C57BL/6 wild-type (WT) mice aged 6–8 weeks were purchased from Shanghai SLAC Laboratory Animal Cooperation. All the mice were housed in a specific pathogen-free and temperature-controlled standard environment in accordance with the National Institutes of Health Guide for Care and Use of Laboratory Animals. The animal experimental protocols were approved by the Animal Review Committee of Zhejiang University School of Medicine and were in compliance with institutional guidelines.

The complete information for all antibiotics and compounds used in this study was listed as followed: disulfiram (MedChemExpress, HY-B0240, in DMSO), tetracyclin (Aladdin, T105493-10 g, in DMSO), colistin (Solarbio, C9810, in DMSO), meropenem (Aladdin, M427157-1 mL, in DMSO), ampicillin (Aladdin, A105483-5 g, in DMSO), sulbactam (Aladdin, E129319-1 g, in DMSO), N-Acetyl-L-cysteine (MedChemExpress, HY-B0215, in DMSO), ZnCl₂ (MedChemExpress, HY-Y0420, in H₂O), Tpen (Aladdin, N159625-250 mg, in DMSO), polymyxin B (Aladdin, P105490-1 g, in DMSO), kanamycin (Aladdin, K331597-200 mg, in DMSO), miconazole (Macklin, M909105-25 g, in DMSO), benzoquinone (Macklin, B802580-5 g, in DMSO), CuGlu (Aladdin, D133471-25 g, in H₂O), TTM (Aladdin, A189030-200 mg, in DMSO), Ferric (III) Citrite (MedChemExpress, HY-B1645, in H₂O), deferoxamine (MedChemExpress, HY-B0988, in DMSO), Dimethyl sulfoxide (MedChemExpress, HY-Y0320), coin oil (MedChemExpress, HY-Y1888).

The laboratory strains used in this study were obtained from the American Type Culture Collection (ATCC) and included Escherichia coli (ATCC 25922), Staphylococcus aureus (ATCC 25923), methicillin-resistant Staphylococcus aureus (MRSA; ATCC 43300), Klebsiella pneumoniae (ATCC 13883), Pseudomonas aeruginosa (ATCC 27853), and Acinetobacter baumannii (ATCC 17978). Clinical strains (E. coli and A. baumanii) were collected from Microbiology Laboratory of Children’s hospital, 2023–2024.

The MIC₅₀ values of antibiotics were determined using the broth microdilution method according to CLSI guidelines. Bacterial suspensions (0.5 McFarland standard) were diluted 1:100 in cation-adjusted Mueller-Hinton broth and dispensed into 96-well plates containing two-fold serial dilutions of antibiotics (0.25–256 μg/mL). Plates were incubated at 37°C for 18–22 h. MIC₅₀ was defined as the lowest concentration of antibiotic that inhibited ≥50% of bacterial growth, assessed by optical density at 600 nm (OD600). Experiments were performed in triplicate.

Bacteria were cultured to log phase and diluted to an OD600 of 0.2–0.3. Disulfiram and colistin were added as indicated. The bacterial cells were subsequently washed 4 h later. The cell numbers were determined by dilution, incubation on Luria–Bertani agar plates at 37°C overnight, and calculation by multiplying the countable bacterial colony units by the number of dilutions.

Bacteria were cultured to log phase, diluted to 5 × 10⁵ colony-forming units (CFUs)/ml in 200 μL of LB and placed in a 96-well microliter plate. The drugs and compounds were added as indicated. The growth curve of bacteria was monitored by measuring the absorbance at 600 nm using a microplate reader (SpectraMax190, Molecular Devices) over a 24 h period.

Drug synergism was tested in checkerboard assays as described (Chen et al., 2023). Briefly, disulfiram and other antibiotics or compounds were twofold serially diluted in an 8 × 8 matrix within a 96-well microliter plate. Bacteria were grown to log phase and diluted in each well to 5 × 10⁵ CFUs/ml in 200 μL of MHB. After 18–22 h of incubation at 37°C, the absorbance of each well at 600 nm was measured with a microplate reader (SpectraMax190, Molecular Devices). The fractional inhibitory concentration index (FICI) was calculated accordingly, and a value less than 0.5 demonstrated synergy.

The membrane permeability rate of the cells was determined by propidium iodide (PI, 2.5 μg/mL) staining and flow cytometry. A minimum of 20,000 events per sample were acquired using the PE fluorescence channel during flow cytometry analysis. The analysis was performed using a DxFLEX or a CytoFLEX LX (Beckman Coulter, United States).

