Ultrasound-triggered platelet vehicles loading fluorescein for targeted sonodynamic therapy of glioblastoma Authors: Yu-Xin Guo, Qing Deng, Ke Li, Quan Zhang, Jie Guo, Meng-Fei Wang, Jia-Lin Wu, Hui-Mei Lei, Yue-Ying Chen, Huan Pu, Yu-fan Gao, Yong-Hong Xu, Xiao Chen, Qing Zhou Published: 2026 ABSTRACT AbstractGlioblastoma (GBM), the deadliest primary brain malignancy, resists treatment because of the blood‒brain barrier (BBB) and intrinsic resistance mechanisms. Fluorescein (FL), a clinically approved fluorophore, shows promise as a sonosensitizer for sonodynamic therapy (SDT), but faces challenges in tumor delivery and incomplete mechanistic understanding beyond ROS-mediated damage. We developed a bioinspired platelet-based delivery platform exploiting the natural tumor-homing properties of platelets for targeted FL delivery. Platelets were loaded with fluorescein diacetate (FDA), a lipophilic prodrug that is converted intracellularly to active FL, combined with probenecid to inhibit MRP1-mediated efflux, generating the ultrasound-responsive FL@plt. In vitro, FL demonstrated sonocytotoxicity; upon low-intensity ultrasound, FL@plt rapidly released FL into GBM cells, producing significant cytotoxicity enhanced by probenecid-mediated retention. In subcutaneous and orthotopic mouse models, intravenous FL@plt with probenecid and sequential ultrasound achieved extensive tumor FL accumulation, massive necrosis, significant growth inhibition, and prolonged survival. Mechanistically, FL-SDT induces DNA damage while promoting HIF-1α degradation via the ubiquitin‒proteasome and autophagy‒lysosome pathways, thus suppressing downstream DNA repair enzymes (SMUG1 and LIG4). This dual mechanism combines direct genotoxicity with impaired DNA repair capacity, underlying the potent anti-GBM efficacy. Our platelet-based FL delivery represents a practical, safe, translational therapy for overcoming BBB limitations by disabling HIF-1α-regulated DNA repair. Introduction Glioblastoma (GBM) is the most aggressive and common primary brain tumor in adults. It presents a formidable challenge in oncology because of its infiltrative nature, rapid progression, and resistance to conventional therapies (Singh et al. 2025). With an incidence of approximately 3–4 cases per 100,000 people each year and a median survival of only 12–15 months even with multimodal treatment regimens, including surgical resection, radiotherapy, and temozolomide chemotherapy, there is an urgent need for novel therapeutic strategies to improve patient outcomes (Omuro 2013; Jezierzański et al. 2024). Although emerging biological treatments, such as dendritic cell vaccines and oncolytic viral immunotherapies (Reddy et al. 2024; Shah et al. 2024), show clinical promise, their efficacy is frequently hindered by the formidable blood‒brain barrier (BBB) and complex tumor heterogeneity. To address these delivery limitations, non-invasive modalities such as sonodynamic therapy (SDT) have garnered significant attention for their potential to target deep-seated tumors, such as GBM, while minimizing damage to surrounding healthy tissue. SDT harnesses low-intensity ultrasound (typically 0.5–3.0 W/cm2 at 1–2 MHz) to activate sonosensitizers that accumulate in tumor cells (Datta et al. 2024). Upon sonication, these sonosensitizers generate reactive oxygen species (ROS), such as singlet oxygen (1O₂) and hydroxyl radicals (•OH), inducing oxidative stress and subsequent cell death. Unlike photodynamic therapy (PDT), which is limited by the depth of light penetration, SDT benefits from the ability of ultrasound to cross deep tissues and the blood‒brain barrier (BBB), making it particularly well suited for intracranial applications (Park et al. 2023). Over the past decade, SDT has progressed from proof-of-concept studies to preclinical demonstrations of its efficacy against intracranial tumors. As of 2025, ongoing clinical trials have reported complete tumor clearance in some cases, underscoring the evolving potential of this approach (Liang et al. 2020; Zhang et al. 2021). These developments highlight the role of SDT in addressing the therapeutic resistance of GBM. Despite progress thus far, there are some major challenges hinder the fulfilment of the potential of SDT in the treatment of GBM. The most prominent one is the difficulty of achieving targeted and adequate sonosensitizer delivery to GBM, largely because the BBB consists of both physical and biochemical barriers (Sweeney et al. 2019). The physical barrier features the endothelial cells that line the interior of the capillaries of the brain and are connected by tight junctions, which prevent the paracellular diffusion of hydrophilic molecules and large molecules. The biochemical barrier comprises the efflux transporter proteins that are highly expressed in BBB endothelial cells (Abdullahi et al. 2018). These transporters, chiefly ATP-binding cassette (ABC) transporters, such as P-glycoprotein (P-gp), multidrug resistance proteins (MRPs), and breast cancer resistance protein (BCRP), actively pump potentially harmful substances (Koehn 2020), both hydrophilic and lipophilic, from the brain and back into the bloodstream. Most sonosensitizers, including widely used porphyrin- and phthalocyanine-based agents, can hardly penetrate the BBB (Huang et al. 2024) and are therefore unable to reach therapeutically sufficient concentrations in the GBM. To address this, innovations in focused ultrasound (FUS) have been explored as a non-invasive strategy to transiently open the BBB. Such approaches, including micro- or nanoparticle-mediated cavitation and optimized ultrasound guidance, have been shown to improve BBB permeability, thereby enhancing the delivery of therapeutic agents, such as sonosensitizers, for both oncological and neurological disorders (Yildirim et al. 2019; Shumer-Elbaz et al. 2025; Ediriweera et