Synergistic chemodynamic-photothermal therapy of super-resistant Enterobacter asburiae-infected sepsis using multifunctional mesoporous nanocomposite

Abstract

Multidrug-resistant bacterial infections pose a severe threat to sepsis, where uncontrolled bacterial proliferation and the accompanying inflammatory response lead to multiorgan dysfunction. Recent clinical isolation of a super-resistant Enterobacter asburiae (E. asburiae) strain co-harboring mcr-10 and bla_NDM−1 underscores the urgent need for more effective antimicrobial strategies. Herein, we develop a synergistic strategy for the chemodynamic and photothermal therapy of super-resistant E. asburiae using a multifunctional mesoporous nanocomposite. This nanocomposite is composed of mesoporous silica nanoparticle loaded with polymyxin B, polydopamine, and palladium nanoparticles, possessing properties including photothermal therapy, chemodynamic therapy, and drug cargo. This synergistic approach achieves bactericidal effects through localized hyperthermia, enhanced generation of reactive oxygen species, and membrane-disrupting antibiotic action at a low dose of 42.1 µg mL⁻¹. In a murine peritonitis-sepsis model, treatment with this multifunctional mesoporous nanocomposite reduces the bacterial burden by 100%, decreases serum levels of inflammatory cytokines to normal levels, and improves the 3-day survival rate to 100%. These findings highlight a promising therapeutic strategy that leverages nanomaterial-mediated synergy to combat super-resistant pathogens.

Graphical Abstract

We report a multifunctional mesoporous nanocomposite that integrates photothermal therapy, chemodynamic therapy, and antibiotic delivery to synergistically eliminate super-resistant Enterobacter asburiae co-harboring mcr-10 and bla_NDM-1. This nanoplatform achieves complete bacterial eradication and inflammation suppression in a murine sepsis model, offering a potent strategy against super-resistant pathogens

Supplementary Information

The online version contains supplementary material available at 10.1186/s12951-025-03996-6.

Introduction

Sepsis is a life-threatening multiorgan dysfunction caused by an uncontrolled host immune response to infection [1, 2]. Despite optimal medical care, sepsis remains a great challenge worldwide and causes more than 8 million deaths annually [3]. Pathologically, sepsis is a deleterious and nonresolving systemic inflammatory response syndrome, mainly induced by bacterial infection, ultimately evolving to acute organ failure through irreversible damage to endothelial, epithelial, and immune cells [4, 5]. The cornerstone of treatment lies in promptly identifying the source of infection and then employing antibiotics. However, the widespread clinical use of antibiotics to treat bacterial infections has led to the emergence and proliferation of antibiotic resistance. Recently, we isolated a super-resistant Gram-negative Enterobacter asburiae (E. asburiae) strain co-harboring mcr-10 and bla_NDM−1 in the clinic, representing the first such strain in China [6]. It showed extensive antibiotic-resistant traits, including resistance to both colistin and carbapenem. Colistin is the last line of defense against multidrug-resistant (MDR) Gram-negative bacterial infections [7], which binds to lipopolysaccharide in the outer membrane of bacteria, and eventually leads to cell death by increasing membrane permeability [8]. The minimum inhibitory concentration (MIC) of colistin against this super-resistant E. asburiae was > 512 µg mL⁻¹, significantly higher than the sensitivity threshold (2 µg mL⁻¹) of polymyxins according to the Clinical and Laboratory Standards Institute standard [6]. The high dose of polymyxins used clinically will further lead to bacterial resistance, nephrotoxicity, and neurotoxicity [9, 10]. Limiting the usage of polymyxins not only reduces the occurrence of these severe side effects, but also delays the development of polymyxin-resistant strains, thereby prolonging their clinical efficacy. More importantly, apart from therapies with high-concentration antibiotics, there are currently no clinical approaches for sepsis infections caused by super-resistant E. asburiae [11]. Therefore, it is urgently needed to develop antimicrobial strategies to eradicate colistin-resistant MDR bacterial infections in the post-antibiotic era.

As representative chemical agents, antibiotics such as polymyxins are widely used against MDR Gram-negative pathogens [12, 13]. Meanwhile, chemodynamic therapy (CDT) based on reactive oxygen species (ROS) [14, 15] and photothermal therapy (PTT) driven by near-infrared (NIR) light [16, 17] provide complementary non-antibiotic bactericidal strategies. To achieve efficient antibacterial therapy, researchers have employed mesoporous silica nanoparticles (MSNs) as drug carriers due to their large surface area, tunable pore size, and excellent biocompatibility [18–21]. Their stimulus-responsive release behavior facilitates targeted delivery of antibiotics to infected tissues while reducing systemic exposure. However, monotherapy often fails to eradicate bacteria due to high MICs and poor antibiotic accumulation at infection sites [22–25]. To address this issue, nanotechnology-implemented synergistic strategies have shown great promise in antibacterial therapy, which can combine chemical and physical modalities to enhance bactericidal efficacy, minimize antibiotic dosage, and alleviate resistance development [26–28]. Polydopamine (PDA) exhibits strong NIR absorption, high photothermal conversion efficiency, and favorable biocompatibility, making it well-suited for mild-temperature PTT [29, 30]. Its abundant phenolic and amine groups allow for surface functionalization and nanoparticle anchoring [31]. Notably, noble metals such as palladium (Pd) can be anchored onto PDA through in situ galvanic replacement, producing nanozymes that exhibit peroxidase (POD)-like catalytic activity for CDT [32]. These Pd-decorated PDA structures can catalyze the decomposition of endogenous H₂O₂ into cytotoxic ROS, inducing oxidative stress in bacteria.

