Engineered Human Ferritin Nanocage Enhances Intracellular Vancomycin Delivery and Activity Against Gram-Positive Bacteria in Macrophages

Abstract

Purpose

The limited cellular penetration of vancomycin (Van) restricts its intracellular availability and represents a major challenge for treating infections caused by intracellular Gram-positive bacteria. We investigated whether the engineered ferritin nanocage HFn-D93G could improve intracellular Van delivery while preserving the intrinsic antibacterial activity of Van.

Methods

HFn-D93G-Van was characterized for drug loading, nanocage integrity, particle size, pH-dependent release, and biocompatibility; molecular docking examined Van-protein interactions. Cellular uptake, potential uptake pathways, and subcellular distribution were examined in RAW264.7 cells, while intracellular Van accumulation, spatial association with bacteria, antibacterial activity, and cytokine responses were assessed in infected macrophages. In vivo efficacy was evaluated in a murine S. aureus peritonitis-sepsis model (n = 6 mice per group) by assessing intracellular bacterial burden in peritoneal macrophages, major-organ bacterial burdens, serum cytokines, and histopathology.

Results

HFn-D93G-Van achieved 28.71 ± 1.82% drug loading while retaining nanocage morphology and α-helical structure. At 48 h, Van release increased from 27.1% at pH 7.4 to 67.4% at pH 6.0 and 84.7% at pH 5.0. MICs remained comparable to free Van, whereas intracellular Van increased 7.34-fold (P < 0.0001). Uptake was energy dependent and involved multiple endocytic pathways, with CD71-associated involvement. HFn-D93G-Van showed greater spatial association with intracellular bacteria and reduced S. aureus, S. epidermidis, and E. faecium burdens by 2.00, 1.62, and 1.76 log₁₀ units, respectively, with lower cytokine levels. In vivo, peritoneal-macrophage intracellular S. aureus was reduced by 0.88 log₁₀ units versus free Van (P < 0.0001), while major-organ burdens were comparable between Van and HFn-D93G-Van. Histopathological alterations appeared less pronounced after both treatments.

Conclusion

HFn-D93G increased intracellular Van accumulation and reduced intracellular bacterial burdens, with the intracellular delivery advantage further observed in peritoneal macrophages in vivo. Further studies of pharmacokinetics, biodistribution, localized infection models, immunogenicity, and long-term safety are required to define its preclinical potential.

Introduction

Intracellular bacterial persistence remains a major obstacle to the treatment of recurrent and difficult-to-eradicate infections. Although Gram-positive pathogens such as Staphylococcus aureus (S. aureus), Staphylococcus epidermidis (S. epidermidis), and Enterococcus faecium (E. faecium) are generally regarded as extracellular bacteria, several of these organisms can survive within host cells, with intracellular persistence of S. aureus being particularly well documented in macrophages and other host-cell populations.^1–3 Within host cells, bacteria are less accessible to immune clearance and to antibiotics that act predominantly in the extracellular space, and may further adapt to nutrient limitation, acidic conditions, and other intracellular stresses that favor reduced metabolic activity and diminished antibiotic susceptibility.^1,2 For S. aureus, the relevance of intracellular survival has been demonstrated across different infection settings, including localized skin and soft-tissue infections as well as systemic and abdominal infection models.^4–7 These intracellular reservoirs may contribute to incomplete bacterial clearance and infection recurrence, while creating a drug-access barrier that is not fully reflected by conventional extracellular susceptibility testing.

Vancomycin (Van) remains an important glycopeptide antibiotic for severe infections caused by susceptible Gram-positive bacteria. Its antibacterial activity depends on recognition of D-Ala-D-Ala-containing peptidoglycan precursors, but its relatively large molecular size and hydrophilic character restrict passive diffusion across mammalian cell membranes.^8,9 Consequently, susceptibility to Van in conventional extracellular assays does not necessarily ensure adequate Van access to bacteria residing within host cells. To address this limitation, several nanocarrier strategies have been explored to improve intracellular Van delivery, including particle-engineered and liposomal formulations, pH-responsive polymeric nanoparticles, Van-loaded membrane vesicles, and, more recently, self-assembled Van nanodrugs.^5,9–12 These approaches generally aim to increase intracellular drug availability through enhanced cellular uptake, retention, or controlled release. Beyond Van-specific formulations, PLGA nanoparticles, metal-organic frameworks, and protein-based carriers have also been investigated for intracellular antimicrobial delivery, illustrating the diversity of nanocarrier platforms explored in this field.^13–15 These systems differ in composition and formulation characteristics, with distinct implications for drug loading, cellular uptake, retention, and release.^16,17 Among these platforms, protein nanocarriers offer structurally defined architectures and opportunities for genetic engineering, providing additional possibilities for precisely tuning drug loading and delivery properties.

Human H-chain ferritin (HFn) represents one such protein nanocarrier. HFn self-assembles from 24 subunits into a hollow nanocage and can be genetically modified while retaining its overall protein architecture.^18–20 HFn also interacts with transferrin receptor 1 (TfR1/CD71), providing a receptor-associated route for cellular uptake.^19,21 These properties have been extensively exploited for the delivery of antitumor cargos, including doxorubicin and paclitaxel.^22–25 Compared with its extensive use for antitumor drug delivery, the application of ferritin to antibiotic delivery, particularly for intracellular bacterial infection, has received relatively limited attention.^18,19,26

Conventional ferritin loading strategies often rely on pH- or denaturant-induced disassembly followed by reassembly to introduce cargo into the internal cavity.^20,22 Various ferritin-engineering strategies have been explored to tune cargo loading, including modification of the inner surface of the nanocage and engineering of intersubunit channel regions.^27–29 Of particular relevance to the present study, previous work identified a drug-entry region at the two-fold intersubunit interface of HFn and demonstrated that engineering this region, including the Asp93-to-Gly substitution (D93G), could facilitate the loading of anticancer cargos without requiring complete nanocage disassembly.²⁷ However, whether ferritin engineering strategies can improve intracellular antibiotic delivery under infection conditions requires further investigation. Addressing this question requires complementary models that can capture cellular-level delivery behavior as well as host-level infection responses.

Macrophage infection models permit direct analysis of cellular uptake, intracellular drug accumulation, subcellular distribution, and viable intracellular bacterial burden, but cannot establish whether an in vitro delivery advantage persists in a living host. Therefore, in vivo infection models are required to determine whether enhanced intracellular antibiotic delivery translates into improved antibacterial effects under physiologically relevant conditions. Among available models, staphylococcal skin and soft-tissue infection models are particularly important for evaluating strategies against recurrent infections, as intracellular bacterial reservoirs may contribute to bacterial persistence and relapse. In addition, systemic infection models, including murine S. aureus peritonitis-sepsis models, enable evaluation of bacterial burden together with organ-level infection outcomes and host inflammatory responses.^4–7

We hypothesized that the previously reported D93G-engineered HFn nanocage could support Van loading while maintaining the assembled nanocage structure, thereby increasing intracellular Van availability and enhancing intracellular bacterial clearance without affecting the intrinsic antibacterial activity of Van. To test this hypothesis, HFn-D93G-Van was evaluated for drug loading, nanocage assembly, molecular interactions, pH-dependent release, cellular uptake, intracellular Van accumulation, and antibacterial activity in macrophages infected with S. aureus, S. epidermidis, or E. faecium. A murine S. aureus peritonitis-sepsis model was further used to examine whether the intracellular advantage observed over free Van in cultured macrophages could also be detected in peritoneal macrophages in vivo.

Materials and Methods

Expression and Purification of HFn-D93G

The D93G mutation was introduced into HFn based on previous structural studies suggesting that Asp-to-Gly substitution in the drug-entry region could enhance local channel flexibility.²⁷ Transformed E. coli cells were grown overnight in LB medium supplemented with 50 μg/mL kanamycin. Protein expression was induced with 0.1 mM isopropyl β-D-1-thiogalactopyranoside (IPTG; Sigma-Aldrich, MO, USA), and the cultures were incubated overnight at 16 °C. The E. coli cells harboring the target protein expression plasmid were harvested by centrifugation at 8,000 rpm for 30 min and resuspended in lysis buffer (20 mM phosphate buffer, 500 mM NaCl, pH 4.5). The cell suspension was lysed by high-pressure homogenization. After lysis, the crude extract was heated at 72 °C for 15 min to denature most of the host proteins, followed by centrifugation at 12,000 rpm for 30 min at 4 °C to collect the supernatant. Subsequently, ammonium sulfate precipitation was performed to enrich the target protein, solid ammonium sulfate was gradually added to the supernatant with gentle stirring at 4 °C to reach the optimal saturation concentration, and the mixture was incubated for 2 h to allow complete precipitation. The protein precipitate was collected by centrifugation at 12,000 rpm for 30 min at 4 °C and then resuspended in reconstitution buffer (20 mM phosphate buffer, pH 7.4) with gentle stirring overnight at 4 °C to ensure complete dissolution. The resuspended protein solution was transferred to a dialysis bag (molecular weight cut-off: 15 kDa) and dialyzed against 50 mM Tris-HCl buffer (pH 7.5) at 4 °C for 48 h, with the dialysis buffer refreshed every 8 h to remove residual ammonium sulfate and other small-molecule impurities. The dialyzed protein sample was filtered through a 0.22 μm syringe filter to remove potential aggregates. The purity of the protein was verified by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), while the integrity of the assembled nanocage was assessed by native polyacrylamide gel electrophoresis (Native-PAGE).

