Lipid droplet–enriched luminogens enable adoptive macrophage transfer for treatment of bacterial sepsis

Abstract

Bacterial sepsis, a life-threatening systemic inflammatory response to infection affecting over 30 million people annually, is exacerbated by antibiotic resistance and immune suppression. Here, we report a small luminescent molecule, TPA2PyPh, as a potent antibacterial agent and its potential for lipid droplet–engineered macrophage transfer strategy in treating bacterial sepsis. Engineered macrophages, created by directly incubating TPA2PyPh with macrophages, enabled the luminogen to precisely bind to intracellular lipid droplets. Upon engulfment of bacteria, these TPA2PyPh-loaded macrophages use natural lipid uptake mechanisms to deliver the luminogen to intracellular bacteria, disrupting their membranes and inserting into the bacterial DNA, thereby inducing bacterial elimination. Our findings show that the adoptive transfer of TPA2PyPh-loaded macrophages substantially diminishes bacterial burden in septic mice and substantially reduces mortality rates. This study demonstrates the potential of TPA2PyPh as an effective antibacterial agent and supports the use of adoptive lipid droplet–engineered macrophage transfer as an effective approach for treating sepsis and managing severe infectious diseases.

TPA2PyPh enables lipid droplet–engineered macrophage transfer, effectively reducing bacterial burden and mortality in septic mice.

INTRODUCTION

Bacterial sepsis represents a life-threatening clinical and biochemical syndrome characterized by acute organ dysfunction due to an aberrant response of the body to microbial invasion (1–3). As a principal cause of death worldwide, it accounts for approximately 48.9 million new cases and 11 million fatalities annually, representing a substantial challenge for international health care systems. Clinically, antibiotic therapy remains the cornerstone of sepsis treatment according to clinical guidelines (4–6). However, the growing prevalence of antibiotic resistance makes selecting effective antibiotics increasingly difficult, creating substantial barriers to the successful treatment of sepsis (7, 8). Clinical observations indicate that although over 60% of patients with sepsis initially survive the acute inflammatory phase, they rapidly progress to a state of extended immunosuppression, which is characterized by dysfunction and apoptosis of immune cells, leading to a paralysis of immune functions that diminishes the capacity to eliminate invading pathogens and further secondary bacterial infections (3, 5, 9). Consequently, the ongoing development of an alternative therapeutic approach for bacterial sepsis is receiving increasing attention, which should eliminate invasive pathogens through a nonantibiotic strategy while simultaneously rebuilding the immune system to address postsepsis immunosuppression, thereby achieving a comprehensive cure for bacterial sepsis.

Over the past decades, extensive efforts have been devoted to identifying potential therapeutic targets to alleviate immune dysregulation due to infection, such as mitigating reactive oxygen species (ROS) such as H₂O₂, O₂^·−, and ^·OH (10, 11), as well as targeting anaphylatoxin C5a (12, 13). However, these approaches face challenges such as inconsistent efficacy, high costs, and complex administration. Recently, adoptive cell transfer (ACT) has emerged as an effective treatment for immunosuppressed patients suffering from severe bacterial sepsis (14–17). This approach involves selecting and genetically modifying immune cells from either the patient or a donor to enhance their infection-fighting capabilities, which enables direct elimination of pathogens, modulation of inflammatory cytokine production to prevent tissue damage, and restoration of immune system balance to counteract the immunosuppression commonly associated with septic episodes. Currently, the most commonly used cells in ACT for sepsis are engineered T cells (17), macrophages (14), natural killer cells (15), and dendritic cells (18). Among these, macrophages are particularly valued for their superior bacterial phagocytic abilities and their capability to activate both innate and adaptive immune systems, making them excellent antibacterial therapeutic cells (19). Several adoptive engineered macrophage strategies have been developed, primarily through nanotechnology to reinforce the lysosome-related antibacterial mechanisms (20), thereby achieving effective bacterial clearance after cell infusion (14). However, there are shortcomings in the current lysosomal-engineered ACT strategies. Many bacteria, including Staphylococcus aureus and Escherichia coli, have evolved immune evasion strategies to escape phagolysosomal killing, leading to intracellular survival and recurrent infections (21–24). Therefore, there is an urgent need to develop innovative antimicrobial agents for overcoming immune escape and improving therapeutic effect of ACT for sepsis.

Recent studies have highlighted the role of lipid droplets (LDs) as innate immune hubs that integrate cellular metabolism and host defense in eukaryotic cells (25). LDs store essential lipids necessary for producing signaling molecules, membrane building blocks, and metabolic energy. In addition, LDs can serve as nutrient sources that attract intracellular pathogens, including parasites, bacteria, and viruses (26, 27), which highlights the potential of LDs in innate immunity beyond their metabolic functions. On the basis of these findings, we proposed that LDs can be used as a promising target for delivering therapeutic agents during ACT, leveraging their role in immune function and pathogen interaction. To achieve these LD-mediated antibacterial functions, an antimicrobial agent with excellent LD targeting effect and potent bactericidal capabilities is desired, which could facilitate the efficient binding of antimicrobials to LDs and the corresponding capture and killing of bacteria. The current design paradigm of LD-targeting molecules typically involves creating compounds with substantial lipophilicity, which allows them to effectively integrate into the lipid-rich environment (28, 29). This strategy has led to the development of several LD-targeted fluorescent probes, such as BODIPY derivatives and Oil Red O, which have been widely used for tracking intracellular LD (28–31). Therefore, if these fluorescent probes could be further functionalized with specific groups, then they could be used to achieve simultaneous targeting and destruction of pathogens associated with LDs.

