Levofloxacin-Catalase-PLGA Nanoparticles Potentiate Sonodynamic Therapy to Alleviate BCG-Induced Tuberculous Granulomas via PI3K/Akt/VEGF Pathway Suppression in Rats

Abstract

Purpose

Hypoxic tuberculous granulomas induce vascular dysfunction and severely impede drug penetration. This study aimed to elucidate the molecular mechanism by which sonodynamic therapy (SDT) induces vascular normalization and synergistic immunomodulatory effects.

Methods

Levofloxacin and catalase co-loaded PLGA nanoparticles (LEV@CAT-NPs) were prepared and characterized. Bacillus Calmette-Guérin (BCG) was used as a surrogate for virulent Mycobacterium tuberculosis due to its high biosafety. Therapeutic efficacy and mechanisms were evaluated in BCG-induced 3D tuberculous spheroids and a rat subcutaneous granuloma model via transcriptomic sequencing, Western blot, immunofluorescence, pharmacokinetic analysis, and bacterial load quantification.

Results

LEV@CAT-NPs exhibited excellent stability and oxygen-generating capacity. SDT effectively alleviated intralesional hypoxia and downregulated HIF-1α and VEGF expression, promoting vascular normalization — evidenced by reduced vascular density, increased pericyte coverage, decreased leakage, improved perfusion, and enhanced tight junctions. Mechanistically, SDT inhibited the PI3K/Akt/VEGF pathway, reduced macrophage VEGF secretion, and induced macrophage apoptosis, thereby suppressing abnormal angiogenesis and improving drug penetration. The treatment also activated innate and adaptive immune pathways, promoting CD3⁺ and CD8⁺ T cell infiltration.

Conclusion

The SDT-driven mechano-chemical synergistic strategy effectively induces vascular normalization in tuberculous granulomas, improves drug delivery, and offers a promising approach for treating drug-resistant tuberculosis through microenvironment remodeling.

Graphical Abstract

Introduction

Tuberculosis (TB), caused by Mycobacterium tuberculosis (Mtb), remains one of the deadliest infectious diseases globally, and the emergence of drug-resistant TB has further complicated an already challenging prevention and control landscape.^1,2 Tuberculosis typically presents with symptoms including fever, cough, and hemoptysis. For drug-sensitive TB, the standard four-drug regimen requires a treatment course of at least 6 months, while drug-resistant TB necessitates treatment for 18 to 24 months or longer.³

The unique structure of tuberculous lesions creates multiple physiological barriers, including a dense extracellular matrix, abnormal vasculature, a hypoxic microenvironment, and acidic pH within the lesion. These factors result in highly heterogeneous drug distribution within lesions, where local drug concentrations frequently fall below the minimum inhibitory concentration (MIC).^4,5 This exacerbates treatment failure and promotes drug-resistant strains, while also compromising patient adherence due to severe adverse effects. Therefore, developing novel strategies to effectively enhance drug delivery and surmount the physiological barriers within lesions represents a core challenge in combating the tuberculosis epidemic.

The tuberculous granuloma formed after Mtb infection is the key pathological basis of the treatment dilemma. Studies show that the three-dimensional (3D) microanatomical environment of the granuloma creates physical barriers through necrotic centers and fibrotic regions, limiting drug distribution.^6,7 This structure not only physically hinders drug penetration via dense encapsulation and abnormal vasculature (central ischemia and peripheral leakage) but also fosters a severely hypoxic microenvironment at its core. Hypoxia promotes the release of inflammatory cytokines (eg, TNF-α, IL-6) via transcription factors such as HIF-1α while suppressing antibacterial immune responses, leading to expanded necrotic areas and chronic inflammation.⁸ Meanwhile, hypoxia drives high expression of VEGF by stabilizing HIF-1α. The upregulated VEGF further induces abnormal angiogenesis characterized by distorted vessel structure, incomplete pericyte coverage, and uneven permeability, thereby perpetuating abnormal angiogenesis and consolidating the drug delivery barrier.⁹ Thus, the granuloma structure essentially forms a pharmacological sanctuary that frequently compromises the efficacy of standard chemotherapy.

As the core mediator of the self-reinforcing cycle between hypoxia and aberrant vasculature, the HIF-1α/VEGF axis represents a promising interventional target.^10,11 Although VEGF-targeted anti-angiogenic agents (eg, bevacizumab) have been evaluated in preclinical models to induce “normalized” vasculature, their clinical benefit is uniformly short-lived, partial vascular remodeling is rapidly followed by excessive vascular pruning that further compromises tissue perfusion and amplifies intralesional hypoxia.¹² Therefore, achieving long-lasting and safe reversal of vascular abnormalities while simultaneously ameliorating the hypoxic and immunosuppressive microenvironment remains an unresolved challenge in the field.

Sonodynamic therapy (SDT) is a non-invasive treatment technique that uses low-intensity ultrasound (US) to activate sonosensitizers, generating reactive oxygen species (ROS). SDT offers deep tissue penetration, spatiotemporal controllability, and synergistic bactericidal and microenvironment-remodeling effects, showing promise in deep drug-resistant infections such as osteomyelitis and diabetic foot.^13–15 However, SDT is inherently oxygen-consuming, whereas the hypoxic core of tuberculous granulomas has extremely low O₂ tension,^16,17 creating a “hypoxia-treatment oxygen consumption” paradox that limits bactericidal efficiency and microenvironment repair.

To simultaneously address SDT’s oxygen dependence, induce vascular normalization, and reverse the immunosuppressive microenvironment, we developed a multifunctional PLGA-based nanoplatform: LEV@CAT-NPs, co-loaded with catalase (CAT) and levofloxacin (LEV) an anti-TB agent with intrinsic sonosensitizer activity. We hypothesize that under US activation, CAT first catalyzes endogenous hydrogen peroxide in the granuloma to generate oxygen in situ, breaking the “hypoxia-treatment oxygen consumption” vicious cycle. The in situ–generated oxygen then enhances the ROS production of LEV to exert direct bactericidal effects, while alleviating hypoxia to remodel the microenvironment and promote vascular normalization.

Bacillus Calmette-Guérin (BCG) is an attenuated Mycobacterium bovis strain sharing 99.9% genetic homology with Mtb, along with conserved pathogenic traits including slow growth, similar cell wall architecture, and latent infection phenotypes. Its superior biosafety makes BCG a valid surrogate host for TB pathogenesis research¹⁸ and has established it as the most widely used model strain for preclinical TB research.^19,20 Macrophages, as key effector cells in TB immunopathology, orchestrate in the tuberculous granuloma microenvironment. We developed a macrophage-derived 3D in vitro BCG tuberculoma model, which recapitulates key features of in vivo granulomas (hypoxia, acidosis, and lipid deposition), along with a rat subcutaneous BCG granuloma model. Using this paired model system, we demonstrated that in 3D tuberculous spheroids, US + LEV@CAT-NPs treatment downregulates the HIF-1α/VEGF axis and inhibits PI3K/Akt phosphorylation to promote vascular normalization and boost drug penetration. In the rat subcutaneous granuloma model, this strategy further drove effector T-cell infiltration, and reversal of the immunosuppressive microenvironment, achieving synergistic bacterial clearance and tissue repair. This work provides a translationally promising TB therapeutic strategy, revealing that oxygen-targeted interventions can reverse infection-related vascular abnormalities. These findings may offer a generalizable paradigm for hypoxia-associated mycobacterial infectious diseases. Additionally, the 1.0 MHz ultrasound frequency commonly used in preclinical SDT has a limited penetration depth of approximately 3–5 cm in soft tissue due to frequency-dependent attenuation, with significant acoustic energy loss in air-rich lung tissue, which may restrict direct translation to deep-seated tuberculous lesions.^21–23 Nevertheless, the subcutaneous BCG-induced granuloma model employed here provides a valid platform for mechanistic validation in superficial lesions.