Preparation of HFn-D93G-Van

HFn-D93G-Van was prepared using a thermal incubation method adapted from a previously reported loading protocol for ferritin nanocages.²⁷ The molar feed ratio of HFn-D93G to Van was optimized by evaluating ratios of 1:200, 1:400, 1:600, 1:800, and 1:1200. All HFn-D93G/Van molar feed ratios were calculated using the molecular weight of the assembled 24-subunit HFn-D93G nanocage. The drug loading content (DLC), loading efficiency (LE), protein recovery, and the number of Van molecules loaded per HFn-D93G nanocage obtained at the different molar feed ratios are summarized in Table S2. Based on a combined assessment of these four parameters, the 1:800 molar feed ratio was selected for subsequent experiments. Briefly, the HFn-D93G solution (0.9 mg/mL, 5 mL) was pre-incubated at 60 °C for 30 min. Van solution (525 μL, 20 mg/mL; Beijing Solarbio Science & Technology Co., Ltd., Beijing, China) was then added to obtain an HFn-D93G/Van molar feed ratio of 1:800, and the mixture was incubated at 60 °C for 4 h. The reaction mixture was passed through a 0.22 μm filter to remove particulate aggregates before dialysis against 20 mM Tris-HCl buffer (pH 7.5) at 4 °C for 40 h using a 15 kDa MWCO dialysis membrane (Solarbio, China) to remove unbound Van. After dialysis, the sample was centrifuged at 12,000 rpm for 10 min and passed through a second 0.22 μm filter to remove aggregates formed during dialysis. The subunit integrity of HFn-D93G and HFn-D93G-Van were assessed by SDS-PAGE. Protein concentration was measured using a bicinchoninic acid (BCA) assay kit. For Van quantification, 1 M HCl (5 μL) was added to the HFn-D93G-Van solution (300 μL) to adjust the pH to approximately 2.0. The acidified sample was incubated at room temperature for 15 min to disrupt the ferritin nanocages and release the loaded Van. The Van concentration was then determined using an Agilent 1100 high-performance liquid chromatography (HPLC) system equipped with a photodiode array (PDA) detector and a MAbPac RP column (4 μm, 3.0×100 mm; Thermo Scientific, MA, USA). The detailed HPLC conditions and experimental methods are listed in Table S1. The LE, DLC, and protein recovery were calculated as follows:

TEM Characterization

HFn-D93G and HFn-D93G-Van solutions (0.2 mg/mL, 20 mM Tris-HCl buffer, pH 7.5) were applied onto carbon-coated copper grids that had been glow-discharged for 1 min in air and negatively stained with 2% phosphotungstic acid (PTA) for 30s. After air drying, images were acquired using a JEM-1400 Plus transmission electron microscope (TEM, JEOL, Japan) operated at 100 kV.

DLS Assay

HFn-D93G and HFn-D93G-Van nanoparticles (1 mL, 0.3 mg/mL) were prepared in 20 mM Tris-HCl buffer (pH 7.5). The hydrodynamic diameter and particle size distribution of the samples were determined by dynamic light scattering (DLS) using a Zetasizer Ultra instrument (Malvern, UK). For colloidal stability analysis, HFn-D93G-Van nanoparticles were stored at 4 °C and measured at predetermined time points (0, 7, 14 and 21 days).

CD Spectroscopy Assay

Circular dichroism (CD) measurements were performed with reference to a previously reported protocol, with modifications.³⁰ CD spectra of HFn-D93G and HFn-D93G-Van were recorded using a J-1500 spectropolarimeter equipped with a temperature-controlled cell holder (JASCO, Japan). Samples were prepared at 0.3 mg/mL in 10 mM Tris-HCl buffer (pH 7.5) and transferred to a quartz cuvette with a 1 mm path length. Spectra were collected at 25 °C from 260 to 190 nm at a scanning speed of 50 nm/min, with a bandwidth of 2.0 nm, a data interval of 0.1 nm, and a data integration time of 2 s. Each spectrum represents the average of three consecutive scans of the same sample after subtraction of the corresponding buffer baseline. Thermal unfolding was monitored at 222 nm while the temperature was increased from 20 to 95 °C at a heating rate of 1 °C/min.

Molecular Docking Analysis

Molecular docking was conducted to predict the interaction of vancomycin (Van) with HFn-D93G. The AlphaFold 3-predicted structure of the HFn-D93G 24-mer was used as the receptor, and vancomycin(1+) as the ligand. Docking was performed using AutoDock Vina 1.2.5, with the search region defined around the previously reported two-fold drug-entry channel of HFn.²⁷ The grid box was centered at (21.648, −30.526, −51.993) Å with dimensions of 30 × 30×30 Å. The exhaustiveness and energy range were set to 80 and 5 kcal/mol, respectively, and up to 100 poses were generated. Docked poses were ranked according to the Vina score and inspected for their spatial distribution within the two-fold channel region. The selected pose was further analyzed for protein–ligand interactions using Discovery Studio Visualizer 2024, and structural representations were generated using PyMOL 3.0.3.

In vitro Van Release Under Different pH and Serum-Containing Conditions

HFn-D93G-Van nanoparticles (380 μM Van equivalents, 900 µL) in 20 mM Tris-HCl (pH 7.5) were loaded into Tube-O-DIALYZER^TM (MWCO 8 kDa, G-Biosciences, MO, USA). With reference to previously reported in vitro release methods, HFn-D93G-Van nanoparticles were dialyzed against phosphate-buffered saline (PBS, 10mM, pH 7.4, Solarbio, China), PBS (10mM, pH 6.0), PBS (10mM, pH 5.0), or PBS containing 50% fetal bovine serum (FBS, Gibco, NY, USA) at 37 °C with gentle stirring in the dark.^27,31 At predetermined time points (0, 2, 4, 8, 12, 24, and 48 h), 500 μL of the external dialysis medium was collected and replaced with an equal volume of fresh medium. For serum-containing samples, 1 mL of acetonitrile (ACN; Thermo Scientific, MA, USA) was added, and the samples were mixed thoroughly and centrifuged at 14,000 rpm at 4 °C for 10 min to pellet the precipitated proteins. All collected dialysis medium samples were concentrated to 250 μL, and the released Van was measured using HPLC (Agilent 1100, USA). For each release condition, the sample was divided among three dialysis units, and the release study was performed in triplicate.

Cell Culture and Bacterial Strains

RAW264.7 cells were obtained from Procell Life Science & Technology Co., Ltd. (Wuhan, China) and cultured in Dulbecco’s modified Eagle’s medium (DMEM, Gibco, NY, USA) supplemented with 10% FBS at 37 °C in a 5% CO₂ incubator (CCL-170B-8, Esco, Singapore). Human umbilical vein endothelial cells (HUVECs) were purchased from Procell Life Science & Technology Co., Ltd. (Wuhan, China) and cultured in the PUMC-HUVEC-T1 medium at 37 °C under 5% CO₂.

The bacterial strains Staphylococcus aureus (S. aureus) ATCC29213, Staphylococcus epidermidis (S. epidermidis) ATCC35984, and Enterococcus faecium (E. faecium) ATCC19434 were purchased from BeNa Culture Collection (Beijing, China). All strains in this study were grown in a constant-temperature shaker (HZQ-X100, Suzhou Peiying Experimental Equipment Co., Ltd., China) at 220 rpm and 37 °C. S. aureus ATCC29213 and S. epidermidis ATCC35984 were cultured in nutrient broth (NB, Guangdong Huankai Microbial Sci. & Tech. Co., Ltd., China), and E. faecium ATCC19434 were grown in brain heart infusion broth (BHI, Oxoid Ltd., UK).

Cytotoxicity Assay

The cytotoxicity of HFn-D93G-Van, HFn-D93G, and free Van was evaluated using an MTT assay.³² RAW264.7 cells and HUVECs were seeded in 96-well plates at densities of 5×10⁴ and 3×10⁴ cells/well, respectively, and incubated at 37 °C under 5% CO₂ for 24 h. After cell attachment, the cells were incubated with HFn-D93G-Van or free Van at two-fold serial concentrations starting from 256 μM Van equivalent for 24 or 72 h, whereas HFn-D93G was applied at carrier protein concentrations matched to those in the HFn-D93G-Van. Afterward, the supernatant was removed, and cells were incubated with 100 µL/well of DMEM containing MTT reagent from an MTT assay kit (Solarbio, China) at 37 °C for 4 h. After discarding the supernatant, 110 μL of formazan dissolving solution was added to each well. After gentle shaking for 10 min, the absorbance at 490 nm was recorded using a Synergy H1 microplate reader (BioTek, USA) with cell-free wells serving as blanks.