In this study, we designed and synthesized a luminescent compound TPA2PyPh, which features a biphenyl unit with high hydrophobicity and rigidity, facilitating efficient penetration of the cell membrane and subsequent loading into macrophage LDs. In addition, TPA2PyPh has a robust bactericidal activity that can efficiently insert into the grooves of double-stranded DNA (dsDNA) and subsequently induce bacterial DNA disruption. After macrophages were labeled with TPA2PyPh, the TPA2PyPh-loaded macrophages (TPP-macrophages) were injected into mice with multidrug-resistant bacteria-induced sepsis. TPP-macrophages can actively capture and phagocytose invading bacteria and then deliver LD-loaded TPA2PyPh to intracellular bacteria via natural lipid uptake mechanisms, which further disturb bacterial membrane integrity and induce DNA disorder to kill bacteria. Our in vivo results demonstrated that this adoptive TPP-macrophage strategy could effectively eliminate bacteria from the peritoneum and blood in mice. Moreover, it can also alleviate the immunosuppressive microenvironment within the body, effectively rescuing mice from bacteria-induced sepsis. This LD-engineering antibacterial strategy reinforces the host defense mechanism for efficient removal of multidrug-resistant bacteria in sepsis.

RESULTS

The design and characterization of TPA2PyPh

TPA2PyPh is a fluorescent molecule specifically designed for simultaneous lipid and bacterial DNA binding. The structure and synthetic route of TPA2PyPh are depicted in Fig. 1 and 2A, and the characterizations using nuclear magnetic resonance and mass spectrometry are detailed in the Supplementary Materials (figs. S1 to S7). TPA2PyPh features a hydrophilic terminus comprising two positively charged pyridinium salt moieties (Fig. 1B). The arrangement of the two positively charged pyridinium salts flanking the hydrophilic terminus enhances the molecular affinity for DNA through electrostatic interactions, which enables the compound to easily insert into DNA grooves (Figs. 1C and 2B and fig. S8). Meanwhile, the hydrophilic terminus significantly improves solubility in aqueous environment, enabling TPA2PyPh to have low background signal and offer turn-on response. At the opposite terminus of TPA2PyPh, biphenyl structures were strategically introduced to provide hydrophobic and rigid characteristics, which endows TPA2PyPh with the capability to accumulate in hydrophobic biomolecules such as phospholipids and liposomes. Meanwhile, the structural rigidity of the biphenyl moiety facilitates efficient insertion and transmembrane crossing of TPA2PyPh into cell and bacterial membranes. In addition, the electron-withdrawing effect of the pyridinium salt moiety induces an effective donor-acceptor interaction within TPA2PyPh, which leads to strong fluorescent signals when irradiated by light, providing excellent molecular imaging capabilities.

Fig. 1. Chemical structure and the experimental approach.

(A) Schematic representation of the experimental approach for adaptive macrophage transfer. (B) Chemical structure of TPA2PyPh. (C) Three-dimensional (3D) structure and description of functional unit. Created in BioRender, X. Liu (2025); https://BioRender.com/f38j855.

Fig. 2. Synthetic route and photophysical properties.

(A) Synthetic route of TPA2PyPh. RT, room temperature. Tol, toluene; NBS, N-Bromosuccinimide; DCM, dichloromethane; Mel, methyl iodide; ACN, acetonitrile (B) Molecular docking model of TPA2PyPh with a bacterial DNA fragment (Protein Data Bank code: 4U8C). (C) Normalized UV-vis absorption and photoluminescence (PL) spectra of TPA2PyPh in an aqueous [containing 0.1% dimethyl sulfoxide (DMSO)] solution with a concentration of 10 μM. (D and E) PL spectra (D) and corresponding column graph (E) of TPA2PyPh or incubation with Raw 264.7 and MRSA USA300 in a PBS solution (1×, pH 7.4; containing 0.2% DMSO) with a concentration of 2 μM. a.u., arbitrary units. (F to K) The PL spectra and line graph of TPA2PyPh (10 μM) upon addition of dsDNA, oleic acid (OA), and dioleoylphosphatidylcholine (DOPC) in a PBS solution (1×, pH 7.4; containing 0.1% DMSO).

The photophysical properties of TPA2PyPh were subsequently investigated. In phosphate-buffered saline (PBS) solution (1×, pH 7.4), TPA2PyPh exhibited good solubility and had a broad absorption spectrum from 350 to 500 nm, peaking at 421 nm (Fig. 2C). The compound displayed an emission maximum at ~612 nm with a large Stokes shift of 191 nm (Fig. 2C). As shown in Fig. 2 (D and E), the fluorescence intensity of the suspensions significantly increased following the addition of TPA2PyPh to Raw 264.7 cells or S. aureus American Type Culture Collection (ATCC) 25923 compared to TPA2PyPh alone (>20-fold for Raw 264.7; >90-fold for S. aureus ATCC 25923), indicating good cell and bacteria tracking capability. To explore the mechanism behind this fluorescence enhancement, we selected several biological substances for further investigation, including dsDNA, the genetic material abundant in the nucleus of mammalian cells and bacterial mimic nuclei; oleic acid (OA), a fatty acid found in cell LDs; and dioleoylphosphatidylcholine (DOPC), a phospholipid commonly found in cell membranes. As depicted in Fig. 2 (F to K), the addition of dsDNA and OA to the TPA2PyPh solution resulted in a marked enhancement in fluorescence. This phenomenon can be attributed to the strong binding affinity of TPA2PyPh toward dsDNA and OA, which limits the intramolecular motions of TPA2PyPh, thereby triggering fluorescence light up. In contrast, only a weak fluorescence enhancement (<2-fold) was observed with DOPC, indicating weak interactions between TPA2PyPh and DOPC. On the basis of these findings, we propose that TPA2PyPh can penetrate Raw 264.7 cell membrane and further interact with OA to induce the fluorescent enhancement due to its excellent affinity toward OA. Given the double-layered structure of mammalian cell nuclear membranes, TPA2PyPh may face challenges in crossing this barrier, thus avoiding direct binding to DNA within the nucleus of Raw 264.7 cells and minimizing potential effects on its DNA integrity. For S. aureus ATCC 25923, TPA2PyPh can traverse the outer membrane of the bacterium; since bacterial cells lack a protective nuclear membrane, TPA2PyPh readily binds to DNA within the bacterial nucleus and resulting in bright fluorescence emission.