al. 2025). Nanocarrier formulations (e.g. liposomes, polymersomes, and dendrimers) have increased circulation time and tumor uptake (Chen et al. 2020; Rawding et al. 2022; Zheng et al. 2023); however, they are subject to clearance by the reticuloendothelial system and off-target accumulation in the liver and spleen (Maruyama 2011). Despite these developments, effective delivery and retention of sonosensitizers in highly heterogeneous and infiltrative GBM remain formidable challenges. Recent advances in biomimetic carriers, notably platelet-based delivery systems, have opened new avenues for targeted sonosensitizer delivery to GBM (Xu et al. 2022). Leveraging platelets as delivery carriers may overcome these barriers. Platelets, which are micron-sized, anucleate cell fragments derived from megakaryocytes, have gained significant attention as natural carriers for tumor-targeted drug delivery. Recent research has demonstrated the many advantages of using platelets as carriers for tumor-targeted delivery (Li et al. 2022). Platelets can be readily harvested from blood using existing clinical infrastructure. They are non-immunogenic and cause fewer adverse effects than synthetic carriers. Resting platelets have a long lifespan in the bloodstream, enabling sustained drug release and reducing the need for frequent dosing. They naturally interact with tumor cells and the tumor microenvironment (TME), enabling precise drug delivery to cancer sites. They can be loaded with various drugs, including chemotherapeutics, immunotherapies, and gene therapies, thus enhancing their applicability to different cancer types. Importantly, platelet activation can be triggered by artificially generated reactive oxygen species (ROS), e.g. through the sonication of platelet-loaded sonosensitizers. Once activated, the loaded platelets rapidly adhere to the vascular endothelium, forming aggregates and releasing the drug payload. In light of the above favorable attributes, we have proposed a strategy of ‘ultrasound-responsive platelets’ to enable highly efficient sonosensitizer delivery to GBM and targeted SDT of GBM (Wang et al. 2025). The main strategy leverages sonosensitizer-loaded platelets that circulate to glioblastoma multiforme (GBM). Initial low-intensity sonication generates ROS, which then activate these platelets, causing them to aggregate in tumor vessels and rapidly release their cargo. This localized oxidative disruption enhances drug penetration into the tumor tissue, where a second sonication then induces sonodynamic destruction of the tumor cells. Crucially, this requires stable and sufficient sonosensitizer loading without perturbing platelet function. However, most available sonosensitizing agents fail to meet this requirement, which either have an inadequate loading capacity, loading stability, or have a negative impact on platelet functions. Therefore, in our previous studies, we had to prove the feasibility and efficacy of this strategy utilizing a nano-formed sonosensitizers based on chlorin e6 attached to decorated nanoparticles (Wang et al. 2023). Platelet-based photodynamic and sonodynamic therapies often utilize complex chemical conjugation or nanocarrier co-loading. However, these methods can compromise platelet function and hinder clinical translation, primarily owing to difficulties in preparation, characterization, and quality control. Fluorescein (FL), a particularly promising sonosensitizer for GBM treatment due to its unique properties and clinical relevance, offered an excellent solution to this dilemma. As a clinically approved agent already used in fluorescence-guided surgery, FL carries several advantages, including an established safety profile and tumor selectivity (Romeo et al. 2025). When activated by ultrasound, FL generates ROS and induces tumor cell death while maintaining selectivity for malignant tissue (Prada et al. 2020). Compared to heavier porphyrin derivatives, FL offers good aqueous solubility, rapid renal clearance, and minimal prolonged sonosensitivity. Safety studies on porcine models further support the intracranial application of FL, with no adverse effects observed on healthy brain tissue under focused ultrasound (Raspagliesi et al. 2021). These attributes position FL as an attractive sonosensitizer for SDT of GBM, particularly for recurrent or advanced GBM, where BBB disruption enhances its uptake (Huang et al. 2024). However, FL per se is unable to be stably loaded into platelets in sufficient quantities because of its negative charge, which impedes membrane diffusion, and the presence of efflux transporters (e.g. MRPs) on the platelet plasma membrane has two limitations (Jedlitschky et al. 2012), and their constant efflux activities prevent stable loading of FL in platelets. In the present study, we hypothesized that a small-molecule prodrug strategy combined with an efflux inhibitor would achieve stable, high-capacity FL loading into platelets, preserving their natural function without complex nanocarriers, for efficient targeted sonodynamic therapy (SDT) against GBM. We prepared platelets stably and profusely loaded with FL using a simple method, i.e. incubating isolated platelets in phosphate-buffered saline (PBS) containing fluorescein diacetate (FDA) and probenecid. The rationale behind this is as follows. FDA is a nonpolar, hydrophobic compound, and thus easily diffuses across the plasma membrane into platelets. Once inside the platelets, FDA undergoes rapid enzymatic hydrolysis by intracellular esterases, converting it into the hydrophilic FL. Probenecid helps to trap FL in platelets by blocking the efflux activity of MRP1 in the plasma membrane. The FL-loaded platelets were responsive to ultrasound activation and were employed together with probenecid to successfully demonstrate the tumor-targeted delivery of FL and the efficacy of FL-SDT in