To evaluate the potential cytotoxic effects of endocytosis inhibitors used in the cellular uptake assay, RAW264.7 cells were incubated with chlorpromazine (CPZ, 10 μg/mL), genistein (100 μg/mL), or wortmannin under the same conditions used for the uptake inhibition assay. CPZ and genistein were maintained at the indicated concentrations for 6.5 h. Wortmannin was evaluated under two conditions: 10 μg/mL for 6.5 h or 10 μg/mL for 30 min followed by 2 μg/mL for 6 h. Cell viability was determined using the MTT assay as described above. Cell viability was calculated as follows:

Hemolysis Assay

Mouse red blood cells (RBCs) were isolated and washed according to a previously reported hemolysis assay method, with minor modifications.³³ Fresh blood was collected from 8-week-old male BALB/c mice into EDTA-containing tubes and gently mixed. After standing for 30 min, the blood samples were centrifuged at 2500 rpm for 10 min at 4 °C and the plasma was discarded. RBCs were washed twice with PBS (10mM, pH 7.4) and diluted with PBS to obtain a 2.5% (v/v) RBC suspension. For the hemolysis assay, 100 μL of serially diluted HFn-D93G-Van and free Van solution was mixed with 100 μL of RBC suspension in a 96-well plate and incubated at 37 °C for 1 h. RBCs treated with 0.1% Triton X-100 and PBS were used as positive and negative controls, respectively. After incubation, the plate was centrifuged at 2000 rpm for 10 min, and 100 μL of the supernatant from each well was transferred to a new 96-well plate. The absorbance of the released hemoglobin was measured at 576 nm using a Synergy H1 microplate reader (BioTek, USA). The hemolysis ratio was calculated as follows:

MIC Assay

The minimum inhibitory concentrations (MICs) of HFn-D93G-Van, free Van, and HFn-D93G were determined using a modified broth microdilution method.^4,34 S. aureus ATCC29213, S. epidermidis ATCC35984 and E. faecium ATCC19434 were grown at 37 °C with shaking at 220 rpm to the logarithmic phase. The bacterial suspensions were then adjusted with Mueller-Hinton broth (MHB, Beijing AoBoX Bio-Tech Co., Ltd., China) medium to obtain a final concentration of 5×10⁵ CFU/mL in each well. Van and HFn-D93G-Van were diluted two-fold in MHB medium to achieve final Van equivalent concentrations of 0.25–64 μM. HFn-D93G was prepared at protein concentrations matched to those in the HFn-D93G-Van formulations. Subsequently, bacteria were added to 96-well plates containing the tested samples. Bacteria without antibiotics served as positive controls, while blank medium was used as a negative control. After incubation at 37 °C for 18 h, absorbance values [optical density (OD) at 600 nm] were determined using a Synergy H1 microplate reader (BioTek, USA). The MIC was defined as the lowest tested concentration with no visible bacterial growth.

Quantification of Intracellular Vancomycin

The intracellular accumulation of Van in RAW264.7 cells was determined using a previously reported method with minor modifications.³⁵ RAW264.7 cells were seeded in 6-well plates at a density of 1 × 10⁶ cells/well and incubated at 37 °C under 5% CO₂ for 24 h. The cells were then infected with S. aureus ATCC 29213 at a multiplicity of infection (MOI) of 10 for 1 h. After infection, extracellular bacteria were removed by washing with PBS, and the remaining extracellular bacteria were eliminated by incubation with gentamicin (200 μg/mL) for 2 h. The cells were subsequently washed with PBS and incubated with HFn-D93G-Van or free Van at 64 μM Van equivalent in culture medium containing gentamicin (50 μg/mL; 2 mL/well) for 6 h. Afterward, the medium was removed, and the cells were washed three times with PBS. Cellular Van was extracted by adding 500 μL/well of 75% (v/v) ACN. For each treatment, extracts from three wells were pooled, yielding three samples per group. The pooled extracts were vortexed for 30s, left to stand for 10 min, and centrifuged at 12,000 × g for 10 min at 4 °C. The supernatants were collected, evaporated to dryness under a nitrogen stream, and reconstituted in 100 μL of the initial mobile phase. After filtration through a 0.22 μm membrane, the Van concentration was measured by HPLC. Total cellular protein was determined in identically treated parallel wells using a BCA assay. Intracellular Van accumulation was normalized to the cellular protein content and expressed as μg Van/mg cellular protein.

Fluorescent Labeling of HFn-D93G and HFn-D93G-Van

HFn-D93G and HFn-D93G-Van were labeled with Alexa Fluor 647 NHS ester (MedChemExpress, NJ, USA) for cellular entry and colocalization assays, with minor modifications from previously reported ferritin labeling methods.^36–38 Briefly, HFn-D93G or HFn-D93G-Van was buffer-exchanged into sodium bicarbonate buffer (100 mM, pH 8.5), and concentrated by ultrafiltration. Alexa Fluor 647 NHS ester stock solution (10 mM in DMSO) was added to the protein solution at a dye-to-protein molar ratio of 5:1. The reaction was carried out at room temperature for 3 h in the dark with gentle mixing. The unreacted dye was removed using Zeba™ dye removal spin columns (Thermo Scientific, MA, USA). The purified fluorescently labeled nanoparticles were designated AF647-HFn-D93G and AF647-HFn-D93G-Van, respectively. The protein concentrations were determined using a BCA assay kit.

Intracellular Localization Assay

With reference to previously reported confocal imaging assays for nanoparticle intracellular localization, RAW264.7 cells were seeded in μ-Slide 8-well chamber slides at 8×10⁴ cells/well and cultured in DMEM containing 10% FBS at 37 °C under 5% CO₂ for 24 h. The cells were pre-incubated in serum-free DMEM for 1 h and then exposed to AF647-HFn-D93G-Van or AF647-HFn-D93G for 6 h.^28,39 AF647-HFn-D93G-Van was used at a final concentration of 32 μM Van equivalent, and AF647-HFn-D93G was used at the corresponding protein concentration. The cells were then washed with PBS, fixed with 4% paraformaldehyde (PFA) for 15 min at room temperature, permeabilized with 0.1% Triton X-100, and blocked with BlockAid blocking buffer (Thermo Scientific, USA). To visualize lysosomal compartments, cells were incubated with an anti-LAMP1 mAb (1:100, clone D4O1S, CST, MA, USA) overnight at 4 °C, followed by incubation with Alexa Fluor 488-conjugated anti-mouse IgG (H+L) F(ab’)₂ fragment (1:300, CST, MA, USA) for 2 h at room temperature. Nuclei were stained with DAPI (2 μg/mL, Wuhan Servicebio Technology Co., Ltd., Wuhan, China) for 10 min at 37 °C, washed with PBS, and imaged using a confocal laser scanning microscope (CLSM; STELLARIS 5, Leica, Germany). To further quantify the colocalization of HFn-D93G-Van and HFn-D93G with lysosomes, confocal images were analyzed using the Coloc 2 add-on in Fiji-ImageJ to calculate Pearson’s correlation coefficients (R) and thresholded Manders’ M1 coefficient (tM1).

Cellular Uptake and Endocytic Pathway Analysis

The cellular entry efficiencies of HFn-D93G-Van and HFn-D93G were quantified using flow cytometry. For this assay, RAW264.7 cells were seeded in 12-well plates at 1×10⁶ cells/well and cultured in DMEM with 10% FBS at 37 °C with 5% CO₂ for 24 h. After pre-incubation in serum-free DMEM for 1 h, the cells were incubated for 6 h with AF647-HFn-D93G-Van at final concentrations of 4, 8, 16, 32, and 64 μM Van equivalent or with AF647-HFn-D93G at the corresponding protein concentrations. Free AF647 was used as the fluorescence control. After incubation, the cells were washed twice with PBS to remove extracellular nanoparticles, gently harvested in DMEM, and centrifuged at 1200 rpm for 3 min. The collected cells were washed with PBS, stained with Zombie Green™ Fixable Viability dye (BioLegend, CA, USA) for 20 min at room temperature in the dark, and resuspended in 1× staining buffer. Samples were analyzed using a flow cytometer (ID7000, Sony, Japan), and cellular uptake was evaluated according to AF647-positive cell percentage and mean fluorescence intensity (MFI). For flow cytometric analysis, cells were first gated based on forward and side scatter profiles to exclude debris. Viable cells were identified as the Zombie Green-low population, and AF647 fluorescence was analyzed within the viable-cell gate. Free AF647 was included as a fluorescence control. The AF647-positive threshold was established using unstained cells as the negative control and then applied consistently across all samples. A representative gating strategy is shown in Figure S5.

The internalization of HFn-D93G-Van by RAW264.7 cells was examined using low-temperature incubation, pharmacological endocytosis inhibitors, and CD71 interference, with reference to previously reported uptake studies.^40–42 For the low-temperature experiment, cells were pre-cooled at 4 °C for 30 min and then incubated with AF647-HFn-D93G-Van at 4 °C for 6 h. For the inhibition experiments, cells were pre-incubated with genistein (100 μg/mL), chlorpromazine (CPZ; 10 μg/mL), wortmannin (10 μg/mL), or anti-CD71 antibody (20 μg/mL; clone YTA 74.4, Santa Cruz Biotechnology, TX, USA) at 37 °C for 30 min. After pretreatment, AF647-HFn-D93G-Van was added and cells were incubated for an additional 6 h at 37 °C. Genistein, CPZ, and anti-CD71 antibody were maintained at the indicated concentrations throughout the uptake period. For wortmannin treatment, the concentration was reduced to 2 μg/mL after the 30 min pretreatment and maintained during the subsequent 6 h incubation to minimize potential toxicity associated with prolonged exposure. Cells incubated with AF647-HFn-D93G-Van at 37 °C without an inhibitor or antibody served as the control. After incubation, the cells were washed, collected, and stained with Zombie Green. AF647 fluorescence was measured by flow cytometry within the viable-cell population, and relative uptake was calculated from the AF647 fluorescence intensity and normalized to the control group.

