resulting supernatants were concentrated using a centrifugal vacuum concentrator (SpeedVac, Thermo Fisher Scientific, Waltham, Massachusetts, USA). Metabolite profiling was performed using an ultrahigh-performance liquid chromatography (UHPLC) system coupled to a high-resolution mass spectrometry (HRMS) instrument (X500B, SCIEX, Framingham, MA, USA). Separation was achieved on both reversed-phase and hydrophilic interaction liquid chromatography columns to comprehensively cover polar and nonpolar metabolites. The raw mass spectrometry data were processed with SCIEX OS 3.3.1 and Progenesis QI 2.3 software for data curation, including baseline subtraction, peak alignment, and feature extraction. Multivariate statistical analyses, including principal component analysis (PCA) and orthogonal projections to latent structures-discriminant analysis (OPLS-DA), were conducted on the MetaboAnalyst 6.0 and EZinfo platforms. Differential metabolites were filtered based on the following criteria: fold change >2, p-value <0.05, coefficient of variation (CV) <30%, and variable importance in projection (VIP) >1. Targeted metabolomic analysis was conducted on a SCIEX 7500 system (SCIEX, Framingham, MA, USA) as previously described. Data acquisition and processing, including peak integration and quantification, were performed with SCIEX OS software (v3.3.1). Detailed LC–MS parameters are summarized in Supplementary Tables S2–S5. Screening of Antibacterial Metabolites The antimicrobial activities of seven endogenous metabolites (uracil, guanosine, 3-methylthiopropionic acid (MTPA), cytidine, D-mannose, melibiose, and tyramine) were assessed by the standard broth microdilution method in 96-well plates. Briefly, metabolites were serially diluted 2-fold in Mueller–Hinton broth and inoculated with bacterial suspensions at a density of 1 × 106 CFU/mL in HEPES buffer (pH 6.8). The minimum inhibitory concentration (MIC) was defined as the lowest metabolite concentration that completely inhibited visible growth after 18 h of incubation at 37 °C. The potential synergy between these metabolites and meropenem was evaluated using checkerboard assays. The fractional inhibitory concentration index (FICI) was calculated and interpreted as described previously. Time-Kill Assay A time-kill assay was performed to evaluate the bactericidal activity of MTPA. Overnight bacterial cultures were diluted 1:100 in fresh LB medium and grown with shaking at 37 °C for 6 h. The bacterial suspensions were adjusted to approximately 1 × 106 CFU/mL in HEPES buffer (pH 6.8) and treated with MTPA at various concentrations (0, 5, 10, 15, and 20 mM, corresponding to 0, 0.6, 1.2, 1.8, and 2.4 mg/mL, respectively) in LB medium. Aliquots (100 μL) were withdrawn at specified time points (0, 4, 8, 12, 16, 20, and 24 h), serially diluted (10-fold) in sterile saline, and spread onto LB agar plates. After 18 h of incubation at 37 °C, the bacterial colonies were enumerated. To compare the bactericidal effects of different organic acids, bacterial suspensions in HEPES buffer (pH 6.8) were treated with formic acid, acetic acid, and propionic acid at concentration ranges identical to those used for MTPA (0–20 mM). Assessment of Resistance Development The resistance development assay was conducted as previously described, with minor modifications. Overnight cultures of K. pneumoniae were diluted 100-fold in fresh MH broth supplemented with MTPA at 0.25×, 0.5×, 1×, 2×, or 4× the MIC, and incubated at 37 °C with shaking for 24 h. Bacterial cultures were serially subcultured for 30 generations in fresh MH medium containing 0.25 × MIC of MTPA, with MICs determined periodically to monitor resistance development. Mouse Infection Model A murine infection model was established using female BALB/c mice (6–8 weeks old; n = 6 per group). Mice were intraperitoneally challenged with a lethal dose of K. pneumoniae Z49 (1.2 × 108 CFU in 200 μL) or Z220 (5 × 106 CFU in 200 μL). At 1 h postinfection, the mice were administered saline (control), MTPA (20 or 40 mg/kg), or Meropenem (5 mg/kg). At 20 h postinfection, all mice were euthanized, and the heart, liver, spleen, lungs, and kidneys were collected for determination of bacterial burden (CFU/g) and histopathological analysis. A separate survival study was conducted under identical infection and treatment conditions (n = 15 per group), and survival was monitored every 24 h for 7 days. All animal experiments were performed in compliance with relevant guidelines and regulations and were approved by the Animal Welfare and Ethics Committee of China Agricultural University (Approval No. Aw50204202-2-3). Outer Membrane Permeability Assay The outer membrane permeability of K. pneumoniae strains harboring bla NDM‑1 or bla NDM‑5 was assessed using the fluorescent probe 1-N-phenylnaphthylamine (NPN). Briefly, exponentially growing cultures were harvested by centrifugation and adjusted to an OD600 of 0.5 in HEPES buffer (pH 6.8). The bacterial suspensions were then treated with MTPA (0, 5, 10, or 15 mM) for 30 min at 37 °C under shaking conditions. After the treatment, the cells were collected by centrifugation, washed, and