An enhanced sonodynamic antibacterial strategy using bismuth porphyrin metal-organic frameworks to inhibit bacterial endogenous antioxidant H₂S metabolism

Abstract

Abstract

Surgical site infections (SSIs) represent a global healthcare burden, a challenge intensified by the increasing prevalence of antibiotic resistance. To address drug-resistant bacterial infections, various non-antibiotic therapeutic strategies have been explored, among which sonodynamic therapy (SDT) has emerged as a promising approach due to its deep tissue penetration and controllability. Nevertheless, bacteria are not defenseless against SDT and can produce endogenous hydrogen sulfide (H₂S) as an antioxidant to resist reactive oxygen species (ROS)-mediated killing, thereby limiting therapeutic efficacy. Herein, we reported a dual-functional and synergistic sonodynamic antibacterial strategy based on two-dimensional bismuth porphyrin metal–organic frameworks (Bi-TCPP). Bi-TCPP simultaneously functioned as an efficient sonosensitizer to generate singlet oxygen (¹O₂) under ultrasound irradiation and as an effective scavenger of bacterial-derived H₂S through strong Bi³⁺–H₂S interactions. By weakening the bacterial antioxidant reduction system and disrupting redox homeostasis, this strategy enabled efficient bacterial eradication via low-level ROS, thereby significantly amplifying SDT-mediated antibacterial efficacy. The effectiveness of this synergistic strategy was systematically demonstrated against multiple bacterial species, including drug-resistant strains and different survival states such as planktonic bacteria and biofilms in vitro. It was further validated in vivo, highlighting its potential as an alternative therapeutic approach for treating drug-resistant bacterial infections.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12951-026-04437-8.

Introduction

Surgical site infections (SSIs), a prevalent and potentially devastating postoperative complication, remain a formidable challenge in clinical medicine [1, 2]. These infections are frequently driven by colonization with specific pathogens, including Gram-positive and Gram-negative bacteria such as Staphylococcus aureus (S. aureus) and Escherichia coli (E. coli), as well as various antimicrobial-resistant strains [3, 4]. Another paramount challenge is the formation of bacterial biofilms within the surgical site, leading to persistent, recalcitrant infections that are extremely difficult to eradicate and may result in severe systemic complications and drug resistance [5, 6]. Conventional management, often involving repeated surgical interventions and prolonged antibiotic regimens, not only heightens the risk of complications but also imposes a significant healthcare burden [7]. Consequently, the development of non-antibiotic antibacterial strategies is imperative for improving clinical outcomes.

Advances in non-antibiotic antibacterial strategies have recently been investigated, with those based on reactive oxygen species (ROS) generation emerging as a research hotspot. Among ROS-based antibacterial strategies, sonodynamic therapy (SDT) has been identified as a promising approach for antimicrobial intervention due to its high responsiveness and permeability [8, 9]. SDT employs low-intensity focused ultrasound to activate sonosensitizing agents, thereby catalyzing the generation of cytotoxic ROS within infected tissues. Mechanistic studies have demonstrated that SDT-derived ROS, particularly singlet oxygen (¹O₂) and hydroxyl radicals (•OH) [10, 11], induce bactericidal effects through peroxidation of membrane lipids, oxidative damage to essential proteins, and degradation of genomic DNA, exhibiting potent activity against both planktonic and drug-resistant microorganisms [12]. Another distinctive therapeutic advantage of SDT lies in its efficacy against bacterial biofilms - a major pathophysiological determinant of chronic infections and antimicrobial treatment failure. The synergistic action of ultrasonic mechanical energy and ROS-mediated chemical attack enables SDT to penetrate the extracellular polymeric matrix, disrupt biofilm architecture, and eradicate embedded pathogens [13]. However, achieving satisfactory antibacterial effects requires high ROS concentrations, which can easily lead to increased oxidative stress at the site of infection, causing serious complications. Hence, effective killing of bacteria at the site of infection at low doses of ROS is a worthwhile research topic.

Under normal physiological conditions, an appropriate amount of ROS can maintain normal bacterial growth and metabolism, such as participating in quorum sensing, biofilm formation, DNA binding, oxidative modification, and gene transcription processes. However, when the production of ROS exceeds the bacterial clearance capacity, an imbalance of oxidation-reduction occurs, leading to oxidative stress and damage to biomolecules such as proteins, lipids, and nucleic acids, as the mentioned SDT sterilization principle [14]. However, to cope with oxidative stress, bacteria have evolved complex antioxidant defense systems [15]. The enzymatic antioxidant system mainly includes superoxide dismutase (SOD), catalase (CAT), peroxidases (such as alkyl hydroperoxide reductase and glutathione peroxidase), the thioredoxin system, and the glutathione system [16]. These enzymes directly clear ROS or repair oxidative damage through catalytic reactions. Previously, we found that bacteria can produce hydrogen sulfide (H₂S), an antioxidant that confers significant resistance to oxidative stress damage [17, 18]. This gaseous mediator exerts protective effects by chelating ferrous iron, thereby inhibiting hydroxyl radical generation, while enhancing the activity of catalase and superoxide dismutase [19]. These coordinated actions promote bacterial survival under antibiotic pressure and facilitate persistent infections, suggesting H₂S inhibition could enhance antimicrobial efficacy against refractory infections.

Bismuth (Bi) exhibits a strong affinity for hydrogen sulfide (H₂S) [20], forming non-toxic bismuth sulfide (Bi₂S₃) through selective chemical interaction. Due to its low toxicity, bismuth has been widely utilized in pharmaceuticals, including gastrointestinal therapeutics such as bismuth subsalicylate and bismuth citrate [21, 22]. Notably, Bi³⁺ ions effectively sequester H₂S—a key virulence factor in pathogenic bacteria—by precipitating it as inert Bi₂S₃. This reaction disrupts bacterial metabolic tolerance under oxidative stress, impairing microbial survival. Furthermore, bismuth-based metal-organic frameworks have garnered significant attention for their structural tunability, high surface area, precise dimensional control, and multifunctional antimicrobial properties. These materials leverage the synergistic effects of Bi³⁺ release and organic ligands, offering enhanced antibacterial efficacy against drug-resistant strains [22, 23].