Colocalization Assay with Intracellular FITC-Labeled Gram-Positive Bacteria

The colocalization of HFn-D93G-Van and HFn-D93G with intracellular Gram-positive bacteria was examined using FITC-labeled bacteria and CLSM. Briefly, S. aureus ATCC29213, S. epidermidis ATCC35984, and E. faecium ATCC19434 were cultured to logarithmic phase and washed twice with PBS. Following established protocols for FITC-based bacterial labeling, the bacterial pellets were resuspended in FITC solution (50 μg/mL) prepared in sodium bicarbonate buffer (100 mM, pH 8.5), followed by incubation at 37 °C for 45 min in the dark.⁴³ The labeled bacteria were then washed three times with PBS to remove free FITC. RAW264.7 cells were seeded in μ-Slide 8-well chamber slides at 8×10⁴ cells/well and cultured overnight. The cells were infected with S. aureus (MOI = 10, 1 h) or S. epidermidis (MOI = 10, 2 h). For E. faecium, the cells were infected at an MOI of 100 for 2 h and centrifuged at 600 × g for 5 min to promote bacterial attachment. After infection, extracellular bacteria were removed by washing with PBS and the remaining extracellular bacteria were eliminated by antibiotic incubation (S. aureus and S. epidermidis: gentamicin, 200 μg/mL, 2 h; E. faecium: gentamicin, 200 μg/mL, plus ampicillin, 100 μg/mL, 2 h). The cells were then washed and incubated with AF647-HFn-D93G-Van or AF647-HFn-D93G in DMEM containing gentamicin (50 μg/mL) for 6 h at 37 °C. HFn-D93G-Van was used at a final concentration of 32 μM Van equivalent, whereas HFn-D93G was applied at the corresponding protein concentration. After incubation, the cells were washed with PBS, fixed with 4% PFA for 15 min at room temperature, and stained with DAPI and cell membrane dye (CA1660, Solarbio, China). Fluorescence images were collected using a CLSM (STELLARIS 5, Leica, Germany). The distribution of AF647-labeled nanoparticles and FITC-labeled intracellular bacteria was analyzed by merged images, magnified views, and fluorescence intensity profiles. Colocalization was quantified using Fiji-ImageJ by calculating Pearson’s correlation coefficient and thresholded Manders’ M1 coefficient (tM1).

Intracellular Antibacterial Activity and Cytokine Quantification

For the intracellular antibacterial assay, RAW264.7 cells were infected with S. aureus ATCC29213, S. epidermidis ATCC35984, or E. faecium ATCC19434 using the infection and extracellular-bacterial removal procedures described above. To assess intracellular antibacterial activity, the infected RAW264.7 cells were further incubated with HFn-D93G, free Van, and HFn-D93G-Van for 12 h at 37 °C under 5% CO₂.⁵ HFn-D93G-Van was added at Van equivalent concentrations of 4, 8, 16, and 32 μM to evaluate its concentration-dependent intracellular antibacterial activity. For comparison at the selected dose, infected cells were treated with free Van, HFn-D93G-Van, or HFn-D93G for 12 h. Van and HFn-D93G-Van were used at 32 μM Van equivalent, and HFn-D93G was added at the corresponding protein concentration contained in the 32 μM HFn-D93G-Van formulation. Untreated infected cells served as controls. After treatment, the cells were washed with PBS and lysed with 0.1% Triton X-100 for 10 min to release the intracellular bacteria. Lysates were serially diluted, plated on agar plates, and incubated for colony enumeration. The assay was performed in triplicate and viable intracellular bacteria were expressed as log₁₀ CFU/well.

For the cytokine release assay, infected RAW264.7 cells were treated with Van, HFn-D93G, or HFn-D93G-Van under the same dosing conditions described above. Uninfected cells were used as the blank group, and untreated infected cells served as controls. After 12 h of treatment, the cell culture supernatants were collected and centrifuged to remove cell debris. The levels of TNF-α, IL-6, and IL-1β were measured using ELISA kits (Elabscience, Wuhan, China), according to the manufacturer’s instructions.

In vivo Efficacy of HFn-D93G-Van in a Murine S. Aureus Peritonitis-Sepsis Model

The in vivo activity of HFn-D93G-Van was examined using a murine S. aureus peritonitis-sepsis model adapted from previously reported protocols.^4,5 Thirty female C57BL/6 mice (6–8 weeks, 17.48 ± 0.38 g; SPF Biotechnology Co., Ltd., Beijing, China) were acclimated for 1 week and randomly assigned to five groups (n = 6): Healthy, Control, HFn-D93G, Van, and HFn-D93G-Van. Mice in the four infection groups were intraperitoneally inoculated with 100 μL of S. aureus ATCC 29213 (5 × 10⁸ CFU/mL), whereas mice in the Healthy group received an equal volume of PBS. At 2 h post-infection, the Healthy and Control groups were treated with PBS, the Van group with free Van (10 mg/kg), and the HFn-D93G-Van group with HFn-D93G-Van at a Van-equivalent dose of 10 mg/kg. The HFn-D93G group received the corresponding amount of HFn-D93G protein. At 24 h post-infection, serum, peritoneal lavage fluid, and the liver, spleen, kidneys, and lungs were collected. Serum TNF-α, IL-6, and IL-1β levels were measured using ELISA kits (Elabscience, Wuhan, China) according to the manufacturer’s instructions. Organ samples were weighed, homogenized in sterile PBS, serially diluted, and plated for colony enumeration, with bacterial burdens expressed as log₁₀ CFU/g tissue. Peritoneal macrophages were isolated from peritoneal lavage fluid by adherence enrichment. After removal of non-adherent cells, the enriched macrophages were incubated with gentamicin (200 μg/mL) for 2 h to eliminate extracellular bacteria, washed with PBS, and lysed with 0.1% Triton X-100. The lysates were serially diluted and plated for colony enumeration, with viable intracellular bacteria expressed as log₁₀ CFU/mL. Portions of the liver, spleen, kidneys, and lungs were used for histological analysis by H&E staining (Wuhan Servicebio Technology Co., Ltd). Sections were examined using a Pannoramic SCAN 150 digital slide scanner (3DHISTECH, Budapest, Hungary), and representative images were captured.

Statistical Analysis

Quantitative data were expressed as mean ± standard deviation (SD). Statistical analyses were performed using GraphPad Prism version 8. Comparisons between two groups were conducted using the Student’s t-test. For comparisons among multiple groups, one-way ANOVA was used, followed by Dunnett’s multiple comparisons test when each treatment group was compared with the control group, or Tukey’s multiple comparisons test when pairwise comparisons among groups were required. P < 0.05 was considered statistically significant. Significance levels are denoted as ns, not significant; *P < 0.05; **P < 0.01; ***P < 0.001; and ****P < 0.0001.

Results

Preparation and Physicochemical Characterization of HFn-D93G-Van

The successful purification of HFn-D93G was evaluated by structural modelling and electrophoretic analysis (Figure S1A–C). Structural modelling showed that HFn-D93G maintained the characteristic ferritin nanocage architecture, consistent with the reported 24-subunit ferritin nanocage structure.^18,22 Electrophoretic analysis showed the expected protein band, supporting successful HFn-D93G purification and nanocage assembly. After vancomycin (Van) loading, HFn-D93G-Van retained the characteristic HFn monomer band at approximately 20 kDa with no obvious degradation bands (Figure 1A), suggesting preserved HFn-D93G subunit integrity.

Figure 1.

Physicochemical characterization of HFn-D93G-Van. (A) SDS-PAGE analysis of HFn-D93G before and after Van loading. (B) Representative TEM images of HFn-D93G (left) and HFn-D93G-Van (right), with the corresponding particle size distributions shown in the insets. Scale bars, 50 nm. (C) Hydrodynamic diameter distributions of HFn-D93G (left) and HFn-D93G-Van (right) measured by DLS. (D) Representative HPLC chromatograms of HFn-D93G-Van and free Van. (E) CD spectra of HFn-D93G and HFn-D93G-Van. (F) Thermal unfolding profiles of HFn-D93G and HFn-D93G-Van from 20 to 95 °C. (G) Cumulative Van release from HFn-D93G-Van in PBS at pH 5.0, 6.0, and 7.4. (H) Cumulative Van release from HFn-D93G-Van in PBS and 50% FBS/PBS. DLS data are presented as mean ± standard deviation (SD) from three instrumental measurements (n = 3), whereas release data are presented as mean ± SD from three technical replicates (n = 3). Each CD spectrum represents the average of three consecutive scans.

The morphology and size distribution of HFn-D93G and HFn-D93G-Van were examined using TEM and DLS.^10,17 TEM images showed that HFn-D93G formed well-dispersed spherical nanocages, consistent with the characteristic architecture of ferritin-based protein cages.^22,44 After Van loading, HFn-D93G-Van retained a similar spherical morphology without obvious aggregation or structural collapse. ImageJ analysis showed that the average diameter increased from 12.70 ± 0.81 nm for HFn-D93G to 15.30 ± 0.94 nm for HFn-D93G-Van (Figure 1B). A corresponding increase in particle size was also observed by DLS, with mean hydrodynamic diameters of 21.09 ± 0.21 nm for HFn-D93G and 44.85 ± 0.22 nm for HFn-D93G-Van. The corresponding PDI values were 0.2043 ± 0.0067 and 0.2371 ± 0.0025, respectively (Figure 1C). For HFn-D93G-Van, the mean hydrodynamic diameter was 44.85 ± 0.22 nm on day 0, 44.91 ± 0.26 nm on day 7, 45.02 ± 0.12 nm on day 14, and 45.70 ± 0.29 nm on day 21. The corresponding PDI values were 0.2371 ± 0.0025, 0.2403 ± 0.0103, 0.2412 ± 0.0087, and 0.2429 ± 0.0135, respectively. The size-distribution profiles remained similar across the four time points (Figure S2).