resuspended in HEPES buffer. Subsequently, 5 μM NPN was added, and the mixtures were incubated for 30 min at 37 °C. Polymyxin B was used as a positive control. Fluorescence intensity was immediately measured at excitation and emission wavelengths of 350 and 420 nm, respectively. Inner Membrane Permeability Assay The integrity of the inner membrane was evaluated using a propidium iodide (PI) uptake assay. K. pneumoniae strains harboring bla NDM‑1 and bla NDM‑5 were grown to the mid-exponential phase and diluted to approximately 1 × 106 CFU/mL in HEPES buffer (pH 6.8). MTPA was added at final concentrations of 0, 5, 10, or 15 mM, followed by incubation at 37 °C for 2 h. Following treatment, PI was added to a final concentration of 20 μg/mL, and the samples were incubated for a further 1 h. Aliquots of the suspensions were mounted on glass slides, and images were captured with a fluorescence microscope. Membrane Fluidity Assay Membrane fluidity was assessed by Laurdan generalized polarization (GP) spectroscopy as described previously. , K. pneumoniae strains harboring bla NDM‑1 or bla NDM‑5 were grown to the exponential phase, harvested, and resuspended to an OD600 of 0.5 in 15 mM Tris-HCl buffer. The cell suspensions were incubated with 10 μM Laurdan for 30 min at 37 °C with shaking (220 rpm) in the dark. After washing with Tris-HCl buffer, the Laurdan-labeled cells were treated with MTPA (0, 5, 10, or 15 mM in HEPES buffer, pH 6.8) for 30 min at 37 °C. Benzyl alcohol was used as a positive control for membrane fluidization. Fluorescence emission spectra were recorded at 435 and 490 nm following excitation at 350 nm. The GP value was calculated using the equation: GP = (I 435 – I 490)/(I 435 + I 490), where I 435 and I 490 represent the emission intensities at the respective wavelengths. Proton Motive Force (PMF) Assessment The PMF and its two components, the membrane potential (Δψ) and the transmembrane pH gradient (ΔpH), were assessed in K. pneumoniae strains harboring bla NDM‑1 or bla NDM‑5. Exponential-phase bacterial cells were harvested, washed, and resuspended in PBS to an OD600 of 0.5. To assess the membrane potential (Δψ), cells were stained with 20 μM DiSC3, a fluorescent dye sensitive to changes in membrane potential. Concurrently, the intracellular pH (ΔpH) was measured using 10 μM BCECF-AM, a ratiometric pH-sensitive fluorescent dye. After incubation at 37 °C for 30 min with shaking (220 rpm) in the dark, the stained cells were washed with PBS and subsequently treated with MTPA at various concentrations (0, 5, 10, or 15 mM) in HEPES buffer (pH 6.8), with CCCP serving as the positive control. Fluorescence measurements were immediately performed using a spectrofluorometer. For DiSC3, the fluorescence was monitored at an excitation wavelength of 622 nm and an emission wavelength of 670 nm. For BCECF-AM, the fluorescence emission was recorded at 535 nm following excitation at 488 nm. Morphological Analyses Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) were employed to observe the ultrastructural changes in K. pneumoniae strains following treatment with MTPA. Bacterial cells in the stationary phase were harvested, washed, and resuspended in PBS. The suspensions were treated with MTPA at concentrations of 0, 5, and 10 mM for 20 min at 37 °C. After treatment, the cells were collected by centrifugation and gently washed twice with PBS. The samples were fixed with glutaraldehyde, dehydrated through a graded ethanol series, and processed according to standard protocols. Finally, specimens were mounted on copper grids and examined with a microscope (Hitachi, Tokyo, Japan). Iodonitrotetrazolium Chloride (INT) Reduction Assay Overnight cultures of K. pneumoniae-bla NDM‑1 and K. pneumoniae-bla NDM‑5 strains were diluted 1:100 in fresh LB broth and incubated at 37 °C. Following incubation, the cells were harvested, washed, and resuspended in buffer to an OD600 of 1.0. The bacterial suspensions were then treated with MTPA at final concentrations of 0, 5, 10, and 15 mM, along with 1 mM INT. After thorough vortexing, the mixtures were incubated at 37 °C for 10 min. Subsequently, the samples were centrifuged, and the resulting cell pellets were resuspended in ethanol. Following another centrifugation step, the absorbance of the supernatants was measured at 490 nm to quantify the reduction of INT. Oxygen Consumption Assay Oxygen consumption was assessed using an oxygen-sensitive fluorescent probe (Bestbio, Shanghai, China). Exponential-phase cells of K. pneumoniae-bla NDM‑1 and K. pneumoniae-bla NDM‑5 were harvested, washed, and resuspended in assay buffer to an OD600 of 0.5. The cell suspensions were loaded with the probe in a 96-well plate. After treatment with MTPA (0–20 mM), an oxygen-blocking reagent was added, and fluorescence (λex 460 nm /λem 600 nm) was measured at 37 °C for 90 min. ATP Determination K. pneumoniae-bla NDM‑1 and K. pneumoniae-bla NDM‑5 strains were grown overnight and subsequently diluted 1:100 in fresh LB medium, followed by incubation at 37 °C. The bacterial cells were then harvested by centrifugation and resuspended in PBS to an OD600 of 