Bacterial imaging and antibacterial activities

To further explore the binding efficiency of TPA2PyPh to the bacteria, we carried out a cell uptake experiment to investigate the internalization of TPA2PyPh in S. aureus ATCC 25923, MRSA (methicillin-resistant Staphylococcus aureus) USA300, and E. coli ATCC 25922. Each bacterium was in a concentration of ~10⁶ colony-forming units (CFU) and incubated with 5 μM TPA2PyPh for different time intervals. Subsequently, the supernatant was analyzed using an ultraviolet-visible (UV-vis) spectrophotometer. As shown in Fig. 3 (A to C), S. aureus ATCC 25923 and MRSA USA300 exhibited a notable decrease in the absorption spectra within 5 min and reached saturation around 10 min, indicating rapid uptake of TPA2PyPh by both bacteria. Conversely, E. coli ATCC 25922 showed a slower cell uptake of TPA2PyPh, with saturation occurring around 30 min. These differences might be attributed to E. coli’s thicker and more intricate outer membrane, which limits the penetration of TPA2PyPh. Meanwhile, confocal laser scanning microscopy (CLSM) was used to monitor the cell uptake by incubating TPA2PyPh with S. aureus ATCC 25923, MRSA USA300, and E. coli ATCC 25922. Figure 3D reveals substantial fluorescence within S. aureus ATCC 25923, MRSA USA300, and E. coli ATCC 25922 cells postincubation with TPA2PyPh, indicating its effective bacterial penetration. To further verify the intracellular localization of TPA2PyPh in bacteria, we costained E. coli ATCC 25922 with Hoechst and TPA2PyPh, followed by confocal imaging across multiple z layers (fig. S9). The results showed a good colocalization between the fluorescence signals of TPA2PyPh and Hoechst in different z layers, indicating that TPA2PyPh can effectively penetrate the bacterial outer membrane and bind to bacterial DNA.These results demonstrated that TPA2PyPh could be taken up by both Gram-positive and Gram-negative bacteria with a high efficiency.

Fig. 3. Bacterial uptake, bacterial imaging, and antibacterial properties.

(A to C) Changes in UV-vis absorption spectra of supernatants after incubation of TPA2PyPh (5 μM) with bacteria for different time intervals (1× PBS solution, pH 7.4; containing 0.1% DMSO). (A) S. aureus ATCC 25923, (B) MRSA USA300, and (C) E. coli ATCC 25922. (D) Cellular uptake efficiency of TPA2PyPh upon incubation with different bacteria. (E) Confocal images of S. aureus ATCC 25923, MRSA USA300, and E. coli ATCC 25922 after incubation with 2 μM TPA2PyPh. Scale bar, 5 μm. (F) Antibacterial activities of TPA2PyPh toward S. aureus ATCC 25923, MRSA USA300, and E. coli ATCC 25922. (G to I) The inhibitory effect of TPA2PyPh against S. aureus ATCC 25923, MRSA USA300, and E. coli ATCC 25922 biofilm formations. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 (Student’s t test). n.s., not significant. (J) 3D confocal images of MRSA USA300 biofilms incubated with TPA2PyPh (2 × MIC) at different time points (interval of x and y axes, 20 μm; z axis, 2 μm).

Subsequently, we evaluated the antibacterial activity of TPA2PyPh by determining the minimal inhibitory concentration (MIC). As depicted in Fig. 3F, TPA2PyPh demonstrates broad-spectrum antibacterial activity, with MIC values ranging from 1 μM (0.759 μg/ml for S. aureus ATCC 25923 and MRSA USA300) to 5 μM (3.795 μg/ml for E. coli ATCC 25922). These performances prompted further investigation into its efficacy against biofilm formation, a persistent challenge in health care due to heightened antimicrobial resistance (32, 33). As shown in Fig. 3J, after incubation with the MRSA USA300 biofilm, the red fluorescence from TPA2PyPh was rapidly increased in intensity within the biofilm until it reached saturation within 60 min. The distinct red fluorescence surrounding Hoechst-labeled bacteria in the biofilm indicates that TPA2PyPh can penetrate the MRSA USA300 biofilm effectively. Moreover, we further evaluated the efficacy of TPA2PyPh against biofilm formation. Culturing S. aureus ATCC 25923, MRSA USA300, and E. coli ATCC 25922 with varying concentrations of TPA2PyPh, followed by MTT incubation, showed that increasing the concentration of TPA2PyPh led to a significant reduction and eventual disappearance of the purple coloration of the biofilms. This indicates that TPA2PyPh strongly inhibits biofilm formation for S. aureus ATCC 25923, MRSA USA300, and E. coli ATCC 25922, respectively (Fig. 3, G to I). In addition, we investigated the biofilm-disrupting ability of TPA2PyPh by treating mature MRSA USA300 biofilms with various concentrations and quantifying disruption via crystal violet staining and MTT assay (figs. S10 and S11). After 24 hours, TPA2PyPh at 16 × MIC, 32 × MIC, and 64 × MIC disrupted >70, ~90, and ~ 95% of the biofilm, respectively, whereas 32 × MIC vancomycin achieved only ~25% disruption of mature MRSA biofilm. These results illustrate the high affinity of TPA2PyPh for bacteria, as well as its effective antimicrobial activity and its potent penetration and disruption of biofilms.