Van loading was optimized by varying the molar feed ratio of HFn-D93G to Van from 1:200 to 1:1200 (Table S2). The DLC increased with increasing Van input, reaching 40.11 ± 0.74% at the 1:1200 ratio, whereas the LE decreased from 18.99 ± 2.77% to 10.68 ± 0.66%. The protein recovery remained relatively stable at molar feed ratios from 1:200 to 1:800 but decreased to 58.32 ± 2.54% at 1:1200. Among the tested conditions, the 1:800 molar feed ratio exhibited a DLC of 28.71 ± 1.82%, an LE of 14.99 ± 1.52%, and a protein recovery of 83.92 ± 2.57%, and was selected for subsequent preparation of HFn-D93G-Van. HPLC analysis was then performed to characterize Van loading in the selected formulation.^24,27 Free Van showed a characteristic chromatographic peak, and HFn-D93G-Van exhibited a corresponding peak at the same retention time (Figure 1D). For the HFn-D93G-Van formulation prepared at the 1:800 molar feed ratio, the loading capacity corresponded to approximately 142 Van molecules per ferritin nanocage.

The effect of Van loading on the structural integrity of HFn-D93G was further evaluated by CD spectroscopy and thermal stability analysis. CD analysis revealed two characteristic negative bands of HFn-D93G at approximately 208 nm and 222 nm, indicative of a predominantly α-helical secondary structure (Figure 1E). After Van loading, HFn-D93G-Van retained a comparable CD profile with only a slight decrease in negative ellipticity at 222 nm. In line with the CD results, thermal stability analysis showed that both HFn-D93G and HFn-D93G-Van maintained high folding ratios over a broad temperature range, followed by a sharp decrease above approximately 80 °C (Figure 1F).

To determine whether acidification along the endosomal-lysosomal pathway affects Van release, HFn-D93G-Van was incubated in PBS at pH 7.4, 6.0, and 5.0. These conditions were selected to represent the extracellular physiological environment, the mildly acidic milieu of early endosomes and bacteria-containing intracellular compartments, and the more acidic conditions encountered during late endosomal and lysosomal trafficking, respectively.^45,46 Van release exhibited pronounced pH dependence over 48 h. At pH 7.4, cumulative release remained limited, reaching approximately 22.0% at 24 h and 27.1% at 48 h. At pH 6.0, release increased to 57.1% at 24 h and 67.4% at 48 h, and increased further at pH 5.0 to 77.0% and 84.7%, respectively (Figure 1G). Release was also evaluated in serum-containing medium. In 50% FBS/PBS, cumulative Van release reached approximately 35.9% at 24 h and 48.3% at 48 h, compared with 22.0% and 27.1%, respectively, in PBS (Figure 1H). Thus, although serum promoted Van release, more than half of the loaded Van remained associated with HFn-D93G-Van after 48 h under protein-containing conditions.

Molecular docking was performed to examine potential interactions between Van and HFn-D93G. AutoDock Vina generated 98 valid poses, and the selected pose had a Vina score of −7.409 kcal/mol. Van was located at the external region of the two-fold intersubunit channel near the two-fold axis of HFn-D93G (Figure 2A). A side view showed that the selected Van pose remained predominantly on the external side of the channel center, with part of the molecule extending toward the channel entrance (Figure 2B). The docked Van molecule contacted residues from adjacent subunits, including ARG44, LYS88, ASP46, ASP90, ASP92, TYR41, CYS91, GLU95, PHE82, and LYS158 (Figure 2C). Hydrogen-bonding and electrostatic interactions involved several residues surrounding the channel, while PHE82 and LYS158 contributed π–π stacking and alkyl interactions, respectively. The corresponding two-dimensional interaction map is shown in Figure S3.

Figure 2.

Molecular docking of Van with HFn-D93G. (A) Overall view of the HFn-D93G nanocage showing Van at the external region of the two-fold intersubunit channel near the two-fold axis. The two subunits forming the channel are highlighted. (B) Side view of the selected docking pose showing the position of Van relative to the two-fold channel entrance, center, and inner region. (C) Enlarged view of the selected Van docking pose and surrounding interaction residues. The selected pose had a Vina score of −7.409 kcal/mol, with hydrogen-bonding, electrostatic, and hydrophobic interactions predicted around Van.

In vitro Biocompatibility of HFn-D93G-Van

The in vitro biocompatibility of HFn-D93G-Van was evaluated using cell viability and hemolysis assays. Cell viability was determined by the MTT assay after 24 and 72 h of incubation. In RAW264.7 cells, HFn-D93G-Van maintained high cell viability over a broad range of Van equivalent concentrations after 24 h, with a moderate decrease observed at 128 and 256 μM (Figure 3A). After 72 h, cell viability was lower across the tested concentration range, particularly at the higher concentrations (Figure 3B). Free Van showed a similar viability profile, with no significant difference from HFn-D93G-Van at the corresponding Van equivalent concentrations. HFn-D93G alone was also well tolerated within the matched protein concentration range, in line with the reported cytocompatibility of ferritin nanocages.²⁶ HUVECs showed a similar but slightly lower viability profile than RAW264.7 cells, with cell viability remaining above approximately 70% across most tested concentrations after 24 h but decreasing further after 72 h, particularly at higher concentrations (Figure S4A and B). No marked difference was observed between HFn-D93G-Van and free Van at the corresponding Van equivalent concentrations.

Figure 3.

In vitro biocompatibility of HFn-D93G-Van. (A and B) Viability of RAW264.7 cells after incubation with HFn-D93G, free Van, or HFn-D93G-Van for 24 h (A) and 72 h (B). HFn-D93G-Van and free Van were tested at equivalent Van concentrations, whereas HFn-D93G was tested at the corresponding protein concentrations. (C and D) Hemolysis results of HFn-D93G-Van (C) and free Van (D) at the indicated concentrations. PBS and 0.1% Triton X-100 were used as negative and positive controls, respectively. Statistical analysis was performed between HFn-D93G-Van and free Van at each corresponding Van equivalent concentration using the Student’s t-test; no significant difference was observed. Data are shown as mean ± SD (n = 3 technical replicates per group).

Hemocompatibility was further evaluated using a hemolysis assay. Triton X-100 (0.1%) induced pronounced hemolysis, whereas PBS showed negligible hemolytic activity (Figure 3C). HFn-D93G-Van showed minimal hemolytic activity across the tested concentration range of 0.5–256 μM Van equivalent, with hemolysis ratios remaining below 1% and comparable to those of free Van under the same conditions (Figure 3D).

Cellular Uptake and Intracellular Distribution of HFn-D93G-Van

Following the biocompatibility assessment, the cellular entry and intracellular distribution of HFn-D93G-Van were examined in RAW264.7 cells, as efficient intracellular delivery is critical for improving the activity of Van against host cell-associated Gram-positive bacteria.^5,16 AF647-labeled HFn-D93G-Van was then used to evaluate cellular uptake, internalization pathways, and subcellular localization. After incubation with AF647-HFn-D93G-Van, the CLSM images showed clear red fluorescence signals inside RAW264.7 cells (Figure 4A). The signals were mainly distributed in the cytoplasm, with limited overlap with the nuclei. HFn-D93G-Van showed partial overlap with lysosomes in the merged images. A comparable lysosomal distribution pattern was observed for AF647-HFn-D93G. Quantitative colocalization analysis based on Pearson’s correlation coefficients (R) and thresholded Manders’ M1 coefficient (tM1) showed no significant differences between HFn-D93G-Van and HFn-D93G (Figure 4B and C). Flow cytometry was used to quantify cellular entry of HFn-D93G-Van and HFn-D93G. The gating strategy used for flow cytometric analysis is shown in Figure S5. Representative histograms showed a progressive rightward shift of the AF647 fluorescence peak as the Van equivalent concentration increased from 4 to 64 μM, indicating increased intracellular accumulation of AF647-HFn-D93G-Van (Figure 4D). The mean fluorescence intensity (MFI) of AF647-HFn-D93G-Van increased from 4 to 32 μM and showed no further increase at 64 μM, suggesting a plateau-like entry profile at higher concentrations (Figure 4E). A similar concentration-dependent trend was observed in the percentage of AF647-positive cells, which exceeded 90% at higher Van equivalent concentrations and remained above the control level in all HFn-D93G-Van-treated groups (Figure 4F). Consistently, AF647-HFn-D93G also showed concentration-dependent cellular uptake in RAW264.7 cells, as reflected by increased MFI and AF647-positive cell percentages with increasing protein concentrations (Figure S6).

Figure 4.

Cellular uptake and intracellular distribution of HFn-D93G-Van in RAW264.7 cells. (A) Representative CLSM images showing intracellular localization of HFn-D93G-Van and HFn-D93G. Scale bars, 10 μm. (B and C) Quantitative analysis of lysosomal colocalization of HFn-D93G-Van or HFn-D93G using Pearson’s R values (B) and tM1 values (C). (D) Representative flow cytometry histograms of RAW264.7 cells incubated with AF647-HFn-D93G-Van at different Van equivalent concentrations. (E and F) Mean fluorescence intensity (MFI) (E) and percentage of AF647-positive cells (F) determined by flow cytometry. AF647-positive cells were defined using a fluorescence threshold established from unstained cells. (G and H) Representative flow cytometry histograms (G) and quantitative analysis (H) of cellular uptake of AF647-HFn-D93G-Van under low-temperature conditions or after incubation with endocytosis inhibitors or anti-CD71 antibody. Pearson’s R values and tM1 values (n = 6 fields of view per group), and flow cytometry data (n = 3 technical replicates per group), are presented as mean ± SD. Statistical analysis was performed using Student’s t-test for colocalization analysis and one-way ANOVA followed by Dunnett’s multiple comparisons test for flow cytometry-based uptake and inhibition assays. For the concentration-dependent uptake analyses, the free AF647 control served as the reference group, whereas the no-inhibitor group at 37 °C served as the reference for the low-temperature, inhibitor, and anti-CD71 analyses. A dash (-) indicates the reference group, for which no P value was assigned. ns, not significant; ***P < 0.001; ****P < 0.0001.