Materials and Methods

All materials used in this study are listed in the Supplementary Material and Methods.

The Preparation and Evaluation of Drug-Loaded PLGA Nanoparticles

PLGA-PEG-COOH-based nanoparticles (Blank-NPs, LEV-NPs, CAT-NPs, and LEV/CAT co-loaded LEV@CAT-NPs) were prepared via the double emulsion method according to our previously published protocol.²⁴

The morphology of LEV@CAT-NPs was characterized by transmission electron microscopy (TEM, Hitachi, Japan), and average particle size, zeta potential and polydispersity index (PDI) were measured using a Malvern Zetasizer 3000HS system. The 7-day storage stability in PBS, drug loading capacity, encapsulation efficiency, and successful encapsulation of active ingredients were evaluated following standard nanomaterial characterization protocols. The drug loading capacity (LC%) and encapsulation efficiency (EE%) of LEV and CAT in LEV@CAT-NPs were determined using ultraviolet spectrophotometry, and calculated as follows:

LC (% w/w) = [Mass of drug in NPs/Total mass of NPs] × 100%;

EE (% w/w) = [Mass of drug in NPs/Total mass of drug used in encapsulation] × 100%.

A series of experiments was conducted to further evaluate the nanoparticles’ functional properties, focusing on drug release, enzymatic catalysis, oxygen generation, protein protection (anti-proteolysis), and sonodynamic efficacy (¹O₂ generation). Detailed experimental and analytical protocols are provided in the Supplementary Information.

Ultrasound Irradiation Protocol

Ultrasound parameters (1 MHz, 1 W/cm², 50% duty cycle, 5 min total time) and irradiation procedures under both in vitro and in vivo conditions are detailed in the Supplementary Information.

BCG Strain, Cell Lines, and Culture Conditions

Details regarding the BCG strain, cell lines, and culture conditions are provided in the Supplementary Information.

3D Tuberculous Spheroid Construction

The in vitro 3D tuberculous spheroids were constructed using Aggrewell™ 400 24-well plates (Stemcell Technologies). Briefly, RAW 264.7 cells were mixed with BCG at a MOI of 5 (1×10⁶ cells/mL, 5×10⁶ CFU/mL) and seeded into the plates. Compact spheroids with diameters of 100–200 μm were obtained after 48 h of culture. Uninfected cell spheroids prepared under identical conditions.

The in vivo Studies Using Animal Model

All animal protocols were approved by the Laboratory Animal Ethics Committee of Chongqing Medical University (Approval No. IACUC-CQMU-2023-0428) and carried out in compliance with the Guideline for Ethical Review of Laboratory Animal Welfare (GB/T 35892–2018) and the Regulations on the Administration of Laboratory Animals of the People’s Republic of China. Female Sprague Dawley (SD) rats aged six to eight weeks (240–280 g) were utilized to establish the granuloma model. After 3 consecutive days of pre-treatment with dexamethasone (2 mg/kg/d), 1 mL of BCG suspension (1×10⁹ CFU/mL) was injected subcutaneously into the right hip. Successful model formation was confirmed on day 8 post-injection. After the treatment was completed, SD rats were euthanized by intraperitoneal injection of sodium pentobarbital, and tissue specimens were collected.

Experimental Groups and Treatment Protocols

Based on our previous study,²⁴ in which the biosafety profile of LEV@CAT-NPs and the ultrasound parameters were thoroughly characterized, including near-infrared fluorescence imaging for biodistribution, Hematoxylin‑and‑Eosin(H&E) staining of major organs, serum biochemical analysis (ALT, AST, BUN, Cr), in vitro hemolysis and cytotoxicity assays, and ultrasound safety validation. The experimental design in the present study was optimized. Interventions that had previously been proven ineffective (such as ultrasound alone, free CAT, and free LEV) were omitted to streamline the experimental groups, improve efficiency, and comply with animal ethics. Furthermore, because the formulation, dosage, and administration route were unchanged between the two studies, the present study focused on therapeutic efficacy and mechanistic characterization.

For in vitro experiments, 3D tuberculous spheroids were divided into five treatment groups: ① Control: cultured in DMEM medium only, without ultrasound or drug intervention; ②LEV-NPs: LEV-NPs (LEV concentration at 4.0 μg/mL); ③LEV-NPs + US: LEV-NPs for 4 h, followed by ultrasound (1.0 MHz, 1.0 W/cm², 50% duty cycle, 5 min); ④LEV@CAT-NPs: LEV@CAT-NPs (LEV 4.0 μg/mL + CAT 4.0 μg/mL²⁵); ⑤ LEV@CAT-NPs + US: LEV@CAT-NPs for 4 h, followed by ultrasound (same parameters as above). All groups were then incubated for 24 h.

For animal experiments, modeled rats were randomly divided into five groups (n = 5 per group) and treated via tail vein injection (5.0 mg/kg LEV,²⁴ 5.0 mg/kg CAT²⁶,300 μL per injection, every other day): ① Control: saline (equal volume); ② LEV-NPs: LEV-NPs injection; ③ LEV-NPs + US: LEV-NPs injection, followed by ultrasound at the lesion site (same parameters, 5 min) 24 h later; ④ LEV@CAT-NPs: LEV@CAT-NPs injection; ⑤ LEV@CAT-NPs + US: LEV@CAT-NPs injection, followed by ultrasound (same parameters, 5 min) 24 h later.

Drug Penetration Assessment

FITC-labeled rifampin (Ex/Em = 495/519 nm) was used as a fluorescent tracer to evaluate drug penetration in both 3D tuberculous spheroids and in vivo granuloma tissues. For in vitro assessment, spheroids were incubated with FITC-rifampin (4 μg/mL) for 4 h after different interventions, and the fluorescence distribution across spheroid layers was visualized by CLSM. For in vivo validation, FITC-rifampin (10 mg/kg) was administered via the tail vein after intervention, and granuloma tissues were collected 1 h post‑injection for frozen sectioning and CLSM imaging.

Rifampin Concentration and Pharmacokinetic Analysis

Rifampin concentrations in tuberculous granuloma tissues after different treatments was quantified by high-performance liquid chromatography-tandem mass spectrometry (HPLC-MS/MS; Agilent 1290–6460, USA), with diphenhydramine (1 μg/mL) as the internal standard, using the standard curve method. Pharmacokinetic parameters were calculated using a two-compartment model in DAS 2.0 software.

Colony-Forming Unit (CFU) Assay

Protocols for quantifying bacterial load in 3D tuberculous spheroids and in resected granuloma tissues are provided in the Supplementary Information.

Preparation of Conditioned Medium (CM)

After different treatments of the tuberculoma, the culture medium was collected and centrifuged at 3000 rpm for 10 min to remove cell debris. The supernatant was then filtered through a 0.22 μm filter. The filtrate was mixed with DMEM at a 2:1 (v/v) ratio to obtain conditioned medium. These conditioned media were designated as follows: CM-control, CM-LEV-NPs, CM-LEV@CAT-NPs, CM-US + LEV-NPs, and CM-US + LEV@CAT-NPs, corresponding to the control, LEV-NPs, LEV@CAT-NPs, US + LEV-NPs, and US + LEV@CAT-NPs treatments, respectively.

Tube Formation Assay

Pre-chilled Matrigel (Beyotime, China) was added to 24-well plates at 120 μL/well and polymerized at 37°C for 30 min. HUVECs were detached with 0.25% trypsin (Gibco, USA), resuspended in the corresponding conditioned medium, and seeded onto Matrigel-coated plates at 1×10⁵ cells/well (500 μL per well). After 6 h of incubation at 37°C with 5% CO₂, tube formation was visualized under an inverted microscope (Echo, USA), and tube length, and coverage area were quantified using the ImageJ Angiogenesis Analyzer plugin.