Herein, we develop a multifunctional antibacterial nanocomposite for the chemodynamic and photothermal therapy of super-resistant E. asburiae. The nanocomposite, composed of MSN loaded with polymyxin B, PDA, and Pd nanoparticles (PMB@MSN@PDA/Pd, termed PMP@Pd NPs) was prepared via a sol-gel method followed by simple stirring (Scheme 1A). Under NIR irradiation, PMP@Pd NPs exhibit a pronounced photothermal effect that increases the local temperature, thereby enhancing their POD-like activity and triggering the release of PMB with bactericidal activity (Scheme 1B). The photothermal performance of the PDA/Pd shells contributes to both the release of PMB and the generation of ROS, offering a synergistic antibacterial effect that maximizes the therapeutic potential of PMP@Pd NPs. The nanocomposite demonstrates strong antibacterial activity against clinically isolated super-resistant E. asburiae. It can disrupt bacterial membranes, induce leakage of intracellular components, and cause metabolic dysfunction through the synergistic effects of heat, ROS, and PMB. In a peritonitis-sepsis infection model, PMP@Pd NPs exhibit excellent therapeutic efficacy, underscoring their promise as a multifunctional synergistic antibacterial strategy for treating MDR bacterial infections (Scheme 1C). Recent CDT-PTT antibacterial platforms based on Ti₃C₂/CuWO₄/Pt, CuFeS₂ and MoS₂/CuO₂ have demonstrated promising synergistic effects, yet many still suffer from limited catalytic stability, suboptimal heat-catalysis coupling, or inadequate control over drug release [33–35]. In contrast, the PMP@Pd nanoplatform developed in this work integrates a highly stable Pd catalytic center with a PDA photothermal shell and a mesoporous silica drug reservoir. This hierarchical architecture enables efficient NIR-enhanced ROS generation, heat-responsive antibiotic release, and ROS- and heat-mediated membrane disruption. Such a tightly coupled mechanism provides a more robust and controllable CDT-PTT-antibiotic synergy, offering distinct advantages for combating super-resistant Enterobacter asburiae infections.

Fig. 1.

Schematic diagram of the synthesis and antibacterial application of PMP@Pd NPs. (A) Synthesis process of PM NPs, PMP NPs, and PMP@Pd NPs. (B) Schematic diagram of the enhanced POD-like activity and drug release of the PMP@Pd NPs nanoplatform under photothermal effects. (C) Schematic diagram of the antibacterial mechanism of PMP@Pd NPs. PMP@Pd NPs had an effective therapeutic effect in the infected sepsis model

Results and discussion

Preparation and characterization of PMP@Pd NPs

We first synthesized MSNs through a sol-gel method combined with a templating method to control their pore structure and size. Transmission electron microscope (TEM) images revealed the spherical shapes of MSNs (Fig. 2A). After PDA modification, a core/shell structure was formed with all mesopores sealed, indicating the formation of PMB@MSNs@PDA (PMP NPs) (Fig. 2B). Pd NPs were decorated on the surface of PMP NPs by an in situ reduction method (Fig. 2C). Quantitative TEM analysis revealed that the Pd NPs grown on the PDA layer possessed a uniform diameter of 12.2 ± 3.5 nm (Fig. S2A). Scanning electron microscope (SEM) images of MSNs, PMP NPs, and PMP@Pd NPs revealed homogeneous spherical structures with an average diameter of around 130 nm (Fig. S1A–C). Energy-dispersive X-ray spectroscopy (EDS) mapping confirmed that PMP@Pd NPs contained C, N, O, Si, and Pd elements (Fig. 2D). Furthermore, the zeta potentials of PMP NPs (−27.3 ± 0.4 mV) and PMP@Pd NPs (−13.3 ± 0.4 mV) were significantly different from those of MSNs (−22.0 ± 1.3 mV) (Fig. 2E). Meanwhile, the particle sizes of MSNs, PMP NPs, and PMP@Pd NPs determined by dynamic light scattering (DLS) were 131.7 ± 3.1 nm, 135.0 ± 3.9 nm, and 143.2 ± 2.3 nm, respectively. This increase in particle size further demonstrated the surface coating of PDA and Pd NPs on MSNs (Fig. 2H).

Fig. 2.

Characterization of PMP@Pd NPs. TEM images of (A) MSNs, (B) PMP NPs, and (C) PMP@Pd NPs. (D) Dark–field image and corresponding elemental mapping of PMP@Pd NPs. (E) Zeta potential of MSNs, PM NPs, PMP NPs, and PMP@Pd NPs. (F) FT-IR spectra of MSNs, PM NPs, and PMP@Pd NPs. (G) XPS spectra of MSNs, PM NPs, and PMP@Pd NPs. (H) Hydrodynamic diameters of MSNs, PM NPs, PMP NPs, and PMP@Pd NPs. (I) XRD spectra of MSNs, PM NPs, and PMP@Pd NPs. (J) Nitrogen adsorption–desorption isotherms of MSNs

We employed Fourier transform infrared (FT-IR) spectroscopy to elucidate the chemical structures (Fig. 2F). The characteristic bands of MSNs included Si–O–Si asymmetric stretching vibration (1085 cm⁻¹), Si–O–Si symmetric stretching vibration (800 cm⁻¹), and Si–OH stretching vibration (960 cm⁻¹) [36]. The characteristic peak at 1510 cm⁻¹ corresponding to the bending vibration of N − H was observed in the FT-IR spectra of PMP@Pd NPs, indicating the PDA coating. In addition, we employed X-ray photoelectron spectroscopy (XPS) to characterize the elemental compositions (Fig. 2G). MSNs were composed of Si and O, while PM NPs and PMP@Pd NPs were composed of C, O, N, Si, and Pd. These findings further suggested the incorporation of PMB and Pd NPs onto MSNs. Further examination of the Pd 3 d region revealed two distinct peaks at 342.88 eV and 337.58 eV, corresponding to the metallic Pd species in the 3d_3/2 and 3d_5/2 states, respectively (Fig. S2B). Additionally, X-ray diffraction (XRD) patterns confirmed the amorphous nature of the MSNs. (Fig. 2I). The XRD patterns of PM NPs and PMP@Pd NPs had similar crystal structures to MSNs although the reflection intensity was slightly attenuated, indicating that the crystal structure of MSNs remained after the loading of PMB, PDA, and Pd NPs. Besides the chemical composition, we employed Brunauer-Emmett-Teller (BET) analysis to show the high specific surface area of MSNs (374.9207 m² g⁻¹), which was beneficial for drug loading (Fig. 2J and S2C). To achieve drug loading with high efficiency, we optimized the MSNs/PMB ratio and chose a 1:1 weight ratio that yielded a drug loading efficiency of 42.1 ± 1.7% (Fig. S3A and S3B).

