Lipopolysaccharide-imprinted magneto-TiO₂ nanoagents harness dopamine charge transfer to drive visible-light photodynamic therapy for sepsis

Jiateng Wu, Jiali Wang, Weige Dong, Yu Wan, Chungu Zhang, Ming-Yu Wu and Shun Feng*
Sichuan Engineering Research Center for Biomimetic Synthesis of Natural Drugs, School of Life Science and Engineering, Southwest Jiaotong University, Chengdu 610031, China. E-mail: fengshunxd@hotmail.com

First published on 5th August 2025

---

---

1. Introduction

The global health crisis precipitated by multidrug-resistant Pseudomonas aeruginosa (P. aeruginosa), a predominant pathogen in life-threatening sepsis cases, underscores the urgent need for innovative antimicrobial strategies that simultaneously achieve pathogen specificity, therapeutic potency, and systemic biocompatibility.^1–3 This pathogen's virulence is fundamentally orchestrated by lipopolysaccharide (LPS), a structurally complex amphiphilic molecule serving as a critical structural component conferring outer membrane integrity and antibiotic resistance, and as a potent endotoxin triggering immune evasion and catastrophic hyperinflammatory cascades.^4–7 Through these mechanisms, LPS actively drives sepsis pathogenesis, positioning it as a prime therapeutic target for precision intervention.^8,9

Photodynamic antibacterial therapy (PDAT) has emerged as a transformative paradigm in antimicrobial stewardship, utilizing spatiotemporally controlled light activation to generate bactericidal reactive oxygen species (ROS), a mechanism circumventing conventional antibiotic resistance while enabling rapid microbial clearance.^10–13 Titanium dioxide (TiO₂), as an inorganic photosensitizer, shows compelling advantages including cost efficiency, earth abundance, and exceptional photochemical stability.^14–16 However, its clinical implementation remains constrained by UV-dependent activation requirements that compromise tissue penetration depth and risk genotoxic side effects through direct DNA photodamage.¹⁷

Recent breakthroughs in photocatalytic engineering have demonstrated that ligand-to-metal charge transfer (LMCT) mechanisms can unlock visible-light activity in traditionally UV-dependent materials like TiO₂.^18–24 Among these ligands polydopamine has distinguished itself through its unique combination of optoelectronic tunability and surface-binding catechol groups, which form stable coordination bonds with titanium atoms at the interface. This shifts the photoactivation threshold of TiO₂ from UV to visible wavelengths (400–650 nm).^25–27 While PDA's optoelectronic adaptability bridges UV limitations of TiO₂ with visible-light functionality, challenges persist in controlling non-selective ROS generation and nanoparticle bioaccumulation, driving research toward smart LMCT interfaces that balance efficacy with biosafety.^21,28

Herein, we develop an LPS-imprinted magneto-photocatalytic nanoplatform (LPS-MIP) that synergizes visible-light activation, pathogen-specific targeting, and rapid magnetic clearance. The nanoplatform was constructed using a surface imprinting technique with Fe₃O₄@SiO₂@TiO₂ (FST) as the core (Scheme 1). The PDA layer not only serves as a biocompatible molecular imprinting matrix but also establishes an LMCT pathway with TiO₂, narrowing the bandgap to enable visible-light-driven ROS generation. Leveraging boronate affinity-guided surface imprinting with P. aeruginosa LPS as the template, selective recognition cavities are craft on the PDA shell. The synergistic interplay between 3-aminophenylboronic acid (3-APBA) crosslinkers and LPS glycans ensures high-fidelity binding to P. aeruginosa via their conserved membrane LPS motifs. The magnetic Fe₃O₄ nanoparticles (NPs) enable rapid post-treatment retrieval under an external field, reducing hepatic nanoparticle retention in murine models, a critical leap toward mitigating systemic toxicity.²⁹ By enabling targeted eradication of resistant pathogens without UV dependency or residual toxicity, our design establishes a translatable blueprint for next-generation antimicrobials in an era of escalating antibiotic resistance.

2. Experimental

2.1 Preparation of the magneto-photocatalyst

Approximately 0.5 g Fe₃O₄@SiO₂ NPs (the preparation is illustrated in the SI) were ultrasonically dispersed in a mixture of 50 mL anhydrous ethanol and 10 mL tetrabutyl titanate for 10 min. The suspension was mechanically stirred at 80 °C, followed by dropwise addition of 10 mL H₂O/ethanol (1:5 v/v). After 4 h of reaction, the composite was magnetically separated, washed alternately with water and ethanol (3× each), and vacuum-dried at 60 °C for 8 h. The product was calcined in air at 500 °C for 2 h (heating rate: 5 °C min⁻¹) using a tubular furnace.