Hence, we synthesized a 2D bismuth-porphyrin MOF (Bi-TCPP) that exhibited dual antibacterial functions under ultrasound: H₂S scavenging via Bi₂S₃ formation and ¹O₂ generation. Remarkably, H₂S depletion sensitized both planktonic and biofilm-embedded bacteria to oxidative killing, requiring a small amount of ¹O₂ doses. Omics analyses revealed that H₂S exhaustion disrupts redox homeostasis, leading to fatal oxidative stress through metabolic pathway dysregulation. As illustrated in Scheme 1, this synergistic mechanism establishes Bi-TCPP as an effective antimicrobial platform against resistant pathogens.

Scheme 1.

The schematic diagram illustrated the synergistic antibacterial mechanism of Bi-TCPP, a bismuth-based 2D-MOFs

Materials and methods

Preparation of Bi-TCPP

Bi(NO₃)₃·5H₂O (110.5 mg), TCPP (66.6 mg), 0.1 mL HCl (12 mol/L), and 30 mL DMF were added to a beaker, and the mixture was stirred gently at room temperature for 1 h. After thorough homogenization, the mixture was transferred to a 100 mL Teflon-lined stainless-steel autoclave, heated to 120 °C, and kept for 6 h. Then, the reaction was stopped and cooled down to room temperature. The obtained product was centrifuged at 10,000 rpm for 5 min, then washed with a DMF (30 mL)/HCl (0.1 mL) mixture and ethanol at least 3 times, respectively. Eventually, it was dried in an oven at 60 °C (Fig. 1a).

Fig. 1.

Structural and Compositional Characterization of Bi-TCPP. (a) Schematic illustration of the Bi-TCPP synthesis process. (b) Transmission electron microscopy (TEM) and (c) Scanning electron microscopy (SEM) images revealing the morphology of the as-synthesized Bi-TCPP. (d) EDS layered image and (e) Corresponding elemental mapping of Bi-TCPP, demonstrating uniform element distribution. Scale bar: 200 nm. (f) Full-range XPS survey spectrum and (g) High-resolution Bi 4f spectrum of Bi-TCPP. (h) XRD pattern of Bi-TCPP, confirming its crystalline structure. (i) Size distribution of Bi-TCPP nanoparticles measured by DLS

Characterization of Bi-TCPP and Bi₂S₃

The microstructure of the samples and their elemental composition were observed and evaluated by SEM (ZEISS Sigma 300, Germany) and TEM (FEI Talos F200S, America). The phase structures of Bi-TCPP and Bi₂S₃ were analyzed by XRD (Rigaku Smart Lab SE, Japan). The elemental oxidation states were determined by X-ray photoelectron spectroscopy (XPS, Shimadzu UV-3600i Plus, Japan). Inductively coupled plasma (ICP, Agilent 5800, America) analysis was employed to quantify the elemental composition of Bi-TCPP and to assess the release profile of bismuth ions in phosphate-buffered saline (PBS). Subsequently, nanoparticle size and zeta potential of Bi-TCPP were evaluated using dynamic light scattering (DLS, Malvern Zetasizer Nano ZS90, England).

Sonodynamic effect of Bi-TCPP

The ¹O₂ release capacity of the samples was assessed using the DPBF method. A mixed solution was obtained by dissolving 50 µM DPBF in 3 mL of ethanol. Subsequently, 1 mL of the material suspension was added to the DPBF solution and exposed to ultrasound (1 MHz, 1.5 W cm^{− 2}) [24]. Subsequently, the absorbance values of the mixed solution following ultrasonic irradiation were determined by a UV spectrophotometer (Shimadzu UV-3600i Plus, Japan). The absorbance of the DPBF solution at 410 nm was measured using a microplate reader (Tecan, Switzerland) to quantitatively evaluate the generation of singlet oxygen (¹O₂). The reactive oxygen species (ROS) elicited by ultrasonic excitation of diverse materials was further measured using an ESR spectrometer.

Biocompatibility of Bi-TCPP and Bi₂S₃

The cytotoxicity of Bi-TCPP at various concentrations was evaluated using the CCK-8 kit, and its potential hemolytic effect on BALB/c mouse erythrocytes was assessed via a hemolysis assay. For in vivo evaluation, 50 µg of Bi₂S₃ was implanted subcutaneously in mice, and the implantation sites were photographed at 0, 1, 3, 7, 12, and 24 days to monitor tissue response. Subsequently, major organs, including the heart, liver, spleen, lung, and kidney, were harvested for hematoxylin and eosin (HE) staining to examine histopathological changes. In parallel, after treating a subcutaneous infection model in mice with 100 µg mL⁻¹ Bi-TCPP, HE staining was performed to assess the microstructures of the same organs.

Cell and bacterial culture

L929 cells and RAW264.7 cells were cultivated in DMEM cell culture medium supplemented with 10% fetal bovine serum. Post-inoculation, S. aureus (ATCC 25923), E. coli (ATCC 25922), MRSA (ATCC 43300), and CRE-E. coli (Carbapenem-Resistant Escherichia coli, CCUG 58540) were incubated in Luria-Bertani (LB) medium at 37 °C until the bacterial density reached 10⁸ colony-forming units (CFU) mL^− 1.