0.5. The cell suspensions were treated with MTPA at final concentrations of 0, 5, 10, and 15 mM in HEPES buffer (pH 6.8) and incubated statically at 37 °C for 30 min. Following incubation, the cells were lysed using ice-bath ultrasonication, and the resulting supernatants were collected. Intracellular ATP levels were determined using a commercial ATP assay kit (Nanjing Jiancheng, Nanjing, China). An aliquot of the supernatant was mixed with the working solution from the kit according to the manufacturer’s instructions. After incubation at room temperature for 15 min, luminescence was measured using a microplate reader. Determination of Respiratory Chain Complex I Activity Overnight cultures of K. pneumoniae-bla NDM‑1 and K. pneumoniae-bla NDM‑5 strains were treated with MTPA (0–10 mM) at 37 °C for 5 min. Bacterial cells were harvested by centrifugation and lysed by ultrasonication on ice. Respiratory chain complex I activity was assessed in these lysates using a commercial assay kit (Genmed, Shanghai, China) according to the manufacturer’s instructions. Analysis of NADH/NAD+ Ratio Overnight cultures of K. pneumoniae-bla NDM‑1 and K. pneumoniae-bla NDM‑5 strains were grown and treated with MTPA (0–20 mM) for 5 min at 37 °C. The cells were harvested by centrifugation, washed twice with saline, and lysed by ultrasonication. Supernatants obtained after centrifugation were immediately used to determine the NADH/NAD+ ratio using a commercial assay kit (Beyotime, Shanghai, China). Measurement of Intracellular Reactive Oxygen Species (ROS) Exponentially growing bacterial cells (OD600 = 0.5) were treated with MTPA (0, 5, 10, and 15 mM) for 5 min. After treatment, the cells were harvested by centrifugation and incubated in saline containing 10 μM 2′,7′-dichlorodihydrofluorescein diacetate at 37 °C for 30 min. Fluorescence intensity was then measured at excitation and emission wavelengths of 488 and 525 nm, respectively. Assessment of H+-ATPase Activity K. pneumoniae-bla NDM‑1 and K. pneumoniae-bla NDM‑5 strains were cultured to mid-exponential phase and harvested. The cell pellets were washed and resuspended in PBS containing 0, 5, 10, or 15 mM MTPA. After incubation at 37 °C for 30 min, the cells were collected and lysed by ultrasonication on ice. The clarified lysates, obtained by centrifugation, were used to determine H+-ATPase activity with a commercial kit (Beibo, Shanghai, China) according to the manufacturer’s instructions. Transcriptomic Analysis Exponentially growing cultures of K. pneumoniae-bla NDM‑1 and K. pneumoniae-bla NDM‑5 (OD600 = 1.0) were treated with 0 or 10 mM MTPA for 20 min at 37 °C. Total RNA was extracted using the HiPure Bacterial RNA Kit (Megan Biotechnology). RNA quality was verified using an Agilent 2100 Bioanalyzer (RIN > 8.0) and a NanoDrop 2000 spectrophotometer. Sequencing was performed on an Illumina PE150 platform after library preparation. After quality filtering and annotation, gene expression was quantified using FPKM. Differentially expressed genes (DEGs) between MTPA-treated and control groups were identified (p < 0.05) using DESeq (v1.38.3). RNA Extraction Midlog phase cultures of K. pneumoniae-bla NDM‑1 and K. pneumoniae-bla NDM‑5 were treated with MTPA (0–20 mM) for 20 min at 37 °C. Cells were harvested by centrifugation, and total RNA was extracted using the HiPure Bacterial RNA Kit (Magen, Guangzhou, China). RNA integrity was confirmed by electrophoresis on a 2% agarose gel (Gel-Doc GO, Bio-Rad). Results Impacts of bla NDM‑1 and bla NDM‑5Genes on the Bacterial Metabolic Profile To investigate the metabolic alterations associated with the expression of bla NDM‑1 and bla NDM‑5, we cloned these genes into the pACYC184 plasmid vector, generating recombinant plasmids pACYC184-bla NDM‑1 and pACYC184-bla NDM‑5. The empty pACYC184 vector was used as a negative control. These constructs were subsequently transformed into competent K. pneumoniae ATCC 13882 cells. Successful transformation was confirmed by PCR amplification and gel electrophoresis (Figure S1). The plasmids were then extracted and sequenced, and the results demonstrated that the sequences remained consistent with those of the original plasmids, with no mutations or deletions detected. Details on the plasmid map and PCR conditions are provided in Supporting Information Figure S2 and Tables S6 and S7. Meropenem susceptibility testing demonstrated that the expression of these genes conferred a high level of resistance, with MIC increases of 2048-fold (bla NDM‑1) and 8192-fold (bla NDM‑5) compared to the K. pneumoniae-pACYC184 control strain, confirming functional expression of the resistance genes (Table S8). Growth curve analysis showed no significant differences in the growth kinetics among the K. pneumoniae-bla NDM‑1, K. pneumoniae-bla NDM‑5, and the control K. pneumoniae-pACYC184 strains under the experimental conditions (Figure S3). This indicated that the genetic manipulation and expression of bla NDM‑1 and bla NDM‑5 did not impose a general fitness cost on bacterial growth, thereby validating the constructed strains as suitable models for subsequent metabolic studies. Untargeted