2.2 Preparation of LPS-MIP NPs

LPS (1 mg) and 3-APBA (4 mg) were dissolved in 20 mL water under stirring (500 rpm) in an ice bath for 1 h. Separately, FST NPs (200 mg) were ultrasonically dispersed in 20 mL water (10 min) and mixed with the LPS/3-APBA solution. The suspension was stirred at 25 °C for 20 min. Dopamine (DA, 20 mg in 5 mL water) was then added dropwise, followed by 20 min of stirring. Hexamethylenetetramine (HMTA, 0.4 g in 5 mL deionized water) was incorporated, and the mixture was heated to 90 °C under stirring for 3 h. The resulting LPS-MIP NPs were magnetically separated, alternately washed with water and ethanol (3× each), and vacuum-dried at 60 °C for 8 h. The non-imprinted polymer (NIP) was prepared in parallel except that no LPS was added.

2.3 Kinetic and isothermal adsorption studies

For kinetic adsorption studies, approximately 4 mg of LPS-MIP NPs were activated with 1 mL methanol, magnetically separated, and washed three times with water. The activated LPS-MIP was then dispersed in 4 mL of LPS solution (100 μg mL⁻¹) under continuous rotation. Aliquots were collected at predetermined intervals (10–150 min), magnetically separated, and analyzed via the phenol–sulfuric acid method. Residual LPS concentration was determined by measuring absorbance at 490 nm using a UV-vis spectrophotometer and quantified against a standard calibration curve. The adsorbed amount (Q) was calculated using

	(1)

where C₀ and C_e are the initial and equilibrium LPS concentrations (μg mL⁻¹), V is the solution volume (mL), and m is the adsorbent mass (mg). The binding constant (K_d) and maximum apparent adsorption capacity (Q_max) were derived from the Scatchard equation

	(2)

where Q_e and C_t represent equilibrium adsorption (mg g⁻¹) and residual concentration at time t, respectively. For isothermal adsorption studies, activated LPS-MIP (4 mg) was incubated with 4 mL of LPS solutions at varying concentrations (10–200 μg mL⁻¹) for 60 min. Post-incubation, supernatants were analyzed similarly to determine equilibrium parameters. Adsorption kinetics were further modelled using pseudo-first-order

ln(Qe − Qt) = lnQe − K1t	(3)

and pseudo-second-order

	(4)

where K₁ (min⁻¹) and K₂ (g (mg min⁻¹)) denote rate constants, and Q_t is the adsorption capacity at time t.

2.4 Specificity and selectivity of LPS-MIP NPs

The bacterial strains, including methicillin-resistant Staphylococcus aureus (MRSA), Escherichia coli (E. coli), and P. aeruginosa, obtained from the China General Microbiological Culture Collection Center (CGMCC), were initially revived from −80 °C frozen stocks by inoculation into Luria–Bertani (LB) medium and subjected to a two-stage activation process involving sequential liquid culture (37 °C, 150 rpm, 12 h) and quadrant streaking on LB agar. Centrifugation (8000 × g, 3 min, 4 °C) of the activated cultures yielded pellets that were standardized to 10⁸ CFU mL⁻¹ in fresh medium, with viability >95% confirmed by plate counts. For experiments, bacterial suspensions were prepared via triple-washing with PBS (pH 7.4), 10-fold dilution to ∼10⁷ CFU mL⁻¹, and incubation with LPS-MIP/NIPs under rotation. Magnetic separation every 10 min enabled absorbance-based capture efficiency calculations, while zeta potential shifts were analyzed via dynamic light scattering. Concurrently, fluorescein isothiocyanate (FITC)-labelled nanoparticles were synthesized by ultrasonicating 20 mg LPS-MIP/NIP in 8 mL ethanol, conjugating with FITC solution (800 μL, 100 μg mL⁻¹) under dark conditions (6 h, 25 °C, 500 rpm), and rigorous washing until fluorescence-free supernatants were achieved (λ_ex/em = 490/520 nm). Lyophilized FITC-NPs (1 mg mL⁻¹) were incubated with bacteria (37 °C, 30 min, 60 rpm), followed by magnetic separation, PBS resuspension, and flow cytometric analysis of bacterial binding.

2.5 Electrochemical and photo-response properties of LPS-MIP

Cyclic voltammetry (CV), electrochemical impedance spectroscopy (EIS), and photoelectrochemical measurements were conducted using a CHI660E electrochemical workstation (Chenhua, China) with a three-electrode system: a modified glassy carbon electrode (GCE, 3 mm diameter) as the working electrode, an Ag/AgCl (saturated KCl) reference electrode, and a platinum wire counter electrode. For electrode modification, 10 μL of FST, LPS-MIP, or NIP suspensions (2 mg mL⁻¹ in ethanol) was mixed with an equal volume of 1% Nafion™, drop-cast onto the GCE, and dried at 37 °C for 1 h. CV scans (−0.2 V to +0.6 V vs. Ag/AgCl, 50 mV s⁻¹) and EIS (0.1 Hz–100 kHz, 10 mV amplitude) were performed in 0.1 M PBS (pH 7.4) containing 5 mM [Fe(CN)₆]^3−/4−. Photoelectrochemical tests used 0.2 M Na₂SO₄ electrolyte, with current–time (i–t) curves recorded at 0 V bias under alternating 20 s white-light irradiation (100 mW cm⁻²) and dark cycles after stabilization to steady baseline. Data were acquired at 10 Hz to ensure temporal resolution.