H₂S detection assay

The quantification of intracellular hydrogen sulfide (H₂S) in bacterial cells was performed using the cell-permeable fluorescent probe Sulfidefluor-7 acetoxymethyl ester (SF7-AM; Cayman Chemical, Ann Arbor, MI, USA) according to established protocols [25]. Following pretreatment with 100 µM L-cysteine (Sigma-Aldrich, St. Louis, MO, USA), 10 µM pyridoxal 5’-phosphate (PLP; Aladdin, Shanghai, China), and 2.5 µM SF7-AM, bacterial suspensions were incubated at 37 ℃ under light-protected conditions for 30 min. After two cycles of centrifugation (5000 × g, 5 min) and washed with phosphate-buffered saline (PBS), the bacterial pellets were resuspended in prewarmed PBS supplemented with 100 µM L-cysteine and 10 µM PLP for an additional 30 min incubation. For microscopic analysis, processed samples were transferred to confocal culture dishes and examined by confocal laser scanning microscopy (CLSM) using excitation/emission wavelengths of 488/526 nm.

Antibacterial capacity in vitro

1 mL of bacterial suspension (10⁷ CFU mL^{− 1}) was incubated with 1 mL of Bi-TCPP (100 µg mL⁻¹) suspension in a 2 mL EP tube [26]. For the purpose of evaluating the antibacterial effect of the material, the CTR (Ctrl) group and the BiT (Bi-TCPP) group were co-incubated at room temperature for 10 min, while the US (ultrasound irradiation alone) group and the BiT-US (exposed to ultrasound with Bi-TCPP) group had the EP tubes placed on the ultrasonic machine for 5 min (1 MHz, 1.5 W cm^{− 2}). After 2 h, the bacterial solution was withdrawn from the EP tube, uniformly smeared on the LB agar plate, and incubated at 37 °C for 18 h. At least 3 parallel samples were fabricated for each group. When the bacteria reached an appropriate colony size, photographs were taken, and the number of colonies in each group was counted. Then, the antibacterial efficiency was computed based on the number of colonies. To examine the bacterial microscopic morphology, the bacteria of S. aureus (ATCC 25923), E. coli (ATCC 25922), MRSA(ATCC 43300), and CRE-E. coli (CCUG 58540) in different treated groups were fixed in 4% formaldehyde solution and washed with PBS. Moreover, the bacteria were dehydrated with ascending concentrations of ethanol solutions (30%, 50%, 70%, 90% and 100%), air-dried, and then SEM images were captured. Then, the samples were stained with uranyl acetate and lead citrate and placed on copper grids for TEM image acquisition. Subsequently, after being grouped and treated, they were stained with a bacterial Live/Dead staining kit for 15 min, then photographed and recorded under a laser confocal microscope (CLSM; Leica TCS SP8). The intracellular ROS levels of bacteria after group treatment were determined and recorded using CLSM (Leica TCS SP8) with a ROS detection kit.

After treatment, bacterial ATP and protein leakage were measured using ATP Assay Kit (Beyotime, China) and BCA Protein Assay Kit (Beyotime, China). Besides, the intracellular GSH levels were determined using an Elabscience kit (Elabscience, China).

Biofilm elimination capacity in vitro

The metal discs (Ti, 1 cm in diameter) were placed in 2 mL of the bacterial solution and incubated for 24 h to form biofilms. Grouping was as follows: the CTR group served as the control; BiT was added to the bacterial solution and incubated for 5 min; and the US group and the BiT-US group received 5-minute US irradiation (1 MHz, 1.5 W cm^− 2). Then, the metal sheet was retrieved and subjected to ultrasonic treatment with an ultrasonic water bath (B3500S-MT, Branson Ultrasonics Co) at 10 Hz for 10 min. After incubation at 37 °C, the number of bacteria on the agar plate surface was enumerated. Furthermore, the surface biofilm structure of the treated metal sheet was observed through SEM. The live/dead staining method was used to assess biofilm formation. Eventually, the biofilm mass was quantified using the crystal violet staining method.

RNA-Seq analysis

Collect S. aureus and E. coli after treatment with Bi-TCPP and ultrasonic processing (1 MHz, 1.5Wcm^− 2, 5 min), as well as naturally growing S. aureus and E. coli. For each condition, obtain three independent RNA sequencing samples. After establishing a DNA library from multiple samples, PE150 mode sequencing was performed on the Illumina sequencer. Differential gene expression analysis was conducted using DESeq (v1.38.3). Genes were considered differentially expressed if they showed a |log₂FC| > 1 and a p value < 0.05.

Subcutaneous infection model in mice

All experimental procedures and protocols were reviewed and approved by the Animal Ethics Committee of Shanghai Pudong Hospital (Approval No. KCZX20240928-002). A murine dorsal subcutaneous infection model was established to assess the antibacterial capacity and biocompatibility of Bi-TCPP under sonodynamic therapy (SDT). Briefly, 6-week-old SPF male mice (BALB/c) were selected. After anesthesia, their dorsal hair was shaved with a gold-titanium plate (diameter of 5 mm, 1 mm thick) implanted, and 50 µL of S. aureus (10⁶ CFU mL^− 1) was added simultaneously to simulate an implant-associated infection. After one day, the mice were categorized into four groups: CTR (PBS only), BiT (200 µL of 100 µg mL⁻¹ Bi-TCPP solution injected into the abscess), US (treated with ultrasound only at 1 MHz, 1.5 W cm⁻², 5 min), and BiT-US (Bi-TCPP added, followed by the same ultrasound treatment).

Therapeutic efficacy in vivo

Under anesthesia (45 mg/kg, intraperitoneal), the metal sheets were retrieved on days 7 and 12 for bacterial analysis via sonication and plating. The mice were then euthanized by cervical dislocation. The collected abscess and soft tissue samples were fixed in 4% paraformaldehyde for processing with H&E, Masson trichrome, Giemsa, and immunohistochemical (IL-6, TNF-α) staining.