metabolomic analysis was performed to compare the metabolic profiles of K. pneumoniae-bla NDM‑1 and K. pneumoniae-bla NDM‑5 with those of the susceptible control K. pneumoniae-pACYC184. Multivariate analysis revealed significant metabolic reprogramming associated with the carriage of the bla NDM genes. PCA showed a clear separation between the resistant and susceptible strains (Figure A), a distinction that was further supported by hierarchical clustering of the identified metabolites also revealed distinct profiles that clustered the drug-resistant and drug-susceptible bacteria separately (Figure B). Compared with the control, we identified 103 and 114 differentially abundant metabolites in K. pneumoniae-bla NDM‑1 and K. pneumoniae-bla NDM‑5, respectively. Specifically, K. pneumoniae-bla NDM‑1 exhibited 32 downregulated and 71 upregulated metabolites, while K. pneumoniae-bla NDM‑5 exhibited 58 downregulated and 56 upregulated metabolites (Figure C). These findings demonstrate that the acquisition and expression of bla NDM genes significantly reprogram the metabolic network of K. pneumoniae. On the basis of the magnitude of their change (high fold-change), seven key metabolites (three upregulated and four downregulated) were prioritized for further functional validation of their antimicrobial potentiation effects (Figure D). Metabolomic profiling reveals distinct metabolic landscapes between carbapenem-susceptible and carbapenem-resistant K. pneumoniaestrains. (A) Multivariate analysis using PCA and OPLS-DA showing clear separation between the susceptible strain (K. pneumoniae-pACYC184) and the resistant strains (K. pneumoniae-bla NDM‑1 and K. pneumoniae-bla NDM‑5). (B) Heatmap visualization of metabolite abundance profiles across the three strains. (C) Volcano plots highlight significantly upregulated and downregulated metabolites in resistant strains compared to the susceptible control. (D) Fold-change analysis of key differentially abundant metabolites in K. pneumoniae-bla NDM‑1 and K. pneumoniae-bla NDM‑5 relative to K. pneumoniae-pACYC184. (E) Screening and selection of differentially abundant metabolites based on antimicrobial potential (n = 6 biological replicates). MTPA Exhibits Bactericidal Potential Among the tested metabolites, only MTPA showed selective antimicrobial activity against resistant strains (K. pneumoniae-bla NDM‑1 and K. pneumoniae-bla NDM‑5), with MIC values of 10 mM under the assay conditions (Figure E). Time-kill kinetic assays were further performed to evaluate the bactericidal efficacy of MTPA against both engineered (K. pneumoniae-bla NDM‑1 and K. pneumoniae-bla NDM‑5) and wild-type (K. pneumoniae-Z49 and K. pneumoniae-Z220) strains. The results demonstrated concentration-dependent antimicrobial activity: 10 mM MTPA exerted bacteriostatic effects by completely inhibiting bacterial growth, while 20 mM MTPA exhibited potent bactericidal activity, with few or no surviving clones (Figure A). In contrast, under the same HEPES buffer system, no bactericidal activity was observed for formic acid, acetic acid, or propionic acid against the tested strains, whereas MTPA maintained a robust bactericidal effect. We further assessed the in vitro antibacterial activity of MTPA against additional clinical K. pneumoniae isolates harboring bla NDM‑1 (Z152, Z186, and Z310) and bla NDM‑5 (Z223) and obtained results comparable to those observed for Z49 and Z220. Serial passage experiments over 30 generations showed no development of acquired resistance (Figure B), indicating that MTPA maintains a stable antimicrobial profile and a low risk of resistance. Subsequent checkerboard synergy assays yielded FICI values consistently greater than 0.5 for all combinations, suggesting no synergistic effect with meropenem, ertapenem, imipenem, ampicillin, and ceftazidime (Figures 2C, S4). To evaluate the therapeutic potential of MTPA, a lethal murine model of peritonitis was established using clinically isolated NDM-producing K. pneumoniae strains Z49 (bla NDM‑1) and Z220 (bla NDM‑5). Although Meropenem, used as a positive control, showed limited efficacy in reducing bacterial burdens in tissues after 18 h of treatment, MTPA administration produced a pronounced dose-dependent antibacterial response. Specifically, a dose of 40 mg/kg reduced bacterial loads by more than 10-fold in the Z49 model, while 20 mg/kg resulted in greater than 103-fold reductions in bacterial loads in major organsincluding the heart, liver, spleen, lungs, and kidneysin the Z220 model (Figure D). Histopathological analysis further confirmed the protective effect of MTPA, demonstrating significant alleviation of infection-induced tissue damage compared with untreated controls (Figure S5). Notably, although all control animals (saline- or Meropenem-treated) succumbed to infection within 2–4 days, MTPA treatment provided substantial survival benefits, with 53.3% and 60.0% survival rates in mice infected with Z49 and Z220, respectively (Figure E). Taken together, these results underscore the potent antibacterial efficacy of MTPA against NDM-producing K. pneumoniae in both in vitro and in vivo settings. In vitro