2.6 ROS generation efficiency

The production capacity of reactive oxygen species (ROS) was evaluated using 2′,7′-dichlorodihydrofluorescein diacetate (DCFH-DA) as a fluorescent probe. Fluorescence measurements were performed a microplate reader with λ_ex = 488 nm and λ_em = 525 nm. The increase in the fluorescence intensity at 525 nm was quantitatively correlated with ROS generation rates. The production rate of singlet oxygen (¹O₂) was determined using 9,10-anthracenediyl-bis(methylene)dimalonic acid (ABDA) as the indicator probe. The reaction of ABDA with ¹O₂ generates endoperoxides, resulting in characteristic absorbance decay at 378 nm. The decrease of absorbance at 378 nm was quantitatively correlated with ¹O₂ yield. The production capacity of hydroxyl radical (˙OH) was evaluated using HPF as a selective fluorescent probe. Fluorescence measurements were conducted using a spectrofluorometer with λ_ex = 490 nm and λ_em = 515 nm. The increase in fluorescence intensity at 515 nm was quantitatively correlated with ˙OH generation rates.

2.7 In vitro antibacterial evaluation of LPS-MIP NPs

P. aeruginosa suspensions (10⁸ CFU mL⁻¹) were treated with LPS-MIP (200 μg mL⁻¹) in PBS (pH 7.4) under 30 min of dark adsorption (37 °C, 150 rpm), followed by white light irradiation (100 mW cm⁻², 30–150 min). Post-treatment, serial dilutions were plated on LB agar for CFU enumeration after 24 h incubation (37 °C), with survival rates calculated as (CFU_treated/CFU_initial) ×100%. For strain-specific activity, live/dead staining was performed using calcein AM (λ_ex/em = 336/447 nm) and propidium iodide (PI, λ_ex/em = 493/636 nm): bacteria (10⁸ CFU mL⁻¹) were treated with LPS-MIP (200 μg mL⁻¹, 30 min, 37 °C), and irradiated (2 h light/dark), stained (20 min), and imaged via confocal microscopy. Scanning electron microscope (SEM) analysis involved fixing treated bacteria in 2.5% glutaraldehyde, PBS rinsing, ethanol gradient dehydration (30–100%), and sputter-coating (COXEM EM-30) for morphological assessment.

2.8 In vivo sepsis treatment

All animal procedures were conducted in accordance with the guidelines set by the Institutional Animal Care and Use Committee of Sichuan Province, and the project protocols received approval from the Animal Ethics Committee of Southwest Jiaotong University (No. SWJTU-2403-NSFC (014)). Thirty-six 6-week-old female SPF BALB/c mice (Chengdu Dashuo Experimental Animal Co., Ltd) were acclimatized for 3 days and randomized into four groups (n = 9): blank (control), PBS, LPS-MIP, and Gen (gentamicin, 30 mg kg⁻¹). Sepsis was induced via tail vein injection of P. aeruginosa (1 × 10⁷ CFU mL⁻¹, 200 μL in saline), while controls received PBS. The LPS-MIP group underwent extracorporeal blood treatment: 100 μL blood collected via tail vein was incubated with LPS-MIP nanoparticles (200 μg mL⁻¹, 1 h, 37 °C), irradiated with white light (100 mW cm⁻², 2 h), and reinfused. PBS or Gen was administered daily via the tail vein to respective groups. Survival rates, weight changes, and behavioral responses were monitored daily. Blood samples collected via retro-orbital puncture at specified intervals (PBS group: days 1, 2, 4; others: days 1, 4, 7) were analyzed for bacterial load (plate counting), complete blood count (CBC, Chengdu Aochuang Biotechnology Co., Ltd), and cytokines (ELISA). Terminally, major organs (heart, liver, spleen, lungs, and kidneys) were harvested for histopathological assessment (Chengdu Lilai Biotechnology Co., Ltd).

2.9 Biosafety assays

Cell counting kit-8 (CCK-8) and hemolysis assays were performed to evaluate biocompatibility. For CCK-8 assay, TC-1 cells (5 × 10³ cells well⁻¹) in the logarithmic growth phase were seeded into 96-well plates, incubated for 24 h, and treated with LPS-MIP or NIP solutions (0–200 μg mL⁻¹, six replicates per concentration) for another 24 h. After PBS washing, 10% CCK-8 in DMEM was added (100 μL well⁻¹), incubated for 30 min, and OD450 was measured to calculate viability. For hemolysis testing, a 2% mouse RBC suspension was prepared via sequential centrifugation (1000 rpm, 10 min) and triple-washing with PBS. Aliquots (180 μL) were mixed with LPS-MIP solutions (0.1–200 μg mL⁻¹, 20 μL), PBS (negative control), or Triton X-100 (positive control), incubated at 37 °C for 1 h, and centrifuged (1200 × g, 5 min). Supernatant absorbance at 545 nm was measured in triplicate to calculate hemolysis percentage using