We used an E. coli murine intra-abdominal infection model to analyze the synergistic effect of disulfiram and colistin in vivo. Briefly, E. coli bacteria were seeded on Luria–Bertani (LB) agar plates and cultured overnight. A single clone was selected for culture in liquid LB medium and incubated with shaking at 200 rpm at 37°C for another 12 h. The bacterial suspension was then prepared at a concentration of 7 × 10⁶ colony-forming units (CFUs) per 100 μL of PBS. To induce intra-abdominal infection, the mice were intraperitoneally administered 100 μL of live E. coli suspension. Thirty minutes later, disulfiram (resolved in corn oil), colistin (resolved in corn oil) or control solvent (corn oil) was administered to the mice intraperitoneally as indicated. Mortality was assessed hourly for 72 h.

Bacteria were cultured to log phase and diluted to an OD600 of 0.2–0.3. Disulfiram and other antibiotics were added as indicated. Four hours later, the cells were inoculated on a slide and subjected to microscopic observation. Images were taken with an Olympus IX73 inverted microscope, and the cell lengths were measured using ImageJ software.

Motility assay was tested as described (Ejim et al., 2011). Briefly, swimming motility was observed on 0.3% (w/v) agar medium composed of 10 g/L trypticase peptone, 10 g/L NaCl and 5 g/L yeast extract. The bacterial cells were cultured to log phase, standardized to 2–3 × 10⁶ CFUs/ml, point inoculated onto a swim plate and incubated for 20 h at 37°C. The swimming areas were calculated using ImageJ software.

E. coli were grown to log phase and cultured with disulfiram or the solvent DMSO (n = 3 biologically independent samples) for 2 h. The cells were collected, total RNA was extracted using TRIzol^® Reagent according to the manufacturer’s instructions (Invitrogen), and the genomic DNA was removed using DNase I (TaKara). The RNA was then quantified with an ND-2000 (NanoDrop Technologies) and subjected to library construction, sequencing and data analysis. The RNA-seq strand-specific libraries were prepared with a TruSeq RNA sample preparation kit from Illumina (San Diego, CA) using 5 μg of total RNA. In brief, rRNA was removed using a RiboZero rRNA removal kit (Epicenter) and fragmented in fragmentation buffer. cDNA synthesis, end repair, A-base addition and ligation of the Illumina-indexed adaptors were performed according to Illumina’s protocol. Paired-end libraries were sequenced on an Illumina NovaSeq 6000 (Shanghai BIOZERON Co., Ltd.). The raw paired-end reads were trimmed and quality controlled in Trimmomatic with the following parameters: SLIDINGWINDOW: 4:15 MINLEN:75 (version 0.36; http://www.usadellab.org/cms/uploads/supplementary/Trimmomatic). Then, the clean reads were separately aligned to the reference genome in orientation mode using Rockhopper (http://cs.wellesley.edu/~btjaden/Rockhopper/) software. The differentially expressed mRNAs with a fold change > 2 or < 0.5 and p value < 0.05 were selected via the R package edgeR (https://bioconductor.org/packages/release/bioc/html/edgeR.html).

The transcriptomes from each chemical compound-treated E. coli experiment were downloaded from NCBI (colistin-treated E. coli: GSE220559; kanamycin-, H₂O₂-treated E. coli: GSE56133). Differentially expressed genes were analyzed using the bioconductor package limma (https://www.bioconductor.org/packages/release/bioc/html/limma.html). The clustering of transcriptomic changes was performed using the R package pheatmap (https://cran.r-project.org/web/packages/pheatmap/index.html) with the clustering method “ward. D.”

Cellular ROS levels were measured using CellROX (Thermo Fisher Scientific, United States, C10444) probes. Briefly, bacteria were cultured to log phase, diluted to an OD600 of 0.2–0.3, cultured with the indicated compounds for 1 h, and then, the probes were loaded at 5 μM at 37°C for 30 min. The cells were washed twice and resuspended in PBS, and then flow cytometry was performed using a DxFLEX or a CytoFLEX LX (Beckman Coulter, United States). A minimum of 20,000 events per sample were acquired using the FITC fluorescence channel during flow cytometry analysis.

The data are shown as the means ± SDs. Differences were analyzed by Student’s t test or one-way ANOVA with Tukey’s post hoc analysis. Kaplan–Meier survival analysis was performed for survival experiments. A p value < 0.05 was considered statistically significant. Statistical analysis was carried out in GraphPad Prism 9.3 (GraphPad Software).