and in vivo antibacterial activity of MTPA against carbapenem-resistant K. pneumoniae. (A) Time-kill curves of MTPA against K. pneumoniae carrying bla NDM‑1 and bla NDM‑5. (B) Assessment of resistance development: changes in the MIC of MTPA after serial passaging. (C) Synergistic antibacterial effect between MTPA and Meropenem against K. pneumoniae carrying bla NDM‑1 and bla NDM‑5 (n = 3 biological replicates). (D) Bacterial loads in mouse tissues infected with K. pneumoniae-Z49 (bla NDM‑1) and K. pneumoniae-Z220 (bla NDM‑5) following treatment with MTPA (****p < 0.0001). (E) Survival rates of infected mice treated with MTPA, Meropenem, or saline control (n = 15 per group). MTPA Disrupts Bacterial Membrane Integrity and Function Cellular membrane integrity is crucial for maintaining normal physiological function, and membrane disruption is frequently a cause of rapid microbial death. On the basis of the rapid bactericidal activity of MTPA, we examined its effects on bacterial membrane integrity. We found that MTPA treatment resulted in concentration-dependent membrane disruption in both K. pneumoniae-bla NDM‑1 and K. pneumoniae-bla NDM‑5 strains, as evidenced by significantly increased inner (Figures A and S6) and outer (Figure B) membrane permeability. PI staining further confirmed pronounced concentration-dependent membrane damage, with the extent of disruption and outer membrane permeability escalating with higher MTPA concentrations. Moreover, Laurdan-based fluorescence polarization assays indicated that MTPA induced significant increases in membrane fluidity (Figure C), suggesting substantial alteration of membrane biophysical properties. These results collectively indicate that MTPA exerts antibacterial effects through membrane destabilization, impacting both the structural integrity and dynamic physical characteristics of bacterial membranes. MTPA alters the membrane composition of carbapenem-resistantK. pneumoniae. (A) Inner membrane permeability and (B) outer membrane permeability of K. pneumoniae strains following MTPA treatment. (C) Membrane fluidity. (D) Transmembrane pH gradient (ΔpH) and (E) electrical potential (ΔΨ). (F) Representative SEM and TEM images showing morphological alterations in MTPA-treated cells. (n = 3 biological replicates, ***p < 0.001, ****p < 0.0001). The PMF, which comprises both the ΔpH and the membrane potential (Δψ), is essential for bacterial membrane bioenergetics. Our results indicate that MTPA treatment induced a concentration-dependent disruption of PMF in K. pneumoniae-bla NDM‑1 and K. pneumoniae-bla NDM‑5 strains, leading to a significant decrease in intracellular pH (Figure D) and a concurrent increase in Δψ (Figure E). These disruptive effects on membrane energetics, combined with direct structural damage to the cell envelope, resulted in a severe loss of bacterial integrity, extensive cytoplasmic leakage, and ultimately cell death. SEM and TEM imaging visually confirmed these alterations: MTPA-treated cells displayed pronounced morphological defectssuch as cytoplasmic condensation, membrane rupture, loss of cell wall integrity, and abundant vacuolationcompared to untreated controls (Figure F). Collectively, these data demonstrate that the potent antibacterial activity of MTPA against resistant pathogens arises from its dual targeting of both the structural and functional integrity of the bacterial membrane. MTPA Disrupts Cellular Bioenergetics and Induces Redox Dysregulation Having established that MTPA disrupts bacterial membrane integrity and function, we next investigated the downstream consequences on cellular bioenergetics and redox homeostasis. The PMF is central to bacterial respiration and ATP synthesis. , Given that PMF dissipation can cripple these core energetic processes, we systematically assessed the impact of MTPA on bacterial bioenergetics by quantifying ATP dynamics, electron transport chain (ETC) function, and H+-ATPase activity. Treatment with MTPA strongly inhibited H+-ATPase, reducing its activity by >65% at 15 mM relative to controls (Figure A). This failure in proton extrusion directly coincided with a sharp decline in intracellular ATP, depleting pools by 40.8–56.7% in K. pneumoniae -bla NDM‑1 and 64.2–77.9% in K. pneumoniae -bla NDM‑5. Concomitantly, extracellular ATP rose significantly (p < 0.0001; Figure B), culminating in a 36.7–65.0% net loss in total cellular ATP. Furthermore, MTPA severely disrupted respiratory electron flow, as measured by the conversion of INT to formazan. A 5 mM dose reduced INT reduction by ≥72% in both strains (75.0% and 72.7%, respectively; Figure C), indicating profound ETC inhibition. Given that impaired respiratory function is closely associated with perturbations in NADH/NAD+ balance, complex I activity, oxygen consumption, and ROS generation, − we further characterized MTPA-induced dysregulation of the cellular redox state. Consistent with these connections, MTPA treatment resulted in a pronounced increase in the NADH/NAD+ ratio compared with untreated controls (p < 0.0001; Figure D). This shift was accompanied by markedly suppressed complex I activity (p < 0.0001; Figure