	(5)

2.10 Statistical analysis

All quantitative data are presented as mean ± standard deviation (SD) from at least three independent experiments. Statistical analyses were performed using GraphPad Prism software (version 9.0). One-way analysis of variance (ANOVA) with Tukey's post hoc test was employed for multiple group comparisons, with significance levels denoted as follows: ns (not significant) for P ≥ 0.05, * for P <0.05, ** for P < 0.01, *** for P < 0.001, and **** for P < 0.0001.

3. Results and discussion

3.1 Synthesis and structural characterization of hierarchical LPS-MIP NPs

Transmission electron microscopy (TEM) and SEM revealed confirmed spherical morphologies (∼400 nm diameter) for both FST and LPS-MIP NPs with granular surface nanostructures, confirming TiO₂ immobilized successfully on the NPs (Fig. 1(a), (b) and Fig. S1a–d).

Fourier-transform infrared spectroscopy (FT-IR) verified molecular-level structural evolution through characteristic vibrations: persistent Fe–O (584 cm⁻¹) confirming magnetic core stability; O–Si–O (1096 cm⁻¹) demonstrating silica encapsulation; Ti–O–Ti/Ti–O (400–700 cm⁻¹) evidencing TiO₂ formation; and B–OH vibrations (707 cm⁻¹ bend, 1340 cm⁻¹ stretching) with 1640 cm⁻¹ peak confirming boronate-affinity imprinting (Fig. 1(c)).

X-ray diffraction analysis quantified phase composition and crystallinity (Fig. 1(d)). Retention of Fe₃O₄ diffraction patterns (JCPDS 19-0629) confirmed structural stability after encapsulation, while emerging anatase TiO₂ peaks at 25.2° (101), 37.8° (004), and 48.0° (200) (JCPDS 21-1272) demonstrated successful titania coating. Crucially, LPS-MIP maintained identical diffraction peak positions to Fe₃O₄@SiO₂@TiO₂, demonstrating that molecular imprinting preserves core crystallinity without structural perturbation.

Complementary characterization further validated the nanostructures. Magnetic saturation decreased progressively from 80.5 emu g⁻¹ (Fe₃O₄) to 9.2 emu g⁻¹ (FST) and 9.1 emu g⁻¹ (LPS-MIP) due to diamagnetic shell contributions, yet still ensured rapid magnetic separation (≤5 min) (Fig. 1(e)). Zeta potential shifts (−28.7 mV → −34.2 mV) after imprinting confirmed surface hydroxyl group exposure (Fig. 1(g)). Energy dispersive spectroscopy (EDS) data showed that LPS-MIP mainly contained C, O, Si, Ti and Fe, accounting for 6.53%, 32.07%, 22.16%, 16.77% and 22.47% (Table S1), respectively. Collectively, these analyses verify the successful synthesis of LPS-MIP with stable multilayered functionality.

3.2 Adsorption kinetics and thermodynamic binding profiling of LPS-MIP

As depicted in Fig. S2a, LPS-MIP exhibited rapid adsorption kinetics with a steep uptake within 20 min, followed by gradual equilibration at 90 min. In contrast, NIP achieved equilibrium within 30 min due to nonspecific binding via surface boronic acid–LPS interactions. Notably, pseudo-second-order modeling (R² > 0.99, Fig. S2b and c) confirmed chemisorption-dominated mechanisms for both materials, implied template-selective recognition through geometrically matched cavities.

Isothermal adsorption profiles (Fig. S2d) revealed LPS-MIP superior capacity (Q_max = 13.6 mg g⁻¹), 2.09-fold higher than NIP (4.4 mg g⁻¹), with Scatchard analysis (Fig. S2e and f) further demonstrating its higher dissociation constant (K_d = 26.5 μg mL⁻¹ vs. 11.8 μg mL⁻¹ for NIP). This distinct behavior highlights cavity-enabled multivalent hydrogen bonding/electrostatic interactions for selective LPS recognition^30–32 and boronic acid-assisted dynamic covalent capture of LPS-MIP.^33–35 The synergy between molecular memory and functional monomers establishes a biomimetic “lock-key” mechanism, positioning LPS-MIP as a promising platform for Gram-negative bacterial endotoxin removal in biomedical applications.

3.3 LMCT-optimized photoelectrochemical activity and reactive oxygen species

The enhanced photoelectrochemical performance of LPS-MIP stems from the synergistic LMCT effect between TiO₂ and PDA, which optimizes both visible-light absorption and charge separation. This enhancement was characterized through complementary spectroscopic and electrochemical methods.