Photothermal performance of PMP@Pd NPs

We investigated the PTT performance of PMP@Pd NPs since the efficacy of PTT is crucial for the antibacterial performance (Fig. 3A). Under 808-nm NIR irradiation, PMP@Pd NPs exhibited exceptional photothermal capabilities, with their temperature rapidly rising to 54.0 °C within 360 s. This temperature significantly exceeded that of PMP NPs (45.1 °C) and PM NPs (29.7 °C), confirming that the enhanced photothermal effect originated from the synergistic contributions of the PDA coating and Pd NPs (Fig. 3B and C). The photothermal response of PMP@Pd NPs was time- and concentration-dependent. At a concentration of 150 µg mL⁻¹, the temperature increased from 26.4 °C to 58.1 °C within 360 s, whereas no significant temperature change was observed in the absence of PMP@Pd NPs (Fig. 3D). In addition, the photothermal conversion of PMP@Pd NPs was dependent on power intensity. As the power density increased from 0.5 to 2.0 W cm^− 2, the temperature rose from 34.5 °C to 65.5 °C, consistent with the thermal imaging data (Fig. 3E and S4A). PMP@Pd NPs displayed outstanding photothermal stability, retaining consistent performance over four consecutive irradiation cycles without significant efficiency loss (Fig. 3F). Quantitative analysis revealed a photothermal conversion efficiency (η) of 35.4% for PMP@Pd NPs, surpassing previously reported nanomaterials (Fig. S4B, S4C and Table S1). These findings underscore the potential of PMP@Pd NPs as a robust and reliable platform for photothermal antibacterial applications.

Fig. 3.

Photothermal and POD-like properties of PMP@Pd NPs. (A) Schematic illustration of photothermal-enhanced POD-like activity and drug release of PMP@Pd NPs. (B) Thermal images and (C) temperature curves of PM NPs, PMP NPs, and PMP@Pd NPs under NIR irradiation. (D) Temperature curves of different concentrations (0, 30, 50, 80, 100, and 150 µg mL⁻¹) of PMP@Pd NPs under NIR irradiation (808 nm, 1.5 W cm⁻²). (E) Temperature curves of PMP@Pd NPs under 808-nm NIR laser irradiation with various laser power densities (0.5, 1.0, 1.5, and 2.0 W cm⁻²). (F) Heating and cooling profiles of PMP@Pd NPs (100 µg mL⁻¹) over continuous on/off cycles of NIR irradiation (1.5 W cm⁻²). (G) POD-like property of MSNs, PM NPs, PMP NPs, and PMP@Pd NPs. (H) Michaelis-Menten curves of ROS generation velocity versus the concentration of TMB. (I) Michaelis-Menten curves of ROS generation velocity versus the concentration of H₂O₂ (pH 5.5). (J) EPR spectra of H₂O₂, PMP@Pd, PMP@Pd NPs/H₂O₂. (K) POD-like property of PMP@Pd NPs with or without NIR light irradiation. (L) Release of PMB in phosphate buffer (pH 5.5) with or without NIR irradiation

Catalytic activity and drug-release properties of PMP@Pd NPs

In addition to the excellent photothermal performance, we further studied the POD-like activity of PMP@Pd NPs (Fig. 3G). The POD-like property of PMP@Pd NPs was monitored using TMB and H₂O₂, given that the oxidation of TMB produced a blue-colored product with a characteristic absorbance at 652 nm. MSNs, PM NPs, and PMP NPs had no characteristic peak, indicating that the POD-like activity of PMP@Pd NPs originated from Pd NPs. To achieve optimal POD-like activity, PMP@Pd NPs were synthesized with an optimized ratio (Na₂PdCl₄: PMP NPs = 2:1) (Fig. S5). The optimal pH and temperature for POD-like activity of PMP@Pd NPs were approximately 5.5 and 45 °C, respectively (Fig. S6A-S6D). Meanwhile, the POD-like activity was enhanced with increasing concentration of PMP@Pd NPs (Fig. S7A). Based on the Michaelis-Menten curves (Fig. 3H), the production of ROS increased as the TMB concentration rose, and the Michaelis–Menten constant (K_m) and the maximal reaction velocity (V_max) of PMP@Pd NPs were calculated as 0.1909 mM and 2.790 × 10⁻⁸ M s⁻¹ using Lineweaver-Burk plots (Fig. S7B). For H₂O₂, the K_m was 1.4193 mM and the V_max was 1.58228 × 10⁻⁷ M s⁻¹ (Fig. 3I, S7C and Table S2). We employed electron paramagnetic resonance (EPR) with spin traps of 2,2,6,6-tetramethyl-4-piperidinone (TEMP) to verify the generation of ¹O₂ (Fig. 3J) [37, 38]. PMP@Pd NPs exhibited enhanced POD-like activity and generated ¹O₂ under 808-nm NIR light irradiation, indicating the potential of NIR-enhanced enzymatic activity (Fig. 3K). NIR light irradiation promoted catalytic efficiency through the photothermal effect, accelerating electron transfer and reaction kinetics at the active sites. Under an acidic environment, PMP@Pd NPs demonstrated sustained drug-release capability. NIR irradiation further enhanced the drug release, achieving superior controlled release efficacy through photothermal-responsive behavior (Fig. 3L). NIR irradiation increased the local temperature of PMP@Pd NPs, which weakened the noncovalent interactions between PDA and polymyxin B, induced thermal loosening of the PDA coating, and accelerated molecular diffusion inside the mesopores. These temperature-dependent effects collectively resulted in enhanced, on-demand drug release under NIR irradiation.