S7)the primary enzyme responsible for NADH oxidation. Oxygen consumption assays further confirmed the respiratory depressive effect of MTPA, showing dose- and time-dependent decreases in aerobic respiration relative to controls (Figure S8). In parallel, intracellular ROS levels rose dramatically in both bla NDM‑1 (378%) and bla NDM‑5 (258%) strains (Figure E). Together, these results demonstrate that beyond disrupting PMF generation, MTPA triggers a systemic collapse of bioenergetic homeostasis, characterized by coordinated impairments in ATP synthesis, electron transport, and NADH oxidation capacity, ultimately culminating in profound redox imbalance and oxidative stress. MTPA disrupts respiration and induces oxidative stress in K. pneumoniae. (A) H+-ATPase activity and (B) intra- and extracellular ATP levels in K. pneumoniae strains after MTPA treatment (n = 3 biological replicates). (C) Bacterial respiratory activity (n = 3 biological replicates). (D) NADH/NAD+ ratio (n = 3 biological replicates). (E) Intracellular ROS levels (n = 3 biological replicates). (F) Multivariate analysis (PCA and OPLS-DA) of metabolomic profiles from susceptible and carbapenem-resistant K. pneumoniae strains (n = 6 biological replicates) (**p < 0.01, ***p < 0.001, ****p < 0.0001). Metabolic Pathway Alterations Reveal the Antibacterial Mechanism of MTPA Comparative metabolomic profiling was conducted to evaluate the concordance of MTPA-induced metabolic perturbations. Multivariate statistical analyses of drug-resistant K. pneumoniae revealed substantial metabolic reprogramming upon MTPA exposure, as evidenced by PCA-derived segregation of treatment groups. OPLS-DA further identified statistically significant metabolite alterations (VIP > 1.0, FC > 2, p < 0.05; Figure F), which were subsequently mapped to biochemical pathways (Figure S9). A subset of critical pathways was prioritized to elucidate MTPA’s bactericidal mechanism (Figure ). The metabolic landscape indicated a primary collapse in cellular bioenergetics. Marked reductions in central carbon metabolites within the galactose pathway, including lactose, d-galactose, and d-glucose, pointed to a critical shortage of substrates for ATP generation. This energy crisis was exacerbated by a parallel disruption in pantothenate and CoA biosynthesis, where the accumulation of intermediates pantetheine 4′-phosphate and dephospho-CoA suggested a synthesis bottleneck. Given CoA’s indispensable role in the TCA cycle and fatty acid oxidation, this impairment further crippled energy production. The ensuing bioenergetic failure was amplified by dysregulation of NAD+ biosynthesis; the accumulation of precursors alongside depleted NAD+ pools impaired electron transport chain function and compromised NADPH-dependent antioxidant defenses. Concurrently, MTPA instigated a catastrophic collapse of redox homeostasis. An observed increase in riboflavin signaled an initial adaptive response, which was ultimately overwhelmed by the concomitant accumulation of its dysfunctional derivatives, flavin adenine dinucleotide (FAD) and flavin mononucleotide. The failure to recycle these cofactors critically disrupted the electron transport chain. Furthermore, this FAD deficiency directly incapacitated the FAD-dependent enzyme glutathione reductase, paralyzing the glutathione cycle. This was evidenced by the pathognomonic coaccumulation of both reduced (GSH) and oxidized (GSSG) glutathione, a hallmark of a disabled antioxidant system that culminated in lethal oxidative stress and ROS accumulation. Beyond energy and redox catastrophe, MTPA concurrently compromised structural integrity. In the amino sugar pathway, depletion of N-acetyl-d-glucosamine and accumulation of UDP-N-acetylmuramate disrupted the substrate cycle for peptidoglycan biosynthesis, severely weakening the cell wall. This assault on structural components extended to the genetic level. Pyrimidine metabolism was halted by a critical depletion of dTDP, a key DNA precursor, while the accumulation of salvage metabolites like CMP, uracil, adenine, and dAMP indicated widespread nucleic acid damage and failed repair mechanisms. Diminished levels of the signaling molecules cAMP and cGMP further implied a breakdown in global stress response networks. MTPA induces global reprogramming of metabolic pathways in carbapenem-resistant K. pneumoniae. MTPA Alters the Bacterial Transcriptome To elucidate the antibacterial mechanism of MTPA at the transcriptional level, we analyzed its effects on the global transcriptomes of K. pneumoniae -bla NDM‑1 and K. pneumoniae -bla NDM‑5. Initial quality control confirmed the reliability of the RNA-seq data, including assessments of gene alignment distribution, uniformity of gene coverage, and expression abundance (Figures S10 and S11), confirming that all libraries met construction standards and possessed sufficient sequencing depth for downstream analysis. PCA of transcriptomic profiles revealed clear separation between MTPA-treated and control groups (Figure S12A), indicating profound transcriptome-wide alterations in response to MTPA. We identified a substantial number of differentially