Antibacterial mechanism

Having demonstrated the antimicrobial efficacy of TPA2PyPh, we further investigated its biological mechanism. We used flow cytometry to monitor changes in bacterial membranes treated with TPA2PyPh. DiOC₂(3) and TO-PRO-3 were used to gain key insights into changes in membrane potential and permeability, respectively. DiOC₂(3) is a ratiometric probe that emits red fluorescence when aggregated within intact cell membranes and shifts to green fluorescence when the membrane potential decreases. TO-PRO-3 is a nucleic acid dye that only permeates cells with damaged membranes. Dimethyl sulfoxide (DMSO), the solvent for drug dissolution, and ciprofloxacin, an antibiotic that does not interact with bacterial membranes, were used as the control groups. As shown in Fig. 4A, incubation with DMSO and ciprofloxacin did not result in a significant reduction in the DiOC₂(3) fluorescent signal or a notable increase in TO-PRO-3 fluorescence, indicating that neither DMSO nor ciprofloxacin affected the membrane potential or permeability of the bacteria. As positive controls, carbonyl cyanide 3-chlorophenylhydrazone (CCCP), a known membrane decoupler, and nisin, a known pore-forming peptide, were used to monitor changes in bacterial signals. Treatment with CCCP resulted in a significant alteration in DiOC₂(3) fluorescent signals of bacteria, indicating that CCCP successfully disrupted bacterial membrane potential. Conversely, treatment with nisin led to significant shifts in both DiOC₂(3) and TO-PRO-3 signals, revealing that nisin disrupted both membrane potential and permeability. Furthermore, following a 30-min treatment with TPA2PyPh, we observed significant changes in DiOC₂(3) and TO-PRO-3 signals, similar to those induced by nisin, compared to the untreated group. These results indicate that TPA2PyPh could disrupt both membrane polarization and permeability of bacterial membrane.

Fig. 4. Effect of TPA2PyPy treatment on mRNA expression and bacterial membrane integrity.

(B) Volcano plot of the transcriptome results between TPA2PyPh-treated MRSA USA300 and untreated MRSA USA300. (C and D) GO (C) analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway (D) analysis of the up-regulated and down-regulated DEGs. For each group in (B) to (D), n = 3 biological replications. FC, fold change. (A) Flow cytometry analysis of the membrane potential and permeability of E. coli ATCC 25922 cells after 30 min of incubation with 1% DMSO (solvent control), 5 μM CCCP, nisin (25 μg/ml; 2 × MIC), 1 μM ciprofloxacin (2 × MIC), and 10 μM TPA2PyPh (2 × MIC).

Given that TPA2PyPh can strongly bind to DNA, potentially affecting multiple bacterial functions and biological processes, we conducted a comparative transcriptome analysis on MRSA USA300 to elucidate the gene expression differences between TPA2PyPh-treated and untreated groups, thereby gaining insight into its antibacterial mechanism. As shown in Fig. 4B, the analysis revealed 137 differentially expressed genes (DEGs) in the TPA2PyPh-treated groups compared to the untreated groups, with 69 up-regulated genes and 68 down-regulated genes. Notably, most of these DEGs (68.6%) were associated with biological processes, indicating a substantial impact of TPA2PyPh on bacterial biological activities. The remaining DEGs were primarily related to cellular components (14.6%) and molecular functions (4.4%). To elucidate the biological implications of these DEGs, we conducted gene ontology (GO) enrichment analysis (Fig. 4C), revealing associations with key bioprocesses such as cytolysis, carbohydrate transport, transmembrane transporter activity, ion binding, and plasma membrane functions. These changes at the gene level suggest adaptation mechanisms of MRSA USA300 in response to TPA2PyPh treatment. Furthermore, the Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis (Fig. 4D) highlighted significant enrichment in pathways including adenosine 5′dtriphosphate (ATP)–binding cassette (ABC) transporters, galactose metabolism, and purine metabolism. Particularly noteworthy was the enrichment in ABC transporters, known for their crucial roles in bacterial metabolic activities, mediating ATP-dependent solute uptake and efflux across cell membranes (34–36). In addition, ABC transporters serve as virulence factors, influencing nutrient uptake and toxin/antimicrobial agent efflux (34, 35, 37). The dysregulation of ABC transporters underscores the profound impact of TPA2PyPh on bacterial metabolic pathways at the genetic level. These findings collectively suggest that TPA2PyPh enhances bactericidal activity through its influence on regulatory networks and metabolic pathways in MRSA USA300.

Cell uptake of Raw 264.7 and in vitro intracellular bacteria

Inspired by the excellent antimicrobial properties and strong LD affinity of TPA2PyPh, we further investigated the feasibility of loading TPA2PyPh into the LDs of Raw 264.7 cells and assessed its potential in treating challenging systemic infections with good selectivity between bacteria and normal cells through adoptive LD-engineered macrophage transfer. To determine whether TPA2PyPh can be transported into the LDs of immune cells, we incubated TPA2PyPh with Raw 264.7 cells to generated LD-engineered macrophages (TPP-Raw). Fluorescence images revealed that TPA2PyPh precisely colocalized with BODIPY 493/503, a widely used LD tracker, demonstrating that TPA2PyPh can be efficiently loaded into the LDs within Raw 264.7 cells (Fig. 5, A and B). Notably, the fluorescence pattern of TPA2PyPh did not overlap with nuclear staining by Hoechst dye (figs. S12 and S13), indicating its inability to penetrate the nuclear envelope, which consists of two lipid bilayer membranes, thus preventing any potential damage to host cellular chromosomes. To further investigate the biosafety of TPA2PyPh, we assessed its cytotoxicity toward various cells. As depicted in Fig. 5C and fig. S14, TPA2PyPh exhibited low cytotoxicity toward Raw 264.7, NIH 3T3, 4T1, CT26, GSE-1, HIEC-6, and bone marrow cells within the concentration range of 0 to 20 μM. Minimal growth inhibition was observed even at concentrations exceeding 50 μM, which is significantly higher than its MICs against MRSA USA300 and E. coli ATCC 25922 (a 50-fold and 10-fold difference, respectively). To further evaluate the impact of TPA2PyPh on the physiological activity of macrophages, we analyzed cell surface markers and intracellular ROS levels after incubation with TPA2PyPh. Quantitative real-time polymerase chain reaction analysis was performed to assess the relative expression of key macrophage markers, including inducible nitric oxide synthase, CD86, interleukin-18 (IL-18), tumor necrosis factor–α (TNF-α), IL-6, and IL-1β (fig. S15). Some proinflammatory markers exhibited a slight upward trend; however, the magnitude of these changes was substantially smaller than that observed in lipopolysaccharide- and interferon-γ–induced M1 polarization of macrophages, suggesting that TPA2PyPh does not significantly alter macrophage polarization or activation state. Furthermore, intracellular ROS levels were evaluated using DCFH-DA, revealing that TPA2PyPh treatment did not induce a significant ROS increase, in contrast to H₂O₂-treated cells (fig. S16). These findings indicate that TPA2PyPh does not disrupt the normal physiological function of Raw 264.7 cells. In addition, hemolysis assays demonstrated low hemolytic activity of TPA2PyPh (Fig. 5, D and E). These findings collectively indicate that TPA2PyPh can be efficiently loaded into the LDs of Raw 264.7 cells with high biosafety, making it a promising candidate for further in vivo applications.