In vitro Vascular Permeability Assay

HUVECs (2 × 10⁵ cells) were seeded into the upper chamber of a Transwell insert. The lower chamber was filled with 500 μL of DMEM supplemented with 10% FBS. After overnight incubation at 37°C under 5% CO₂, the culture medium was removed and replaced with conditioned medium for another 24 h.

Once a confluent monolayer had formed, the liquid was removed, and 300 μL of DMEM was added to the lower chamber. The cell monolayer was gently washed twice with pre-warmed serum-free DMEM. FITC-dextran (40 kDa, 1 mg/mL, 150 μL) was then added to the upper chamber. The plate was kept in the dark and incubated at 37°C for 1 h.

After incubation, 100 μL of medium from the lower chamber was collected and transferred to a light-shielded 96-well plate. Vascular permeability was calculated using the following formula:²⁷

P = (C_t – C₀) × V/(A × t × C₀), where:

C_t: fluorescence intensity in the lower chamber after 1 h of incubation;

C₀: initial fluorescence intensity (at time zero);

V: volume of the solution in the lower chamber (μL);

A: surface area of the upper chamber membrane (cm²);

t: permeability time (h).

TEM Examination of Endothelial Intercellular Junctions

Vascular tissues isolated from rat subcutaneous granulomas were fixed in 2.5% glutaraldehyde and processed according to standard TEM sample preparation protocols. Ultrathin sections were double-stained with uranyl acetate and lead citrate and then observed under a transmission electron microscope (JEM-1400PLUS, Hitachi High-Technologies, Japan).

Tissue Hypoxia Assessment by Pimonidazole Staining and Photoacoustic Imaging

Hypoxia was assessed by both pimonidazole immunofluorescence staining and photoacoustic imaging. For immunofluorescence: 60 mg/kg pimonidazole was injected intravenously into rats on day 9 post-treatment, and rats were euthanized 1 h later. Paraffin sections of granuloma tissues were incubated overnight with a rabbit anti-pimonidazole primary antibody (PAb2627AP, Hypoxyprobe, Inc., USA) for staining. For photoacoustic imaging, photoacoustic signals from the granuloma region were acquired using a photoacoustic imaging system (VisualSonics, Canada) for quantitative analysis.

Immunofluorescence Staining of Tuberculous Spheroids

Treated tuberculous spheroids were fixed with 4% paraformaldehyde and processed according to standard immunofluorescence protocols. The only modification was that the primary antibody incubation at room temperature for 2 h (instead of conventional 4°C overnight incubation). After secondary antibody incubation and DAPI counterstaining, samples were imaged using a confocal laser scanning microscope (CLSM; Nikon, Japan).

Immunofluorescence Staining and EdU Proliferation Assay

HUVECs were seeded in 15 mm confocal dishes and incubated with conditioned medium for 24 h. Following standard immunofluorescence staining protocols, cells were incubated overnight at 4°C with primary antibodies targeting VEGF, tight junction markers (ZO-1, Occludin), the proliferation marker Ki-67, and the inflammatory cytokine IL-33, and then visualized by CLSM. Cell proliferation was further assessed by EdU assay according to the manufacturer’s protocol, details are provided in the Supplementary Information.

IHC and Immunofluorescent Staining of Tissue Sections

Paraffin sections of tuberculous granuloma tissues were subjected to immunohistochemical (IHC) and immunofluorescence (IF) co-localization analysis. For IHC, sections were incubated overnight with primary antibodies against HIF-1α, VEGF, and VEGFR. For IF, two staining panels were used: CD31 co-stained with α-SMA, and CD31 alone.

Western Blot Analysis

Total protein was extracted from tuberculous spheroids and quantified by BCA assay, then separated on 10% or 12.5% SDS-PAGE gels and transferred to PVDF membranes according to standard WB protocols. Membranes were incubated overnight at 4°C with primary antibodies against PI3K (CST, USA, 1:1000), p-PI3K (CST, USA, 1:1000), Akt (CST, USA, 1:1000), p-Akt (CST, USA, 1:1000), HIF-1α (Abcam, UK, 1:1000), VEGF (Proteintech, China, 1:1000), Bax (Proteintech, China, 1:1000), and Bcl-2 (Proteintech, China, 1:1500). After incubation with an HRP-conjugated secondary antibody (1:5000), bands were visualized with an ECL-Plus kit and imaged using the Azure Biosystems C400 system.

ELISA

Levels of VEGF, IL-6, TNF-α, IL-10, and TGF-β in the samples were quantified using commercial ELISA kits according to the manufacturers’ instructions. Absorbance was read at 450 nm using a microplate reader.

Quantitative Real-Time PCR (qRT-PCR)

Total RNA was extracted from cell spheroids, tuberculous spheroids and tissue samples, and reverse transcribed into cDNA for qRT-PCR analysis. The relative mRNA expression of target genes was calculated using the 2-^ΔΔCt method, with rat β-actin or mouse Gapdh as internal references. Primer sequences are listed in Table S1.

Transcriptomic Sequencing and Bioinformatic Analysis

Transcriptomic sequencing was performed on 3D tuberculous spheroids (Illumina platform, aligned to mouse reference genome GRCm38) and rat subcutaneous granuloma tissues (BGISEQ-500 platform, aligned to rat reference genome Rn6), respectively. Standard transcriptomic analysis workflows were used for differential gene expression screening, GO/KEGG functional enrichment, and unsupervised clustering with visualization (heatmap and PCA). Functional annotation of rat tissue samples was performed using the BGI Dr. Tom system.

Statistical Analysis

Experimental data are presented as the mean ± standard deviation (Mean ± SD). Statistical analysis and graph plotting were performed using GraphPad Prism 8 (GraphPad Software; La Jolla, CA, USA). Each experiment was independently repeated in triplicate. Comparisons between two groups were analyzed using Student’s t-test, while differences among multiple groups were assessed by one-way ANOVA. P value < 0.05 was considered statistically significant. Significance levels in the figures are denoted as follows: * p < 0.05, ** p < 0.01, and *** p < 0.001.

Results

Synthesis, Characterization, and in vitro Release Profile of LEV@CAT-NPs

As observed by TEM, the LEV@CAT-NPs exhibited a spherical shape with uniform size and good dispersibility (Figure 1A). Dynamic light scattering (DLS) showed that the mean hydrodynamic diameter of LEV@CAT-NPs was 328.9 ± 11.4 nm (Figure 1B). As shown in Supplementary Table S2, PDI of LEV@CAT-NPs was 0.173 ± 0.017, indicating good homogeneity. The zeta potential of LEV@CAT-NPs was −10.4 ± 0.493 mV (Figure 1C). The nanoparticles showed no significant change in particle size after seven days of storage in PBS (Figure 1D). Furthermore, the absorption spectra of free LEV, free CAT, LEV-NPs, CAT-NPs, and LEV@CAT-NPs were measured using a UV-Vis spectrophotometer across 250–500 nm (Figure 1E). LEV@CAT-NPs exhibited maximum absorption peaks at 280 nm and 303 nm, confirming the successful loading of both LEV and CAT into the nanoparticles. The drug loading capacities were 3.66 ± 0.04% for LEV and 3.61 ± 0.08% for CAT, respectively. Under ultrasound triggering, the cumulative drug release of LEV and CAT from LEV@CAT-NPs reached 69.18% and 70.37%, respectively, within 72 h (Figure 1G). Thus, LEV@CAT-NPs possessed desirable physicochemical properties and controlled release, making them suitable for subsequent experiments.

Figure 1.

Synthesis and characterization of LEV@CAT-NPs. (A) TEM image of LEV@CAT-NPs (scale bar, 200 nm); (B) DLS size distribution of LEV@CAT-NPs; (C) Zeta potential of different formulations (pH 7.4 PBS); (D) Colloidal stability of LEV@CAT-NPs in pH 7.4 PBS over 7 days.; (E) UV-Vis absorption spectra of free drugs and nanoparticle formulations (250–500 nm); (F) Dissolved oxygen generation of different formulations in H₂O₂ solution; (G) 72 h cumulative release of LEV and CAT from LEV@CAT-NPs with or without ultrasound (US) irradiation; (H) Initial CAT activity of different formulations (ammonium molybdate assay, n = 3); (I) Relative CAT activity after proteinase K treatment (n = 3). Relative CAT activity was normalized to the initial activity at 0 h (set as 100%). Data are presented as mean ± SD.