To understand the effects of disulfiram on bacteria, we first monitored the 50% minimal inhibitory concentration (MIC₅₀) of disulfiram on the 6 common clinical bacteria including E. coli, S. aureus, K. pneumoniae, A. baumannii, P. aeruginosa and methicillin-resistant S. aureus (MRSA). The results revealed that disulfiram had relatively high MIC₅₀ values against these bacteria, with values of 256 μg/mL against E. coli and P. aeruginosa and 32 μg/mL against S. aureus, A. baumannii and MRSA (Figure 1a). These concentrations are higher than typical clinical plasma levels, suggesting in vitro activity may not fully translate in vivo. Next, we tested whether disulfiram killed bacteria at high concentrations. We used tetracycline, a bacteriostatic drug (Maier et al., 2021), as a control. The results revealed that while tetracycline killed E. coli at 1 × MIC₅₀ (0.5 μg/mL), disulfiram had nearly no killing effect on E. coli at 1 × MIC₅₀ (256 μg/mL; n = 3, 95.53 ± 6.93% vs. 68.90 ± 10.18%, p = 0.0389), indicating that disulfiram had no bactericidal effect (Figure 1b). A microplate reader was then used to monitor the bacterial growth curve to further observe the effect of disulfiram on the growth of E. coli. Disulfiram delayed the exponential growth phase of E. coli in a concentration-dependent manner (Figure 1c). Moreover, disulfiram halted E. coli growth instantly at the exponential phase (Figure 1d), indicating that disulfiram has a fast bacteriostatic effect on E. coli. These results demonstrated that disulfiram is a fast bacteriostatic drug with no bactericidal effect.

Antibiotic sensitizers or adjuvants play important roles in preventing microbial infections through the collaboration of antibiotics (Wang et al., 2024). Fast bacteriostatic drugs are often used as adjuvants for stationary-phase bactericidal drugs. A recent study showed that disulfiram could bind to New Delhi metallo-β-lactamase 1 (NDM-1) and resensitize drug-resistant bacteria to colistin and meropenem (Chen et al., 2023). However, we found that 32 μg/mL disulfiram could reduce the MIC of colistin on E. coli by 4-fold, which did not involve antibiotic resistance genes (FICI = 0.375, Figure 2a). The combination of disulfiram and colistin amplified the bactericidal effect of colistin alone (Figure 2b; Supplementary Figure S2a). Similarly, disulfiram also synergized with polymyxin B (FICI = 0.25) or kanamycin (FICI = 0.375) against E. coli or with colistin (FICI = 0.375) or kanamycin (FICI = 0.5) against A. baumannii (Supplementary Figure S1). The clinical isolates that generally cause sepsis or Healthcare-Associated Infections (HAIs) may have different sensitivity patterns than lab strains. We have also included clinical isolated strains for assay. The result showed that disulfiram synergized with colistin against clinical isolated E. coli (FICI = 0.047) and A. baumannii (FICI = 0.125, Supplementary Figure S1). A cell permeability assay revealed that although disulfiram alone did not permeabilize the cell membrane, it augmented the cell membrane damage caused by colistin (n = 4, 19.48 ± 2.74% vs. 8.31 ± 1.13%, p = 0.0286, Figure 2c). Moreover, we applied disulfiram to an animal model of E. coli intraperitoneal infection and found that combination therapy with disulfiram and colistin resulted in better outcomes than treatment with colistin alone for the disease (n = 7–8, 70.83 ± 20.41 h vs. 51.02 ± 34.85 h, p = 0.013, Figure 2d; Supplementary Figure S2b). These results indicated that disulfiram could act as a potential antibiotic sensitizer or adjuvant with stationary-phase bactericidal antibiotics, including colistin, polymyxin B or kanamycin.

Next, we sought to investigate the effect of disulfiram on E. coli. Swimming motility is vital to bacteria, as this process supports their movement toward resources. We found that under disulfiram treatment, the motility of E. coli was inhibited (At 64 μg/mL, n = 3, 100 ± 0% vs. 0 ± 0%, p = 0.0092, Figure 3a). This may result from decreased swimming ability or potential viability loss. Studies have shown that motility patterns are associated with bacterial cell shape and that the antibacterial cellular pathways of chemical compounds are associated with bacterial cytology profiles (Nonejuie et al., 2013; Wadhwa and Berg, 2022). Under microscopic observation, while sulbactam, ampicillin and meropenem, which are penicillin-binding protein (PBP)-targeting antibiotics, changed the length of E. coli, disulfiram did not have an effect (n = 30, 3.78 ± 1.04 μm vs. 3.77 ± 0.94 μm, p > 0.9999, Figure 3b), indicating that the target of disulfiram may not be bacterial PBPs (Penwell et al., 2015).

