Decoy Nanozymes Enable Multitarget Blockade of Proinflammatory Cascades for the Treatment of Multi-Drug-Resistant Bacterial Sepsis

Sepsis is a life-threatening organ dysfunction characterized by severe systemic inflammatory response to infection. Effective treatment of bacterial sepsis remains a paramount clinical challenge, due to its astonishingly rapid progression and the prevalence of bacterial drug resistance. Here, we present a decoy nanozyme-enabled intervention strategy for multitarget blockade of proinflammatory cascades to treat multi-drug-resistant (MDR) bacterial sepsis. The decoy nanozymes (named MCeC@MΦ) consist mesoporous silica nanoparticle cores loaded with CeO2 nanocatalyst and Ce6 photosensitizer and biomimetic shells of macrophage membrane. By acting as macrophage decoys, MCeC@MΦ allow targeted photodynamic eradication of MDR bacteria and realize simultaneous endotoxin/proinflammatory cytokine neutralization. Meanwhile, MCeC@MΦ possess intriguing superoxide dismutase and catalase-like activities as well as hydroxyl radical antioxidant capacity and enable catalytic scavenging of multiple reactive oxygen species (ROS). These unique capabilities make MCeC@MΦ to collaboratively address the issues of bacterial infection, endotoxin/proinflammatory cytokine secretion, and ROS burst, fully cutting off the path of proinflammatory cascades to reverse the progression of bacterial sepsis. In vivo experiments demonstrate that MCeC@MΦ considerably attenuate systemic hyperinflammation and rapidly rescue organ damage within 1 day to confer higher survival rates (>75%) to mice with progressive MDR Escherichia coli bacteremia. The proposed decoy nanozyme-enabled multitarget collaborative intervention strategy offers a powerful modality for bacterial sepsis management and opens up possibilities for the treatment of cytokine storm in the COVID-19 pandemic and immune-mediated inflammation diseases.


Introduction
Sepsis is a life-threatening organ dysfunction caused by dysregulated host immune response to infection, which involves an initial overwhelming proinflammatory stage and subse-quent immunosuppressive phase [1,2]. Sepsis is reported to affect more than 49 million people every year and account for nearly 20% of all death globally and has been recognized as a worldwide health priority by the World Health Organization (WHO) [3,4]. Bacteria are the primary pathogens of sepsis, and current treatments extensively used in clinical practice include empirical administration of broadspectrum antibiotics, fluid resuscitation, and organ support [5,6]. These approaches can support end-organ functions and may help manage bacterial infection. However, they have a minor sepsis therapeutic effect especially in the context of the prevalence of bacterial drug resistance [7,8]. In addition to the standard care of sepsis, dozens of clinical trials of immunomodulators concerning various aspects of sepsis conditions have been undertaken in recent years [9,10]. Despite conspicuous advantages and progress of these well-designed immunomodulators, limited success has been met as evidenced by persistent sepsis mortality [11]. These dilemmas in sepsis management desperately call for the innovation of medical intervention strategy.