UV-Vis diffuse reflectance spectroscopy (DRS) provided direct evidence of improved visible-light harvesting. LPS-MIP exhibited significantly enhanced absorption across the 410–560 nm range compared to FST (Fig. S3). While FST displayed a continuous decrease in absorbance with increasing wavelength, LPS-MIP manifested a characteristic absorption plateau in this region. This spectral feature confirms the successful enhancement of visible-light absorption enabled by PDA modification.

Electrochemical characterization validated the improved charge separation and transfer efficiency. Cyclic voltammetry (CV) revealed a 6.6-fold increase in current response for LPS-MIP compared to FST (Fig. 2(a)), indicating enhanced electrochemical activity.^36,37 Electrochemical impedance spectroscopy (EIS) demonstrated significantly reduced charge transfer resistance (R_ct), evidenced by a decreased Nyquist semicircle radius for LPS-MIP (Fig. 2(b)).³⁸ Crucially, under 100 W cm⁻² white light irradiation, LPS-MIP achieved a photocurrent density of 0.34 μA cm⁻², representing a 6.6-fold enhancement over the FST core (Fig. 2(c)). This directly links the improved light absorption and charge separation to enhanced photoelectrochemical activity. Furthermore, LPS-MIP demonstrated a >40-fold enhancement in fluorescence intensity within 5 min (Fig. S4a), signifying a substantial increase in the generation of ROS, a key indicator of photoelectrochemical efficiency. Collectively, the DRS plateau, enhanced CV currents, reduced EIS resistance, significantly boosted photocurrent, and rapid ROS generation conclusively validate the synergistic LMCT effect in optimizing the photoelectrochemical performance of LPS-MIP.

At a DA/3-APBA molar ratio of 20:1 (Fig. S4B), LPS-MIP indicated the highest ROS generation efficacy, while it showed concentration-dependent ROS generation (up to 200 μg mL⁻¹) via DCFH-DA (2′,7′-dichlorodihydrofluorescein diacetate) assays. ABDA (9,10-anthracenediyl-bis(methylene)dimalonic acid) and HPF (hydroxyphenyl fluorescein) probes further confirmed rapid ¹O₂ production (I/I₀ = 0.329 at 70 min) and time-dependent ˙OH production, underscoring its capability for photoelectrochemical optimization and cytotoxic ROS generation in PDAT (Fig. 2(d) and (f)).

3.4 Specific pathogen recognition and selective capture

To validate specificity, FITC-labeled LPS-MIP/NIP was incubated with bacterial strains. Flow cytometry revealed LPS-MIP bound P. aeruginosa with 83.5% fluorescence-positive cells, significantly surpassing NIP (42.6%) and non-target pathogens: E. coli (22.6%) and MRSA (28.2%) (Fig. 3(a)). Pre-saturation with free LPS reduced LPS-MIP fluorescence intensity by 4.1-fold (Fig. 3(b)). This attenuation arises from the competitive occupation of imprinting cavities by free LPS molecules, which blocks their availability for binding to bacterial surface LPS. Consequently, the pre-saturated LPS-MIP loses its target recognition capability, directly validating the molecular imprinting mechanism as the basis for its selective bacterial binding.³⁹ Capture efficiency assays showed that LPS-MIP achieved 44.3% P. aeruginosa removal within 10 min (vs. <10% for E. coli or MRSA), escalating to >70% at 70 min while maintaining <20% non-target binding (Fig. 3(c)). Zeta potential shifts in P. aeruginosa post-LPS-MIP interaction (Fig. 3(d)) further supported selective binding via complementary ligand–receptor interactions. These results demonstrate robust specificity and selectivity, and magnetic separation capability of LPS-MIP for targeted pathogen.

3.5 Targeted photodynamic antibacterial activity

LPS-MIP exhibited potent and selective photodynamic antibacterial activity under white LED light (OPPLE, LTG0120143001, 100 mW cm⁻²). Time-dependent assays (200 μg mL⁻¹) showed that LPS-MIP achieved complete P. aeruginosa eradication within 120 min, outperforming non-imprinted NIP (Fig. S5). Concentration-dependent studies revealed LPS-MIP reduced P. aeruginosa viability to <30% at 150 μg mL⁻¹ and induced total elimination at ≥200 μg mL⁻¹, while NIP retained >55% survival at equivalent doses (Fig. 3(e) and (f)). Selectivity tests demonstrated near-complete P. aeruginosa eradication (>99%) versus high survival rates for E. coli (66.7%) and MRSA (48.3%) under identical conditions (Fig. S6). NIP exhibited non-specific activity (50% P. aeruginosa survival). Mechanistic validation via live/dead staining (Fig. S7) and SEM imaging (Fig. S8) revealed LPS-MIP selectively induced membrane damage and structural disintegration in P. aeruginosa, with minimal effects on non-target bacteria. This specificity stems from boronate affinity-driven recognition of LPS glycans^40–42 and spatially confined ROS at imprinting sites, enabling localized oxidative damage to the target pathogen while sparing non-target species.