Antibacterial efficacy in vitro

The clinically isolated mcr-10 and bla_NDM−1 co-occurrence E. asburiae showed super resistance compared with the antibiograms of common strains (Fig. 4A). The colistin-resistant mcr-10 and bla_NDM−1 co-harboring E. asburiae exhibited resistance to PMB, with a MIC > 512 µg mL⁻¹. This strain exhibited extensive drug resistance because it harbored two resistance plasmids, p5549_mcr-10 and p5549_NDM-1, which carried the mcr-10 and bla_NDM−1 genes, respectively. The multidrug resistance phenotype of this strain was attributed to these plasmids together with multiple chromosomal resistance genes (qnrE1, oqxB9, oqxA10, and fosA). Given the inherent antimicrobial properties of H₂O₂, we investigated its MIC against mcr-10 and bla_NDM−1 co-occurrence E. asburiae (Fig. S8A and S8B). To assess the intrinsic photothermal effect, we examined bacterial viability under mild heating (45 °C). Only limited inactivation was observed (Fig. S8C and S8D), indicating that mild hyperthermia alone was ineffective in killing E. asburiae. We investigated the antibacterial activity of NIR-irradiated PMP@Pd NPs against mcr-10 and bla_NDM−1 co-occurrence E. asburiae by using colony counting assay and the live/dead cell assay. The mcr-10 and bla_NDM−1 co-occurrence E. asburiae treated with PMP@Pd NPs/H₂O₂/NIR irradiation exhibited negligible colony formation compared with that treated with PM or PMB (Fig. 4B). Neither free PMB nor PMB-loaded PM NPs showed significant antibacterial activity. In the absence of H₂O₂ and NIR irradiation, both MSNs@PDA/Pd (MP@Pd NPs) and PMP@Pd NPs exhibited limited antibacterial effects. In contrast, under H₂O₂ treatment without NIR irradiation, MP@Pd/H₂O₂ reduced bacterial viability to 45%, primarily attributed to the POD-like activity of Pd NPs. Furthermore, PMP@Pd/H₂O₂ demonstrated enhanced antibacterial performance, reducing bacterial viability to 37%, suggesting the antibacterial effect of PMB and ROS generation. Upon NIR irradiation, the antibacterial efficacy of MP@Pd/H₂O₂ was further enhanced, with bacterial viability decreasing to 28%, highlighting the antibacterial effect of ROS generation and PTT. Importantly, the treatment of PMP@Pd with H₂O₂ and NIR irradiation exhibited bacterial viability below the detection limit, underscoring the exceptional synergistic antibacterial effect of PMB, ROS generation, and PTT (Fig. 4F). We characterized the morphological changes of mcr-10 and bla_NDM−1 co-occurrence E. asburiae using SEM, which showed wrinkles and crumples on the cell membrane after treatment (Fig. 4C and D). The live/dead bacterial staining images further confirmed the bactericidal efficacy. Live E. asburiae labeled with calcein-AM displayed green fluorescence, while dead E. asburiae stained with propidium iodide (PI) showed red fluorescence after treatment, indicating the excellent synergistic antibacterial effect (Fig. 4E and G and S8E).

Fig. 4.

Antibacterial activity of PMP@Pd NPs. (A) Antibacterial spectrum of antibiotics including colistin, ampicillin, oxacillin, vancomycin, cefazolin, and ceftriaxone (µg mL⁻¹). (B) Photographs of bacterial colonies grown on agar plates after different treatments. (C and D) SEM images of bacteria before and after treatment. (E) Fluorescence images of bacteria after different treatments, where green and red represent live and dead bacteria, respectively. (F) Quantitative analysis of E. asburiae after different treatments. Data were shown as mean ± S.D. (n = 3, ***p < 0.001, ****p < 0.0001, ns, not significant). (G) Quantification of live/dead bacteria after different treatments. (PMB: 42.1 µg mL⁻¹, PM NPs: 100 µg mL⁻¹, MP@Pd NPs: 100 µg mL⁻¹, PMP@Pd NPs: 100 µg mL⁻¹, H₂O₂: 0.39 mM, and the incubation time was 6 h)

Exploration of the antibacterial mechanism

To understand the antibacterial mechanism of PMP@Pd NPs, we explored the mode of action using the mcr-10 and bla_NDM−1 co-occurrence E. asburiae (Fig. 5A). PMP@Pd NPs were attached to bacteria, which led to membrane disruption, given that PDA exhibited adhesive properties towards bacterial surfaces [29]. Due to the attachment of PMP@Pd NPs to the bacterial membrane, they could enhance the binding of PMB to bacteria. In colistin-resistant bacteria, the outer membrane lipopolysaccharides are modified with L-Ara4N or PEtN, and the lipid A is hydroxylated or palmitoylated. Such modifications are typical in colistin-resistant Enterobacter spp [39, 40]. We measured the uptake of PMB in colistin-resistant bacteria after PMP@Pd NP treatment using high-performance liquid chromatography mass spectrometry (HPLC-MS). The colistin-resistant bacteria exposed to PMP@Pd NPs absorbed significantly more PMB, approximately 6.4-fold compared with bacteria treated with PMB (Fig. 5B). The enhanced PMB uptake increased bacterial membrane permeability, as evidenced by a β-galactosidase-mediated o-nitrophenyl-β-D-galactopyranoside (ONPG) hydrolysis assay that released yellow o-nitrophenol (ONP) with an absorbance at 415 nm. The PMP@Pd/H₂O₂/NIR treatment resulted in a significant increase in membrane permeability of mcr-10 and bla_NDM−1 co-occurrence E. asburiae (Fig. 5C). Meanwhile, the POD-like activity of PMP@Pd synergistically contributed to bacterial membrane disruption through ROS generation and PTT [41].

Fig. 5.