Pathologically, the initial proinflammatory stage of sepsis provides a relatively feasible intervention window, based on the helpless clinical fact that once sepsis enters the immunosuppressive phase, its progression is very tricky to reverse accompanied by a sharp increase in patient mortality [12]. The proinflammatory stage of sepsis is initiated by the massive invasion of pathogens, which introduce pathogenassociated molecular patterns (PAMPs) such as endotoxins into the host system [13]. The PAMPs can then be recognized by Toll-like receptors expressed on the extracellular surface of immune cells to activate host innate immunity, triggering the secretion of proinflammatory cytokines such as tumor necrosis factor α (TNF-α), interleukin-6 (IL-6), and interleukin-1β (IL-1β) [14]. Subsequently, these proinflammatory cytokines continuously recruit more inflammatory immune cells to the persistent infection sites, where as daunting "root" elicit excessive immune system activation and fuel systemic inflammation [15,16]. Concomitantly, immoderate production of proinflammatory cytokines and the burst of reactive oxygen species (ROS) occurred, resulting in cytokine storm and inevitably causing tissue damage and multiple organ dysfunction [17,18]. Researchers have attempted to intervene sepsis by targeting specific proinflammatory mediators, such as counteracting the burst of ROS with nanozymes [19][20][21]. Many nanozymes have been found to exhibit excellent antioxidant capacity, either by themselves or after surface modification [22][23][24][25]. They have shown significant ROS scavenging capacity in the treatment of proinflammatory sepsis [26,27]. In addition to this, photodynamic eradication of multi-drug-resistant (MDR) bacteria is also commonly used as a therapeutic option [28,29]. However, previous clinical trials have demonstrated that intervention means individually targeting certain proinflammatory mediators cannot inhibit the progression of the proinflammatory stage of sepsis in time to reduce the mortality rate [30,31]. Therefore, from the perspective of the full path of the occurrence and progression of the proinflammatory stage, we hypothesize that multitarget (pathogens, PAMPs, proinflammatory cytokines, and ROS) collaborative intervention may be more effective in tackling sepsis. However, an integrated intervention system with combined capabilities of pathogen eradication, PAMP neutralization, proinflammatory cytokine sequestration, and ROS scavenging remains elusive.
Herein, we present a multitarget collaborative intervention strategy for rescuing overwhelming inflammation to treat multi-drug-resistant (MDR) bacterial sepsis based on decoy nanozymes. The decoy nanozymes (termed MCeC@MΦ) are fabricated by incorporating mesoporous silica nanoparticles (MSN) with cerium oxide nanocatalyst (CeO 2 NC) and photosensitizer of chlorin e6 (Ce6) and then encapsulating it with macrophage membranes (MΦ) (as depicted in Figure 1(a)). The MCeC@MΦ possess an antigenic exterior the same as macrophages due to MΦ camouflage and inherit a remarkable ability to bind endotoxins and proinflammatory cytokines. By acting as macrophage decoys, MCeC@MΦ allow simultaneous endotoxin neutralization and proinflammatory cytokine sequestration, protecting host immune cells from overactivation and cutting off the chain of inflammatory actions. Meanwhile, the loaded CeO 2 NC bestows the MCeC@MΦ with intriguing superoxide dismutase-(SOD-) and catalase-(CAT-) like activities as well as hydroxyl radical antioxidant capacity (HORAC), enabling rapid scavenging of multiple ROS to relieve systemic oxidative stress and mitigate tissue damage. More importantly, intrinsic bacteria targeting of MΦ and the coexistence of Ce6 photosensitizer and CeO 2 NC scavenger empower MCeC@MΦ to realize targeted photodynamic therapy (PDT) of MDR bacteria without PDT-aggravated inflammation, digging out the "root" of sepsis inflammatory fluxes. By collaboratively addressing the issues of bacterial infection, endotoxin and proinflammatory cytokine secretion, and ROS burst to block proinflammatory cascades, the MCeC@MΦ-based "weeding and uprooting" intervention strategy may offer a powerful paradigm for clinical bacterial sepsis management (Figure 1(b)).

Preparation and Characterization of Decoy Nanozymes.