Statistical analysis

All the data in this research are expressed as the mean ± standard deviation (SD), where n ≥ 3. Images and fluorescence intensities were analyzed using Imagej software. One-way analysis of variance and T-tests were used to assess significance; p-values < 0.05 indicated statistical significance.

Results and discussion

Synthesis and characterization of Bi-TCPP

Transmission electron microscopy (TEM) and scanning electron microscopy (SEM) images revealed that Bi-TCPP consists of rectangular nanosheet-like particles (Fig. 1b, c). Elemental mapping images confirmed the homogeneous distribution of Bi, C, O, and N elements within the Bi-TCPP nanosheets (Fig. 1d, e). The X-ray photoelectron spectroscopy (XPS) spectra of our material demonstrate its elemental composition. Meanwhile, the emergence of the Bi 4f peak confirmed the successful incorporation of bismuth (Fig. 1f, S2). In the Bi 4f spectrum of Bi-TCPP, two peaked at 164.9 eV and 159.6 eV were observed, which correspond to Bi³⁺ 4f₅/₂ and Bi³⁺ 4f₇/₂ [22], respectively. This indicated that the bismuth existed predominantly in the + 3 oxidation state (Fig. 1g). Furthermore, analysis of the O 1s spectrum after carbon correction revealed a characteristic peak at 530.30 eV, assigned to the Bi–O bond, confirming the chemical interaction between bismuth and the organic framework (Figure S3). The X-ray diffraction (XRD) pattern of Bi-TCPP revealed a prominent diffraction peak below 10°, indicative of its internal TCPP crystal structure (Fig. 1h). DLS analysis indicated that Bi-TCPP exhibited an average hydrodynamic diameter of 395.3 nm with a zeta potential of approximately − 6.88 mV (Fig. 1i, S1). What’s more, Transmission electron microscopy (TEM), scanning electron microscopy (SEM), and DLS size distribution measurements revealed that Bi-TCPP consisted of rectangular, nanosheet-like particles approximately 400 nm in size.

ICP analysis showed that the mass fraction of bismuth in Bi-TCPP was 33.6%, while TCPP accounted for approximately 66.7% (Figure S4). The stability of Bi-TCPP in ethanol and deionized water was further evaluated by monitoring particle size and TEM morphology. No significant change in particle size was observed within the first 3 days. However, slight aggregation with increased particle size was detected on day 7, and a pronounced increase in size with extensive nanosheet aggregation was observed by day 14, indicating a gradual decrease in stability over time (Figure S5).

Sonodynamic effect and ion release

As depicted in Fig. 2a, Bi-TCPP could be excited by ultrasonic irradiation to produce toxic singlet oxygen (¹O₂). The ROS probe 1,3-diphenylisobenzofuran (DPBF) undergoed rapid oxidation to a colorless product upon contact with ¹O₂[27]. Therefore, its absorbance could be employed to monitor the variations in ¹O₂ concentration (Fig. 2b). Ultrasound alone could not affect the absorbance of the DPBF reagent over time. Under ultrasonic excitation, Bi-TCPP exhibited stronger capacity to generate ¹O₂ compared to TCPP, with the concentration of ¹O₂ increasing over time. (Fig. 2c, S6). Furthermore, following 5 min of ultrasonic stimulation, the amount of singlet oxygen (¹O₂) generated increased with the mass of Bi-TCPP (Fig. 2d, S7). The electron spin resonance (ESR) results of TCPP and Bi-TCPP under ultrasonic irradiation, when compared with those of its ligand, also corroborate the findings mentioned above (Fig. 2e, f). ESR analysis revealed robust ¹O₂ generation, while no discernible ·OH signal was observed. Although a weak O₂·⁻ signal was detected, it did not exhibit the characteristic ESR pattern, indicating that ¹O₂ is the dominant ROS responsible for the observed effect. (Figure S8).

Fig. 2.

In Vitro Sonodynamic Performance. (a) Schematic diagram illustrating the proposed mechanism of Bi-TCPP-mediated sonodynamic therapy (SDT). (b) Working principle of the DPBF probe for singlet oxygen (¹O₂) detection. c, d) Time-dependent decomposition of DPBF by TCPP (100 µg/mL) and Bi-TCPP (100 µg/mL) under ultrasound irradiation, reflecting ¹O₂ generation kinetics. e, f) Electron spin resonance (ESR) spectra of ¹O₂ generated from Bi-TCPP (e) and TCPP (f) and using TEMP as the trapping agent. g) Release of Bi³⁺ from 500 µg/mL Bi-TCPP in PBS with or without ultrasound stimulation (US: ultrasound)

Inductively coupled plasma (ICP) analysis further revealed the release behavior of bismuth ions from Bi-TCPP. When 500 µg mL⁻¹ Bi-TCPP was dispersed in PBS, bismuth ions were gradually released over time, showing a slow and sustained release within the first 8 h. No significant difference in release rate was observed between the ultrasound-treated and untreated groups during this period. However, beyond 8 h, ultrasonic stimulation markedly accelerated bismuth ion release compared to the group without external stimulation (Fig. 2g).

Biocompatibility in vitro

Before in vivo application, the biosafety of Bi-TCPP was systematically evaluated. After co-incubation with L929 cells for three days, Bi-TCPP showed negligible cytotoxicity at concentrations up to 100 µg mL⁻¹ (Figure S9). Hemolysis assays further confirmed its excellent blood compatibility, with minimal erythrocyte damage even at 400 µg mL⁻¹ (Figure S10). In RAW cells subjected to SDT, Bi-TCPP induced a concentration-dependent increase in ROS generation within 200 µg mL⁻¹ (Figure S11). Corresponding CCK-8 assays indicated a slight reduction in cell viability at 50–100 µg mL⁻¹ and a significant decrease at 200 µg mL⁻¹ (Figure S12). Based on these findings, a concentration of 100 µg mL⁻¹ was selected for subsequent experiments to balance biosafety and functional efficacy.