The antibacterial mechanism of PMP@Pd NPs. (A) Schematic diagram of the antibacterial mechanism of PMP@Pd NPs. (B) Intracellular accumulation of PMB (C) membrane permeability, (D) extracellular K⁺ concentrations, (E) extracellular protein concentrations, (F) extracellular nucleic acid concentrations, (G) grayscale analysis of agarose gel electrophoresis of plasmids, (H) intracellular ROS levels, (I) intracellular MDA levels, and (J) intracellular ATP levels of mcr-10 and bla_NDM−1 co-occurrence E. asburiae after different treatments

Since potassium ions (K⁺), proteins, and nucleic acids are important intracellular components, their discharge serves as a hallmark of cell membrane deterioration. We explored changes of these intracellular contents within bacteria. After PMP@Pd/H₂O₂/NIR treatment, bacteria exhibited the highest leakage of K⁺, proteins, and nucleic acids (Fig. 5D-5F). Meanwhile, gel electrophoresis of plasmids extracted from different treatment groups indicated that the drug resistance genes had been destroyed (Fig. 5G and S9). Furthermore, PMP@Pd NPs induced significant biochemical alterations in intracellular metabolism, particularly through the accumulation of highly deleterious ROS. To assess intracellular ROS levels, we employed fluorescent 2’,7’-dichlorofluorescein diacetate (DCFH-DA) as a probe that showed increased fluorescence intensity (Fig. 5H). This elevation in ROS levels correlated with bacterial cell death, suggesting activation of oxidative damage-mediated pathways. To further elucidate the metabolic consequences of the PMP@Pd/H₂O₂/NIR treatment, we quantified lipid peroxidation levels using malondialdehyde-thiobarbituric acid (MDA-TBA) assays and adenosine 5’-triphosphate (ATP) concentrations. The PMP@Pd/H₂O₂/NIR treatment resulted in a significant increase in lipid peroxidation concurrent with a substantial depletion of intracellular ATP compared with control groups (Fig. 5I and J). Mechanistically, PDA-mediated photothermal heating (808 nm) accelerates Pd-catalyzed peroxidase-like reactions and boosts ROS generation, while ROS-induced lipid peroxidation together with mild hyperthermia increases bacterial membrane permeability and facilitates intracellular delivery of polymyxin B. Concurrently, NIR-driven heating loosens the PDA gatekeeper and accelerates drug diffusion from the mesopores, enabling on-demand release. These interdependent processes form a mutually reinforcing network that produces a genuine CDT-PTT-antibiotic synergy rather than a simple co-occurrence.

Transcriptome analysis of E. asburiae treated with PMP@Pd NPs

We conducted a transcriptomic analysis to investigate gene expression in E. asburiae after treatment with PMP@Pd NPs. Following treatment, RNA was extracted for transcriptomic sequencing. The total amount of RNA was the lowest after the combined treatment, consistent with the quantitative results of DNA damage (Fig. S10A and S10B). Correlation coefficients for both intragroup and intergroup samples were determined and visualized as heat maps. A large correlation coefficient indicated a strong correlation between two samples and a similar pattern in gene expression [42]. The correlation coefficients and principal component analysis showed high intragroup consistency and clear intergroup differences in gene expression between the control and PMP@Pd/H₂O₂/NIR groups (Fig. S11A and S11B). Differentially expressed genes (DEGs) were subsequently identified between the two groups.

To visualize the distribution and abundance differences of DEGs, we generated a heatmap and dendrogram through the Bray–Curtis index (Fig. 6A). The clustering analysis revealed that many DEGs and a large Bray–Curtis distance between combined treatment and the control group, indicating that the combined treatment significantly affected the transcription levels of E. asburiae and further effectively led to bacterial death. Venn diagram analysis revealed that 4085 genes were expressed between the control and the PMP@Pd/H₂O₂/NIR group, with 7 genes exclusive to the control and 108 genes exclusive to the PMP@Pd/H₂O₂/NIR group (Fig. 6B). We further compared the DEGs of E. asburiae between the combined treatment and the control group. A total of 2579 genes were expressed differentially (P value < = 0.05, |log₂FoldChange| >= 1.0), of which 1368 genes were upregulated and 1211 genes were down–regulated (Fig. 6C). We performed gene ontology (GO) analyses and Kyoto encyclopedia of genes and genomes (KEGG) enrichment analyses. GO annotations showed that DEGs were correlated with biological processes (e.g. cellular process, metabolic process, and single-organism process), cellular components (e.g. membrane, membrane part, cell, and cell part), and molecular functions (e.g. catalytic activity, binding and transporter activity) (Fig. 6D). KEGG enrichment analysis highlighted that upregulated genes were predominantly enriched in fatty acid degradation, sulfur metabolism, ABC transporters, two-component systems, and bacterial chemotaxis (Fig. 6E). In contrast, down-regulated genes were in pathways such as biosynthesis of secondary metabolites, carbon metabolism, biosynthesis of amino acids, nucleotide metabolism, and lipopolysaccharide biosynthesis (Fig. 6F).

Fig. 6.

Antibacterial mechanisms of PMP@Pd NPs revealed by transcriptomic analysis. (A) Heatmap of DEGs between the control and PMP@Pd/H₂O₂/NIR groups. (B) Venn diagram of DEGs between the control and PMP@Pd/H₂O₂/NIR groups. (C) Volcano plot of total DEGs between the control and PMP@Pd/H₂O₂/NIR groups. (D) GO enrichment analysis of DEGs between the control and PMP@Pd/H₂O₂/NIR groups. (E) Upregulated and (F) downregulated DEGs enriched in the KEGG pathways. Selected differentially expression genes involved in (G) membrane homeostasis, (H) ion transport/drug metabolism genes, (I) redox homeostasis, (J) lipid homeostasis, (K) ATP homeostasis, and (L) DNA biosynthesis and repair

The plasmid integrity is critical for bacterial adaptation, particularly in antibiotic resistance and metabolic flexibility [43]. Transcriptomic analysis revealed that treatment with PMP@Pd/H₂O₂/NIR altered the expression of genes related to plasmid stability and replication. Folate, as a key methyl donor, plays an essential role in nucleotide biosynthesis [44]. In this study, genes involved in folate metabolism (e.g. queF, metH, folD, and thyA) were significantly downregulated following PMP@Pd/H₂O₂/NIR treatment, suggesting folate deficiency in E. asburiae that might contribute to plasmid damage. Notably, key plasmid replication initiator repB was also downregulated, indicating potential inhibition of plasmid propagation. In contrast, stress response regulators such as lexA and recA were upregulated, implying activation of the Save Our Souls (SOS) response due to plasmid-related DNA damage. These results collectively indicated that the PMP@Pd/H₂O₂/NIR treatment disrupted plasmid integrity and replication, thereby impairing bacterial adaptability and resistance (Fig. S12).