The decoy nanozymes of MCeC@MΦ were prepared by consecutively decorating MSN (size around 65 nm) with CeO 2 NC (size around 4 nm) and Ce6 molecule through a nucleophilic substitution reaction and chemical covalent coupling method [32,33], respectively, and then encapsulating it with MΦ (Figures 1(a) and 2(a)). The conjugation of CeO 2 NC to MSN was confirmed by the X-ray powder diffraction (XRD) pattern ( Figure S1) and Fourier transform infrared (FTIR) spectrum ( Figure S2) of CeO 2 NCdecorated MSN (MCe), as well as energy-dispersive X-ray (EDX) elemental mapping of MCeC@MΦ in which Si and Ce elements (the ratio of Si and Ce was around 15) were clearly observed (Figure 2(b) and Figure S3). UV-vis absorption spectra showed that Ce6-decorated MCe (MCeC) and Ce6-decorated MSN (MC) exhibited the characteristic absorption peaks of Ce6 at around 400 and 650 nm [34], indicating the successful linkage of Ce6 (Figure 2(c) and Figure S4). The covalent attachment of Ce6 could be confirmed by the negligible Ce6 release in MCeC during one-day storage ( Figure S5). The amount of Ce6 attached in MCeC was determined to be 10 μg/mL by calculating the concentration of free Ce6 in the reaction solution before and after covalent coupling based on the 2 Research standard absorption curve of Ce6 ( Figure S6). N 2 adsorption-desorption experiments showed that the Brunauer-Emmett-Teller (BET) surface area and median pore volume of MSN became significantly decreased after CeO 2 NC and Ce6 decoration (Figure 2(d) and Figure S7). From the transmission electron microscopy (TEM) images of the decoy nanozymes, the MCeC@MΦ were found to possess regular and well-defined core-shell structures with the average size of 71:2 ± 1:9 nm, in which the MCeC cores were wrapped with thin shells (red arrows), revealing the encapsulation of MΦ (Figure 2(a)). In addition, the whole preparation process of MCeC@MΦ could also be monitored and confirmed by the reversed zeta potential and increased hydrodynamic size results ( Figure S8 and S9). Through MΦ coating, the MCeC@MΦ are expected to inherit the same antigenic exterior of macrophages. To demonstrate this, sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was performed to analyze the protein composition profile of MCeC@MΦ. As shown in Figure 2(e), the overall protein profile of MCeC@MΦ was nearly identical to that of purified MΦ, while no protein signal was detected in MCeC, indicating the effective translocation of MΦ proteins onto the MCeC. Furthermore, the expression of typical cell membrane receptors including Toll-like receptor 4 (TLR-4), tumor necrosis factor receptor 1 (TNF-R1), interleukin-6 receptor α (IL-6Rα), and interleukin-1 receptor 1 (IL-1R1) [35][36][37] was clearly identified in MCeC@MΦ based on western blot analysis ( Figure 2(f)). Moreover, the prepared MCeC@MΦ were demonstrated to exhibit long-term size stability and outstanding in vitro biocompatibility based on the results of dynamic light scattering measurement and cell antiproliferation assay (Figures 2(g) and 2(h) and Figure S10). This excellent size stability and biocompatibility might be ascribed to the stabilizing effect of hydrophilic surface glycans on MΦ [38].

In Vitro Endotoxin Neutralization and Proinflammatory
Cytokine Sequestration. The robust expression of cell membrane receptors (TLR-4, TNF-R1, IL-1R1, and IL-6Rα) in MCeC@MΦ could empower the decoy nanozymes to bind endotoxins and proinflammatory cytokines [39], enabling endotoxin neutralization and proinflammatory cytokine sequestration. To demonstrate this, the separation of typical endotoxin and proinflammatory cytokine suspensions was carried out. Here, two kinds of decoy nanozymes that fuse with cell membranes derived from normal macrophages (MCeC@n-MΦ) and LPS-stimulated macrophages (MCeC@s-MΦ), respectively, were employed ( Figure S11). The endotoxin-and proinflammatory cytokine-binding      Research ability of decoy nanozymes was first assessed. As shown in Figure S12 and S13, the