Oxygen Generation and Enzymatic Stability of LEV@CAT‑NPs

The oxygen generation capacity of LEV@CAT-NPs was determined by measuring the dissolved oxygen concentration over time using a portable dissolved oxygen meter. The oxygen production of free LEV and LEV-NPs was similar to that of the blank control group, indicating that neither formulation could catalyze the decomposition of H₂O₂ into O₂. In contrast, free CAT, CAT-NPs, and LEV@CAT-NPs were able to generate significantly more O₂ (Figure 1F). LEV@CAT-NPs exhibited enhanced acoustic sensitivity, generating significantly more singlet oxygen (¹O₂) under US excitation (Figure S1).

The ammonium molybdate assay demonstrated that CAT-NPs and LEV@CAT-NPs retained detectable catalase activity (Figure 1H). CAT stability against proteinase K degradation was evaluated using the ammonium molybdate assay. As shown in Figure 1I, free CAT retained only 24.20 ± 2.29% of its initial activity after 24 h of proteinase K treatment, while LEV@CAT-NPs retained 56.27 ± 1.63% of CAT activity (p < 0.001). This result is consistent with the protective effect of nanocarriers reported in the literature,²⁸ confirming that encapsulation within the nanoparticle matrix significantly improves CAT resistance to proteolytic degradation.

BCG-Induced Granuloma-Like Structures Activate HIF-1α/VEGF Signaling and Elevate Endogenous H₂O₂ Levels

Time-lapse imaging showed that macrophages began to aggregate within 24 h after BCG infection, forming compact tuberculous spheroids with a mean diameter of 142.86 ± 66.84 μm by 48 h (Figure 2A). Hypoxia staining confirmed that the BCG-infected spheroids exhibited a significantly more hypoxic microenvironment than uninfected control spheroids. Consistent with this, immunofluorescence staining showed markedly upregulated HIF-1α and VEGF fluorescence intensity in the tuberculous spheroids (Figure 2B). H₂O₂ content in the spheroids reached 17.62 ± 0.89 μM, an 88.72% increase relative to the control group (p < 0.001; Figure 2C).

Figure 2.

Microenvironmental features of 3D tuberculous spheroids. (A) Bright-field imaging of spheroid formation (scale bars: 360 μm (first three panels), 20 μm (right)); (B) Confocal immunofluorescence of HIF-1α and VEGF in control and tuberculous spheroids (scale bar: 50 μm); (C) Intracellular H₂O₂ levels (n = 3); (D and E) qRT-PCR quantification of HIF-1α (D) and VEGF (E) mRNA expression in tuberculous spheroids and cell spheroids (n = 3); (F and G) Western blot (F) and grayscale analysis (G) of HIF-1α and VEGF protein levels by Image J (n = 3). *p < 0.05,**p < 0.01,***p < 0.001.

At the molecular level, quantitative real-time qRT-PCR showed that HIF-1α and VEGF mRNA expression was upregulated by 1.29-fold (P < 0.01) and 5.53-fold (P < 0.001), respectively, relative to controls (Figures 2D and E). Western blot analysis confirmed corresponding increases in protein levels of 1.44-fold and 1.72-fold, respectively (Figures 2F and G). Collectively, these results demonstrate that BCG-induced 3D tuberculous spheroid model recapitulates core pathological features of in vivo TB granulomas, including a hypoxic microenvironment, activated HIF-1α/VEGF signaling, and elevated oxidative stress.

US + LEV@CAT-NPs Downregulate HIF-1α and VEGF Expression in Tuberculous Spheroids

Immunofluorescence revealed intense HIF-1α (red) and VEGF (green) signals in the untreated group, whereas these signals were markedly reduced in the group receiving US with LEV@CAT-NPs (Figure 3A and B). qRT-PCR confirmed that HIF-1α and VEGF mRNA levels in 3D tuberculous spheroids were significantly reduced after the combined treatment (P < 0.001 vs control; Figure 3C and D). Consistent with the reduced protein expression, VEGF secretion into the cell culture supernatant was also significantly suppressed in the co-treatment group (Figure 3E). Taken together, these findings indicate that the application of US with LEV@CAT-NPs alleviates hypoxia in tuberculous spheroids, leading to downregulation of HIF-1α and subsequent inhibition of its downstream target VEGF at both the mRNA and protein levels.

Figure 3.

US + LEV@CAT-NPs treatment downregulates HIF-1α/VEGF expression in BCG-induced tuberculous spheroids. (A and B) Immunofluorescence images of HIF-1α (A) and VEGF (B) in tuberculous spheroids following various treatments; scale bar: 100 μm. (C and D) Relative mRNA levels of HIF-1α and VEGF in tuberculous spheroids across different groups (n = 3). (E) Secreted VEGF concentration in the culture supernatant of tuberculous spheroids (n = 3).***P < 0.001.

Synergistic Effects of US + LEV@CAT-NPs on Macrophage Polarization, Microenvironment Remodeling, and Bacterial Clearance in a 3D Tuberculoma Model

We first assessed macrophage polarization after different treatments: qRT-PCR showed that US + LEV@CAT-NPs co-treatment significantly upregulated mRNA levels of M1 markers iNOS (3.73-fold) and CD86 (5.09-fold), while downregulating M2 markers Arg-1 (to 92.19% of control levels) and CD206 (to 83.51% of control levels) (all P < 0.001 vs untreated control, Figure 4A), indicating robust polarization toward the pro-inflammatory M1 phenotype. Consistently, ELISA results showed that the co-treatment significantly increased the secretion of pro-inflammatory cytokines (IL-6, TNF-α) and reduced the levels of anti-inflammatory cytokines (IL-10, TGF-β) in the culture supernatant (Figure 4B), confirming reprogramming of the immunosuppressive tuberculoma microenvironment toward a pro-inflammatory state.

Figure 4.

Synergistic effects of US + LEV@CAT-NPs on macrophage polarization, apoptosis induction, and intracellular BCG clearance in 3D tuberculous spheroids.(A) mRNA expression levels of M1 marker (iNOS, CD86) and M2 marker (Arg-1, CD206) determined by qRT-PCR (n = 3). (B) The secretion levels of pro-inflammatory cytokines (TNF-α, IL-6) and anti-inflammatory cytokines (IL-10, TGF-β) in the culture supernatant measured by ELISA. (C and D) Analysis of apoptosis-related factors. (C) Protein expression levels of the pro-apoptotic protein Bax and the anti-apoptotic protein Bcl-2 assessed by Western blot, with the Bax/Bcl-2 ratio quantified. (D) mRNA expression levels of Bax and Bcl-2 detected by qRT-PCR (n = 3). (E) Photographs of BCG growth within 3D tuberculous spheroids on 7H10 agar plates (left) and quantification of colony counts (right) (n = 3). **P < 0.01,***P < 0.001.

We further evaluated the pro-apoptotic effects of the co-treatment: Western blot showed that the co-treatment upregulated the pro-apoptotic Bax protein and downregulated the anti-apoptotic protein Bcl-2, leading to a significantly elevated Bax/Bcl-2 ratio (Figure 4C). This trend was consistent with mRNA-level changes (Figure 4D), confirming that the co-treatment induced apoptosis of BCG-infected macrophages.

Finally, intracellular BCG burden was measured by CFU assay. Compared to the control, treatment with LEV-NPs + US reduced BCG load by 66.61%, while US + LEV@CAT-NPs further reduced bacterial load by 90.08% (Figure 4E).