Given the high sensitivity of the bacterial membrane to temperature, the photothermal effect of PMP@Pd NPs at elevated temperatures contributed to the eradication of E. asburiae. PMP@Pd/H₂O₂/NIR treatment resulted in intracellular accumulation of PMB in E. asburiae and thereby inhibited bacterial viability, accompanied by a downregulation of membrane homeostasis, ion transport, and drug metabolism genes (Fig. 6G and H). Notably, PMP@Pd/H₂O₂/NIR treatment generated ROS, leading to the disruption of lipid homeostasis and redox homeostasis within bacteria and the downregulation of related gene expression (Fig. 6I and J). Genes related to ATP metabolic biosynthetic processes, DNA biosynthesis, and DNA repair were significantly downregulated, consistent with the decreased ATP levels and plasmid disruption after the combined treatment (Fig. 6K and L). These results indicated that the antibacterial mechanism of PMP@Pd NPs relied on disrupting membrane homeostasis and drug metabolism, thus triggering biochemical processes including the intracellular ROS generation, plasmid damage, and ATP level reduction to kill E.asburiae.

Antibacterial therapeutic efficacy in a mouse peritonitis-sepsis model

Sepsis peritonitis is a disease with a high incidence rate. We investigated the antibacterial efficacy of PMP@Pd NPs in vivo in a mouse peritonitis-sepsis infection model. We first assessed the hemolysis of PMP@Pd NPs using red blood cells from mice. Even at a high concentration of 300 µg mL⁻¹, the hemolysis percentage remained at 1.9%, demonstrating excellent blood compatibility (Fig. S13A). We evaluated the cytotoxicity of PMP@Pd NPs using RAW 264.7 macrophage cells. The cell viability remained over 94.4% after incubation with 300 µg mL^{− 1} of PMP@Pd NPs over 48 h, indicating good biocompatibility at the cellular level (Fig. S13B). Mice were infected with clinically isolated mcr-10 and bla_NDM−1 co-occurrence E. asburiae via intraperitoneal injection. PMB, PMP@Pd NPs, and PMP@Pd NPs/NIR were administered intraperitoneally 30 min postinfection, with a saline-treated group serving as the control (Fig. 7A). The temperatures of the mice treated with saline, PMB, or PMP@Pd NPs were recorded using a thermal imaging camera both before and after NIR irradiation (Fig. 7B and S13C). A slight temperature change was observed in the health, control, and PMB groups, whereas a significant temperature increase was detected in the peritoneal cavity of the PMP@Pd NPs group after NIR irradiation, indicating the effective photothermal conversion in vivo. We maintained the intraperitoneal temperature at approximately 45 °C to ensure the effectiveness of mild photothermal therapy while simultaneously protecting abdominal tissues from thermal damage [45]. Following various treatments, the mice were monitored every 12 h for relevant parameters. Meanwhile, macroscopic abdominal photographs of mice showed severe bacterial infection (orange dotted circle) (Fig. 7C). The clinical scores of the control and PMB groups were significantly higher than those of the other groups at 12 h, reflecting a significant deterioration in physiological condition. Notably, in the PMP@Pd NPs/NIR group, the clinical scores reverted, indicating the restoration of normal physiological conditions (Fig. 7D) [46]. Meanwhile, in the control and PMB groups, more than 90% of the mice succumbed within 24 h. The mortality rate of mice treated with PMP@Pd NPs was approximately 83.3%, indicating limited therapeutic efficacy. In contrast, the PMP@Pd NPs/NIR treatment showed improved therapeutic efficacy against peritonitis-sepsis, as evidenced by 100% survival rate in infected mice, the same as the health group (Fig. 7E). Furthermore, the body weight of mice decreased after bacterial infection but returned to normal levels after PMP@Pd NPs/NIR treatment (Fig. S13D).

Fig. 7.

Therapeutic efficacy of PMP@Pd NPs in a mouse peritonitis-sepsis infection model induced by mcr-10 and bla_NDM−1 co-occurrence E. asburiae. (A) Schematic of establishment of the peritonitis-sepsis mouse model and treatment schedule. (B) Thermal images and (C) digital images of the macroscopic abdominal views of mice after different treatments. The orange dotted circle indicates bacteria-infected sites. (D) Clinical scores and (E) survival rate of infected mice after treatment. (F) Bacterial viability rate in blood, peritoneum, heart, liver, spleen, lung, and kidney after different treatments. (n = 6, ***p < 0.001, ****p < 0.0001, ns, not significant)

After treatment, fresh blood samples of mice were collected for blood routine examination including white blood cell (WBC), lymphocyte, granulocyte, monocyte and other common blood indicators, as well as liver function biomarkers such as alanine aminotransferase (ALT), aspartate aminotransferase (AST), and alkaline phosphatase (ALP). Renal function indicators including creatinine (CREA) and urea, and the cardiac function marker creatine kinase (CK) were also examined (Fig. S14 and S15). Unlike the control and PMB groups, the treatment groups of PMP@Pd NPs and PMP@Pd NPs/NIR displayed parameters consistently falling within the normal range, demonstrating the exceptional in vivo safety and biocompatibility of PMP@Pd NPs. To gain deeper insight into the impact of PMP@Pd NPs/NIR treatment on the immune system, we assessed the serological cytokine profiles of mice post-treatment. The control and PMB groups exhibited varying degrees of elevated cytokine levels, including tumor necrosis factor-α (TNF-α), interleukin-6 (IL-6), IL-1β, and IL-10. After PMP@Pd NPs/NIR treatment, a significant reduction in inflammatory cytokine secretion was observed (Fig. S16). These findings suggested that the combined treatment significantly mitigated the inflammatory cytokine storm in mice. To elucidate whether the improved survival rates and normalized blood biomarkers resulted from the antibacterial efficacy of PMP@Pd NPs/NIR therapy, we quantitatively analyzed bacterial colonization through colony-forming unit (CFU) enumeration in peritoneal fluid, blood, and major organs following therapeutic intervention. A comparative analysis of treatment modalities revealed that PMP@Pd NPs/NIR demonstrated superior bactericidal efficacy compared with both control groups and PMB monotherapy. This combined therapy significantly reduced bacterial loads in all examined sites, including peritoneal cavity, systemic circulation, and vital organs, particularly in cardiac, hepatic, splenic, renal, and pulmonary tissues (Fig. 7F and S17). The comprehensive bacterial eradication capacity of PMP@Pd NPs/NIR therapy effectively mitigated infection-induced multiorgan dysfunction, suggesting its potential for preventing sepsis-related systemic complications.