mouse GBM models. Furthermore, we explored the therapeutic mechanism of FL-SDT beyond direct oxidative cytotoxicity, which critically involved impact on the hypoxia-inducible factor 1 alpha (HIF-1α), which is a key regulator of survival mechanisms, particularly the DNA damage repair machinery, in tumor cells. This work presents a simple, novel targeted SDT strategy for GBM with potential for clinical translation and provides insight into its mechanism. Methods Materials Fluorescein diacetate (FDA) and sodium fluorescein (Aladdin Industrial Co., Shanghai, China); probenecid (HY-B0545, MCE, Monmouth Junction, NJ); trisodium citrate 3.8% (Yuanye Biotechnology, China); cobalt (II) chloride (CoCl₂) and 9,10-anthracenediyl-bis(methylene) dimalonic acid (ABDA) were purchased from Sigma-Aldrich, Inc. (St. Louis, MO). Antibodies included CD62p (bs-0561R, Bioss, Beijing), β-actin (66009-1-Ig), Bax (50599-2-Ig), Bcl-2 (26593-1-AP), cleaved-Caspase 3 (25546-1-AP), β-tubulin (10094-1-AP), ubiquitin (10201-2-AP), and LAMP1 (65050-1-Ig) from Proteintech; γ-H2AX (AP0687) and SMUG 1 (A10166) from Abclonal; Caspase 3 (14220S) and LC3B (2775S) from Cell Signaling Technology; p-ATR (GTX132146, GeneTex); HIF-1α (20960-1-AP, Proteintech); and LIG 4 (CPA7172, Cohesion Bioscience, Suzhou), Multicolor Prestained Protein Ladder (WJ103) was bought from Epizyme. Additional reagents included the Annexin-V Apoptosis Detection Kit (E-CK-A218, Elabscience, China), CCK-8 (Biosharp, China), DHE probe (US Everbright® Inc.) and comet assay kit (C2041M, Beyotime, Shanghai, China). An immunoprecipitation kit with protein A + G magnetic beads was obtained from Beyotime (P2179S, Beyotime, Shanghai, China). Cell culture The GL261-luciferase (GL261-luc) and parental GL261 cell lines were sourced from the Cell Bank of the Shanghai Institutes for Biological Sciences (Shanghai, China). The cells were grown in DMEM with 10% fetal bovine serum (FBS; HYcezmbio, Wuhan, China) in a humidified, 5% CO₂ incubator at 37 °C and passaged regularly. Animals Female BALB/c nude mice and male C57BL/6 mice (4–6 weeks old) were obtained from the Laboratory Animal Center of Hubei Province (Hubei, China). The animals were housed under specific pathogen-free conditions at 22 °C​​​​​​ ± 1 °C, 50% ± 5% humidity with 12:12-h light/dark cycles and ad libitum access to autoclaved food and water. The study protocols were approved by the Ethics Committee at Renmin Hospital of Wuhan University (WDRM Animals Welfare No. 20230907 C), which are consistent with IACUC guidelines. The animal study was conducted from December 1, 2023 to September 5, 2024. Ultrasound irradiation time screening Based on prior work, an ultrasound irradiation intensity of 1 W/cm2 was applied (Wang et al. 2025). The selection of irradiation durations was guided by the dual considerations of the thermal and mechanical effects of ultrasound. The thermal effects were evaluated by an infrared thermal imager (HM-TPH21-3AXF, Hangzhou Microimage Software Co., Ltd., China). Concurrently, the mechanical damage induced by ultrasound was determined through cellular morphological alterations under a microscope. FL-SDT-induced cytotoxicity GL261 cells (5 × 103 cells/well in 96-well plates) were treated with fluorescein diacetate or probenecid at various concentrations for 12 h and then assessed for viability using CCK-8 assay. For intracellular fluorescein (FL) retention, the cells were incubated in complete DMEM at 37 °C, 5% CO₂ for 4 or 8 h and analyzed by flow cytometry (FL1 channel: 488/525 nm). Following protocol optimization (1 MHz, 1.0 W/cm2, 1 min), GL261 cells were incubated with FL ± probenecid for 1 h. After replacing FL medium with fresh medium (retaining probenecid in the designated groups), the cells underwent ultrasound exposure (1 MHz, 1.0 W/cm2, 1 min), followed by a 4 h incubation (37 °C, 5% CO₂). Cell viability (CCK-8), apoptosis markers (Bax, Bcl-2), and DNA damage indicators (γH2AX, p-ATR; TEM ultrastructure) were then assessed, and DNA damage was also evaluated by comet assay. Reactive oxygen species (ROS) generation was measured by dihydroethidium (DHE) staining, and 10 μM DHE probe solution was added 15 min post-sonodynamic therapy and after irradiation use, fluorescence microscopy (Olympus IX73, Japan) and flow cytometry to detect ROS generation. Singlet oxygen generation was detected by the degradation of 9,10-anthracenediyl-bis(methylene) dimalonic acid (ABDA). The reactions containing 100 μM probe and fluorescein were monitored with UV–vis spectroscopy by recording the decrease in absorbance at 400 nm. Preparation and characterization of FL@plt Mouse platelets were isolated from male C57BL/6 mice (5–6 weeks of age, 22–24 g) by density gradient centrifugation. Briefly, whole blood was collected with 300 μL of 3.8% trisodium citrate (1:9 v/v) and centrifuged at 200 × g for 15 min (22 °C, brake disabled) to obtain platelet-rich plasma, followed by 1000 × g for 10 min to pellet the platelets (Xu et al. 2022). Platelet counts were determined by hemocytometer under phase-contrast microscopy. For fluorescein loading (FL@Plt), platelets (1 × 106) were incubated with FDA (100 μmol/L) in 1 mL of PBS for 1 h at room temperature, with or without probenecid (PRO, 100 μmol/L), to inhibit efflux. After being washed twice with 1 mL of PBS (1000 × g, 10 min), the platelets were resuspended for analysis. The stability of FL@plt was assessed by a nanoparticle sizer (Zetasizer Nano ZSP, Malvern, England). Viability was assessed by CCK-8 assay after 4 h and 8 h (absorbance: 630 nm), while FL retention was quantified by flow cytometry (FITC channel: 530/30 nm). Platelet activation was determined by CD62P immunofluorescence using PB450-conjugated antibodies (PB450 channel: 450/45 nm). FL@Plt, either ultrasound irradiated or not, and platelets were also observed with scanning electron microscopy (SEM) and transmission electron microscopy (TEM). Ultrasound-triggered FL release from FL@Plt to in vitro GBM cells FL@Plt (1 × 106 of each in 1 mL of PBS in EP tubes containing probenecid 100 μmol/L or not depending on the experimental design) was either irradiated with ultrasound (1 MHz, 1 W/cm2, 1 min) or incubated with GBM cell-conditioned medium (GC-CM) at 5% CO2 and 37 °C for 30 min before getting assayed by flow cytometry for FL fluorescence, a free fluorescein sodium group (FLS concentration: 100 μmol/L) as a control is needed to investigate whether ultrasound promotes GL261 cells uptake of free fluorescein. Immunofluorescent staining of surface CD62P to validate that intracellular fluorescein in platelets does not interfere with ultrasound-induced activation behavior. To verify FL release from loaded platelets to glioblastoma (GBM) cells, FL-loaded platelets were co-cultured with GL261 cells at a 5:1 ratio and immediately irradiated with ultrasound (1 MHz, 1.0 W/cm2, 1 min). After 30 min, flow cytometry was used to analyze drug fluorescence in the GL261 cells. To assess sonodynamic toxicity, the co-culture system was first ultrasound-irradiated, After incubation for 1 h, the cells were irradiated again, cell viability and apoptosis were analyzed 4 h later. The protein bands were quantified using Image J, the band intensity for each target protein was normalized to the corresponding loading control band intensity. The relative expression level was calculated as the ratio of the target protein's integrated optical density (IOD) to that of the loading control. In vitro hemolysis assay of FL@plt The hemocompatibility of the final formulated FL@plt was evaluated using a standard hemolysis assay. Briefly, a 5% (v/v) red blood cell (RBC) suspension was prepared from fresh mouse blood by repeated washing with 0.9% NaCl solution. The RBC suspension was then incubated with FL@plt and probenecid at 37 °C for 1 h. 0.9% of NaCl and pure water were used as the negative and positive controls, respectively. After centrifugation (3000 × g, 5 min), the absorbance of the supernatant was measured at 540 nm using a microplate reader. The hemolysis ratio was calculated as follows: hemolysis (%) = (ODsample−ODnegative)/(ODpositive−ODnegative) × 100%. FL delivery by ultrasound activated FL@Plt to GBM grafts in mice Intracranial and subcutaneous glioblastoma (GBM) mouse models were established separately using male C57BL/6 mice and female BALB/c nude mice at four weeks of age (14–18 g) to evaluate the tumor-targeting efficiency of sono-activated FL@Plt. For the intracranial model, GL261-luciferase cells in the logarithmic growth phase were resuspended in serum-free medium and implanted intracranially following published protocols (Guo et al. 2025). Tumor growth was monitored via bioluminescence imaging (AnitView100, BLT, China), and experiments commenced on day 15 post-implantation. For the subcutaneous model, BALB/c nude mice were injected with 1 × 106 GL261 cells in 100 μL of serum-free medium in the right flank. The experiments began when the tumors reached ~500 mm3. Mice bearing subcutaneous GBM tumors were randomized into six groups (n = 5) based on tumor volume to ensure an equivalent baseline burden: ① PBS, ② FDA, ③ FL@Plt, ④ LIFU, ⑤ FL@Plt + LIFU, and ⑥ FL@Plt + PRO + LIFU. The LIFUSTIM® focused ultrasound modulation system was specifically applied to subcutaneous GBM mouse models. The system operates at a frequency of 1 MHz, an intensity of 1.0 W/cm2, and a 50% duty cycle. The equipment, designed by SXULTRASONIC (Shenzhen, China), featured a focal length of 1.725 cm and a focus zone area of 0.043 cm2. The mice were randomized into four groups (n = 5) based on bioluminescence signal intensity to ensure an equivalent baseline tumor burden for the intracranial GBM mouse models: ① PBS, ② FL@Plt, ③ FL@Plt + UI, and ④ FL@Plt + PRO + UI. Each mouse received an intravenous injection of PBS, FDA, FL@Plt, or FL@Plt + PRO (200 μL; FL dosage: 2.0 mmol/kg b.w. for a 20 g mouse; PRO dosage: 0.18 mmol/kg b.w. for a 20 g mouse (Töllner et al. 2015). Probenecid was injected intravenously 1 h prior to achieving adequate plasma levels. To determine the optimal timing for the second ultrasound irradiation, intra-tumoral fluorescein distribution was analyzed at specific intervals. For subcutaneous tumors, tissues were harvested and sectioned at 1, 3, and 5 h post-injection, Ex vivo fluorescence imaging was performed on vital organs after their isolation at the same time points. For intracranial tumors, the time points were 0.5, 1, and 2 h post-injection. The second ultrasound irradiation was performed at the peak of fluorescein fluorescence. Twenty-four hours after the second irradiation, the animals were sacrificed following CO₂ anesthesia and euthanasia by cervical dislocation, and the tumor tissues were harvested and fixed in 4% paraformaldehyde, brain or tumor tissues were collected for TUNEL staining and hematoxylin and eosin (H&E) staining. Major organs were also harvested for H&E staining to assess systemic effects. The mouse carcasses were moved to the Animal Center of Renmin Hospital of Wuhan University for weighing, storage, and burial. GBM-targeted SDT mediated by FL@Plt Intracranial grafts of GL261-luc cells were established as described previously. On day 14 post-implantation, the tumor-bearing mice were randomly divided into three groups (n = 5) based on bioluminescence signal intensity: ① FL@Plt, ② FL@Plt + UI, and ③ FL@Plt + PRO + UI, all mice received an intravenous injection of the respective formulations (200 μL in serum-free medium), with the dosage normalized to FL@Plt content as previously calculated. The mice underwent six extracranial ultrasound irradiation (UI) sessions (1 MHz, 1 W/cm2, 1 min) targeting the tumor site. The first UI was performed 1 h after drug administration to activate