In summary, the combination of US and LEV@CAT-NPs provides a solid experimental basis for enhancing anti-tuberculosis immunotherapy strategies by synergistically promoting M1 macrophage polarization, shaping a pro-inflammatory microenvironment, and inducing apoptosis.

US + LEV@CAT-NPs Inhibit Angiogenesis in 3D Tuberculous Spheroids via the PI3K/Akt Signaling Pathway

To explore the molecular mechanism underlying the anti-angiogenic effect of US + LEV@CAT-NPs, we performed RNA-seq-based transcriptomic analysis of tuberculous spheroids before and after co-treatment. A total of 364 differentially expressed genes (DEGs) were identified, among which 315 genes were downregulated and 49 genes were upregulated (|log₂FC| ≥ 0.5). KEGG enrichment analysis showed that these DEGs were significantly enriched in the PI3K/Akt signaling pathway (Figure 5A). Based on these findings and previous literature,²⁹ we hypothesized that the PI3K/Akt signaling pathway plays a key regulatory role in angiogenesis within tuberculous granulomas.

Figure 5.

Transcriptomic and functional validation that US + LEV@CAT-NPs suppress VEGF expression via PI3K/Akt axis inhibition. (A) KEGG enrichment plot from RNA-seq analysis identifying the PI3K/Akt pathway as one of the top enriched pathways; (B and C) Western blot analysis (B) and its quantitative results (C) confirming the downregulation of HIF-1α, VEGF, and reduced phosphorylation levels of PI3K and Akt; (D and E) Pharmacological inhibition with LY294002 demonstrating that VEGF suppression is dependent on PI3K/Akt signaling, as shown by representative blots (D) and quantitative analysis of p-PI3K/PI3K, p-Akt/Akt ratio and secreted VEGF levels (E). *P < 0.05, **P < 0.01, ***P < 0.001.

Western blot results further validated this hypothesis: untreated tuberculous spheroids exhibited significantly elevated phosphorylated-to-total protein ratios of PI3K (p-PI3K/PI3K) and Akt (p-Akt/Akt), as well as higher HIF-1α and VEGF protein levels, compared to uninfected control spheroids. These effects were markedly reduced after US + LEV@CAT-NPs co-treatment, showing a significantly stronger inhibitory effect than the US + LEV-NPs (CAT-free) control group (p < 0.01; Figure 5B and C). To verify pathway specificity, we treated the tuberculous spheroids with the PI3K inhibitor LY294002. As shown in Figure 5D and E, LY294002 alone reduced the protein expression of p-PI3K/PI3K, p-Akt/Akt, and VEGF, while combined treatment with LY294002 and US + LEV@CAT-NPs further suppressed these proteins.

In summary, these results demonstrate that US combined with LEV@CAT-NPs inhibit angiogenesis in 3D tuberculous spheroids by suppressing the PI3K/Akt signaling pathway and downregulating the expression of HIF-1α and VEGF.

Conditioned Medium from Tuberculous Spheroids Treated with US + LEV@CAT-NPs Inhibits Angiogenesis via VEGF Downregulation

To test the hypothesis that US combined with LEV@CAT-NPs inhibits angiogenesis by modulating the tuberculous spheroid microenvironment, we cultured HUVECs with conditioned medium (CM) collected from tuberculous spheroids after different treatments. The CM preparation procedure is illustrated in Figure 6A. Compared with CM-control, HUVECs treated with CM-US + LEV@CAT-NPs showed significantly impaired tube-forming ability. Specifically, tube formation was disrupted (Figure 6B), with a 39.22% decrease in total tube length (Figure 6D), 71.17% reduction in coverage area (Figure 6E), 53.3% decrease in junction number, and 68.29% reduction in mesh count. VEGF immunofluorescence staining confirmed that the fluorescence intensity of the CM-US + LEV@CAT-NPs group was significantly lower than that of the control group (Figure 6C), with only 10.40% of the control level (Figure 6F).

Figure 6.

Conditioned medium from US + LEV@CAT-NPs-treated tuberculous spheroids inhibits angiogenesis in HUVECs.(A) Schematic diagram of the experimental workflow for evaluating the effects of conditioned media (CM) from US + LEV@CAT-NPs-treated tuberculous spheroids on HUVEC angiogenesis; (B) Representative images of tube formation by HUVECs cultured with different CM; scale bar: 50 μm; (C) Immunofluorescence staining of VEGF in HUVECs; scale bar: 20 μm; (D and E) Quantitative analysis of total tube length (D) and coverage area (E) for HUVEC tube‑formation assays (n = 3). Both tube length and coverage area were normalized to the CM‑Control group (set as 100%). Data are shown as mean ± SD. (F) qRT-PCR analysis of VEGF mRNA levels in HUVECs (n = 3). * P < 0.05, ***P < 0.001.

In summary, US combined with LEV@CAT-NPs inhibits in vitro angiogenesis by modulating the tuberculous spheroid microenvironment and downregulating VEGF expression.

LEV@CAT-NPs with Ultrasound Promote Vascular Normalization by Inducing a Quiescent Endothelial Phenotype

To evaluate the effect of LEV@CAT-NPs under US irradiation on endothelial cell function within the tuberculous granuloma microenvironment, we treated HUVECs with CM obtained from tuberculous spheroids subjected to different treatments. Immunofluorescence results revealed that, compared to CM-control and other CM groups, HUVECs treated with CM from the LEV@CAT-NPs + US group showed significantly upregulated expression of the tight junction protein ZO-1 (Figure 7A). Furthermore, this CM significantly inhibited HUVEC proliferation, as confirmed by Ki-67 (Figure 7B) and EdU assays (Figure 7C). These results indicate that US combined with LEV@CAT-NPs enhance endothelial barrier function and suppress excessive endothelial proliferation in vitro.

Figure 7.

Effects of different treatments on the expression of quiescence markers, proliferation, and vascular permeability in HUVECs and on the tight junction of vascular endothelial cells in tuberculous granuloma. (A–D) Representative immunofluorescence images of HUVECs in each treatment group (nuclei stained with DAPI, blue): (A) Tight junction protein ZO-1 (red), scale bar: 25 μm; (B) Proliferation marker Ki-67 (green), scale bar: 25 μm; (C) EdU-labeled proliferating HUVECs (red); scale bar: 50 μm; (D) IL-33 immunofluorescence (red); scale bar: 25 μm; (E and F) Immunofluorescence images of ZO-1 (green) and Occludin (red) in tissue sections; scale bar: 50 μm; (G) TEM image of endothelial cell tight junctions, 20,000× magnification; (H) Quantification of in‑vitro endothelial barrier permeability using 40‑kDa dextran across confluent CM-treated HUVECs monolayers in a Transwell system; (I and J) qRT-PCR analysis of ZO-1 (I) and Occludin (J) mRNA expression (n = 3). ** P<0.01, *** P<0.001.

To further investigate whether this treatment induced a quiescent state in HUVECs, we examined functional markers. CM from the LEV@CAT-NPs + US group downregulated the proliferation marker Ki67 (Figure 7B) and upregulated the quiescence-associated marker IL-33 (Figure 7D). Functionally, Transwell assays confirmed that this treatment group exhibited the lowest permeability to 40 kDa dextran, with a 44.85% reduction vs the control group (P < 0.001; Figure 7H). Together, these findings indicate that the treatment induced a quiescent phenotype in HUVECs, characterized by low proliferation, enhanced barrier function, elevated IL-33, and reduced Ki67 expression. This study provides key in vitro evidence elucidating the mechanism by which US and LEV@CAT-NPs promote vascular normalization.