We further conducted histopathological analysis of major organs. In the control or PMB group, hematoxylin and eosin (H&E)–stained sections revealed severe inflammatory cell infiltration in heart, liver, spleen, lung, and kidney. H&E staining images showed typical multiple organ injuries, including interalveolar septum thickening in lung, tubular epithelial cell swelling in kidney, and widespread inflammatory cell infiltration in these organs. The PMP@Pd NPs/NIR treatment significantly alleviated organ damage (Fig. 8A and S18A). Immunohistochemistry analyses of these major organs further supported these findings. TNF-α production was significantly elevated in the control or PMB group, indicating an enhanced inflammatory infiltration within the tissue. In contrast, after PMP@Pd NPs/NIR treatment, tissues demonstrated a significant decrease in TNF-α levels accompanied by an elevation in IL-10 content, indicating a substantial reduction in inflammatory infiltration (Fig. 8B-E and S18B-S18E).

Fig. 8.

Histopathological analysis of major organs in a mouse peritonitis-sepsis infection model. (A) H&E staining and (B) inflammatory cytokine TNF-α staining and (C) anti-inflammatory cytokine IL-10 staining of heart, liver, and kidney after different treatments. Black arrows indicate inflammatory cell infiltration. Scale bars: 100 μm. Quantitative analyses of (D) TNF-α and (E) IL-10 levels in heart, liver, and kidney of mice. (n = 3, ***p < 0.001, ****p < 0.0001, ns, not significant)

Long-term biocompatibility and biodegradation behavior of the PMP@Pd nanocomposite

To determine whether PMP@Pd could be safely utilized for the clinical treatment of super-resistant E. asburiae, we investigated its long-term in vivo toxicity through intraperitoneal injection in mice. Toxicity-related indicators and organ morphology were examined on Day 14. Whole-blood samples were collected to assess common hematological parameters, including WBC, lymphocytes, granulocytes, and monocytes. No significant differences were observed among the four groups (Fig. S19). Serum biochemical analysis was further performed to evaluate renal function (UREA, CREA) and liver function (ALT, AST). Compared with the PBS control, the values of UREA, CREA, ALT, and AST in the PMP@Pd/NIR group remained almost unchanged and were within normal physiological ranges (Fig. S20A-D). We performed H&E staining of major organs including heart, liver, spleen, lungs, and kidneys to assess potential histopathological effects. No signs of inflammation, fibrosis, or necrosis were detected in the PMB, PMP@Pd, or PMP@Pd/NIR groups relative to the PBS group, suggesting long-term biosafety and minimal systemic toxicity or adverse side effects (Fig. S20E).

Elemental analysis of major organs revealed that the nanocomposite accumulated predominantly in kidneys, liver, and lungs within the first 24 h post-administration. As time progressed, Pd levels in these tissues declined steadily. By Day 14, Pd concentrations had nearly returned to baseline, indicating efficient clearance from the body (Fig. S20F). These findings demonstrate that the PMP@Pd nanocomposite underwent effective biodegradation in vivo.

Conclusions

In summary, we developed a multifunctional PMP@Pd NPs integrating MSNs, PMB, and PDA/Pd shells for the synergistic chemodynamic and photothermal therapy of super-resistant E. asburiae. This design leveraged the high drug-loading capacity of MSNs, the photothermal conversion efficiency of the PDA/Pd shell, and the POD-like activity of Pd NPs to address the limitations of single-mode antibacterial therapies. Under NIR irradiation, PMP@Pd NPs exhibited a synergistic antibacterial effect by combining PTT, CDT, and controlled PMB release. The nanocomposite demonstrated potent activity against clinically isolated E. asburiae, a super-resistant strain co-harboring mcr-10 and bla_NDM−1, by disrupting bacterial membranes, inducing intracellular component leakage, and causing metabolic dysfunction. In a murine peritonitis-sepsis model, intraperitoneal administration of PMP@Pd NPs achieved a 100% survival rate without observed toxicity, highlighting its efficacy in treating deep-seated infections. The eradication of super-resistant E. asburiae by PMP@Pd NPs demonstrates the significance of integrating catalytic and photothermal effects with targeted antibiotic delivery. This strategy offers not only a timely response to super resistance but also a design blueprint for constructing programmable and multi-modal antibacterial therapies. Future efforts toward infection-specific targeting, immune modulation, and adaptable nanoformulations may further expand the therapeutic landscape against evolving bacterial threats.

Materials and methods

Cytotoxicity evaluation by a CCK-8 assay

For the cytotoxicity assay, RAW 264.7 macrophage cells were used and cultured in standard DMEM medium. The cells (5000 cells per well) were seeded into a 96-well plate and incubated overnight. They were then incubated with different concentrations of PMP@Pd (0, 30, 50, 80, 100, 150, 200, and 300 µg/mL) for 24 h. Afterwards, 10 µL of CCK-8 reagent was added to each well and incubated for an additional 2 h. The absorbance at 450 nm was measured using a microplate reader [47].