Endogenous H₂S scavenging ability

Bacteria employ hydrogen sulfide (H₂S) as a protective mechanism against oxidative stress [28]. When S. aureus and E. coli were incubated with Bi-TCPP (100 µg mL⁻¹) without ultrasound, only very small amounts of H₂S were detected: 7.7 ± 1.2% and 5.2 ± 0.6% for S. aureus at 1 and 2 h, respectively, and 15.0 ± 2.9% and 5.9 ± 2.9% for E. coli. Upon ultrasonic stimulation, limited H₂S production was observed post-treatment. After 5 min of sonication followed by 1 h of incubation, the H₂S level in S. aureus treated with Bi-TCPP dropped to 46.3 ± 8.3% of that in the TCPP group. By 2 h, bacterial viability was significantly diminished, and H₂S was nearly depleted (9.9 ± 2.9% of the TCPP 1 h value). Similarly, in E. coli, H₂S levels in the Bi-TCPP group were 26.0 ± 7.4% (1 h) and 5.9 ± 2.1% (2 h) relative to the TCPP control (Fig. 3a–h, S13). Notably, Bi-TCPP at 100 µg mL⁻¹ exhibited strong H₂S-scavenging efficacy. X-ray photoelectron spectroscopy (XPS) analysis of bacterial precipitates after sonication revealed binding energies of Bi and S consistent with Bi₂S₃ crystals (Fig. 3i–j) [28], confirming the formation of bismuth sulfide. These results indicated that under oxidative stress, bacteria generated endogenous H₂S as a defense, and Bi-TCPP disrupted this tolerance mechanism by reacting with H₂S to form Bi₂S₃.

Fig. 3.

H₂S Scavenging and in Situ Bi₂S₃ Formation by Bi-TCPP. (a) H₂S levels in S. aureus at time points of 1 h and 2 h following SDT. Scale bar: 100 μm. (b) H₂S levels in S. aureus at 1 h and 2 h following TCPP-mediated SDT. (c) H₂S levels in S. aureus at 1 h and 2 h following Bi-TCPP-mediated SDT. (d) H₂S levels in E. coli at time points of 1 h and 2 h following SDT. Scale bar: 100 μm. (e) H₂S levels in E. coli at 1 h and 2 h following TCPP-mediated SDT. (f) H₂S levels in E. coli at 1 h and 2 h following Bi-TCPP-mediated SDT. (g) Quantitative analysis of H₂S fluorescence intensity in S. aureus measured using SF7-AM probe. (h) Quantitative analysis of H₂S fluorescence intensity in E. coli. (i) High-resolution XPS spectra of the Bi 4f and S 2p regions from the precipitates of S. aureus after SDT treatment with Bi-TCPP (100 µg mL⁻¹). (j) High-resolution XPS spectra of the Bi 4f and S 2p regions from the precipitates of E. coli after SDT treatment with Bi-TCPP (100 µg mL⁻¹). All quantitative data were expressed as mean ± SD (n = 3). Statistical significance was determined by one-way ANOVA with Tukey’s post-hoc test (****p < 0.0001)

Antibacterial ability in vitro

S. aureus and E. coli infections are common causative pathogens of surgical site infections (SSIs). We evaluated the antibacterial performance of Bi-TCPP under ultrasound stimulation through colony counting and Live/Dead staining assays using these strains and their drug-resistant variants (MRSA and CRE-E. coli). Figures 4a-b demonstrated that after 2 h of ultrasound-activated co-incubation, Bi-TCPP showed significant bactericidal activity with a killing efficiency of 98.5 ± 0.4% against S. aureus and 99.8 ± 0.2% against E. coli (Fig. 4a-b). Given the clinical importance of resistant strains in SSIs [29], we further examined the material’s efficacy against MRSA (98.0 ± 1.5% inhibition) and CRE-E. coli (99.2 ± 0.3% inhibition) (Figure S14).

Fig. 4.

Antibacterial and Anti-Biofilm Performance of Bi-TCPP under ultrasound. (a) Representative colony formation unit (CFU) images of S. aureus and E. coli under different treatments: control, BiT (Bi-TCPP only), US (ultrasound only), and BiT-US (Bi-TCPP + ultrasound). (b) Quantitative analysis of antibacterial efficiency derived from (a). (c) SEM images of bacterial morphology under different treatments, with insets showing magnified views of local membrane structures. Scale bars: 1 μm (overview), 400 nm (magnified views). (d) Fluorescence imaging of intracellular ROS levels in S. aureus and E. coli. Scale bars: 100 μm. (e) Quantitative analysis of DCFH fluorescence intensitcy from (d). (f) 3D confocal Laser scanning microscopy (CLSM) images (Live/Dead staining) and corresponding SEM images of S. aureus biofilms after different treatments. Scale bars: 100 μm and 5 μm. (g) 3D CLSM images (Live/Dead staining) and corresponding SEM images of E. coli biofilms after different treatments. Scale bars: 100 μm and 5 μm. (h) Quantitative analysis of anti-biofilm efficacy against S. aureus and E. coli under different treatments, as determined by the spread plate method. (i) Quantitative analysis of bacterial biofilm activity for S. aureus and E. coli under different treatments. All quantitative data were presented as mean ± SD (n = 3). Statistical significance was determined by one-way ANOVA with Tukey’s post-hoc test (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001)

Live/dead staining revealed exclusive red fluorescence (dead cells) in BiT-US groups compared to control groups showing green fluorescence (live cells), consistent with colony counting results (Figure S15, S16). SEM imaging revealed intact bacterial membranes in control groups. In contrast, severe membrane disruption was observed in the BiT-US groups, as confirmed by TEM, with complete structural disintegration, including cell wall separation and cytoplasmic leakage. (Fig. 4c, S17) Comparative studies with TCPP at 100 µg mL⁻¹ demonstrated Bi-TCPP’s superior sonocatalytic activity (Figure S18). ROS measurements showed significantly elevated levels in BiT-US groups (E. coli : 2241 ± 161.4 a.u.; S. aureus: 1821 ± 366.8 a.u.) versus control (Figs. 4d-e). These results indicated optimal biosafety at 100 µg mL⁻¹, while maintaining potent antibacterial effects through combined ROS generation and disruption of H₂S tolerance.