FL@Plt in the tumor vasculature and release FL into the intracranial GBM. A second UI session was performed 1.5 h later to induce the sonodynamic effect of fluorescein. A third ultrasound was performed 1 h later to further enhance the sonodynamic effect within the tumor. Subsequent UI sessions were conducted weekly, with the full protocol repeated twice to achieve sonodynamic therapy (SDT) efficacy. The mice were monitored daily and sacrificed upon reaching a moribund state for histopathological analysis of the brain and vital organs. Body weight was recorded every other day, and tumor growth during the first two weeks was tracked via weekly fluorescence imaging. For imaging, the mice were intraperitoneally injected with luciferin (150 mg/kg) and imaged 10 min post-injection under gas anesthesia using isoflurane (R.W.D, Shenzhen, China). MRI 3.0 T (SIGNAL Architect, GE healthcare) was used to evaluate tumor progression, and T2 sequence was selected in order to rule out complications such as cerebral edema, hemorrhage, ischemia, and hematomas and to evaluate tumor progression and the efficacy of GBM-targeted SDT. All the remaining animals were sacrificed following CO₂ anesthesia and euthanasia by cervical dislocation, and the tumor tissues were harvested and fixed in 4% paraformaldehyde, brains were harvested for H&E staining and immunohistochemical (IHC). Quantitative analysis of IHC staining was performed using ImageJ, and the integrated optical density (IOD) of the DAB-positive signal and the total analyzed area were measured. The relative protein expression level was expressed as the average optical density (AOD), which was calculated by dividing the IOD by the corresponding area (AOD = IOD/Area). Effects of FL-SDT on HIF-1α and downstream targets GL261 cells were treated with fluorescein diacetate (0.041638 g/L) alone or combined with probenecid (100 µM) for 1 h, followed by optional ultrasound exposure (1 MHz, 1 W/cm2, 1 min). As a positive control, the cells were incubated with 400 µM CoCl₂ for 5 h. To investigate the degradation pathways of HIF-1α, we employed two FDA-approved drugs with well-characterized mechanisms: bortezomib, a proteasome inhibitor used in multiple myeloma treatment (Befani et al. 2012; Sogbein et al. 2024), and ritonavir, an HIV protease inhibitor that also functions as an autophagy inhibitor (Suh et al. 2012). The protein levels of HIF-1α, LIG4, and SMUG1 were subsequently analyzed by WB. Protein interactions and localization were assessed using confocal microscopy to examine HIF-1α co-localization with LC3B and LAMP1 or ubiquitin and PSMB5, and co-immunoprecipitation (Co-IP) to detect HIF-1α-ubiquitin and HIF-1α-autophagy interactions. Gene and protein expression were quantified by qPCR and Western blotting, respectively. The primer sequences are provided in Supplementary Table 1. Statistical analysis The investigators were aware of the allocation during treatment, but the outcome assessment and data analysis were performed blindly using numerical codes. All the data are given as the mean ± standard deviation. One-way analysis of variance (ANOVA) with Tukey's post hoc test was used for single-factor comparisons, and two-way ANOVA with Bonferroni correction was used for analyzing interactions between two independent variables. Survival data were evaluated through the Kaplan–Meier methodology with the log-rank test for survival curve comparisons. Analysis of variance was used to assess the significance among ≥ 3 samples. A p < 0.05 was the threshold of statistical significance. Results Validation of FL's sono-cytotoxicity Preliminarily, the 1-min sonication duration was determined to be optimal, as it achieved a controlled and safe thermal plateau without inducing significant mechanical damage to cellular morphology (Figure S-17A‒C). Incubation with FDA at concentrations up to 100 µM for 24 h was non-toxic to the GBM cells (GL261), and the FDA-treated GBM cells displayed concentration-dependent toxicity following low-intensity ultrasound irradiation (UI) (1 MHz, 1 W/cm2, 1 min), as shown by the cell viability assay (Figure 1A, B). The combination of FDA (100 µM) and probenecid (100 µM) exhibited no toxicity to the GBM cells (Figure 1C), but probenecid significantly slowed the efflux of FDA-derived FL in GBM cells over a period of 8 h (Figure 1D). The GBM cells rapidly displayed bright green fluorescence following FDA treatment, indicating the loading of FL in the cells, as well as good cell viability, since only cells with healthy esterase activities and intact cell membranes can produce and retain FL in quantity (Figures 1E, S-1C, D). Sonication generated superoxide anions in the FL-loaded GBM cells, as indicated by DHE staining (Figure 1E, F), and this effect was intensified in the presence of probenecid. Notably, the decrease in FL (Figure 1E) is concurrent with DHE staining, indicating impaired esterase activity and/or compromised cell membrane integrity. Singlet oxygen was also detected by a decrease in ABDA absorbance at 400 nm (Figure S-18B). Furthermore, TEM revealed swollen, deformed mitochondria and increased autophagosome-like structures in FL-loaded GBM cells following sonication (Figure 1J). Consistently, sonication of the FL-loaded GBM cells resulted in a surge of apoptosis as indicated by a marked increase in the ratio of Bax / Bcl-2 (Figure 1G,H) and annexin V expression (Figure 1I). Overall, these observations confirm FL's capacity to elicit toxicity to GBM cells under low-intensity sonication and that this activity can be augmented by probenecid. Validation of FL's sono-toxicity to mouse GBM (GL261) cells. (A) CCK-8 assay of cell viability following loading with FDA up to 100 µM for 24 h. (B) Viability of GL261 cells loaded with FDA (100 µM) following ultrasound irradiation (UI: 1 MHZ, 1 W/cm2, 1 min). (C) Viability of GL261 cells following