Hemolysis assay

The hemolysis assay was performed according to literature reports [48]. The red blood cells (RBCs) from mice were collected through centrifugation at 3000 rpm for 15 min, washed with PBS three times, and formulated into 5% erythrocyte solution. Different concentrations (0, 30, 50, 80, 100, 150, 200, and 300 µg/mL) of PMP@Pd NPs were mixed with RBCs at 37 ℃ for 2 h (n = 3). As negative and positive controls, RBCs were respectively incubated with PBS and deionized water. After incubation, the supernatant was obtained by centrifugation and measured at 450 nm with a microplate reader. The hemolysis ratio was calculated by the following formula: Hemolysis (%) = (A_sample − A_{negative control})/(A_{positive control} − A_{negative control}) × 100%, where “A” represents the average value at 450 nm.

Antibacterial efficacy of PMP@Pd NPs in vivo

Animal experiments were carried out with the approval of the Laboratory Animal Center of Hangzhou Medical College (Approval No. ZJCLA-IACUC-20011022). Male BALB/c mice (6–8 weeks old) were used to establish an infection model and evaluate the in vivo antibacterial efficacy of PMP@Pd NPs. Freshly isolated monoclonal E. asburiae strains co-harboring mcr-10 and blaNDM-1 (MIC > 512 µg/mL) grown on LB agar were cultured overnight in LB broth at 37 °C. The bacterial pellet was collected by centrifugation at 6000 rpm for 6 min and resuspended in physiological saline to reach an OD₆₀₀ of 0.5. Each mouse was intraperitoneally injected with 100 µL of the diluted bacterial suspension.

Survival curve and clinical scores

The infected mice were randomly divided into five groups (n = 6 per group) and treated with different drug formulations: healthy group, physiological saline (control), PMB alone, PMP@Pd NPs alone, and PMP@Pd NPs + NIR. Intraperitoneal drug administration was performed 30 min after bacterial infection. The health status of the mice was monitored for 72 h, and survival rates were recorded for survival curve analysis.

Mice were scored every 12 h using the following criteria: score 0, no symptoms; score 1, piloerection and huddling; score 2, piloerection, huddling, and diarrhea; score 3, lack of interest in surroundings and severe diarrhea; score 4, decreased movement and listlessness; and score 5, loss of self-righting reflex. When mice reached score 5, they were humanely euthanized.

Bacterial reduction in mouse organs

After the same drug treatment as described in the previous paragraph, blood samples were collected via orbital bleeding at 24 h post-infection, and the mice were sacrificed and dissected to collect the peritoneal lavages, heart, liver, spleen, lungs, and kidneys. The peritoneal lavages, blood, and harvested organs were homogenized and serially diluted with sterile saline. Aliquots of 100 µL from the diluted homogenates were plated onto LB agar and incubated at 37 °C overnight, and colonies were counted to determine the bacterial loads. Each experiment was performed in triplicate.

Histological and immunohistochemical evaluation

Histopathological evaluation was performed to assess sepsis-associated inflammation in mice. The mice were sacrificed at 24–72 h post-infection, and blood samples were collected for biochemical analysis, including alanine aminotransferase (ALT), aspartate aminotransferase (AST), creatine kinase (CK), alkaline phosphatase (ALP), creatinine (CREA), urea, and routine hematological analysis. The heart, liver, spleen, lungs, and kidneys were collected from mice at the same time points and immediately fixed in 4% paraformaldehyde. The specimens were embedded in paraffin and sectioned at a thickness of 4 μm for hematoxylin–eosin (H&E) staining and immunohistochemical staining of TNF-α and IL-10. Images were captured using a microscope (ECLIPSE E100, Nikon, Japan).

In vivo long-term biocompatibility and biodegradation of the PMP@Pd nanocomposite

To evaluate the in vivo biocompatibility of the PMP@Pd nanocomposite, BALB/c mice (6–8 weeks) were randomly assigned to four groups (n = 6) and administered PMB, PMP@Pd, PMP@Pd/NIR, or PBS through intraperitoneal injection. After 14 days, blood samples were collected for routine hematological examination and serum biochemical analysis, including hepatic function (ALT and AST) and renal function (UREA and CREA). Major organs including the heart, liver, spleen, lungs, and kidneys were harvested and examined by H&E staining to evaluate potential histopathological alterations.

To characterize the biodegradation profile of the nanocomposite in vivo, BALB/c mice received daily intraperitoneal injections of PMP@Pd for 14 days. Following the final dose, mice were sacrificed at predetermined time points (12 h, 24 h, Day 3, Day 7, Day 9, and Day 14) (n = 3). The heart, liver, spleen, lungs, and kidneys were collected, weighed, and digested in concentrated nitric acid using microwave digestion. The digests were subsequently heated at 110 °C for 2 h to remove residual nitric acid until the solutions became clear, and then diluted to 10 mL with ultrapure water. Inductively coupled plasma mass spectrometry (ICP-MS) was employed to quantify Pd concentrations in each organ, enabling the determination of the biodistribution and clearance kinetics of the PMP@Pd nanocomposite.

Statistical analysis

Data for all bar charts were plotted via mean ± standard deviation of the mean (SD). Statistical analyses were performed using Prism 8.4 (GraphPad Software), Origin Pro 2023 (OriginLab), and SPSS 20.0 (Statistical Package for the Social Sciences). One-way analysis of variance (ANOVA) and t test were used to determine the significance of the indicated comparisons (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001).

Author contributions

Anlai Zou: Writing-original draft, Software, Methodology, Investigation. Xiaoxue Zhu: Methodology, Investigation. Feng Zhao: Software, Methodology. Jinjie Hou: Methodology, Investigation. Mengliu Zhao: Software, Investigation. Ting Yu: Software, Investigation. Junzheng Zhang: Software, Investigation. Ying Fu: Supervision, Investigation. Zhi Ruan: Supervision, Conceptualization. Yunlei Xianyu: Writing-review & editing, Supervision, Investigation, Funding acquisition, Conceptualization. Jun Zhang : Writing -review & editing, Supervision, Funding acquisition, Conceptualization.

Funding

This study was supported by National Key Research and Development Program of China (2022YFC2603800) and National Natural Science Foundation of China (22104128).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.