Antibiofilm ability of Bi-TCPP in vitro

Surgical site infections (SSIs) are frequently caused by bacterial colonization of implanted medical devices, during which pathogens form dense biofilms that confer antibiotic resistance [30]. To evaluate the biofilm-disrupting capability of Bi-TCPP under ultrasound stimulation, we performed comprehensive analyses using confocal laser scanning microscopy (CLSM) and scanning electron microscopy (SEM). Three-dimensional reconstruction of Live/Dead-stained biofilms demonstrated that combined Bi-TCPP and ultrasound (BiT-US) treatment effectively eradicated biofilms formed by both S. aureus and E. coli (Fig. 4f, g). SEM observations further confirmed these findings, revealing extensive bacterial shrinkage, detachment, and cell death in BiT-US groups, while treatment with either Bi-TCPP or ultrasound alone only partially reduced bacterial adherence (Fig. 4h). Quantitative assessment of biofilm biomass demonstrated that untreated control (CTR) groups maintained 100% relative biomass for both S. aureus (100 ± 11.8%) and E. coli (100 ± 5.0%) (Fig. 4i). Importantly, similar experiments were conducted with drug-resistant MRSA and CRE-E. Coli biofilms revealed equally potent anti-biofilm effects, with CLSM imaging and quantitative assays confirming near-complete biofilm elimination (Figure S19-S22). The combined results demonstrated that Bi-TCPP-mediated sonodynamic therapy not only exhibited exceptional antibacterial activity against planktonic bacteria but also effectively penetrates and dismantles recalcitrant biofilm, a critical advancement for addressing implant-associated infections where conventional antibiotics often fail due to biofilm-mediated resistance mechanisms. The multimodal approach of combining bismuth-coordinated porphyrin with ultrasound activation represented a promising strategy for preventing and treating device-related infections.

Broad-spectrum antibacterial mechanism of Bi-TCPP-based sonodynamic therapy

RNA sequencing of S. aureus following Bi-TCPP-mediated sonodynamic treatment identified 427 significantly upregulated and 388 downregulated genes (p < 0.05, |log₂FC| > 1) (Fig. 5a). KEGG pathway enrichment analysis demonstrated significant enrichment of these differentially expressed genes in ABC transporters, quorum sensing, cysteine and methionine metabolism, oxidative phosphorylation, the citrate cycle (TCA cycle), and the phosphotransferase system (Fig. 5b). Gene Ontology (GO) analysis further indicated that differentially expressed genes were predominantly involved in biological processes, including sulfur compound metabolism and biosynthesis, and purine ribonucleotide biosynthesis (Figure S23). Gene Set Variation Analysis (GSVA) scores simultaneously revealed activation of both ABC transporters and sulfur metabolism pathways, along with suppression of biofilm formation pathways (Figure S24). Gene Set Enrichment Analysis (GSEA) confirmed significant activation of ABC transporters (NES = 1.51, p = 0.01) in the treatment group (Figure S25). Expression profiling of core genes demonstrated predominant upregulation patterns in pathways related to sulfur metabolism, cysteine and methionine metabolism, oxidative phosphorylation, and the TCA cycle (Fig. 5c). Notably, genes associated with oxidative phosphorylation were predominantly downregulated in the material-treated group, whereas those related to sulfur metabolism, cysteine and methionine metabolism, and the TCA cycle were mainly upregulated (Fig. 5d). CysE was identified as a potential hub gene connecting sulfur metabolism with cysteine and methionine metabolism.

Fig. 5.

Transcriptomic analysis of the antibacterial mechanism of Bi-TCPP-based sonodynamic therapy in S. aureus. (a) Volcano plot of differentially expressed genes (DEGs) between Bi-TCPP treated group (B) and control group (S). (b) KEGG pathway enrichment analysis of DEGs displayed in a meteor plot. (c) Heatmap showing expression patterns of core genes involved in sulfur metabolism, cysteine and methionine metabolism, oxidative phosphorylation, and TCA cycle pathways across groups B and S. (d) Chord diagram illustrating pathway associations and Log₂FC values of core genes from the metabolic pathways analyzed in c). Data represent n = 3 biologically independent samples. Statistical thresholds: |log₂FC| > 1, p < 0.05 for differential analysis. (e) Relative expression of peroxidase (POD)-related genes bsaA and chdC in the CTR (control) and Bi-TCPP plus ultrasound (US) groups. (f) Relative expression of catalase (CAT)-related gene katA in CTR and Bi-TCPP + US groups. (g) ATP leakage and (h) protein leakage in CTR and Bi-TCPP + US groups (n = 3). Significance was determined by one-way ANOVA with Tukey’s post-hoc test (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001)

These transcriptomic findings were consistent with the observed antibacterial effects of Bi-TCPP in S. aureus. Specifically, Bi-TCPP-mediated sonodynamic treatment significantly downregulated peroxidase-related genes (bsaA and chdC) and the catalase gene (katA), accompanied by increased ATP leakage and protein leakage (Fig. 5e-h).