Fig. 5. Raw 264.7 cell labeling, cytotoxicity, biosafety of TPA2PyPh, and in vitro antibacterial effect of TPP-Raw.

(A) Intracellular LDs and lysosome imaging. Confocal fluorescence images of Raw 264.7 cell costained with TPA2PyPh (5 μM) and BODIPY 493/503 (5 μg/ml; top), or LysoTracker (2 × 10⁻⁶ M; bottom). Scale bar, 10 μm. (B) Flow cytometric analysis of TPA2PyPh loading efficiency after incubation with Raw 264.7 cells at different concentrations. (C) The viability of Raw 264.7, NIH 3T3, 4T1, and CT26 cells after incubating with different concentrations of TPA2PyPh for 24 hours using Cell Counting Kit-8 assay. (D and E) Hemolysis of TPA2PyPh on BALB/c red blood cells. (F) Intracellular bacterial imaging. Confocal fluorescence images of TPP-Raw (TPA2PyPh-treated concentration, 2 μM) after the infection with GFP-labeled S. aureus with an MOI of 20. Scale bar, 10 μm. (G) Schematic representation of the experimental approach of intracellular bactericidal killing. (H and I) Antibacterial activities of different TPA2PyPh concentrations treated Raw 264.7 cell against MRSA USA300. ***P < 0.001 (Student’s t test).

To evaluate the affinity of TPA2PyPh for bacteria following cellular uptake, we coincubated TPP-Raw with green fluorescent protein (GFP)–labeled S. aureus at a multiplicity of infection (MOI) of 20, and the interaction was visualized using confocal microscopy. The resulting images (Fig. 5F) demonstrated a significant overlap between the fluorescence pattern of TPA2PyPh and the GFP signal emitted by GFP-labeled S. aureus, indicating efficient transfer of TPA2PyPh from LDs to invading bacteria. We also observed the phagocytosis of MRSA USA300 by TPP-Raw using time-lapse fluorescence confocal imaging, revealing that TPP-macrophages retained their effective phagocytic ability (fig. S17). To assess the bactericidal efficacy of TPP-Raw, we quantified the intracellular survival of MRSA USA300 in Raw 264.7 cells treated with 1× PBS (PBS-Raw) or various concentrations of TPA2PyPh (denoted as TPP-Raw 0.5, TPP-Raw 1, TPP-Raw 2, TPP-Raw 5, and TPP-Raw 10, representing the respective concentrations of TPA2PyPh in the culture medium during TPP-Raw preparation). Among these treatments, TPP-Raw 2 exhibited significant bactericidal activity, resulting in a 99% reduction in MRSA USA300 survival compared to PBS-Raw. TPP-Raw 5 and TPP-Raw 10, which had higher concentrations of TPA2PyPh, showed similar bactericidal effects (Fig. 5, H and I, and fig. S18). Liquid chromatography–mass spectrometry analysis was also performed to quantify the loading of TPA2PyPh in TPP-Raw 10 cells. The results showed that after 4 hours of incubation with TPA2PyPh, ~2.9 μg of TPA2PyPh was loaded per 1 × 10⁷ Raw 264.7 cells (fig. S19). These results indicate that TPA2PyPh can be efficiently loaded into Raw 264.7 cells and enhance their bactericidal ability.

In vivo experiment

Given the robust in vitro bactericidal activity of TPP-Raw, we investigated its therapeutic efficacy in a murine model of sepsis induced by MRSA USA300 infection in immunocompromised mice. To mimic the immunocompromised condition observed in septic patients, intraperitoneal injections of cyclophosphamide (CY) were administered for 3 consecutive days. The immunocompromised state was confirmed by a reduction in white blood cell (WBC) counts (Fig. 6G). Subsequently, the immunocompromised mice were intraperitoneally infected with 0.5 × 10⁸ CFU of MRSA USA300 to simulate a septic condition. Following infection, mice were treated with PBS, TPA2PyPh, vancomycin, PBS-Raw, TPP-Raw, and TPP-BMDM (generated by incubating bone marrow–derived macrophage with TPA2PyPh at a concentration of 10 μM) and were administered both intraperitoneally and intravenously to target peritoneal and systemic bacteria. This dual-administration strategy is particularly relevant to sepsis pathogenesis, where bacteria introduced intraperitoneally can enter the lymphatic system, reach the bloodstream, and disseminate to various organs. By combining intraperitoneal and intravenous administration, we aimed to enhance bacterial clearance across both local and systemic compartments, addressing the dynamic progression of sepsis (14). As shown in Fig. 6B, the survival rates of the TPP-Raw (80%) group and TPP-BMDM (80%) group were significantly improved compared to the TPA2PyPh (60%) group and vancomycin (40%) group. We also monitored changes in inflammation-related factors, including TNF-α and IL-6, before and after treatment (Fig. 6, C to F). During the early stages of infection, inflammatory cytokine levels were significantly elevated in all groups compared to healthy mice (P < 0.0001). After 10 days of treatment, only mice in the TPA2PyPh, vancomycin, TPP-Raw, and TPP-BMDM groups survived. Notably, TNF-α and IL-6 levels in the TPP-Raw– and TPP-BMDM–treated mice were restored to levels comparable to those of healthy mice (P > 0.05). In addition, the WBC counts in these groups returned to normal (Fig. 6G), indicating that TPP-Raw and TPP-BMDM treatments effectively reversed the immunocompromised state. In contrast, TNF-α and IL-6 levels remained significantly elevated (P < 0.05) in the surviving mice treated with TPA2PyPh or vancomycin, suggesting that despite improved survival, systemic infection persisted at a relatively high level.