3.6 In vivo efficacy against P. aeruginosa-induced sepsis

LPS-MIP demonstrated robust therapeutic efficacy in a murine sepsis model induced by tail vein injection of P. aeruginosa (Balb/c mice, 6-week-old, n = 9/group). Compared to PBS and gentamicin (Gen, 20 mg kg⁻¹) controls, LPS-MIP (200 μg mL⁻¹, intravenous) achieved >99% bacterial clearance in the bloodstream by day 7 (Fig. 4(a)), effectively halting pathogen proliferation observed in untreated mice. Daily monitoring revealed LPS-MIP-treated mice regained body weight by day 5, contrasting with rapid weight loss in PBS controls (<12 g by day 4) and delayed recovery in Gen-treated groups (Fig. 4(b)). Survival analysis showed 100% survival in LPS-MIP and gentamicin (Gen) groups after day 4, while all PBS-treated mice succumbed to sepsis by day 4 (Fig. 4(c)). These outcomes highlight the ability of LPS-MIP to synergize light-activated ROS generation with molecular imprinting-driven targeting, enabling selective P. aeruginosa inactivation in vivo. The localized oxidative damage and pathogen-specific binding minimized off-target effects, underscoring its translational potential for treating systemic bacterial infections.

Comprehensive analyses validated capacity of LPS-MIP to mitigate sepsis-induced systemic inflammation and multiorgan dysfunction. CBC revealed LPS-MIP-treated mice maintained normal white blood cell (WBC) levels (≤6.8 × 10⁹ L⁻¹) and neutrophil percentages (≤38.9%) by day 7, contrasting with PBS controls exhibiting severe leukocytosis (WBC: 10.57 × 10⁹ L⁻¹) and neutrophilia (87.17%) (Tables S2–S4). ELISA assays demonstrated LPS-MIP suppressed pro-inflammatory cytokines TNF-α and IL-6 to near-baseline levels by day 7, countering PBS-group escalation (Fig. 4(d)–(i)).⁴³ Histopathological evaluation (Fig. 5(a) and (b)) confirmed LPS-MIP organ-protective effects: treated mice exhibited preserved hepatic/splenic/pulmonary/renal architecture, while PBS controls showed hepatic necrosis, splenic atrophy, pulmonary neutrophilic infiltration, and renal tubular degeneration. These coordinated findings, normalized hematological parameters, controlled cytokine storms, and minimized organ damage, demonstrate that therapeutic mechanism of LPS-MIP: light-activated pathogen clearance disrupts bacterial-driven inflammation, while localized ROS generation prevents collateral tissue injury, collectively resolving sepsis progression.

3.7 Biosafety assay of LPS-MIP

Biocompatibility assessments validated the biosafety of LPS-MIP/NIP. CCK-8 assays demonstrated >80% viability of TC-1 cells after 24-hour exposure (Fig. S9). Hemolysis assays revealed minimal red blood cell damage, with LPS-MIP causing <5% hemolysis at 200 μg mL⁻¹, significantly lower than Triton X-100-induced complete lysis in controls (Fig. S10). Post-centrifugation, LPS-MIP-treated groups exhibited intact RBC sedimentation and clear supernatants, contrasting with hemoglobin-rich supernatants in positive controls. These results confirm LPS-MIP/NIP excellent hemocompatibility and negligible cytotoxicity.

4. Conclusions

In summary, this study established a novel antimicrobial nanoplatform that synergistically addresses the critical challenges in sepsis management through rational integration of boronate-affinity molecular imprinting, LMCT engineering, and magnetic guidance to achieve precision sepsis therapy. The LPS-imprinted cavities enabled glycan-specific recognition of P. aeruginosa through conserved lipopolysaccharide motifs, while the LMCT effect between PDA and TiO₂ unlocked UV-free, visible-light-driven ROS generation confined to pathogen-binding sites, eliminating off-target toxicity. Such a visible-light-powered pathogen trap demonstrated unprecedented therapeutic precision in murine sepsis models, achieving >99% bacterial clearance while resolving hyperinflammation (TNF-α/IL-6 reduction to baseline levels) and preventing multiorgan dysfunction, which was a critical leap beyond conventional antibiotic therapies. The platform's biocompatibility and magnetically guided retrieval minimize systemic nanoparticle retention, addressing critical biosafety challenges. By harmonizing molecular recognition, spatiotemporally controlled ROS delivery, and magnetic clearance, this strategy establishes a paradigm shift in antimicrobial therapy, replacing broad-spectrum approaches with pathogen-specific, light-activated precision. The design's modularity offers adaptability to other Gram-negative pathogens, paving the way for clinical translation in combating antibiotic-resistant systemic infections.