In E. coli, Bi-TCPP sonodynamic treatment induced more substantial transcriptomic reprogramming, with 838 significantly upregulated and 444 downregulated genes (p < 0.05, |log₂FC| > 1) (Fig. 6a). Functional enrichment analysis revealed that these differentially expressed genes (DEGs) were significantly clustered in pivotal pathways including ABC transporters, quorum sensing, oxidative phosphorylation, sulfur metabolism, the TCA cycle, and glutathione metabolism (Fig. 6b). GO analysis further confirmed the primary involvement of DEGs in biological processes such as ABC transporter complex assembly, sulfur compound metabolism, and respiratory chain complex organization (Figure S26). To investigate pathway activation states, we performed Gene Set Enrichment Analysis (GSEA) and Gene Set Variation Analysis (GSVA). GSEA results indicated significant activation of ABC transporters (NES = 1.35, p = 0.02) and sulfur metabolism (NES = 1.63, p = 0.01), alongside strong suppression of oxidative phosphorylation (NES = -2.02, p < 0.001) in the treatment group (Fig. 6d, S27). GSVA scores corroborated this pathway activity pattern (Figure S28). At the transcriptional level, core genes in ABC transporter and sulfur metabolism pathways showed predominant upregulation (Fig. 6c, S29a). The core genes of the oxidative phosphorylation pathway showed divergent expression patterns between the two groups, suggesting a bidirectional regulatory role (Figure S29 b).

Fig. 6.

Transcriptomic profiling revealed the antibacterial mechanism of the sonodynamic material in E. coli. (a) Volcano plot of differentially expressed genes (DEGs) between sonodynamic material-treated group (B) and control group (E). (b) KEGG enrichment analysis of DEGs. (c) Heatmap of core gene expression in sulfur metabolism pathway. (d) Gene Set Enrichment Analysis (GSEA) plots for sulfur metabolism pathway. (e) Bacterial cellular leakage under control (CTR) and Bi-TCPP plus ultrasound treatment: ATP leakage, protein leakage and Intracellular glutathione (GSH) content (n = 3). Significance was determined by one-way ANOVA with Tukey’s post-hoc test (***p < 0.001). (f) Proposed molecular mechanism of Bi-TCPP-mediated sonodynamic therapy against E. coli. Data represent n = 3 biologically independent samples. Statistical thresholds: |log₂FC| > 1, p < 0.05 for differential analysis

Furthermore, Bi-TCPP suppressed the expression of multiple antioxidant enzyme genes in E. coli, including peroxidase-related genes (bcp, btuE, and tpx), the catalase gene (katG), and superoxide dismutase genes (sodA and sodB), as well as the glutathione biosynthesis gene (gshB) (Figure S30). These transcriptional changes were accompanied by decreased intracellular glutathione levels, increased ATP leakage, and elevated protein leakage, indicating compromised membrane integrity and disrupted cellular metabolism (Fig. 6e).

Based on these findings, we proposed a unified mechanistic model: Bi-TCPP generates reactive oxygen species (ROS) upon ultrasound activation, which subsequently triggered compensatory activation of bacterial sulfur metabolism. Concurrently, bismuth ions from the material effectively sequester hydrogen sulfide produced during metabolic processes, leading to disruption of downstream glutathione synthesis and depletion of cellular antioxidant capacity. These cascading events ultimately impaird bacterial oxidative phosphorylation function and exacerbate endogenous ROS accumulation. Concomitant cysteine restriction limits TCA cycle flux, further diminishing ATP production. Stress-induced hyperactivation of ABC transporters established a futile consumption cycle that dramatically accelerated ATP depletion. This multilayered metabolic stress ultimately caused irreversible bacterial death (Fig. 6f).

Although differences in metabolic pathway regulation were observed between S. aureus and E. coli (for instance, upregulation of TCA cycle-related genes in S. aureus may indicate a hypermetabolic state), both pathogens exhibited the exact core mechanism: sulfur metabolism disruption caused a sequential failure of oxidative phosphorylation and the antioxidant system, leading to energy collapse. This study provided solid mechanistic evidence for the broad-spectrum antibacterial properties of Bi-TCPP sonodynamic therapy, based on metabolic disruption.

Evaluation of antibacterial activity in vivo

The subcutaneous infection model, a well-established animal model, enabled direct in vivo observation of the anti-infection efficacy of biomaterials [31]. Following model establishment, daily ultrasound treatment was administered during the first three days post-infection. Titanium implants were harvested on days 7 and 12 for quantitative bacterial analysis. (Fig. 7a). On day 7, the BiT-US group (treated with Bi-TCPP plus ultrasound) exhibited a reduction in viable bacterial counts to 7.8 ± 2.4% of the level observed in the CTR group, with minimal residual bacteria remaining. This was significantly lower than both the BiT group (69.4 ± 6.7%) and the US group (60.7 ± 14.3%). By day 12, bacterial plating results demonstrated near-complete eradication of subcutaneous infection in the BiT-US group compared to controls. In contrast, the BiT and US groups showed partial reductions to 39.1 ± 5.6% and 27.9 ± 2.6%, respectively (Fig. 7b, c, d). Evaluation of abscess size revealed significant wound reduction in the BiT-US group versus CTR by day 7, with essentially complete healing observed by day 12 (Fig. 7e, f, g).

Fig. 7.