Fig. 6. In vivo experiment of a sepsis model.

(A) Schematic representation of the experimental approach for adaptive macrophage transfer. ip, intraperitoneally. (B) Percentage survival of mice with sepsis induced by intraperitoneal bacterial inoculation (0.5 × 10⁸ CFU) upon different treatments. TNF-α (C and E) and IL-6 (D and F) measurements of healthy mice, sepsis mice, and sepsis mice after 12 hours (h) or 10 days (d) of ACT treatment. P > 0.05 (n.s.), *P < 0.05, and ****P < 0.0001 versus control group (Student’s t test). (G) WBCs of healthy mice, immunocompromised mice, and sepsis mice after 10 days of different treatments. P > 0.05 (n.s.) and **P < 0.01 versus CY treatment group (Student’s t test). (H) Bacterial burden in major organs at day 10 after cell transfer. P > 0.05 (n.s.), ***P < 0.001, and ****P < 0.0001 versus vancomycin group (Student’s t test). (I) Bacteria and PL distribution in major organs after 12 hours of cell transfer. (J) IVIS image of PL distribution after 12 hours of cell transfer. (K) Bacterial burden in major organs at 12 hours after bacterial infection or day 10 after different treatments.

To further assess bacterial clearance in septic mice, we quantified bacterial loads in major organs after 10 days of treatment. The TPP-Raw and TPP-BMDM treatment groups exhibited a significant reduction in bacterial burden and recovery of body weight compared to the initial infection stage (Fig. 6, H and K, and figs. S20 and S21). Moreover, compared to free TPA2PyPh and vancomycin, TPP-macrophage–mediated adoptive cell transfer led to a more pronounced decrease in bacterial load, highlighting its superior efficacy in bacterial clearance. To evaluate the targeting ability of TPP-Raw for in vivo bacterial infections, we collected major organs 12 hours postinfection. Bacterial distribution and TPP-Raw localization in major organs were quantified by bacterial CFU and in vivo fluorescence imaging, respectively. As shown in Fig. 6 (I to K) and fig. S21, at 12 hours postinfection, bacterial concentration was notably high in the peritoneal and liver, constituting ~75% of the total bacterial population. This distribution pattern corresponded with the predominant fluorescent signals of TPP-Raw observed in these regions (fig. S22), indicating efficient targeting of TPP-Raw against bacterial infections in vivo. We also investigated the in vivo distribution of free TPA2PyPh and TPP-Raw in mice. As shown in figs. S23 to S27, while both exhibited a similar distribution pattern, TPP-Raw demonstrated significantly prolonged retention in vivo. This is evidenced by the sustained increase in fluorescence signals within the first 24 hours, which remained strong until diminishing after 48 hours. This observation suggests that TPP-Raw has enhanced stability and persistence in the biological environment, potentially contributing to its better therapeutic effect. Building on the exceptional therapeutic efficacy of TPP-Raw in immunosuppressed septic mice, we further evaluated this approach in a bloodstream infection sepsis model. As shown in figs. S20 and S29, at 18 hours postinfection, both TPP-Raw and TPP-BMDM significantly suppressed bacterial burden in the kidneys compared to the untreated control group (P ≤ 0.0001). After 5 days of treatment, nearly all bacteria were eradicated in these groups. Moreover, TPP-Raw– and TPP-BMDM–mediated adoptive cell transfer exhibited superior bacterial clearance in the kidneys compared to the commercially available antibiotic vancomycin, highlighting the enhanced efficacy of this therapeutic strategy. Last, we systematically investigated the in vivo toxicity of TPA2PyPh after intravenous injection. Pharmacokinetic studies revealed that following intravenous administration of free TPA2PyPh (10 μM in 100 μl of 1× PBS, pH 7.4), its plasma concentration remained at ~300 ng/ml within the first hour. A rapid decline was observed after 2 hours, with TPA2PyPh becoming nearly undetectable in plasma after 12 to 24 hours (fig. S30). Histopathological analysis for major organs, including the heart, lung, spleen, liver, and kidney, was carried out, and the hematoxylin and eosin staining images showed no obvious organ damage (fig. S31). For hematological analysis, WBCs and lymphocytes in TPA2PyPh-treated groups were normal in comparison with the control groups (fig. S32). The levels of the important liver and kidney function biomarkers, such as aspartate aminotransferase, alanine aminotransferase, urea nitrogen, and creatinine, are similar to those in the control group, indicating no obvious hepatotoxicity and nephrotoxicity (fig. S33). These findings collectively demonstrate that TPP-macrophages effectively eliminate bacteria in vivo, leading to reduced mortality rates and restoration of immune function under immunocompromised conditions.

DISCUSSION

Bacterial sepsis is considered an extremely difficult-to-treat infectious disease due to its high mortality rate and the lack of effective drugs to alleviate immunosuppression. The prevalent chemical modification of existing antibiotics, which often shares a similar bactericidal mechanism, frequently results in an increased risk of developing resistance, thus limiting their effectiveness in treating sepsis. Although adoptive cell transfer (ACT) therapy holds promise as a potential treatment for sepsis, commonly used lysosomal-engineered ACT strategies face challenges posed by bacterial immune evasion mechanisms that allow bacteria to escape phagolysosomal killing.