expressed genes (DEGs) in both of the strains. Specifically, MTPA treatment resulted in 2316 DEGs (1114 upregulated and 1202 downregulated) in K. pneumoniae -bla NDM‑1 and 1777 DEGs (811 upregulated and 966 downregulated) in K. pneumoniae -bla NDM‑5 (Figure S12B,C). To functionally characterize the differentially expressed genes (DEGs) induced by MTPA treatment, we performed Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses. GO analysis (Figure A) showed significant enrichment in biological processes related to fatty acid β-oxidation and cellular respiration, cellular components including ribosomal and respiratory chain complexes, and molecular functions such as structural constituents and ATP-dependent activity. KEGG pathway analysis (Figure B) further indicated strong enrichment of DEGs in essential metabolic pathways, including propanoate metabolism, oxidative phosphorylation, ABC transporters, pentose phosphate and galactose metabolism, pantothenate and CoA biosynthesis, as well as purine and pyrimidine metabolism. These enrichments suggest major disruptions in energy synthesisparticularly through pentose phosphate and galactose pathwaysand impaired transmembrane transport and stress response, as indicated by ABC transporter alterations. Additionally, perturbations in pantothenate/CoA and glutathione metabolism reflect compromised cofactor biosynthesis and oxidative stress regulation, while changes in pyrimidine metabolism are consistent with inhibited nucleic acid synthesis and reduced bacterial proliferation. Functional enrichment analysis of differentially expressed genes (DEGs) in MTPA-treated K. pneumoniae. (A) GO enrichment analysis of DEGs. (B) KEGG pathway enrichment analysis of DEGs. The heatmap of DEGs reveals extensive transcriptional reprogramming in key metabolic pathways under MTPA treatment (Figure S13). In pyrimidine metabolism, upregulation of pyrH, pyrE, and pyrF suggests compensatory nucleotide biosynthesis in response to DNA damage, while downregulation of nrdE and nrdF indicates impaired DNA replication and repair. Similarly, purine metabolism genes showed stress adaptation, with upregulation of adk, cysD, cysC, nudF, and ndk and downregulation of dgt and yrfG, reinforcing MTPA-induced genomic instability, as corroborated by metabolomic data. Central carbon metabolism was also disrupted. Galactose metabolism genes (pgm, rfbD, galU, glk, galE upregulated; pfkA, pfkB downregulated) indicated reduced ATP yield. Pantothenate/CoA biosynthesis was perturbed with upregulation of coaA, coaBC, and mazG promoting intermediate accumulation, and downregulation of coaE hindering final CoA synthesisconsistent with observed energy failure and membrane damage. Oxidative phosphorylation genes reflected respiratory dysfunction: frdABD upregulation signaled a shift to anaerobic respiration, and ppa downregulation further disrupted ATP homeostasis. In glutathione metabolism, elevated pepD, gshA, and gshB suggested enhanced GSH synthesis against oxidative stress, aligning with in vitro redox imbalances. Additional changes included propanoate metabolism adjustments (accA, mgsA up; fucO down), pentose phosphate pathway reprogramming (rbsK, pgm, rpe up; deoC, phnN, ghrB down), and broad downregulation of ABC transporters, collectively indicating metabolic slowdown and stress adaptation to MTPA. Discussion Although carbapenem antibiotics are a critical “last line of defense” against multidrug-resistant Gram-negative infections, their efficacy is increasingly compromised by the global dissemination of carbapenemases such as NDM-1 and NDM-5. , This challenge has intensified efforts to develop novel therapeutic approaches. Notably, studies reveal that antibiotic-resistant and susceptible strains exhibit markedly different metabolic profiles, , raising the possibility that bacterial metabolism could be exploited therapeutically. Specifically, the metabolic state of resistant bacteriainfluenced either by genetic determinants or antibiotic exposurehas been shown to profoundly affect antibiotic efficacy. Consequently, strategic modulation of metabolism, either by targeting pathways or applying exogenous metabolites, can restore susceptibility to conventional drugs. , These findings position metabolic intervention as a promising strategy to counteract resistance. The strategy of comparing metabolic profiles between drug-resistant and susceptible strains to screen for antimicrobial metabolites holds clear promise for applicability across diverse resistance mechanisms and pathogens, such as the reversal of fluoroquinolone resistance in S. aureus via glucose and amino acid replenishment and the potentiation of nitrofurantoin against E. coli by mannitol and lactic acid. In this study, we employed an untargeted metabolomics approach to investigate the metabolic reprogramming underlying carbapenem resistance in K. pneumoniae strains harboring bla NDM‑1 and bla NDM‑5. Comparative analysis revealed significant alterations in the metabolic landscape of resistant strains compared to the susceptible control (K. pneumoniae-pACYC184), indicating that resistance extends beyond gene