Antibacterial Efficacy of Bi-TCPP-Mediated Sonodynamic Therapy in a Murine Subcutaneous Infection Model In Vivo. (a) Schematic diagram of the experimental timeline for infection establishment and treatment administration. (b) Quantification of wound tissue bacterial burden, as determined by CFU counts from explanted implants via spread plate method at day 7 and day 12. (c) Bacterial load in wound tissues across treatment groups at day 7. (d) Bacterial load in wound tissues across treatment groups at day 12. (e) Representative wound photographs of CTR, BiT, US, and BiT-US groups on days 0, 1, 3, 7, and 12 post-treatments. Scale bar: 0.5 cm. (f) Schematic depiction of abscess size at day 7 and day 12. Scale bar: 0.25 cm. (g) Quantitative analysis of relative abscess/wound size at day 7 and day 12. (h) Representative H&E-stained sections of wound tissues from CTR, BiT, US, and BiT-US groups at day 12. Insets showed magnified views of the boxed regions. Scale bars: 600 μm (overview), 100 μm (magnified). (i) Representative Masson-stained sections from CTR, BiT, US, and BiT-US groups. Lower panels showed magnified views of the boxed regions. Scale bars: 600 μm (overview), 100 μm (magnified). (j) Immunohistochemical staining of TNF-α across CTR, BiT, US, and BiT-US groups. Scale bar: 125 μm. k) Immunohistochemical staining of IL-6 across CTR, BiT, US, and BiT-US groups. Scale bar: 125 μm. All quantitative data were expressed as mean ± SD (n = 3). Statistical significance was determined by one-way ANOVA with Tukey’s post-hoc test (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001)

Assessment of subcutaneous healing in vivo

The healing efficacy of subcutaneously implanted Bi-TCPP under ultrasound (US) therapy was evaluated by H&E and Masson’s staining. On day 12 post-treatment, the Ctrl group exhibited extensive scar tissue, minimal granulation tissue, and substantial inflammatory cell infiltration. In contrast, the Bi-TCPP (BiT) and US groups showed reduced but still considerable scar residue, limited granulation tissue, and lingering inflammation. Strikingly, the BiT-US group demonstrated robust healing, characterized by well-formed thick granulation tissue, nascent hair follicles, and an almost complete absence of scarring. Quantitative analysis of collagen deposition via Masson’s staining revealed a significantly higher coverage in the BiT-US group (42.0 ± 6.0%) compared to the BiT (18.9 ± 6.1%), US (24.5 ± 2.9%), and CTR (6.6 ± 3.5%) groups (Fig. 7h, i, S31a). Furthermore, immunohistochemical analysis showed a marked reduction in cells positive for the pro-inflammatory cytokines IL-6 and TNF-α in the Bit-US group (13.0 ± 3.1% and 12.4 ± 2.9%, respectively), relative to the other groups (Fig. 7j, k, S31 b-c). Collectively, these results indicated that Bi-TCPP nanoparticle implantation combined with ultrasound therapy significantly promotes subcutaneous tissue repair by mitigating inflammation, enhancing collagen deposition, and facilitating architectural restoration. Analysis of the body weight curves showed a slight decrease in all groups on day 3, after which the Bi-TCPP group recovered the fastest. Comparison of body weights among the groups revealed no significant differences, indicating low in vivo toxicity of the material, further supporting the safety of the intervention. Histopathological examination of major organs (heart, liver, spleen, lungs, and kidneys) further confirmed the biosafety profile of Bi-TCPP, with H&E-stained sections showing well-preserved tissue architecture and absence of pathological alterations across all treatment groups (Figure S32-33).

Furthermore, formed bismuth sulfide (Bi₂S₃) appeared as irregularly shaped aggregates under TEM, with its main XRD peak matching standard Bi₂S₃ (PDF#17–0320). A 50 µg dose was subcutaneously implanted in mice (n = 3), and during 24 days no abscess, infection, or allergic reaction was observed (Figure S34). By day 12 only a small amount remained, and by day 24 no material was detectable. Mouse body weight and HE-stained organs were normal, indicating that small amounts of Bi₂S₃ do not cause in vivo toxicity (Figure S35).

Conclusion

This study presented an innovative broad-spectrum antibacterial strategy based on a bismuth-porphyrin metal–organic framework (Bi-TCPP) for enhanced sonodynamic therapy. The material exhibited excellent antibacterial and anti-biofilm activity against both Gram-positive and Gram-negative pathogens through low-dose singlet oxygen generation, while obviously minimizing off-target tissue damage. Cross-species transcriptomic analyses revealed that ultrasound-activated Bi-TCPP induced conserved dysregulation of sulfur metabolic systems, leading to impaired energy metabolism, exacerbated oxidative stress, and ultimately bacterial cell death. This mechanism has been consistently validated in both S. aureus and E. coli, establishing Bi-TCPP as a multimodal sonosensitizer with broad-spectrum efficacy and offering a promising therapeutic alternative for complex surgical site infections, particularly those involving pathogens resistant to conventional antibiotics.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Fanshangmin Zhou contributed to methodology, formal analysis, investigation, conceptualization, visualization, and writing - original draft. Jiani Xu contributed to methodology, formal analysis, investigation, writing and conceptualization. Yun Du and Chang You contributed to methodology, formal analysis, and investigation. Chenyang Xu, Lei Shi, and Keyu Kong contributed to the investigation and visualization. Wenxiang Fu and Dachuan Sun performed formal analysis. Xinchen Zhu, Zhen Fang, and Huilin Wang conducted investigation. Shengbing Yang and Kai Huang performed proofreading. Nie Ping was involved in project administration, writing, review and editing. Wang Xin contributed to methodology, review and writing. Youzhuan Xie was responsible for conceptualization, project administration, resources, supervision, writing, review & editing, and funding acquisition.

Funding

This study was supported by the National Natural Science Foundation of China (82372420), National Key Research and Development Program of China (2018YFC2002303).

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

All the animal care and surgery were approved by the the Animal Ethics Committee of Shanghai Pudong Hospital (Approval No. KCZX20240928-002).

Competing interests

The authors declare no competing interests.