Here, we report an antibacterial luminogen, TPA2PyPh, which exhibits strong lipid-binding ability and broad-spectrum bacteriostatic activity by disrupting bacterial membrane integrity and altering gene expression. Once internalized by macrophages, TPA2PyPh can effectively bind with intracellular LDs, thereby enhancing their innate immune function and anti-intracellular bacterial activity. Our findings demonstrate that adoptive transfer of TPA2PyPh-labeled macrophages, which uses LDs in immune cells as targets for loading antimicrobial agents, can effectively eliminate invading bacteria and restore the innate immune system in septic mice with immunosuppression, offering a promising approach for managing late-stage sepsis with immunosuppression.

Despite the therapeutic potential of this strategy, certain limitations should be acknowledged. First, data show that TPA2PyPh-loaded macrophages significantly reduce the bacterial burden in multiple tissues of septic mice. While some bacteria may have invaded host cells before transplantation, this effect is unlikely to result from direct killing of intracellular pathogens by the transplanted TPP-macrophages. Instead, this effect may be partly attributed to enhanced immune responses following transplantation, which could facilitate the clearance of intracellular infections. However, further studies are needed to clarify the extent and efficiency of these immune-mediated effects. In addition, to improve the clearance of intracellular bacteria, future studies could explore combination strategies such as metabolic labeling (38) or nanoparticle-based antimicrobial approaches (39), which may enable broader elimination of bacteria across diverse host cell types and enhance overall therapeutic outcomes.

Second, the clinical translation of ACT-based antimicrobial therapies remains challenging by the compromised immune status of patients with sepsis. Although TPP-macrophages can, in principle, be generated within hours following immune cell isolation, the rapid progression of sepsis and associated immune exhaustion often hinder the timely acquisition and preparation of a sufficient number of functional patient-derived macrophages. In this context, recent advances in induced pluripotent stem cell (iPSC) technologies offer a promising solution. iPSC-derived macrophages can be generated at scale and engineered for universal compatibility, providing a potential “off-the-shelf” cellular product suitable for the urgent and unpredictable clinical demands of sepsis management (40).

Overall, the potent antimicrobial activity of TPA2PyPh, coupled with its synergistic interaction with immune cells, represents a promising avenue for combating severe infections caused by multidrug-resistant pathogens. This strategy directly addresses the urgent need for new therapies that can overcome the shortcomings of traditional antibiotics. To advance toward clinical application, future research should focus on optimizing cell-loading methods to ensure stable and reliable treatment outcomes. In addition, exploring alternative immune cell sources will be critical to addressing the shortage of functional immune cells in patients with sepsis and making this therapeutic approach more practical for clinical application.

METHODS

Absorption and fluorescence spectra

The stock solution of TPA2PyPh was prepared in DMSO at a concentration of 10 mM. One microliter of the stock solution was diluted in 1 ml of PBS (1×, pH 7.4) to prepare 10 μM test solution. The absorption spectrum was scanned from 300 to 800 nm using Eppendorf Vis Cuvettes. The emission spectra were measured in the range of 450 to 800 nm upon excitation at 421 nm.

Bacterial culture

Bacterial cells of S. aureus ATCC 25923, MRSA USA300, and E. coli ATCC 25922 stored in −80°C frozen tubes were seeded onto LB agar plates and incubated overnight at 37°C. Single colonies of the bacterial cells were further incubated in LB culture medium at 37°C until reaching the logarithmic phase.

For biofilm culture, mid-exponential phase [optical density at 600 nm (OD₆₀₀) ~ 0.6] bacterial cells were inoculated in tissue culture 96-well plate or confocal dish at a density of ~10⁵ CFU/ml in high-glucose Mueller-Hinton (MH) medium (containing 1% glucose and 1% NaCl), followed by culture at 37°C for 24 to 48 hours to allow robust biofilm formation.

Cell uptake

Mid-exponential phase (OD₆₀₀ ~ 0.6) S. aureus ATCC 25923, MRSA USA300, or E. coli ATCC 25922 were diluted in PBS (1×, pH 7.4) containing 5 μM TPA2PyPh at a final density of ~10⁶ CFU/ml. After incubated for various time intervals, the suspension was centrifuged at 3000 rpm for 3 min, and the supernatant was subjected to analysis using a UV-vis spectrophotometer.

MIC assay

Mid-exponential phase (OD₆₀₀ ~ 0.6) S. aureus ATCC 25923, MRSA USA300, or E. coli ATCC 25922 cultures were diluted in LB medium containing various concentrations (0, 0.1, 0.2, 0.5, 1, 2, 5, 10, 20, and 50 μM) of TPA2PyPh at a final concentration of ~10⁵ CFU/ml. The obtained dilutions were inoculated in 96-well plates and incubated at 37°C for 14 hours. MIC was recorded at the lowest concentration of the compound that results in no visible bacterial growth.

Inhibition of biofilm formation

The inhibition of biofilm formation study was carried out according to the previously reported method. Overnight cultured S. aureus ATCC 25923, MRSA USA300, and E. coli ATCC 25922 were diluted in MH culture medium (containing 1% glucose) to ~2 × 10⁵ CFU/ml. TPA2PyPh was diluted with same culture medium at concentrations of 2 × MIC, 4 × MIC, 8 × MIC, 12 × MIC, and 16 × MIC before adding in a tissue culture-treated 96-well plate, and then an equal volume of each bacterial suspension was added into each well. After culture at 37°C for 24 hours, the culture medium was removed and washed with PBS (1×, pH 7.4). One hundred microliters of MTT solution (0.5 mg/ml; in 1× PBS, pH 7.4) was added into each well and incubated for another 4 hours at 37°C. DMSO (100 μl) was added to each well after removing the MTT solution and shaking it on a shaker pad for 10 min to aid in dissolving the purple precipitate. The OD value was measured at 570 nm using microplate reader.