acquisition to a fundamental rewiring of bacterial metabolism. From this reprogrammed metabolic network, we identified and validated seven significantly altered metabolites. Notably, one of these, MTPA, exhibited direct bactericidal activity against NDM-producing strains. This finding suggests that the metabolic reprogramming which confers resistance may also create new vulnerabilities that can be therapeutically exploited. While most previously reported metabolites in this context function as antibiotic adjuvants, our study identified a metabolite with direct bactericidal activity, demonstrating that this strategy can also uncover antimicrobial candidates with intrinsic killing properties. Crucially, MTPA’s intrinsic antimicrobial activity is independent of conventional antibiotics, distinguishing it from metabolite-derived adjuvants that merely potentiate existing drugs. , The therapeutic potential of MTPA was confirmed in vivo, as treatment (40 mg/kg) significantly improved outcomes in infected mice. Importantly, a preliminary toxicological evaluation at a high dose (1500 mg/kg) indicated no adverse effects over 5 days, suggesting a favorable safety profile and underscoring the promise of targeting metabolic reprogramming as a novel antimicrobial strategy. Following MTPA treatment, bacterial membranes exhibited a significant increase in permeability and fluidity, coupled with a marked dissipation of both components (Δψ and ΔpH) of the PMF. Given that the PMF is indispensable for bacterial energy homeostasis, its collapse directly compromised respiratory chain activity and abolished ATP synthesis. ,, Concurrently, MTPA-mediated inhibition of H+-ATPase activity further exacerbated the depletion of intracellular ATP. This ATP deficit resulted not only from disrupted biosynthesis but also from enhanced leakage through the compromised membrane, culminating in a catastrophic collapse of cellular energy equilibrium. Once intracellular ATP levels declined below a critical threshold, bacterial cell death was triggered. − This energy crisis was intrinsically linked to a severe redox imbalance. Beyond ATP production, cellular respiration is also the primary source (∼80%) of microbial ROS. , MTPA-induced disruption of the ETC led to a decline in respiratory rate and a pronounced accumulation of NADH, as reflected by an increased NADH/NAD+ ratio. This aberrant redox state accelerated the intracellular accumulation of ROS, where excessive levels caused widespread oxidative damage to essential biomoleculesincluding nucleic acids, lipids, and proteinsultimately leading to irreversible cellular dysfunction and bacterial death. − Antimicrobial exposure often triggers sequential metabolic disturbances that drive functional and structural failure. , Consistent with this paradigm, integrated metabolomic and transcriptomic analyses confirmed that MTPA exerts its bactericidal effects through coordinated disruption of nucleotide metabolism, membrane integrity, redox homeostasis, and energy production systems. The discovery of MTPA’s antibacterial activity, while encouraging, represents an early step in the drug development pipeline. Several key limitations must be overcome to transition this promising lead compound toward clinical application. A major gap is the lack of detailed pharmacokinetic studies to define its bioavailability, metabolic stability, and elimination pathways, which are critical for designing effective dosing regimens. Furthermore, the preliminary safety assessment requires expansion to include rigorous hematological, histological, and biochemical analyses to rule out organ toxicity. Finally, standard investigations into formulation feasibility and synthetic scalability have not yet been conducted, which are necessary to derisk further development. The present study establishes proof of concept for MTPA’s antibacterial activity but does not address pharmaceutical optimization. Consequently, structural analogs with improved potency, formulation strategies to enhance bioavailability, and combination regimens to lower effective concentrations have not been explored, representing critical gaps for clinical development. These crucial aspects will constitute the primary focus of our future research. Conclusions In summary, we demonstrate that carbapenem resistance in K. pneumoniae, conferred by bla NDM genes, is underpinned by a profound metabolic reprogramming. Targeting this vulnerability led to the identification of the endogenous metabolite MTPA as a potent, resistance-breaking antibacterial agent. MTPA exerts its bactericidal effect through a concerted chemical mechanism that disrupts membrane architecture and dissipates the proton motive force. This primary damage triggers a cascade of bioenergetic collapsecharacterized by ATP depletion and lethal oxidative stresswhich was systematically validated by integrated multiomics analyses. Our findings establish MTPA as a promising lead compound and, more broadly, exemplify a novel antimicrobial strategy rooted in the rational exploitation of pathogen-specific metabolic rewiring with small-molecule metabolites. Supplementary Material