To further explore the impact of LEV@CAT-NPs + US on the vascular endothelial barrier in tuberculous granulomas, we evaluated changes in endothelial tight junctions at the protein, transcriptional, and ultrastructural levels. Immunofluorescence staining showed that the expression of ZO-1 and Occludin was significantly higher in the LEV@CAT-NPs + US group than in other groups (Figure 7E and F). Meanwhile, qRT-PCR results indicated that this treatment significantly upregulated the mRNA levels of ZO-1 and Occludin in granulomatous tissues by 2.75-fold and 7.26-fold, respectively (P < 0.001; Figure 7I and J). TEM (Figure 7G) revealed ultrastructural differences: the control group showed disrupted tight junctions and widened intercellular spaces; the LEV-NPs group exhibited swollen endothelial cells with open tight junctions; the LEV@CAT-NPs group displayed partially loose intercellular junctions; the US + LEV-NPs group had blurred and partially interrupted junction structures; whereas the LEV@CAT-NPs + US group showed flattened, spindle-shaped endothelial cells with continuous, compact, and well-defined tight junctions. These multi-level results confirm that US combined with LEV@CAT-NPs upregulates tight junction protein expression and maintains junction structural integrity, thus promoting functional recovery of the endothelial barrier in vivo.

Vascular Normalization and Hypoxia Alleviation Induced by LEV@CAT-NPs + US Therapy

To further investigate the effect of US combined with LEV@CAT-NPs on vascular structure and function in tuberculous granulomas in vivo, we examined the expression of the endothelial marker CD31 and the pericyte marker α-SMA in granuloma tissues. Immunofluorescence results revealed that the control group had disorganized vascular structures in granulomas, with CD31-positive endothelial cells arranged in a scattered, discontinuous punctate pattern. In contrast, the US + LEV@CAT-NPs co-treatment group showed more orderly arranged CD31-positive endothelial cells forming continuous vascular lumens. Meanwhile, the control group had low α-SMA-positive pericyte coverage, while the co-treatment group showed significantly increased pericyte coverage, resulting in structurally intact vessels (Figure 8A). Enhanced continuity of CD31 and increased coverage of α-SMA together indicated the restoration of normal vascular structure, which is a core characteristic of vascular normalization.³⁰

Figure 8.

Effects of different treatments on vascular structure and function, hypoxic microenvironment, and related signaling molecules in tuberculous granulomas in vivo. (A) Vascular density and maturity (pericyte coverage) assessed by immunofluorescence co-localization of α-SMA (green) and CD31 (red); scale bar: 50 μm; (B) Vascular leakage assessed by co-staining of 70 kDa FITC-dextran (green) and CD31 (red); scale bar: 100 μm; (C) Vascular perfusion assessed by co-localization of FITC-lectin (green) and CD31 (red); scale bar: 25 μm; (D) Immunohistochemical staining of HIF-1α; scale bar: 40 μm; (E) Immunofluorescence detection of hypoxic regions using pimonidazole (PIMO);scale bar: 100 μm; (F) Analysis of oxygen saturation (sO₂) in granulomas by photoacoustic imaging. Left panel: Representative photoacoustic imaging of granulomas; Right panel: Quantitative comparison of sO₂ levels in granulomas between the Control and US + LEV@CAT-NPs groups. (G) Semi-quantitative analysis of PIMO⁺ reas in (E). (H) Immunohistochemical staining images of VEGF and VEGFR2; scale bar: 40 μm.***P<0.001.

We further evaluated vascular function by intravenous injection of FITC-dextran (70 kDa, for permeability assessment) and Dylight^® 488-labeled lectin (for perfusion assessment). Immunofluorescence staining of CD31-marked vascular regions showed that the US + LEV@CAT-NPs group had significantly reduced dextran leakage and enhanced lectin perfusion (Figure 8B and C). These findings suggested that the combined therapy improved vascular function by restoring structural integrity.

Our previous in vitro experiments confirmed that US + LEV@CAT-NPs can inhibit hypoxia-induced abnormal activation of endothelial cells; we therefore further evaluated its regulatory effect on the hypoxic microenvironment in a rat tuberculous granuloma model. Immunohistochemistry results showed that HIF-1α-positive staining in the co-treatment group was significantly lower than in the control group (Figure 8D). Pimonidazole hypoxia staining showed that the hypoxic area decreased significantly from 55.46 ± 9.85% in the control group to 7.88 ± 3.86% in the co-treatment group (P < 0.001, Figures 8E and G). Photoacoustic (PA) imaging further revealed that the oxygen saturation (sO₂) within the granulomas in the treatment group was twice as high as in the control group (Figure 8F). These results demonstrated that US combined with LEV@CAT-NPs alleviated hypoxia in tuberculous granulomas in vivo.

As hypoxia-induced HIF-1α/VEGF signaling is the core driver of pathological angiogenesis, we further detected the expression of pathway-related molecules in granuloma tissues. Immunohistochemistry results showed that the expression of VEGF and its receptor VEGFR2 was significantly lower in the treatment group than in the control group (Figure 8H). Taken together, these in vivo findings indicated that US combined with LEV@CAT-NPs modulated the HIF-1α/VEGF signaling axis, inhibited VEGF/VEGFR2 overexpression, and ultimately promoted vascular normalization in tuberculous granulomas.

Combination Therapy with US and LEV@CAT‑NPs Enhances Drug Penetration and Accumulation in 3D Tuberculous Spheroids and in vivo Lesions

This study evaluated the enhancement of drug penetration and diffusion by US combined with LEV@CAT-NPs in a 3D tuberculoma model. After different treatments, spheroids were incubated with FITC-labeled rifampin. CLSM with layered scanning (top, middle, bottom) revealed that in the Control group, the FITC fluorescence signal was enriched in the top layer (Figure 9A), but gradually decreased toward the core, forming a characteristic green ring structure, indicating limited drug penetration into the central region. In contrast, US + LEV@CAT-NPs treatment resulted in strong green fluorescence not only in the top layer but also in deeper layers. Quantification analysis showed that, compared to the Control group, the fluorescence intensity in the top, middle, and core layers increased by 23.61%, 93.94%, and 101.42%, respectively, in the US + LEV@CAT-NPs group (Figure 9B). This treatment promoted deep drug delivery into the tuberculoma.

Figure 9.

Rifampin penetration and distribution in vitro 3D tuberculous spheroids and in vivo tuberculous granulomas after different treatments. (A) Representative CLSM images showing the distribution of FITC-labeled rifampin (green) across different layers (upper, middle, lower) of 3D tuberculous spheroids; scale bar: 100 μm; (B) Quantification of fluorescence intensity in each spheroid layer using ImageJ software;(C) Spatial distribution of FITC-labeled rifampin (green) in granulomatous tissues; scale bar: 100 μm; (D) Mean rifampin concentration-time curve in granulomatous tissues; (E) Pharmacokinetic parameters of rifampin in granuloma tissues calculated from the concentration-time data.

To assess changes in drug distribution and concentration after vascular normalization, we used CLSM to observe the distribution of FITC-labeled rifampin in animal tuberculous granulomas and HPLC-MS/MS to measure drug concentrations in the lesion tissues. CLSM showed that in the Control group, the drug was mainly distributed at the periphery of the granuloma, with weak fluorescence in the central region. In contrast, the US + LEV@CAT-NPs group exhibited increased drug distribution in the central area (Figure 9C). HPLC-MS/MS analysis showed that the rifampin concentration in the Control group decreased significantly 1 h after administration, whereas in the US + LEV@CAT-NPs group, the drug concentration remained 1.40 to 1.98 times higher than that of the Control group from 2 h to 6 h (Figure 9D). Pharmacokinetic parameters (Figure 9E) revealed that the US + LEV@CAT-NPs group had significantly increased AUC _0–t, AUC _0–∞, and C_max (P < 0.01), while t_1/2 showed no significant change (P > 0.05), suggesting that the treatment enhances efficacy primarily by increasing drug exposure and peak concentration rather than altering the clearance rate. These results demonstrated that US + LEV@CAT-NPs enhanced the delivery and retention of rifampin in tuberculous lesions by improving drug penetration and distribution.