To study the detailed mechanisms by which disulfiram inhibits bacterial growth in detail, we subjected disulfiram-treated E. coli to RNA sequencing (RNA-seq) to study the transcriptome changes. Compared with the control E. coli group, disulfiram-treated E. coli expressed higher levels of antioxidant-related genes, including hmpA, cysI, nrdH, and frmA; higher levels of the iron–sulfur cluster repair-related gene ytfE; lower levels of zinT and higher levels of zntA, two genes associated with zinc binding and zinc translocating; higher levels of the pyruvate uptake-related gene btsT; and lower levels of the nucleotide synthesis-related gene ygeW (Figure 4a). The top genes among the differentially expressed genes (DEGs) were hmpA, which encodes flavohemoglobin and is linked to the cell response to oxidative or nitrosative stress (Bonamore and Boffi, 2008). Azoles and quinones are reported to be inhibitors and substrates of flavohemoglobin (Moussaoui et al., 2018; Nobre et al., 2010). We first tested whether these compounds could augment the effect of disulfiram. The results of the checkerboard assay suggested that inhibiting or monitoring flavohemoglobin only had a very mild effect on disulfiram inhibition of E. coli (Supplementary Figures S2c,d), indicating that drugs or chemicals that target flavohemoglobin cannot efficiently enhance the effect of disulfiram. Pathway enrichment analysis further revealed that the differentially regulated genes were enriched in the iron-sulfer cluster binding, oxidoreductase activity and RNA binding pathways (Figure 4b). These results indicate that disulfiram may affect the growth of E. coli through oxidant activity, leading to iron-sulfer cluster damage and the suspension of genetic material synthesis. To provide a preliminary overview of potential similarities in gene expression profiles, we performed clustering analysis by comparing our transcriptomic data with publicly available datasets that examined RNA expression changes in E. coli in response to various bacteriostatic and bactericidal agents. The results revealed that the disulfiram-induced transcriptomic changes were closely associated with those caused by H₂O₂, which cause oxidative damage to pathogens (Schairer et al., 2012). In contrast, bactericidal drugs, including kanamycin or colistin, had relatively distinct transcriptomic changes (Figure 4c). Furthermore, we verified that the ROS inhibitor N-acetyl-L-cysteine (NAC) fully prevented the bacteriostatic effect of disulfiram on both E. coli and S. aureus (Figures 4d,e). These results confirmed that disulfiram inhibits bacterial growth by inducing ROS.

The coeffect of metal ions is proposed as a biological mechanism of disulfiram (Skrott et al., 2017; Gao et al., 2022). Indeed, disulfiram and copper kill Mycobacterium tuberculosis in a synergetic manner (Dalecki et al., 2015). However, when copper gluconate was used as a Cu²⁺ supply in E. coli growth medium, we found the opposite effect that copper ions counteracted the inhibitory effect of disulfiram on E. coli growth (Supplementary Figure S3a). We proposed that copper ions and the metabolites of disulfiram (ditiocarb and diethyldithiocarbamate, which have bacteriostatic effects similar to those of disulfiram) could form chemical complexes (bis (diethyldithiocarbamate)–copper), which may cover the effective bacteriostatic group of disulfiram. Moreover, the copper chelator tetrathiomolybdate (TTM) had a very weak synergistic effect with disulfiram on the growth of E. coli (Supplementary Figure S3a). These results indicated that the disulfiram-mediated inhibition of bacterial growth was independent of copper ions.

As shown by the transcriptomic results, Fe-S cluster damage and repair participate in the mechanism of disulfiram. Ferric ions may induce ROS through the Fenton reaction. However, upon cotreatment with disulfiram, the ferric chelator DFO (deferoxamine) only weakly inhibited the growth of E. coli, whereas ferric citrate had no effect (Supplementary Figure S3b).

Zinc binding and zinc translocation may play a role in this process, as the zinT and zntA genes were upregulated in disulfiram-treated E. coli. Surprisingly, we found that while zinc ions augmented the inhibitory effect of disulfiram on both E. coli (Figure 5a) and S. aureus (Figure 5b), the zinc chelator TPEN (N, N, N′, N′-tetrakis- (2-pyridylmethyl) ethylenediamine) counteracted the inhibitory effect of disulfiram, fully restored the growth of E. coli (Figure 5c) and partly restored the growth of S. aureus (Figure 5d). These results suggested that zinc ions play important roles in the bacteriostatic mechanisms of disulfiram. Therefore, disulfiram has a ROS-dependent bacteriostatic effect. We next tested the relationship between ROS and zinc. We loaded CellROX, a fluorescent ROS probe, into bacteria as an indicator of ROS and found that while zinc ions augmented the disulfiram-induced increase in ROS level in E. coli (n = 3, 74.47 ± 4.98 vs. 65.80 ± 1.13, p = 0.0049) and S. aureus (n = 3, 129.0 ± 9.54% vs. 94.53 ± 8.63%, p = 0.0021), TPEN treatment fully suppressed this increase in both E. coli (n = 3, 64.13 ± 0.47% vs. 65.80 ± 1.13%, p = 0.9356) and S. aureus (n = 3, 47.97 ± 5.24% vs. 94.53 ± 8.63%, p = 0.0001, Figures 5d,e). These results suggested that the bacteriostatic effect of disulfiram is mediated by zinc ions, and that ROS levels depend on zinc.