MCeC@n-MΦ and MCeC@s-MΦ possessed the superior binding ability for the selected endotoxin of lipopolysaccharide (LPS) and proinflammatory cytokines (TNF-α, IL-6, and IL-1β) compared to control group (PBS and MCeC). The difference in the binding amount of decoy nanozymes toward LPS (ng level) and proinflammatory cytokines of TNF-α, IL-6, and IL-1β (pg level) is closely related to the expression of receptors of TLR-4, TNF-R1, IL-6Rα, and IL-1R1 in the membrane of macrophages (Figure 2(f)). Subsequently, the performance of decoy nanozymes for endotoxin neutralization and proinflammatory cytokine sequestration was evaluated. As shown in Figure S14, the decoy nanozymes exhibited an obvious concentration-dependent neutralization/sequestration capability for endotoxin and proinflammatory cytokines, and remarkable neutralization/sequestration rates were achieved for the mixed solution of LPS, TNF-α, IL-6, and IL-1β. In addition, MCeC@s-MΦ were found to obtain relatively higher neutralization/sequestration rates than MCeC@n-MΦ, which is consistent with the fact of enhanced expression of cell membrane receptors in LPS-stimulated macrophages. Theoretically, the robust endotoxin neutralization and proinflammatory cytokine sequestration capability of decoy nanozymes can protect host immune cells from overactivation and ameliorate their inflammatory states. To demonstrate this, the secretion of proinflammatory cytokines (TNF-α, IL-6, and IL-1β) by J774 macrophage cells upon LPS stimulation in the absence/presence of decoy nanozymes was investigated. Here, two different experimental paths were carried out to better simulate the inflammatory conditions of macrophages in bacterial sepsis. Specifically, J774 macrophage cells were coincubated with LPS and decoy nanozymes in the path I to simulate the pathological response of early sepsis, while J774 macrophage cells were first stimulated by LPS for 12 h followed by decoy nanozyme treatment in the path II to evaluate their performance in progressive sepsis (Figure 3(a)). As shown in Figures 3(b) and 3(c), the secretion of proinflammatory cytokines of TNF-α, IL-6, and IL-1β was both significantly inhibited in the two experimental conditions (path I and II) by decoy nanozymes (MCeC@n-MΦ and MCeC@s-MΦ), indicating the mitigation of inflammatory states in LPS-stimulated J774 macrophage cells after decoy nanozyme treatment. In addition to the inhibition of proinflammatory cytokine secretion, decoy nanozymes were found to enhance the secretion of antiinflammatory cytokine of interleukin-10 (IL-10) and the expression of Arg-1 (M2 macrophage marker) in LPSstimulated J774 macrophage cells, showing a positive effect on the polarization of macrophage phenotype from M1 to M2 ( Figure S15 and S16) [40]. Compared to MCeC@n-MΦ, MCeC@s-MΦ exhibited relatively better performance for the inhibition of proinflammatory cytokine secretion as well as the enhancement of macrophage M1/M2 polarization (Figures 3(b) and 3(c) and Figure S16) and was chosen in the following experiments.
2.3. In Vitro ROS Scavenging. As an effective antioxidant, CeO 2 NC possesses intrinsic SOD-and CAT-like activities as well as HORAC due to the shift between Ce 3+ and Ce 4+ valence state [41,42]. In principle, MCeC@MΦ will inherit these unique enzymatic properties of CeO 2 NC and can catalyze the conversion of superoxide anion (•O 2 -), hydrogen peroxide (H 2 O 2 ), and hydroxyl radical (•OH) into water and oxygen, enabling effective scavenging of multiple ROS (Figure 4(a)). X-ray photoelectron spectroscopy (XPS) analysis identified the mixed Ce 3+ /Ce 4+ valence state in MCeC@MΦ, heralding the potential multienzyme characteristics of decoy nanozymes (Figure 4(b) and Figure S17). To demonstrate this, the SOD-and CAT-like activities as well as HORAC performance of MCeC@MΦ were systematically evaluated. As shown in Figure 4 Subsequently, MCeC@MΦ were incubated with J774 macrophage cells, and their cell uptake ability and the performance for intracellular