loading with FDA (100 µM) in the absence or presence of probenecid (PRO, 100 µM) for 24 h. (D) FL levels in GL261 cells at 0, 4, and 8 h after loading with FDA (100 µM) in the absence or presence of PRO. Representative flow cytometric density plots are presented in Figure S-1B. (E) Microscopy of DHE staining (red) and FL fluorescence (green) in GL261 cells loaded with FDA (100 µM) in the absence or presence of PRO (100 µM) following UI. (F) Quantification of DHE staining in E by flow cytometry. Representative flow cytometric histograms are presented in Fig. S-1E. The quantitative data of FL fluorescence in E are presented in Figure S-1C. (G) Protein levels of Bcl-2 and Bax in GL261 cells loaded with FDA in the absence or presence of PRO following UI. (H) Ratio of Bax to Bcl-2 in G. (I) Immunofluorescence staining of annexin V in GL261 cells loaded with FDA in the absence or presence of PRO following UI. Representative flow cytometric density plots are presented in Figure S-1F. (J) TEM of GL261 cells loaded with FDA in the absence or presence of PRO following UI. The arrowheads (blue) indicate the normal mitochondria, arrowheads (yellow) indicate stress-swollen mitochondria, and the arrowheads (white) mark autophagosomes. The values are the means ± SD (n = 3, *p < 0.05, **p < 0.01, ***p < 0.001). Preparation and characterization of ultrasound-responsive FL-loaded platelets FL-loaded platelets (FL@plt) were prepared by incubating isolated mouse platelets with FDA (100 µM) and probenecid (100 µM) in PBS at room temperature for 30 min. The FL@plt displayed bright green fluorescence, confirmed by confocal microscopy, indicating the loading of FL (Figure 2A). The Zeta potential of FL@plt showed no significant alteration compared with pristine platelets (Figure S-19B). The FL@plt remained viable and in a resting state for 8 h, as determined by the CCK-8 assay and the expression of surface CD62P, a marker of platelet activation (Figure 2B, C). Notably, the addition of probenecid markedly delayed the efflux of FL in the FL@plt over 8 h (Figure 2D). These data are compelling evidence that incubation with FDA and probenecid is an effective and efficient method for preparing FL@plt. Characterization of ultrasound-responsive FL-loaded platelets and ultrasound-triggered fluorescein discharge to co-cultured GBM cells. (A) Confocal microscopic imaging of FL@plt. (B) Viability of FL@plt in the absence or presence of PRO. (C and D) Surface expression of CD62P, an indicator of platelet activation, and FL levels in FL@plt following incubation in PBS with or without PRO. Representative flow cytometric density plots for C and D are presented in Figure S-2A & 2B. The values are the means ± SD (n = 3, **p < 0.01). (E) Experimental design for demonstrating the sono-activation of FL@Plt. (F and G) Surface expression of CD62P and FL levels in FL@plt in the absence or presence of PRO (100 µM) following sonication (UI: 1 MHz, 1 W/cm2, 1 min). (H): Experimental design for demonstrating FL@Plt discharging FL to co-cultured GL261 cells under sono-activation. (I) FL levels in in vitro GL261 cells in a mixture with FL@Plt following one sonication. (J–L) Viability, surface annexin V, and cleaved caspase 3 in in vitro GL261 cells in mixture with FL@Plt following two sonications. Cell viability was assayed using the CCK-8 test; surface expression of CD62P and annexin V were assayed using immunofluorescent staining and flow cytometry; and cleaved caspase 3 was assayed by western blotting. Representative flow cytometry density plots for F, G, I, and L are presented in Figure S-3 A, B, E, and F. Values are means ± SD (n = 3, *p < 0.05, **p < 0.01, ***p < 0.001). Quantitative analysis of the Western blot data in H is shown in Figure S-10 in the supplementary information. Upon exposure to low-intensity UI (1 MHz, 1 W/cm2, 1 min) (Figure 2E), the FL@plt exhibited immediate and pronounced activation, as indicated by a sharp increase in cell-surface CD62P, and probenecid enhanced this effect (Figure 2F). At the same time, the FL@plt exposed to sonication experienced a rapid and substantial loss of FL, regardless of probenecid (Figure 2G). In the co-incubation setting (Figure 2H), the GBM cells mixed with FL@plt exhibited pronounced FL loading only after sonication of the mixture, and probenecid markedly enhanced FL loading in the GBM cells (Figure 2I). Note that incubation with fluorescein sodium (FLS, 100 µM) did not result in significant FL loading in GBM cells, even when followed by sonication (Figure S-3C, D). Additionally, SEM revealed that the FL loading maintained the normal morphology of the platelets. TEM clearly showed internal granules within the platelets. After ultrasound irradiation, SEM images depicted platelet aggregation with intertwining pseudopodia, while TEM exhibited indistinct cell boundaries and extended pseudopodia (Figure S-19A). A second sonication of the mixture (1 h after the initial sonication) induced marked toxicity to the GBM cells, as evidenced by loss of viability (Figure 2J) and induction of apoptosis (Figure 2K, L), and probenecid augmented the cytotoxicity. Notably, sonication alone or probenecid did not cause significant platelet activation and loss of FL in FL@plt, and GBM cell-conditioned culture medium (GC-CM) could also induce significant loss of FL (Figure 2F, G). These observations compellingly demonstrate the sono-triggered drug release property of FL@plt and provide initial proof of concept for a targeted GBM-SDT strategy utilizing FL@plt and probenecid. FL@plt combined with low-intensity sonication achieved efficient tumor-targeted delivery of FL and massive tumor damage An in vitro hemolysis assay was conducted prior to animal experiments to verify hemocompatibility. As shown in (Figure S-20), the hemolysis rates of both FL@plt and FL@plt with probenecid were strictly below 2%, demonstrating