Author contributions

Jiateng Wu: conceptualization, data curation, methodology. Jiali Wang: data curation, formal analysis, project administration. Weige Dong: data curation, methodology, validation. Yu Wan: data curation, methodology, project administration, writing – review & editing. Chungu Zhang: data curation, formal analysis, methodology. Ming-Yu Wu: formal analysis, methodology. Shun Feng: funding acquisition, project administration, writing – review & editing. The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data supporting this article have been included as part of the SI. Supplementary information available: Additional materials, instruments, experimental details, SEM, dynamic adsorption curves, UV-vis diffuse reflection spectroscopy, fluorescence intensity changes, survival rate and corresponding plate photographs, live-dead fluorescence microscopy images, hemolysis analysis, EDS analysis and comparative hematological parameters (PDF). See DOI: https://doi.org/10.1039/d5tb01349f

Acknowledgements

The authors declare no competing financial interest. This work was financially supported by the National Natural Science Foundation of China (22174117). The authors would like to thank the Analytical and Testing Center of Southwest Jiaotong University for the laser confocal microscope test.

References

1. G. Mancuso, A. Midiri, E. Gerace and C. Biondo, Pathogens, 2021, 10, 1310–1323 CrossRef CAS PubMed .
2. S. J. Wood, T. M. Kuzel and S. H. Shafikhani, Cells, 2023, 12, 199–240 CrossRef CAS PubMed .
3. S. Qin, W. Xiao, C. Zhou, Q. Pu, X. Deng, L. Lan, H. Liang, X. Song and M. Wu, Signal Transduction Targeted Ther., 2022, 7, 199 CrossRef CAS PubMed .
4. Y. X. Li, L. X. Luo, Y. Q. Kong, Y. J. Li, Q. S. Wang, M. Q. Wang, Y. Li, A. Davenport and B. Li, Biosens. Bioelectron., 2024, 249, 116018 CrossRef CAS PubMed .
5. S. S. Shivatare, V. S. Shivatare and C.-H. Wong, Chem. Rev., 2024, 124, 6693–6696 CrossRef CAS PubMed .
6. F. D. Lorenzo, K. A. Duda, R. Lanzetta, A. Silipo, C. D. Castro and A. Molinaro, Chem. Rev., 2021, 122, 15767–15821 CrossRef PubMed .
7. W. P. J. Smith, B. R. Wucher, C. D. Nadell and K. R. Foster, Nat. Rev. Microbiol., 2023, 21, 519–534 CrossRef CAS PubMed .
8. S. K. Vanaja, A. J. Russo, B. Behl, I. Banerjee, M. Yankova, S. D. Deshmukh and V. A. K. Rathinam, Cell, 2016, 165, 1106–1119 CrossRef CAS PubMed .
9. K. S. Pahil, M. S. A. Gilman, V. Baidin, T. Clairfeuille, P. Mattei, C. Bieniossek, F. Dey, D. Muri, R. Baettig, M. Lobritz, K. Bradley, A. C. Kruse and D. Kahne, Nature, 2024, 625, 572–577 CrossRef CAS PubMed .
10. M. M. Kim, A. A. Ghogare, A. Greer and T. C. Zhu, Phys. Med. Biol., 2017, 62, R1–R48 CrossRef CAS PubMed .
11. E. M. Darby, E. Trampari, P. Siasat, M. S. Gaya, I. Ala, M. A. Webber and J. M. A. Blair, Nat. Rev. Microbiol., 2023, 21, 280–295 CrossRef CAS PubMed .
12. L. Paulowski, A. Donoghue, C. Nehls, S. Groth, M. Koistinen, S. O. Hagge, A. Böhling, M. Winterhalter and T. Gutsmann, Front. Cell Dev. Biol., 2020, 8, 586 CrossRef PubMed .
13. R. Gao, C. Hao, L. Xu, X. Xu, J. Zhao, M. Sun, Q. Wang, H. Kuang and C. Xu, ACS Nano, 2024, 18, 641–651 CrossRef CAS PubMed .
14. M. Yang, S. Qiu, E. Coy, S. Li, K. Załęski, Y. Zhang, H. Pan and G. Wang, Adv. Mater., 2021, 34, 2106314 CrossRef PubMed .
15. X. Han, J. Huang, X. Jing, D. Yang, H. Lin, Z. Wang, P. Li and Y. Chen, ACS Nano, 2018, 12, 4545–4555 CrossRef CAS PubMed .