Confocal imaging

Bacteria with logarithmic growth periods as obtained by the method above were centrifugated at 3000 rpm for 3 min and resuspended in PBS (1×, pH 7.4) to a density of 10⁸ CFU/ml (OD₆₀₀ ~ 0.1). After incubated with TPA2PyPh for 30 min at 37°C, 20 μl of TPA2PyPh-treated bacterial suspension were dropped onto the confocal dish and kept still for 5 min to allow the bacteria to settle and adhere to the bottom before taking the CLSM images.

For biofilm imaging, mature MRSA USA300 biofilm was pretreated by Hoechst (1 μg/ml) at 37°C in PBS (1×, pH 7.4) for 30 min. After washing with PBS (1×, pH 7.4), 2 × MIC TPA2PyPh in PBS (1×, pH 7.4) was added into the confocal dish and allowed to adhere at room temperature for 1 min before CLSM imaging. Images were taken at the various time points for two sequences, sequence 1: λ_ex = 405 nm (Hoechst) and sequence 2: λ_ex = 470 nm (TPA2PyPh), respectively.

For CLSM imaging of Raw 264.7 cells, the cells in eight-well chambers with a number of 5 × 10⁴ cells per well were prepared and incubated in the incubator with 5% CO₂ at 37°C overnight. Then, the culture medium was removed and incubated with Dulbecco’s modified Eagle medium (DMEM) containing indicated concentrations of TPA2PyPh for 4 hours. The cells were also incubated with LysoTracker and BODIPY 493/503, following the standard protocols of the manufacturer, and imaged immediately by confocal microscope.

Membrane potential and permeability assay

Both E. coli ATCC 25922 and S. aureus ATCC 25923 with a concentration of 10⁸ CFU/ml were incubated with the desired concentration of antibiotics, TPA2PyPh, and CCCP for 30 min. Cells were then stained with the BacLight Bacterial Membrane Potential Kit. This kit uses DiOC₂(3) to measure cell membrane potential as a ratio of green (λ_ex = 488 nm and λ_em = 525/50 nm) to red (λ_ex = 488 nm and λ_em = 610/20 nm). Membrane integrity was measured by staining cells with TO-PRO-3, a dye that is excluded from cells with an intact membrane (λ_ex = 640 nm and λ_em = 670/30 nm). The fluorescent intensities of DiOC₂(3) and TO-PRO-3 were measured by the CytoFLEX flow cytometry platform (Beckman Coulter) to assess the effects of TPA2PyPh, CCCP, and antibiotics on bacteria. Each data file recorded 10,000 events, and subsequent analysis was conducted using FlowJo v10 software.

Intracellular bactericidal

The intracellular bactericidal assay was conducted according to the reported method. First, TPP-Raw (denoted as TPP-Raw 0.5, TPP-Raw 1, TPP-Raw 2, TPP-Raw 5, and TPP-Raw 10, representing the respective concentrations of TPA2PyPh in the culture medium during TPP-Raw preparation) was generated by incubating Raw 264.7 cells with different concentrations of TPA2PyPh in DMEM for 4 hours. After washing with PBS (1×, pH 7.4), TPP-Raw cells incubated with MRSA USA300 at an MOI of 20 for 120 min in serum-free DMEM culture medium. Afterward, the culture medium was removed, and TPP-Raw cells were washed by PBS (1×, pH 7.4). Serum-free DMEM containing gentamicin (100 μg/ml) was added to culture dish and incubated for an additional 60 min to clear the extracellular bacteria. Following additional 60 min of incubation, the cells were lysed with 0.1% Triton X-100 (in 1× PBS, pH 7.4). Last, the lysates were serially diluted 10-fold, and each serial dilution was plated onto LB agar for counting bacterial CFUs. Each group was assayed in triplicate.

In vivo experiment

All animal studies were performed in accordance with the China Animal Protection Law and approved by the Institutional Animal Care and Use Committee, Guangzhou Seyotin Biotechnology (approval no. SYT2024094). BALB/c mice (5 to 6 weeks) were intraperitoneally injected with CY (100 mg/kg per day) for 3 days, and the immunocompromised states were assessed by monitoring the body weights. After the CY injection, the mice were infected with 0.5 × 10⁸ CFU of MRSA USA300 (in 100 μl of 1× PBS, pH 7.4) through intraperitoneal injection to establish immunosuppressed model of sepsis. Following infection, PBS (1×, pH 7.4), vancomycin (20 μg/ml; 1× PBS, pH 7.4), TPA2PyPh (10 μΜ; 1× PBS, pH 7.4), PBS-Raw (10⁶ cells in 100 μl of 1× PBS, pH 7.4), TPP-Raw (10⁶ cells in 100 μl of 1× PBS, pH 7.4), and TPP-BMDM (10⁶ cells in 100 μl of 1× PBS, pH 7.4) were administered both intraperitoneally (100 μl) and intravenously (100 μl). After 12 hours of infection, major organs were collected to quantify bacterial CFUs and imaged using IVIS. After 10 days, the mice were euthanized, and major organs (heart, liver, spleen, lungs, and kidneys) were aseptically homogenized to quantify the bacterial CFUs.

For bloodstream infection model, BALB/c mice (5 to 6 weeks) were intravenously injected with 0.5 × 10⁸ CFU of MRSA (in 100 μl of 1× PBS, pH 7.4) to establish sepsis. Following infection, PBS (1×, pH 7.4), vancomycin (20 μg/ml; 1× PBS, pH 7.4), TPA2PyPh (10 μΜ; 1× PBS, pH 7.4), PBS-Raw (10⁶ cells in 100 μl of 1× PBS, pH 7.4), TPP-Raw (10⁶ cells in 100 μl of 1× PBS, pH 7.4), and TPP-BMDM (10⁶ cells in 100 μl of 1× PBS, pH 7.4) were administered intravenously (100 μl). After 18 hours of infection, major organs were collected to quantify bacterial CFUs. After 5 days, the mice were euthanized, and major organs (heart, liver, spleen, lungs, and kidneys) were aseptically homogenized to quantify the bacterial CFUs.

Supplementary Materials

This PDF file includes:

Supplementary Text

Figs. S1 to S33