US + LEV@CAT-NPs Therapy Promotes Immune Cell Recruitment and Activates Innate and Adaptive Immunity

Abnormal vasculature is a core driver of the immunosuppressive microenvironment in tuberculous granulomas. Based on the evidence that vascular normalization promotes immune cell recruitment and activation, we further evaluated immune responses in the tuberculous granuloma model.

We first performed transcriptomic analysis on tuberculous granuloma tissues from each group. Transcriptomic analysis identified 399 differentially expressed genes (DEGs) in the US + LEV@CAT-NPs group compared with the control group (259 upregulated, 140 downregulated; |log₂FC| > 1, P < 0.05) (Figure 10A). Reactome analysis showed that 55 DEGs were enriched in immune-related pathways (Figure 10C). Further KEGG enrichment analysis confirmed that the co-treatment simultaneously activated multiple innate immune pathways (including antigen processing and presentation, natural killer cell-mediated cytotoxicity) and adaptive immune pathways (including TCR signaling, BCR signaling) (Figure 10F). The treatment significantly upregulated chemokines involved in immune cell chemotaxis (CCL5, CXCL9, CXCL10, CXCL11; Figure 10B) and T cell marker genes (CD3D, CD8A; Figure 10D). Among the immune-related DEGs, eight genes involved in both innate and adaptive immune regulation were significantly upregulated (Figure 10E and S2), consistent with the pathway enrichment results.

Figure 10.

US combined with LEV@CAT-NPs enhances immune responses in rat tuberculous granulomas by activating innate and adaptive immune pathways (transcriptomic analysis). (A) Volcano plot of differentially expressed genes; (B) Expression of chemokine-related genes; (C) Reactome pathway enrichment; (D) Expression of CD3D and CD8A genes in tuberculous granulomas; (E) Venn diagram of immune-related genes; (F) KEGG pathway enrichment analysis of immune-related DEGs. *P<0.05,***P<0.001.

We further validated the transcriptomic findings at the protein and functional levels. Immunofluorescence staining showed that CD3⁺ and CD8⁺ T cell infiltration was significantly increased in the treatment group compared with the control group (Figure 11A). ELISA showed that pro-inflammatory cytokines (TNF-α, IL-6) were significantly upregulated, while immunosuppressive cytokines (IL-10, TGF-β) were significantly downregulated in the co-treatment group (Figure 11B).

Figure 11.

US + LEV@CAT-NPs activates immune responses in tuberculous granulomas. (A) Immunofluorescence images showing CD3⁺ and CD8⁺ T cell infiltration in tuberculous granulomas; scale bar: 50 μm; (B) Quantitative analysis of inflammatory cytokine expression in tuberculous granulomas across different groups by ELISA (n = 3). *P < 0.05,**P < 0.01,***P < 0.001.

Together, these results demonstrated that US + LEV@CAT-NPs reversed the immunosuppressive microenvironment of tuberculous granulomas by promoting vascular normalization and enhanced the recruitment and activation of anti-tuberculosis immune cells.

US + LEV@CAT-NPs Therapy Exerts Potent Granuloma Inhibitory and Bactericidal Effects in a Rat Tuberculous Granuloma Model

We first confirmed model establishment: H&E and pimonidazole staining showed that on day 8 after BCG infection, subcutaneous granulomas had central necrotic and hypoxic regions with acid‑fast bacilli (AFB) accumulation, presenting typical features of tuberculous granulomas (Figure S3). Subsequently, we treated subcutaneous tuberculous granulomas in rats using LEV@CAT-NPs combined with US, as illustrated in Figure 12A. Granuloma volume in each group was measured every two days (Figure 12C and D). After 8 days of treatment, compared to the granuloma volume in the control group (5827.69±150.70 mm³), the granuloma volumes in the LEV-NPs group (5171.27±218.80 mm³), the US + LEV@CAT-NPs group (1900.59 ± 132.86 mm³), and the US + LEV-NPs group (4522.23±243.87 mm³) were all reduced. Among these, the combination of LEV@CAT-NPs and US resulted in the most significant reduction in granuloma volume during the treatment period. Meanwhile, CFU counts showed that the bacterial load in the LEV@CAT-NPs combined with US group was (3.83 ± 0.29) × 10³ CFU/g tissue, which was significantly reduced by 99.99% compared to the control group (5.55 ± 2.09) × 10⁷ CFU/g tissue) (P < 0.001; Figure 12B). These results indicated that the combined treatment of LEV@CAT-NPs and US had significant inhibitory effects on granuloma growth and bactericidal efficacy. H&E staining of pathological sections further revealed that the necrotic area within granulomas was significantly reduced in the LEV-NPs and LEV@CAT-NPs groups compared to the control group. The US + LEV-NPs group showed further reduction in necrosis and inflammation, while after treatment with LEV@CAT-NPs combined with US, almost no residual necrosis and marked disappearance of the original granuloma structure (Figure 12E).

Figure 12.

Comparison of therapeutic effects of different treatment regimens on tuberculous granulomas in rats. (A) Schematic diagram of the SDT regimen using LEV@CAT-NPs combined with ultrasound for the BCG-infected subcutaneous tuberculous granuloma model in rats; (B) Bacterial load in granuloma tissues was determined by CFU assay after different treatments (data presented as Log₁₀ CFU/g tissue); (C) Photographs of granuloma tissues in rats after different treatments; (D) Granuloma volume changes over time in each group; (E) H&E staining images of granuloma tissues after different treatments; scale bar: 500 μm. ***P<0.001.

Discussion

Tuberculosis (TB), caused by Mtb, is a chronic infectious disease posing a serious threat to global public health. Upon infecting the host, Mtb triggers the formation of a tuberculous granuloma, a characteristic pathological structure featuring a necrotic core surrounded by dense fibrotic encapsulation.³¹ Current data indicate that the cure rate for drug-resistant TB is only 60%, with the main obstacle being insufficient drug penetration into the necrotic foci.^32,33 Similar to tumors, tuberculous granulomas have vascular abnormalities, such as collapsed and deformed vessels, enhanced permeability, and reduced pericyte coverage.^34,35 Insufficient oxygen supply to the granuloma stabilizes hypoxia-inducible factor 1α (HIF-1α), subsequently promoting the upregulation of VEGF expression, driving aberrant vascularization.^36,37 These abnormal vessels impede anti-TB drug delivery to the granuloma core, severely impairing treatment efficacy.³⁸ Although VEGF-targeted anti-angiogenic therapies (eg, bevacizumab) can suppress angiogenesis and normalize vascular permeability, their application is limited by low delivery efficiency and the risk of aggravated tissue hypoxia due to narrow therapeutic windows.^35,39 To address this need, we developed an ultrasound-augmented strategy based on levofloxacin/catalase-co-loaded oxygen-generating nanoparticles (LEV@CAT-NPs), which overcomes the limitations of conventional therapies and provides a new paradigm for tuberculous granuloma microenvironment remodeling.

In vitro experiments confirmed that LEV@CAT-NPs effectively catalyze the decomposition of H₂O₂ to produce oxygen (Figure 1F) and generate ROS under ultrasound stimulation, demonstrating oxygen-generation capacity and sonosensitizing properties. As a critical preclinical model bridging 2D cell culture and in vivo animal models, 3D spheroids recapitulate cell-cell interactions and molecular diffusion barriers in granulomatous tissues.⁴⁰ We therefore established 3D tuberculous spheroids to evaluate therapeutic efficacy (Figure 2A), which replicated typical in vivo granuloma features:²⁵ a hypoxic microenvironment, upregulation of HIF-1α and VEGF at both mRNA (Figure 2D and E) and protein levels (Figure 2F and G), and H₂O₂ accumulation (Figure 2C).