To determine whether HFn-D93G-Van increased intracellular Van availability under infection conditions, intracellular Van levels were quantified in S. aureus-infected RAW264.7 cells after incubation with HFn-D93G-Van or free Van at equivalent Van concentrations. HFn-D93G-Van resulted in an intracellular Van level of 99.09 ± 1.67 μg/mg protein, significantly higher than that of free Van (13.50 ± 3.65 μg/mg protein), corresponding to an approximately 7.34-fold increase (P < 0.0001; Figure S7).

To examine the endocytic processes involved in HFn-D93G-Van internalization, cellular uptake was measured at low temperature and following pharmacological inhibition. Incubation at 4 °C served to assess the energy dependence of uptake. Genistein, chlorpromazine, and wortmannin were applied as probes of caveolae-associated uptake, clathrin-associated endocytosis, and PI3K-associated macropinocytic processes, respectively.^40,47 Given the reported interaction between HFn nanocarriers and TfR1/CD71, an anti-CD71 antibody was also included to examine receptor-associated uptake.^21,41,48 Relative to the no-inhibitor group at 37 °C, AF647-HFn-D93G-Van uptake decreased to approximately 9.3% at 4 °C, 14.1% following chlorpromazine treatment, and 6.6% following wortmannin treatment (all P < 0.0001; Figure 4G and H). Anti-CD71 treatment reduced uptake to approximately 46.7% of the control level (P < 0.001), supporting the involvement of CD71 in HFn-D93G-Van internalization. By comparison, approximately 86.4% of the control uptake was retained after genistein treatment (ns). To assess whether inhibitor-associated cytotoxicity could affect these results, RAW264.7 cell viability was examined under the corresponding exposure conditions by MTT assay. Continuous exposure to wortmannin at 10 μg/mL reduced cell viability, whereas viability was improved under the modified regimen of 10 μg/mL for 30 min followed by 2 μg/mL for 6 h. CPZ and genistein had limited effects on cell viability under the tested conditions (Figure S8).

Spatial Association of HFn-D93G-Van with Intracellular Gram-Positive Bacteria

Having confirmed the cellular uptake of HFn-D93G-Van by RAW264.7 cells, we next examined its spatial association with intracellular bacteria. In S. aureus-infected RAW264.7 cells, both AF647-HFn-D93G-Van and AF647-HFn-D93G were detected within the cells (Figure 5A). Compared with HFn-D93G, HFn-D93G-Van showed greater spatial overlap with FITC-labeled S. aureus in the merged and enlarged images (Figure 5B and C). This difference was also evident in the line-scan profiles, where the red and green fluorescence signals showed greater concordance for HFn-D93G-Van than for HFn-D93G (Figure 5D and E). Quantitative colocalization analysis showed a higher Pearson’s R for HFn-D93G-Van than for HFn-D93G (0.62 vs 0.41; P < 0.0001; Figure 5F). tM1 analysis showed the same trend, with a higher fraction of the nanoparticle-associated signal overlapping with S. aureus in the HFn-D93G-Van group (0.68 vs 0.50; P < 0.001; Figure 5G).

Figure 5.

Spatial association of HFn-D93G-Van with intracellular S. aureus in RAW264.7 cells. (A) Representative CLSM images showing the intracellular distribution of HFn-D93G-Van or HFn-D93G in RAW264.7 cells infected with FITC-labeled S. aureus. Nuclei, (S) aureus, nanoparticles, and cell membranes are shown in blue, green, red, and magenta, respectively. Scale bars, 10 μm. (B and C) Representative full-field and magnified merged CLSM images showing the spatial association between intracellular S. aureus and HFn-D93G-Van (B) or HFn-D93G (C). White arrows indicate representative colocalized signals, and yellow lines indicate the regions selected for fluorescence intensity profiling. Scale bars, 10 μm. (D and E) Line-scan fluorescence intensity profiles of S. aureus and nanoparticles along the indicated regions in the magnified images. (F and G) Quantitative analysis of colocalization between intracellular S. aureus and HFn-D93G-Van or HFn-D93G using Pearson’s R values (F) and tM1 values (G). Pearson’s R values and tM1 values are presented as mean ± SD (n = 6 fields of view per group). Statistical analysis was performed using Student’s t-test. ***P < 0.001; ****P < 0.0001.

Similar results were obtained in RAW264.7 cells infected with S. epidermidis or E. faecium. AF647-HFn-D93G-Van was detected in bacteria-containing regions in both infection models (Figure S9A), and enlarged images showed overlap between the nanoparticle and bacterial fluorescence signals (Figure S9B and C). Corresponding line-scan profiles showed overlapping red and green signals along the selected regions (Figure S9D and E). Both Pearson’s R and tM1 were higher for HFn-D93G-Van than for HFn-D93G in the S. epidermidis and E. faecium models (Figure S9F and G).

Intracellular Antibacterial Activity and Infection-Associated Cytokine Responses of HFn-D93G-Van

The MIC assay showed that HFn-D93G-Van preserved the antibacterial potency of Van after loading. The MIC of HFn-D93G-Van was 0.5 μM against S. aureus ATCC29213 and 1 μM against both S. epidermidis ATCC35984 and E. faecium ATCC19434, comparable to free Van (Table S3). HFn-D93G showed no detectable antibacterial activity within the tested concentration range. With its extracellular antibacterial activity retained, HFn-D93G-Van was examined for activity against Gram-positive bacteria residing within RAW264.7 cells.

RAW264.7 cells infected with Gram-positive bacteria were used to evaluate the intracellular antibacterial activity of HFn-D93G-Van. After extracellular bacteria were removed by antibiotic treatment, infected cells were exposed to the indicated formulations for 12 h, followed by cell lysis and colony counting (Figure 6A). In the S. aureus model, numerous colonies were recovered from the infected control, HFn-D93G, and free Van groups, whereas markedly fewer colonies were observed after HFn-D93G-Van treatment (Figure 6B). Quantitative analysis showed that HFn-D93G-Van reduced the intracellular S. aureus burden by 2.00 log₁₀ units relative to the infected control, corresponding to 99.0% bacterial clearance (Figure 6C). Compared with free Van, HFn-D93G-Van produced an additional reduction of approximately 1.92 log₁₀ units, corresponding to an approximately 82-fold lower intracellular bacterial burden (P < 0.0001). HFn-D93G alone produced no appreciable reduction in intracellular CFU.

Figure 6.

Intracellular antibacterial activity and TNF-α responses following HFn-D93G-Van treatment in infected RAW264.7 cells. (A) Schematic illustration of the intracellular infection model and CFU counting workflow. The numbered steps (1–6) indicate the sequential workflow: (1) RAW264.7 cell preparation, (2) bacterial infection, (3) removal of extracellular bacteria, (4) HFn-D93G-Van treatment, (5) cell lysis, and (6) CFU counting. (B–D) Intracellular antibacterial activity of HFn-D93G-Van against S. aureus. (B) Representative colony images, (C) intracellular bacterial burden, and (D) TNF-α levels. (E–G) Corresponding analyses in S. epidermidis-infected RAW264.7 cells. (H–J) Corresponding analyses in E. faecium-infected RAW264.7 cells. Data are presented as mean ± SD (n = 3 technical replicates per group). Statistical analysis was performed using one-way ANOVA followed by Tukey’s multiple comparisons test. ns, not significant; *P < 0.05; ****P < 0.0001.

The same analysis was extended to RAW264.7 cells infected with S. epidermidis and E. faecium. HFn-D93G-Van reduced the intracellular bacterial burden by 1.62 and 1.76 log₁₀ units relative to the respective infected controls, corresponding to 97.6% and 98.3% bacterial clearance, respectively (Figure 6E, F, H and I). Relative to free Van, the intracellular bacterial burden was approximately 23-fold lower after HFn-D93G-Van treatment in the S. epidermidis model and 49-fold lower in the E. faecium model (both P < 0.0001). Across the three infection models, intracellular bacterial burden generally decreased with increasing HFn-D93G-Van concentration from 4 to 32 μM Van equivalent, with the lowest CFU levels observed at 32 μM (Figure S10A, D and G).