their suitability for in vivo application. To demonstrate the FL@plt's ability to enable sono-controlled, GBM-targeted delivery, we used both subcutaneous and intra-cranial grafted GBM models grafted in laboratory mice. Low-intensity focused ultrasound (LIFU) was used in the subcutaneous tumor setting, as the shallow location and well-defined size of subcutaneous tumors make them suitable for LIFU. Low-intensity diffuse ultrasound (LIDU) was adopted in the intra-cranial tumor setting, as the precise location and extent of intra-cranial tumors, particularly those in the early stages, are difficult to identify in mice. The experimental design for the subcutaneous tumor setting is shown in Figure 3A. In brief, the tumor-carrying mice primed with probenecid (0.18 mmol/kg b.w.) were each injected intravenously with FL@plt (2.0 mmol/kg b.w. of FL in 106 platelets) in 200 μL of PBS. LIFU (1 MHz, 1 W/cm2, 50% duty cycle, 1 min) was applied once to each tumor at 1 h post-injection. Then, one mouse was sacrificed at 1, 3, and 5 h post-sonication, respectively, and the tumors were harvested for cryosectioning and fluorescent staining. For the control, mice without probenecid priming were also given PBS, FDA in PBS, and FL@plt in PBS, respectively. We found that FL@plt plus LIFU resulted in substantial bleeding (Figure 3B) and extensive FL distribution (Figure 3C) in the tumor tissue 1 h post-sonication. Thereafter, tumor accumulation of FL declined rapidly (Figure 3D, E). Importantly, probenecid priming markedly increased FL accumulation and delayed FL elimination in the tumor tissue (Figure 3C–E). In light of these observations, a second LIFU sonication was applied to the tumors 1 h after the initial sonication to maximize the sonodynamic effect (Figure 5A). As shown in Figure 5B, C, the tumor grafts that received FL@plt plus two LIFU sonications exhibited massive hemorrhaging, extensive necrosis, DNA damage, and apoptosis in the tumor tissue, and these observations were much more intense in the mice primed with probenecid. Notably, LIFU alone had little impact on the tumors. Targeted delivery of FL to subcutaneous GBM grafts by LIFU-actuated FL@Plt. (A) Experimental design. Briefly, subcutaneous GBM-bearing mice were intravenously injected with FL@Plt, and 1 h later, they received LIFU irradiation at the tumor site. Then, one mouse was sacrificed at 1, 3, and 5 h post-sonication. Animals injected with PBS, FDA, FL@plt without UI were used as controls. (B) H&E staining of tissue sections from tumors harvested 1 h post-sonication. (C–E) Fluorescence microscopy of tissue sections from tumors harvested 1, 3, and 5 h post-sonication. Green fluorescence indicates FL. The blue fluorescence indicates nuclear counterstaining. Higher-magnification images of (C)–(E) are presented in Figure S-4A–C. Similar observations were made in the intracranial GBM grafts to which LIDU instead of LIFU was applied (Figure 4A). FL@plt plus LIDU caused substantial intra-tumor bleeding (Figure 4B) and achieved effective FL delivery to intracranial tumors, and probenecid priming enhanced FL accumulation and delayed FL elimination in the tumor tissue (Figure 4E, F). It should be noted that the intracranial tumors exhibited pronounced FL accumulation 1 h post-sonication, but the FL distribution appeared more extensive 2 h post-sonication (Figure 4C–E). Taking these observations into account, a second LIDU sonication was applied to the tumor sites 1.5 h after the initial sonication to maximize the sonodynamic effect (Figure 5D). As shown in Figure 5E and F, tumor grafts that received FL@plt plus two LIDU sonications exhibited massive hemorrhaging, extensive necrosis, DNA damage, and apoptosis in the tumor tissue, and these observations were enhanced in the mice primed with probenecid. TUNEL staining consistently showed that FL-SDT induced the most extensive apoptosis among all the treatment conditions (Figure S-21 & 22). Targeted delivery of FL to intracranial GBM grafts by LIDU-actuated FL@Plt. (A) Experimental design. Briefly, intracranial GBM-bearing mice were intravenously injected with FL@Plt or FL@Plt + PRO, and 1 h later, they received LIDU irradiation at the tumor site. Then, one mouse was sacrificed at 0.5, 1, and 2 h post-sonication. Animals injected with PBS, FL@plt without UI were used as controls. (B) H&E staining of tissue sections from brains harvested 1.0 h post-sonication. (C–E) Fluorescence microscopy of tissue sections from brains harvested 0.5, 1, and 2 h post-sonication. Green fluorescence indicates FL. The blue fluorescence is nuclear counterstaining. Higher-magnification images of (C)–(E) are presented in Figure S-5A–C. Sono-toxicity to subcutaneous and intracranial GBM grafts induced by LIFU- and LIDU-actuated FL@Plt, respectively. (A) Experimental design. Briefly, subcutaneous GBM-bearing mice were intravenously injected with FL@Plt, and 1 h later, they received LIFU irradiation, later received twice UI at the tumor site at an interval of 1 h. The animals were sacrificed 24 h after the second UI. Animals injected with PBS, FDA, LIFU × 2, and FL@Plt without UI were used as controls. (B) H&E staining of tissue sections from tumors harvested 24 h post-2nd sonication. (C) IHC staining of γH2AX and caspase 3 in tissue sections from tumors harvested 24 h post-2nd sonication. (D) Experimental design. Briefly, intracranial GBM-bearing mice were intravenously injected with FL@Plt or FL@Plt + PRO, and 1 h later, they received LIDU irradiation, 1.5 h later, they received twice at the tumor site. The animals were sacrificed 24 h after the second UI. Animals injected with PBS, FDA and FL@Plt without UI were used as controls. (E) H&E staining of tissue sections from brains harvested 24 h post-2nd sonication. (F) IHC staining of γH2AX and caspase 3 in tissue sections from brains harvested 24 h post-2nd