ROS scavenging were evaluated. Figure 4(f) shows the confocal laser fluorescence microscopy (CLSM) images of J774 macrophage cells incubated with MCeC@MΦ at various time points. The fluorescence of MCeC@MΦ (coming from loaded Ce6) in J774 macrophage cells was clearly observed and continued to increase with the extension of incubation time, demonstrating wonderful macrophage uptake efficiency and paving the way for intracellular ROS scavenging. To perform intracellular ROS scavenging experiment, J774 macrophage cells with ROS hyperactivity state were treated by MCeC@MΦ, and dichlorofluorescein (DCF, a fluorescent marker generated from dichlorodihydrofluorescein dye by ROS) staining was carried out [43]. As shown in Figure 4(g), J774 macrophage cells in the treatment group of MCeC@MΦ showed much lower DCF fluorescence than the positive control (PBS), and their fluorescence intensities were further found to be indistinguishable from that of J774 cells without ROS hyperactivity state (negative control), indicating that the excessive intracellular ROS was effectively eliminated by MCeC@MΦ.   Research surface collapse as well as propidium iodide dye (red color, only penetrate microbes with destroyed structure) staining were clearly observed in MDR E. coli treated by MCeC@MΦ/Ir, suggesting a cell wall and membrane disruption-involved bactericidal mechanism. In addition to planktonic bacteria, the PDT performance of MCeC@MΦ toward biofilm (the predominant form of bacteria in vivo) was also assessed. Crystal violet staining assay demonstrated that MCeC@MΦ not only inhibited the formation of MDR E. coli biofilm but also effectively destroyed mature MDR E. coli biofilm under laser irradiation, exhibiting robust biofilm eradication capability (Figure 5(d) and Figure S19).
The residual ROS after photodynamic bacterial elimination is a nonnegligible risk issue, which inevitably cases damage to surrounding tissue and constitutes the main bottleneck of conventional PDT [44]. The coexistence of Ce6 photosensitizer and CeO 2 NC ROS scavenger in MCeC@MΦ offers a promising way to break this bottleneck by deracinating PDT-aggravated inflammation. To demonstrate the feasibility of MCeC@MΦ for the elimination of PDT-aggravated inflammation, the decay of ROS in MCeC@MΦ after irradiation was investigated using singlet oxygen sensor green (SOSG) [45]. As shown in Figure S20, ROS generated in the photodynamic process was continuously attenuated over time, and enhanced ROS decay was found in MCeC@MΦ compared to that of Ce6 and MC after irradiation, respectively, indicating the critical role of CeO 2 NC scavenger. Quantitatively, around 75% of ROS was scavenged at 10 min postirradiation in MCeC@MΦ, while the ROS decay rate in Ce6 and MC at 10 min postirradiation was just 32% and 30%, respectively, showing the outstanding ability for the elimination of PDT-aggravated inflammation ( Figure 5(e)). The excellent bactericidal performance and PDT-aggravated inflammation deracination talent of MCeC@MΦ provide a balanced antimicrobial PDT paradigm, laying a solid foundation for subsequent in vivo bacterial elimination.
2.5. In Vivo MDR Bacterial Sepsis Treatment. The multiple abilities (endotoxin neutralization and proinflammatory cytokine sequestration, ROS scavenging, and bacterial elimination) of decoy nanozymes make them an ideal candidate for in vivo   (Figure 6(a)). Four treatment groups were divided including PBS (positive control), MCeC, MCeC@MΦ, and MCeC@MΦ/Ir, and the group of normal healthy mice was used as the negative control. Figure 6(b) shows the survival profiles of septic mice treated 0.5 h after bacterial inoculation (early sepsis). Compared with the positive control (PBS) with a 100% mortality rate, the survival was greatly enhanced in the other three treatment groups, and an extremely high survival rate (80%) was obtained in the treatment group of MCeC@MΦ/Ir. To ascertain the mechanism behind the wonderful result of survival rate, the inflammation level of