Changes in intracellular bacterial burden were accompanied by lower inflammatory cytokine levels. Relative to the infected control, HFn-D93G-Van reduced TNF-α, IL-6, and IL-1β in S. aureus-infected cells from 2269.0 ± 147.3, 1483.3 ± 125.3, and 233.6 ± 6.8 pg/mL to 767.8 ± 41.4, 688.0 ± 52.7, and 73.6 ± 4.2 pg/mL, respectively (TNF-α and IL-1β, P < 0.0001; IL-6, P < 0.001; Figures 6D, S10B and C). Corresponding reductions were observed in the S. epidermidis model, from 3446.4 ± 189.0, 2540.3 ± 55.5, and 348.9 ± 12.7 pg/mL to 1172.2 ± 71.3, 1073.0 ± 30.0, and 146.1 ± 9.4 pg/mL, respectively (P < 0.0001; Figures 6G, S10E and F). In E. faecium-infected cells, the respective concentrations decreased from 1225.2 ± 55.2, 859.0 ± 27.2, and 118.2 ± 6.9 pg/mL to 573.2 ± 72.0, 476.9 ± 21.5, and 38.9 ± 6.0 pg/mL (P < 0.0001; Figures 6J, S10H and I). Across all three infection models, cytokine concentrations after HFn-D93G-Van treatment were also significantly lower than those in the corresponding free Van groups.

In vivo Efficacy of HFn-D93G-Van

The in vivo efficacy of HFn-D93G-Van was further evaluated in a murine S. aureus peritonitis-sepsis model (Figure 7). At 24 h post-infection, high bacterial burdens were detected in peritoneal macrophages and major organs of the Control group, whereas HFn-D93G alone showed no appreciable antibacterial effect (Figure 8A–E). In peritoneal macrophages, HFn-D93G-Van reduced the intracellular bacterial burden from approximately 4.75 to 3.72 log₁₀ CFU/mL, corresponding to a 1.03-log reduction relative to the Control group (P < 0.0001; Figure 8A). Compared with free Van, HFn-D93G-Van further reduced the intracellular burden by approximately 0.88 log₁₀ units, corresponding to an approximately 7.6-fold reduction (P < 0.0001). In contrast, free Van did not significantly reduce the intracellular bacterial burden relative to the Control group.

Figure 7.

Schematic of the in vivo efficacy study design in a murine S. aureus peritonitis-sepsis model. Mice were infected intraperitoneally with S. aureus ATCC 29213 and treated 2 h later. At 24 h post-infection, bacterial burdens in major organs and peritoneal macrophages were determined by CFU counting, serum TNF-α, IL-6, and IL-1β levels were measured by ELISA, and major organs were collected for histopathological examination.

Figure 8.

In vivo efficacy of HFn-D93G-Van treatment. (A) Intracellular bacterial burden in peritoneal macrophages. (B–E) Bacterial burdens in the liver, spleen, kidney, and lung, respectively. (F–H) Serum TNF-α, IL-6, and IL-1β levels, respectively. (I) Representative H&E-stained sections of the liver, spleen, kidney, and lung from the Healthy, Control, HFn-D93G, Van, and HFn-D93G-Van groups. In the liver, green and yellow arrows indicate focal hepatocellular degenerative changes and vascular/sinusoidal congestion, respectively. In the spleen, yellow arrows indicate red-pulp congestion and green arrows indicate irregularity of the white-pulp-red-pulp interface. In the kidney, cyan, green, and yellow arrows indicate tubular dilatation, tubular epithelial degeneration, and vascular congestion, respectively. In the lung, red, cyan, and yellow arrows indicate inflammatory cell infiltration, alveolar septal thickening, and vascular congestion, respectively. Scale bars = 100 μm. Box-and-whisker plots in (A–E) show the median, interquartile range, and minimum-to-maximum values; data in (F–H) are presented as mean ± SD. For (A-H), n = 6 mice per group. Statistical analysis was performed using one-way ANOVA followed by Tukey’s multiple comparisons test. ns, not significant; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

Reduced bacterial burdens were also observed in major organs. Both Van and HFn-D93G-Van significantly reduced bacterial burdens in the liver, spleen, kidney, and lung compared with the Control group (Figure 8B–E), while no significant differences were observed between the two treatments. Relative to the infected control group, HFn-D93G-Van significantly reduced serum TNF-α, IL-6, and IL-1β levels (Figure 8F–H), from 940.1 ± 149.6, 1333.3 ± 246.3, and 541.0 ± 43.5 pg/mL to 506.6 ± 105.8, 801.9 ± 90.2, and 448.9 ± 38.2 pg/mL, respectively (P < 0.0001, P < 0.001, and P < 0.001, respectively). Free Van also decreased these cytokine levels, whereas HFn-D93G alone produced little change.

Histological findings were consistent with the microbiological and inflammatory changes. The Control and HFn-D93G groups showed infection-associated alterations in the liver, spleen, kidney, and lung, including hepatocellular degenerative changes, red-pulp congestion, renal tubular injury, and pulmonary inflammatory changes (Figure 8I). Compared with the Control and HFn-D93G groups, these histopathological alterations appeared less pronounced in both the Van and HFn-D93G-Van groups.

Discussion

Vancomycin remains active against susceptible Gram-positive bacteria, but its poor penetration into mammalian cells limits drug exposure to bacteria persisting within intracellular niches.^1,8,9 HFn-D93G appears to address this limitation primarily by increasing intracellular Van availability. HFn-D93G-Van showed MIC values comparable to those of free Van, whereas its activity against intracellular bacteria was markedly greater. Under matched exposure conditions, the intracellular Van level in RAW264.7 cells was approximately 7.3-fold higher with HFn-D93G-Van than with free Van. Because this measurement represents total intracellular Van, it does not distinguish released drug from Van that remains associated with the nanocage. It nevertheless shows that HFn-D93G substantially increases the amount of Van delivered into macrophages, making improved intracellular drug availability a more likely basis for the enhanced antibacterial effect than any change in the intrinsic activity of Van.

This delivery function was achieved while largely preserving the ferritin architecture. Ferritin provides a well-defined and genetically modifiable nanocage scaffold, whereas conventional loading approaches often rely on pH- or denaturant-induced disassembly and reassembly.^{18–20,22,27} In the present protocol, Van loading was achieved without an imposed nanocage disassembly/reassembly step. Following loading, HFn-D93G retained its nanocage morphology and predominantly α-helical secondary structure, and the thermal unfolding profiles of HFn-D93G and HFn-D93G-Van remained broadly comparable. Particle size increased after loading by both TEM and DLS, although these techniques measure particles under different physical conditions and do not identify the origin of this increase. Hydration, subtle structural rearrangement, and limited interparticle association therefore cannot be distinguished from the present measurements. The hydrodynamic diameter and PDI changed only slightly over 21 days, with no indication of substantial aggregation during this period, although longer-term storage stability remains to be established.

Molecular docking provides additional structural context for Van association with HFn-D93G. The selected pose was located at the external region of the two-fold intersubunit channel, predominantly on the external side of the channel center, with part of the molecule extending toward the channel entrance. Multiple hydrogen-bonding, electrostatic, and hydrophobic interactions were predicted with residues surrounding this region. The selected pose is consistent with a surface-proximal intersubunit interaction, but it does not define the distribution of the entire loaded Van population or demonstrate passage of Van through the ferritin shell. HPLC establishes Van loading experimentally, whereas docking suggests one possible interaction configuration within the selected channel region.

The release data further indicate that nanocage-associated Van remained releasable under the conditions examined. Cumulative release increased progressively as the pH decreased, reaching approximately 27.1%, 67.4%, and 84.7% after 48 h at pH 7.4, 6.0, and 5.0, respectively. These conditions span the neutral extracellular environment and the increasingly acidic conditions encountered during endosomal–lysosomal trafficking.^45,46 In 50% FBS/PBS, approximately 48.3% of the loaded Van was released at 48 h, leaving more than half associated with HFn-D93G-Van under protein-containing conditions. The formulation thus showed a graded pH-dependent release profile rather than an abrupt pH-triggered response. For intracellular delivery, such behavior may help limit rapid extracellular drug loss while permitting greater Van release under acidic intracellular conditions.^9,10,16

The relatively slow release at neutral pH may initially appear inconsistent with the comparable MIC values of HFn-D93G-Van and free Van, but the two assays measure different properties. Dialysis records cumulative drug release under defined physicochemical conditions, whereas the MIC assay determines whether sufficient microbiologically available drug is present to prevent bacterial growth. Complete release of the loaded Van is therefore not required to obtain a comparable MIC. The predicted docking pose suggested that Van may associate with a surface-proximal region of the HFn-D93G nanocage. Although this does not define the distribution of the entire loaded Van population, such a configuration may allow partial accessibility of nanocage-associated Van, while gradual release during incubation may provide an additional source of active drug. The present experiments cannot determine which contribution predominates. More directly, the MIC results show that association with HFn-D93G did not compromise Van activity against the susceptible strains examined.

Alongside these physicochemical properties, HFn-D93G-Van showed acceptable compatibility under the conditions used for the cellular studies. Van loading did not produce an obvious increase in cytotoxicity relative to free Van at matched Van-equivalent concentrations, and HFn-D93G alone was also well tolerated within the corresponding protein concentration range. Cell viability declined with prolonged exposure and at higher concentrations, particularly in HUVECs, indicating that compatibility remained concentration- and exposure-dependent. Hemolysis was minimal under the tested conditions. These observations supported the concentrations used in the subsequent cellular experiments without extending the in vitro results to systemic safety.

Within the compatible concentration range, HFn-D93G-Van showed concentration-dependent cellular uptake in RAW264.7 cells, supporting its enhanced intracellular delivery capability. Uptake of HFn-D93G-Van was strongly reduced at 4 °C, indicating an energy-dependent process. Chlorpromazine and wortmannin markedly reduced uptake, supporting contributions from clathrin-associated endocytosis and PI3K-associated macropinocytic processes, whereas genistein had only a limited effect under the tested conditions. Anti-CD71 antibody reduced uptake to approximately 47% of the untreated control, consistent with the established interaction between H-ferritin and TfR1/CD71.²¹ However, these effects should not be interpreted as independent quantitative fractions of total uptake. Rather, HFn-D93G-Van internalization is likely mediated by multiple endocytic mechanisms rather than a single dominant pathway inferred from inhibitor effects alone.