16. Z. Cao, G. Yuan, L. Zeng, L. Bai, X. Liu, M. Wu, R. Sun, Z. Chen, Y. Jiang, Q. Gao, Y. Chen, Y. Zhang, Y. Pan and J. Wang, ACS Nano, 2022, 16, 10608–10622 CrossRef CAS PubMed .
17. S. Akhtar, K. Shahzad, S. Mushtaq, I. Ali, M. H. Rafe and S. M. Fazal-ul-Karim, Mater. Res. Express, 2019, 6, 105409 CrossRef CAS .
18. Q. Guo, C. Y. Zhou, Z. B. Ma and X. M. Yang, Adv. Mater., 2019, 31, 1901997 CrossRef CAS PubMed .
19. L. W. Zhang, H. H. Mohamed, R. Dillert and D. Bahnemann, J. Photochem. Photobiol., C, 2012, 13, 263–276 CrossRef CAS .
20. J. Schneider, M. Matsuoka, M. Takeuchi, J. Zhang, Y. Horiuchi, M. Anpo and D. W. Bahnemann, Chem. Rev., 2014, 114, 9919–9986 CrossRef CAS PubMed .
21. L. Caselli, E. Parra-Ortiz, S. Micciulla, M. W. A. Skoda, S. M. Häffner, E. M. Nielsen, M. J. A. van der Plas and M. Malmsten, Small, 2024, 20, 2309496 CrossRef CAS PubMed .
22. H. J. Nam, B. Kim, M. J. Ko, M. Jin, J. M. Kim and D. Y. Jung, Chem. – Eur. J., 2012, 18, 14000–14007 CrossRef CAS PubMed .
23. R. Lakhdar, F. S. Freyria, G. A. Mousdis, B. Bonelli and K. Elghniji, J. Phys. Chem. C, 2024, 128, 13445–13457 CrossRef CAS PubMed .
24. A. Khan, L. W. Evans and D. B. C. Martin, Green Chem., 2025, 27, 4664–4678 RSC .
25. W. Mao, X. Lin, W. Zhang, Z. Chi, R. Lyu, A. Cao and L. Wan, Chem. Commun., 2016, 52, 7122–7125 RSC .
26. S. Y. Feng, S. Wang, C. S. Lei, S. H. Liu and Q. Y. Jin, Catal. Lett., 2023, 153, 3783–3794 CrossRef CAS .
27. L. F. Soto-Garcia, I. D. Guerrero-Rodriguez, L. Hoang, S. L. Laboy-Segarra, N. T. K. Phan, E. Villafuerte, J. Lee and K. T. Nguyen, Int. J. Mol. Sci., 2023, 24, 13272 CrossRef CAS PubMed .
28. U. Riaz, S. M. Ashraf and J. Kashyap, Polym.-Plast. Technol., 2015, 54, 1850–1870 CrossRef CAS .
29. J. Wang, Y. Chen, J. Song, B. Guo, F. Xia, Y. Wan, W. Wu, C. Zhang, S. Feng and M. Wu, Adv. Funct. Mater., 2023, 34, 2312162 CrossRef .
30. Y. Long, Z. Li, Q. Bi, C. Deng, Z. Chen, S. Bhattachayya and C. Li, Int. J. Pharm., 2016, 502, 232–241 CrossRef CAS PubMed .
31. S. N. Shao, S. Gao, Y. Li and Y. Q. Lv, ACS Appl. Mater. Interfaces, 2023, 15, 16408–16419 CrossRef CAS PubMed .
32. Q. C. Zhang, M. Zhang, Z. Huang, Y. Sun and L. Ye, ACS Appl. Polym. Mater., 2023, 5, 3055–3064 CrossRef CAS .
33. L. Li, Y. Lu, Z. J. Bie, H. Y. Chen and Z. Liu, Angew. Chem., Int. Ed., 2013, 52, 7451–7454 CrossRef CAS PubMed .
34. R. R. Xing, S. S. Wang, Z. J. Bie, H. He and Z. Liu, Nat. Protoc., 2017, 12, 964–987 CrossRef CAS PubMed .
35. M. Arabi, A. Ostovan, J. H. Li, X. Y. Wang, Z. Y. Zhang, J. Choo and L. X. Chen, Adv. Mater., 2021, 33, 2100543 CrossRef CAS PubMed .
36. E. Güzel, RSC Adv., 2019, 9, 10854–10864 RSC .
37. H. Maltanava, N. Belko, A. Lugovski, N. Brezhneva, E. Bondarenko, P. Chulkin, G. Gusakov, N. Vileishikova, M. Samtsov and S. Poznyak, Dyes Pigm., 2022, 205, 110599 CrossRef CAS .
38. V. Novakova, M. P. Donzello, C. Ercolani, P. Zimcik and P. A. Stuzhin, Coord. Chem. Rev., 2018, 361, 1–73 CrossRef CAS .
39. J. M. Pan, W. Chen, Y. Ma and G. Q. Pan, Chem. Soc. Rev., 2018, 47, 5574–5587 RSC .
40. I. Delidovich and V. Toussaint, Green Chem., 2024, 26, 720–738 RSC .
41. Q. Zhang, M. Zhang, Z. Huang, Y. Sun and L. Ye, ACS Appl. Polym. Mater., 2023, 5, 3055–3064 CrossRef CAS .
42. S. Shao, S. Gao, Y. Li and Y. Lv, ACS Appl. Mater. Interfaces, 2023, 15, 16408–16419 CrossRef CAS PubMed .
43. D. Wu, Y. X. Shi, H. Zhang and C. H. Miao, Cell Death Dis., 2023, 14, 112 CrossRef PubMed .

This journal is © The Royal Society of Chemistry 2025