septic mice treated 0.5 h after bacterial inoculation in the four treatment groups was first quantitatively assessed. As shown in Figures 6(d)-6(f), the amount of typical proinflammatory factors (TNFα, IL-6, and IL-1β) in blood and peritoneal exudate of septic mice under the treatment of MCeC@MΦ/Ir was efficiently reduced to near-normal levels (negative control) on the 1 st day of treatment. Apart from proinflammatory factor measurement, protein permeability in peritoneal exudate that represents vasculature integrity and is associated with inflammation was also evaluated [46]. As shown in Figure 6(g), a much lower protein concentration in the peritoneal exudate of septic mice was found in the treatment group of MCeC@MΦ/Ir, indicating extremely weak protein permeability. The crossvalidating proinflammatory factor and protein permeability results demonstrated that the hyperinflammation of septic mice was rapidly and effectively ameliorated upon decoy nanozyme treatment. In addition, MDR E. coli bacteria collected from blood, peritoneal exudate, and major organs (kidney, spleen, and liver) of septic mice treated 0.5 h after bacterial inoculation were counted on the 1 st day of treatment to evaluate the actual antimicrobial efficacy. It was found that the number of bacterial colony was positively correlated with the survival of septic mice, and bacterial burden was remarkably reduced in the treatment group of MCeC@MΦ/Ir ( Figure S21). Alleviated proinflammatory factor secretion and reduced bacterial burden in turn circumvent organ damage of septic mice. Figure S22 and Figure S23 show the histological analysis (H&E staining) of organs of septic mice treated 0.5 h after bacterial inoculation under different treatment conditions. It was found that the damage of the liver and kidney of septic mice featuring severe inflammatory cell infiltration in perivascular area and spotty hepatocellular necrosis accompanied by lymphocytic infiltration was tremendously alleviated in the treatment group of MCeC@MΦ/Ir. These multidimensional experimental results described above strongly prove the feasibility of decoy nanozymes for in vivo treatment of MDR bacterial sepsis.
2.6. Biocompatibility Investigation of Decoy Nanozymes. The potential biotoxicity of nanomaterials is a key obstacle to its clinical transformation [47]. By considering the good biocompatibility of the components of MSN, CeO 2 NC, and Ce6 as well as the biomimetic coating of MΦ, the decoy nanozymes of MCeC@MΦ have predictable outstanding biosafety in vitro and in vivo. The human umbilical vein endothelial cells (HUVEC) incubated with MCeC@MΦ were found to possess high viability based on the MTT result shown in Figure 2(h) and Figure S10, confirming their wonderful biocompatibility in vitro. To assess the biosafety effect of decoy nanozymes in vivo, healthy mice were intraperitoneally injected with MCeC@MΦ, and the tissue distribution of Ce element and blood biochemical assay as well as organ histopathological analysis were performed. As shown in Figure S24, a relatively higher amount of Ce element was observed in the liver and spleen, indicating possible liver-and spleen-based metabolic pathway of decoy nanozymes. Figure S25 and S26 show the blood routine and blood biochemical results of healthy mice on the 5 th day of decoy nanozyme postinjection. It was found that there was no obvious difference in blood biochemical indicators detected between the mice injected with MCeC@MΦ, MCeC@MΦ/Ir, and PBS (control), indicating the negligible damage of decoy nanozymes to the metabolism of the liver and kidney of mice. In addition, no lesions and inflammation were found in the main organs of the mice injected with MCeC@MΦ compared to the control from the histopathological staining images, exhibiting outstanding in vivo biocompatibility ( Figure S27). This confirmed that excellent in vitro and in vivo biocompatibility of MCeC@MΦ decoy nanozymes lays a solid foundation for its future clinical transformation.