Once inside RAW264.7 cells, HFn-D93G and HFn-D93G-Van showed comparable lysosome-associated distribution, suggesting that Van loading did not substantially redirect the overall intracellular trafficking of the ferritin scaffold. A clearer difference emerged in their relationship with intracellular bacteria. Across the S. aureus, S. epidermidis, and E. faecium models, HFn-D93G-Van showed greater spatial overlap with intracellular bacteria than empty HFn-D93G, as indicated by confocal imaging, Pearson’s R, tM1, and line-scan analysis. Because AF647 was conjugated to ferritin rather than to Van, this signal reflects the localization of the Van-associated nanocage itself. Van recognizes D-Ala-D-Ala-containing peptidoglycan precursors, and the predicted surface-proximal localization of a fraction of nanocage-associated Van may contribute to its accessibility to Gram-positive bacterial targets.⁴⁹ The imaging data themselves, however, demonstrate enhanced spatial association rather than direct nanocage–bacterium binding or specific bacterial targeting.

The functional consequences of this intracellular delivery were evident in the bacterial clearance assays. HFn-D93G-Van reduced intracellular bacterial burdens by approximately 2.00, 1.62, and 1.76 log₁₀ units in the S. aureus, S. epidermidis, and E. faecium models, respectively, and consistently produced lower intracellular CFU than free Van at matched Van-equivalent concentrations. Empty HFn-D93G showed little antibacterial activity. These findings are consistent with a model in which nanocage uptake increases intracellular Van availability and favors spatial association with intracellular bacteria. The pH-dependent release observed in vitro is also consistent with greater Van liberation under acidic conditions, although Van release from the nanocage was not measured directly inside cells.

The cytokine response broadly paralleled the microbiological changes. HFn-D93G-Van markedly reduced TNF-α, IL-6, and IL-1β across all three macrophage infection models, whereas empty HFn-D93G caused only modest changes and had little effect on intracellular bacterial burden. The parallel reductions in CFU and inflammatory cytokines support an association between improved infection control and a lower inflammatory response. A minor direct effect of the ferritin carrier on macrophage signaling cannot be excluded, but the current data do not indicate that a strong carrier-specific immunomodulatory effect is required to account for the observed cytokine changes.

A similar intracellular advantage was observed in vivo in the murine S. aureus peritonitis-sepsis model. HFn-D93G-Van significantly reduced viable intracellular bacteria recovered from peritoneal macrophages compared with both the infected control and free Van, whereas free Van alone did not significantly reduce the intracellular macrophage burden relative to the infected control. The agreement between the cultured macrophage and peritoneal macrophage results is consistent with the intended role of HFn-D93G in improving cellular access to Van. Intracellular S. aureus reservoirs have similarly been implicated in persistence despite conventional antibiotic exposure, and systemic or abdominal infection models have been used to examine strategies directed at intracellular antibacterial delivery.^4,5,7

At the whole-organ level, however, HFn-D93G-Van and free Van showed comparable antibacterial effects. Both treatments significantly reduced bacterial burdens in the liver, spleen, kidney, and lung, with no significant difference between them. Serum TNF-α, IL-6, and IL-1β were also reduced in both the Van and HFn-D93G-Van groups, and infection-associated histopathological alterations appeared less pronounced in both groups. The additional benefit of HFn-D93G-Van was therefore concentrated in the intracellular macrophage compartment rather than across whole-organ bacterial burdens. This pattern fits a formulation designed to overcome a cellular access barrier while leaving the intrinsic antibacterial potency of Van unchanged. The animal results extend the macrophage findings to an in vivo infection setting without suggesting a general enhancement of Van activity across all systemic endpoints.

This intracellular delivery pattern can be viewed alongside other nanocarrier strategies developed to overcome the limited intracellular availability of Van. Particle-engineered Van formulations, Van-loaded membrane vesicles, pH-responsive polymeric nanoparticles, and liposomal Van have used enhanced cellular uptake or controlled drug release to improve Van activity against intracellular staphylococci.^5,9–11 Other intracellular antibiotic delivery systems, including protein-based nanoparticles, polymersomes, and PLGA nanoparticles, have similarly been developed to increase antibiotic exposure within infected cells.^13,14,50 Ferritin-based delivery offers a distinct protein nanocarrier approach in which cargo loading can be tuned through natural drug-entry channels or engineering of the inner cavity and protein surface.^27–29 HFn-D93G builds on this concept by modifying the ferritin drug-entry region while retaining the assembled protein cage during Van loading. In addition to its structurally defined and genetically modifiable scaffold, the formulation combines the cellular uptake properties of H-ferritin with pH-dependent Van release. Direct quantitative comparison with previously reported Van nanocarrier systems should be interpreted with caution, as carrier composition, bacterial strains, dosing regimens, infection models, and efficacy endpoints vary substantially across studies. Rather than establishing superiority over other platforms, the present findings position HFn-D93G as a distinct protein-based strategy for intracellular Van delivery.

The present in vivo study provides initial validation of the intracellular delivery benefit of HFn-D93G-Van, although several limitations remain. The peritonitis-sepsis model allowed viable intracellular bacteria to be quantified directly from peritoneal macrophages, while organ bacterial burdens, systemic cytokines, and tissue pathology were assessed in the same animals. However, this model does not reproduce localized cutaneous, wound, or implant-associated infection, and the current study does not address pharmacokinetics, quantitative biodistribution, infected-tissue drug accumulation, immunogenicity, or repeated-dose safety. Histological changes were therefore interpreted in the context of acute infection and treatment rather than as a dedicated toxicity assessment. For a protein nanomedicine, reproducible manufacture, formulation consistency, and control of product quality would also be important for subsequent development and regulatory evaluation.⁵¹

The delivery advantage of HFn-D93G does not alter the susceptibility limits of Van itself. Because HFn-D93G-Van retained MIC values comparable to those of free Van, the formulation should not be interpreted as overcoming vancomycin resistance. Instead, its value lies in improving intracellular exposure to susceptible bacteria rather than expanding the antibacterial spectrum of Van. Whether repeated exposure to HFn-D93G-Van alters the selection or emergence of vancomycin resistance was not examined and requires dedicated longitudinal evaluation. Beyond Van, this delivery concept may be applicable to other antibacterial molecules for which intracellular accessibility represents a therapeutic barrier. Structure-based studies have identified candidate compounds directed against S. aureus proteins including FmtA, LLM, and YsxC.^30,52,53 These compounds were not examined here and should not be considered validated cargos for HFn-D93G, but they illustrate alternative antibacterial targets that may be relevant when intracellular delivery is also limiting. Whether a particular molecule can be accommodated by HFn-D93G will depend on its physicochemical properties and requires direct evaluation rather than extrapolation from the present Van formulation.

Conclusion

HFn-D93G improved the intracellular delivery of Van while largely preserving the structural features of the ferritin nanocage and the intrinsic antibacterial activity of Van. HFn-D93G-Van increased intracellular Van accumulation and reduced intracellular S. aureus, S. epidermidis, and E. faecium burdens in RAW264.7 cells. This intracellular advantage was also observed in vivo, where HFn-D93G-Van reduced viable intracellular S. aureus recovered from peritoneal macrophages more effectively than free Van, while Van and HFn-D93G-Van produced comparable reductions in whole-organ bacterial burdens. The principal benefit of HFn-D93G therefore appears to lie in improving the cellular access of Van rather than increasing its intrinsic antibacterial potency. These findings provide proof-of-concept support for HFn-D93G as a structurally defined and genetically modifiable protein-based nanocarrier for intracellular Van delivery. Further preclinical evaluation should address pharmacokinetics, biodistribution, immunogenicity, long-term safety, and efficacy in localized or recurrent infection models.

Funding Statement

This study was funded by the Prevention and Control of Emerging and Major Infectious Diseases-National Science and Technology Major Project(2025ZD01904100), Foundation of State Key Laboratory of Pathogen and Biosecurity of China (Grant No. SKLPBS2417).

Ethical Approval

All animal experiments were reviewed and approved by the Animal Experiment Committee of Laboratory Animal Center, Academy of Military Medical Sciences (AMMS), China (approval no. DWZX-IACUC-2026-318). All animal studies were conducted in accordance with the Chinese Regulations of Laboratory Animals and the Laboratory Animal Requirements of Environment and Housing Facilities, as well as the relevant institutional guidelines for the care and use of laboratory animals.

Author Contributions

Hui Wang and Nianzhi Ning conceived and designed the study, supervised the research, and critically reviewed and revised the manuscript. Binwang Cao and Manxian Wang performed the major experiments, contributed substantially to data acquisition, analysis, and interpretation, and drafted the manuscript. Tao Li, Deyu Li, and Liangyan Zhang contributed to experimental work and data acquisition and participated in reviewing and revising the manuscript. All authors made a significant contribution to the work reported, whether in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; participated in drafting, substantially revising, or critically reviewing the article; agreed on the journal to which the article was submitted; reviewed and approved all versions of the manuscript before submission and during revision, and agree to approve the final version accepted for publication and any significant changes introduced at the proofing stage; and agree to take responsibility and be accountable for the contents of the article.

Disclosure

The authors declare no competing interests.