Our findings confirm that US combined with LEV@CAT-NPs effectively disrupts the hypoxia-driven vicious cycle in tuberculous granulomas. As previously reported, the granuloma microenvironment induces severe core hypoxia, which stabilizes HIF-1α and drives high VEGF secretion and leaky vascular hyperplasia.⁴¹ Ultrastructural evidence shows that vessels in untreated TB lesions have typical abnormalities: near-absent vascular density in the granuloma core, and peripheral aberrant neovessels with loose endothelial junctions, disrupted basement membranes and lost pericytes.^9,42 Functionally, this therapy not only reverses hypoxia (Figure 8E–G) but also downregulates HIF-1α and VEGF (Figure 3A–E), indicating successful intervention in the HIF-1α/VEGF signaling axis. We further elucidated that vascular normalization is achieved not by simple VEGF inhibition, but by blocking the positive regulatory loop of VEGF through inhibiting macrophage PI3K/Akt pathway activation (Figure 5A–E). After treatment, vascular morphology and function were improved: vascular endothelial proliferation was inhibited, permeability was reduced, and vessels were shifted to a quiescent state (Figure 7B–D). Molecular and structural changes included upregulated tight junction proteins (ZO-1 and Occludin) (Figure 7A and G), decreased vascular permeability (Figures 7H and 8B), increased pericyte coverage (Figure 8A), and improved blood perfusion (Figure 8C). These changes remodeled the abnormal vessels into structurally intact, functionally mature vessels. Ultimately, this microenvironmental reversal yielded therapeutic benefits:drug (eg, rifampin) penetration into tissues was enhanced (Figure 9A–C), and its pharmacokinetic (PK) profile was improved (Figure 9D and E). This contrasts with the impaired PK caused by hindered drug penetration across abnormal vessels and dense stroma, as well as low bactericidal efficacy stemming from hypoxia‑induced Mtb dormancy.^43,44

Existing anti-VEGF therapies only block downstream effector molecules, facing the paradox of compensatory revascularization and exacerbated hypoxia.⁴⁵ This study, through transcriptomics and Western blotting, confirmed that US combined with nanoparticles inhibits upstream PI3K/Akt phosphorylation while downregulating downstream VEGF, achieving comprehensive regulation via “dual-site clamping”. In tuberculous granulomas, the signaling cascade triggered by host cell recognition of Mtb trehalose dimycolate (TDM) activates NFATC2 to induce VEGF expression.³⁶ This process intersects with PI3K/Akt pathway regulation; for example, in macrophages, PI3K/Akt activation enhances VEGF secretion via HIF-1α upregulation, promoting angiogenesis.⁴⁶ Radiation-induced PI3K/Akt pathway activation also leads to excessive VEGF secretion by inhibiting autophagy, suggesting the universality of this pathway’s role in regulating VEGF under pathological conditions.⁴⁷ This study showed that in Mtb infection, the function of the PI3K/Akt-VEGF signaling axis is reprogrammed by US combined with LEV@CAT-NPs, shifting it from promoting “cell survival” to mediating “microenvironment remodeling”.

The traditional view holds that hypoxic necrosis in tuberculous granulomas is a “byproduct” of immune defense,⁴⁸ but recent research reveals it is essentially a result of pathogen-host co-adaptation—hypoxia not only induces Mtb dormancy,⁴⁹ but also drives vascular abnormalities via the HIF-1α/VEGF axis, forming a “dual drug-immunity shield”. Macrophage polarization toward M1 phenotype has been confirmed as a key mechanism of host defense against intracellular pathogens,⁵⁰ and apoptosis of infected macrophages is another pathway for intracellular bacterial clearance.^51,52 This study confirmed that US + LEV@CAT-NPs triggers these immune responses: it not only promotes M1 polarization in 3D tuberculous spheroids (Figure 4A) and increases pro-inflammatory cytokine secretion (Figure 4B) but also induces apoptosis by upregulating the Bax/Bcl-2 ratio (Figure 4C). These effects reshape the anti-TB immune microenvironment, facilitating Mtb clearance in the host.

To verify whether vascular normalization reverses immunosuppression, we performed transcriptomic analysis of granuloma tissues, given that functional vasculature is a prerequisite for immune cell recruitment and infiltration.⁵³ The therapy induced widespread gene expression reprogramming (Figure 10A), upregulating chemokines for T cell recruitment (CCL5, CXCL9) and T cell markers (CD3D, CD8A) (Figure 10B and D), and activating innate and adaptive immune pathways (Figure 10C and F). Functional validation confirmed that the treatment increased CD3⁺ and CD8⁺ T cell infiltration into granulomas (Figures 11A) and reversed the cytokine profile (increased pro-inflammatory and decreased anti-inflammatory cytokines) (Figure 11B), reversing the immunosuppressive microenvironment. Ultimately, US combined with LEV@CAT-NPs exerted antibacterial effects in vivo: it inhibited granuloma growth and reduced bacterial burden (Figure 12B).

The innovation of this study lies in a paradigm that remodels the infected microenvironment through synergistic “in situ oxygen supply” and “sonodynamic sterilization”, transforming dysfunctional vascular networks into drug delivery channels. This discovery aligns with the “vascular normalization window” theory in oncology,^54,55 but differs in that vascular abnormalities in tuberculosis originate from the infectious microenvironment rather than genetic mutations; therefore, they can be reversed through metabolic intervention (oxygen supply) rather than genetic targeting, posing lower clinical translation risks. This strategy of reversing vascular abnormalities to enhance drug delivery and immunity,^56–58 not only provides a promising approach for tuberculosis treatment but also opens avenues for treating other chronic infectious diseases characterized by hypoxia and vascular abnormalities.

Although our core findings have been validated in cellular assays, tuberculous spheroids, and animal models, several limitations should be acknowledged. First, the durable therapeutic efficacy of US + LEV@CAT-NPs warrants further validation in non-human primate models with extended longitudinal monitoring, given the interspecies differences in granuloma structure and the need for long-term relapse assessment. Second, ultrasound penetration varies across lesions at distinct anatomical sites (eg, pulmonary and bone TB), and parameters require optimization for site-specific treatment. Third, the present study used BCG as a surrogate for Mtb; although BCG recapitulates key pathological features of Mtb infection, validation using virulent H37Rv or drug-resistant clinical isolates is required to confirm translational applicability. Fourth, while transcriptomic profiling, Western blotting, and pharmacological inhibition provide convergent evidence for PI3K/Akt pathway involvement, definitive causal validation would require genetic perturbation approaches (eg, lentivirus-mediated knockdown/overexpression and rescue experiments). Fifth, comprehensive immunophenotyping of CD4⁺ T-cell subsets, regulatory T cells, dendritic cells, natural killer cells, immune checkpoint molecules (PD-1, TIM-3, LAG-3), and memory T-cell responses is warranted in future studies to fully elucidate the immune mechanisms underlying the therapeutic benefit. Finally, the long-term safety of the nanoparticle formulation and the stability of large-scale production require further evaluation to support clinical translation.

Conclusion

This study presents an ultrasound-augmented, oxygen-generating nanoplatform that addresses a critical bottleneck in tuberculosis therapy, namely the hypoxia-driven vicious cycle that fuels both drug resistance and immunosuppression. By simultaneously correcting hypoxia, normalizing aberrant vasculature, and reactivating anti-tuberculosis immune responses, this strategy achieves a triad of therapeutic benefits that single-target approaches cannot provide. The key advantage lies in its dual-site regulatory mechanism consisting of upstream PI3K/Akt inhibition combined with hypoxia correction. This design avoids compensatory angiogenesis and extends the vascular normalization window. However, the current findings are limited to preclinical models, and long-term safety, large-scale manufacturing, and site-specific ultrasound optimization require further investigation. Future studies should validate this strategy in higher-order models and explore its applicability to other hypoxia-driven chronic infectious diseases.

Funding Statement

This work was supported by the National Natural Science Foundation of China (No. 82570147 and No. 82270115), the Chongqing Graduate Research Innovation Project Fund (Nos. CYS25371), and 2024 Graduate Mentor Team Program, First Clinical College of Chongqing Medical University (No. CYYY-DSTDXM-02408).

Disclosure

The authors report no conflicts of interest in this work.