Discussion
From the perspective of cutting off the full path of the occurrence and progression of the proinflammatory phase induced jointly by pathogens, PAMPs, proinflammatory cytokines, and ROS, we propose a multitarget collaborative intervention strategy for rescuing bacterial sepsis based on decoy nanozymes. The decoy nanozymes are fabricated by wrapping CeO 2 nanocatalyst and Ce6 photosensitizer-loaded MSN cores with cell membranes derived from macrophages. Regarding the culprit of the pathogen, the decoy nanozymes allow targeted photodynamic eradication of bacteria without PDT-aggravated inflammation, digging out the "root" of sepsis inflammatory fluxes. In addition, by acting as macrophage decoys, the decoy nanozymes realize simultaneous endotoxin neutralization and 9 Research proinflammatory cytokine sequestration, protecting host immune cells from overactivation. Meanwhile, the decoy nanozymes enable rapid scavenging of multiple ROS via enzymatic reaction, relieving systemic oxidative stress and mitigating tissue damage. By collaboratively addressing the virulence factors of bacteria, endotoxins, proinflammatory cytokines, and ROS, the decoy nanozymes rapidly attenuate and reverse in vivo systemic hyperinflammation and ultimately confer considerably higher survival rates (>75%) to mice with progressive MDR E. coli bacteremia. The decoy nanozyme-based multitarget

Preparation of Decoy
Nanozymes. MSN was synthesized according to the previously reported method [32]. In brief, 2 g of CTAC and 0.02 g of TEA were dissolved in 20 mL of deionized water and stirred vigorously at 95°C for 1 h. Then, 1.5 mL of mesitylene and 1.5 ml of TEOS were added to the mixed solution. After 1 h reaction, the MSN precipitate was collected, repeatedly cleaned by the mixed solution of hydrochloric acid and ethanol to remove CTAC, and finally dispersed in ethanol. 200 μL of APTES was subsequently added into the obtained MSN solution and refluxed at 65°C for 4 h to prepare aminated MSN. CeO 2 NC was synthesized according to the previously reported method [50]. Briefly, 0.43 g of cerium acetate and 3.25 g of oleylamine were dissolved in 15 mL of xylene and reacted overnight at room temperature. Then, the solution was heated to 90°C in argon atmosphere, and 1 mL of deionized water was quickly injected and aged for 3 h. CeO 2 NC was collected through acetone precipitation and dispersed in chloroform. 10 mL of citric acid and BMPA-mixed DMF solution was subsequently added into the prepared CeO 2 NC solution and stirred vigorously overnight at room temperature to prepare carboxylated CeO 2 NC. 5 mL of aminated MSN and 5 mL of carboxylated CeO 2 NC were mixed and stirred overnight at room temperature to synthesize the MCe. 1 mg of Ce6 was then incubated with the obtained MCe solution in the presence of 1.5 mg of EDC and 1.6 mg of NHS. After overnight reaction at room temperature, the MCeC were prepared and finally dispersed in deionized water. J774 macrophages were cultured, and cell membranes were extracted using the method previously reported [37]. In brief, mouse J774 monocyte macrophages were cultured in DMEM high-sugar medium with/without LPS (50 ng/mL) for 48 h. Then, the cells were digested with 2 mM EDTA solution followed by centrifugation, and cell precipitate was dispersed in membrane protein buffer solution. After 15 min ice bath, the cell suspension was frozen in a liquid nitrogen tank for 5 min and then thawed at room temperature. Freezethawed cells were then centrifuged at 3200 g for 15 min to remove large cell debris, and the collected supernatant was   coli were cultured in lysogeny broth (LB) medium and harvested at the exponential growth phase before use. For antimicrobial experiments, 10 6 CFU of MDR bacteria was incubated with MCeC@MΦ (200 μg/mL of Ce element) under the condition of laser irradiation (660 nm, 0.8 W/ cm 2 , 5 min), and the antimicrobial performance was evaluated by the bacterial growth curve analysis, bacterial ROS determination, live/dead bacterial staining assay, and SEMbased bacterial morphology investigation (see the experimental details in Supporting Information).

4.11
. MDR Biofilm Eradication. The capability of MCeC@MΦ for MDR biofilm formation inhibition and destruction was investigated by crystal violet staining. For MDR biofilm formation inhibition, 10 6 CFU of MDR E. coli suspensions were mixed with MCeC@MΦ (200 μg/mL of Ce element) in 96-well plates and then irradiated by a 660 nm laser (0.8 W/cm 2 ) for 5 min. After 2-day incubation, the plates were gently washed by PBS, and crystal violet ethanol solution was added and reacted for 15 min. The plates after crystal violet staining were then imaged by camera, and the corresponding absorbance of staining solution at 590 nm was measured to indicate the extent of biofilm formation. For mature MDR biofilm destruction, 10 6 CFU of MDR E. coli suspensions were added into 96-well plates and grown 2 days to form integrated biofilm. Then, MCeC@MΦ (200 μg/mL of Ce element) was added onto the surface of mature MDR E. coli biofilm followed by laser irradiation (660 nm, 0.8 W/cm 2 , 5 min), and then, crystal violet ethanol solution was added and reacted for 15 min. Finally, the plates after crystal violet staining were imaged by camera, and the corresponding absorbance of staining solution at 590 nm was measured to indicate the extent of biofilm destruction.
4.12. ROS Decay after PDT. The decay of ROS in MCeC@MΦ after irradiation was investigated using SOSG. In brief, MCeC@MΦ (200 μg/mL of Ce element) was mixed with SOSG (50 μM) and irradiated (660 nm laser, 0.8 W/ cm 2 ) for 5 min. The fluorescence spectra of the solution were collected at different time points after laser irradiation (0, 5, and 10 min) to indicate the level of residual ROS after PDT. The decay of ROS in Ce6 after irradiation (660 nm laser, 0.8 W/cm 2 , 5 min) was used as the control.
4.13. Mouse Model of MDR Bacterial Sepsis. Female mice (BALB/c, 6 weeks) were purchased from Jinan Pengyue Biotechnology Co., Ltd., and allowed to adapt in the laboratory for one week before experiment. All animal experiments were carried out in compliance with the protocols approved by the Shandong University Laboratory Animal Center. In our experiments, the mouse model of MDR bacterial sepsis was established by inoculating lethal doses of MDR E. coli (2 × 10 8 CFU) through intraperitoneal (i.p.) injection to trigger aberrant inflammatory response.
4.14. In Vivo MDR Bacterial Sepsis Treatment. The mice with MDR bacterial sepsis were divided into four treatment groups including PBS (positive control), MCeC, MCeC@MΦ, and MCeC@MΦ plus irradiation (MCeC@MΦ/Ir), while the group of normal healthy mice was used as the negative control. Each treatment group contained 10 mice. To carry out in vivo bacterial sepsis treatment, septic mice were treated by intraperitoneally injecting materials (PBS, MCeC, MCeC@MΦ, or MCeC@MΦ/Ir) at two time points (0.5, and 12 h) after bacterial inoculation. At the 24 th h of bacterial inoculation, 4 mice in each treatment group were sacrificed, and the amount of typical proinflammatory factors (TNF-α, IL-6, and IL-1β) in blood and peritoneal exudate, protein permeability in peritoneal exudate, bacterial burden in blood, peritoneal exudate and major organs (kidney, spleen, and liver), and organ histological condition (H&E staining) were measured and analyzed. In addition, the survival of septic mice in different treatment groups within 120 h after bacterial inoculation was quantitatively analyzed to evaluate the actual therapeutic ability of MCeC@MΦ for in vivo MDR bacterial sepsis treatment.

Data Availability
The data that support the findings of this study are available within the article and its supplementary materials. Raw data are available from the corresponding authors on reasonable request.
conditions. Figure S22: histological analysis of the liver and kidney of septic mice under different treatment conditions. Figure S23: histological analysis of spleen of septic mice under different treatment conditions. Figure S24: the distribution of Ce element in the organs of healthy mice intraperitoneally injected with MCeC@MΦ at different time points. Figure S25: blood biochemistry data of healthy mice on the 5 th day of MCeC@MΦ postinjection. Figure S26: blood routine data of healthy mice on the 5 th day of MCeC@MΦ postinjection. Figure S27: H&E staining images of major organs of healthy mice on the 5 th day of MCeC@MΦ postinjection. (Supplementary Materials)