sonication. Quantitative analysis of the IHC staining data in C&F is shown in Fig. S-6 in the supplemental information. FL@plt combined with low-intensity sonication achieved anti-GBM efficacy Mouse intracranial GBM models were used to demonstrate the anti-GBM efficacy of FL@Plt-mediated SDT. An SDT regimen (Figure 6A) was designed based on the findings presented in (Figures S3–S5). Briefly, GBM-bearing mice were divided into three groups. Group 3 was primed with probenecid (0.18 mmol/kg b.w.). All groups of mice were intravenously injected with FL@Plt (2.0 mmol/kg b.w. of FL in 106 platelets) in 200 μL of PBS per mouse. Animals in groups 2 and 3 were then subjected to 3 LIDU sonications (1 MHz, 1 W/cm2, 1 min) at the tumor site (Figure 6A​​​​​​). The first sonication was performed 1 h after FL@Plt injection to induce FL@Plt activation and FL release in the GBM vasculature. One and a half hours later, once the released FL had distributed to the GBM cells, two further LIDU sonications were conducted 1 h apart to achieve SDT efficacy. This treatment protocol was executed twice at 7-day interval. As shown in Figure 6B–E, two sessions of FL@Plt-mediated SDT significantly slowed GBM growth, prolonged host survival, and prevented the loss of body weight. Notably, two out of the five animals that received FL@Plt-mediated PDT plus probenecid priming were still alive on day 45 after therapy began, whereas all the other animals had already died by this point. MRI scanning performed on day 30 of therapy revealed reduced tumor sizes and significant tissue necrosis in the GBMs that had undergone FL@Plt-mediated SDT, especially when primed with probenecid (Figure 7A). Gross examination of harvested brains showed better preserved brain morphology, as well as much alleviated brain tissue swelling and bleeding, in mice that received FL@Plt-mediated SDT, particularly when primed with probenecid (Figure 7B). Consistently, H&E staining showed a reduced tumor mass and extensive tumor tissue necrosis with no apparent brain tissue injury after FL@Plt-mediated SDT, regardless of probenecid priming (Figure 7C). It should be noted that, in order to minimize animal sacrifice, LIDU sonication alone was not included as a treatment in the therapy experiment, since LIFU sonication with the same parameters did not exert any anti-tumor activity on subcutaneous tumors (Figure 4B). Therapeutic efficacy of GBM-SDT mediated by LIDU-actuated FL@Plt. (A) Experimental design. Briefly, three groups of GBM-bearing animals were intravenously injected with platelets FL@Plt and FL@Plt + PRO. The animals were then subjected to 3 LIDU sonications (1 MHz, 1 W/cm2, 1 min) at the tumor site. The treatment protocol was executed twice at 7-day interval. FL@plt without LIDU was used as a control. (B) In vivo luminescence of brain tumors was taken on days 1, 7, and 14 during therapy. (C) Growth of brain tumor-derived luminescence from day 1 to day 14. GBM-derived luminescence on day 1 and growth rates are listed in Figure S-7A, B. (D) Survival analysis of treated animals. (E) Body weight of treated animals was monitored over the therapy duration. Body weight growth rates were calculated by regression analysis. The values are the means ± SD (n = 5). Morphological and pathological findings of brain tumors following SDT mediated by LIDU-actuated FL@Plt. (A) Brain MRI scan images taken 30 days after therapy began. The areas circled by red dotted lines indicate the location of the brain tumors. Necrotic areas in the tumor tissue are indicated by yellow arrowheads. (B) Gross examination of resected brains at the end of therapy. (C) H&E staining of tissue sections of resected brains at the end of therapy. FL-SDT promoted HIF-1α degradation via the ubiquitin-proteasome system and the autophagy-lysosome pathway, contributing to FL-SDT-induced DNA damage Inflicting DNA damage is a common mechanism of SDT cytotoxicity. However, cancer cells have sophisticated DNA damage repair machinery that can be quickly mobilized to counteract SDT cytotoxicity and promote survival (Jurkovicova et al. 2022). HIF-1α is a key transcription factor that plays a central and multifaceted role in the response to DNA damage in cancer cells, particularly within the hypoxic TME. HIF-1α is quickly upregulated under hypoxia or in response to DNA damage, thereby enhancing the expression and activity of multiple DNA repair components and promoting therapeutic resistance. (Rezaeian et al. 2017). In a previous study, we found that mouse GL261 cells increased the expression of HIF-1α and several key DNA repair enzymes under its regulation, such as single-strand selective monofunctional uracil DNA glycosylase 1 (SMUG1) and DNA ligase 4 (LIG4), in response to photodynamic therapy (PDT)-induced DNA damage (Guo et al. 2025). In the present study, however, we were surprised to find that the same GBM cells, whether in vitro or in vivo, exhibited decreased expression of HIF-1α (Figures 8D, S-9) and its downstream targets, particularly SMUG1 and LIG4 (Figures 8G–I, S-9), despite undergoing pronounced DNA damage following FL-SDT (Figures 4C, F, 8A–C). We also detected the downregulation of the HIF-1 signaling pathway via KEGG pathway enrichment analysis (Figure 8F). We then mimicked a hypoxic response in the in vitro GBM cells using CoCl₂. CoCl₂ stabilizes HIF-1α by blocking its degradation via the ubiquitin‒proteasome pathway, leading to HIF-1α accumulation and subsequent activation of downstream hypoxia-responsive genes (Meng et al. 2025). As shown in Figure 8 J–L, CoCl₂ increased the expression of HIF-1α, SMUG1, and LIG4; this effect could be antagonized by FL-SDT. These observations suggest that FL-SDT might compromise HIF-1α and the DNA repair components under its regulation, thereby contributing to FL-SDT-induced DNA damage. FL-SDT induced DNA damage and suppressed HIF-1α signaling and DNA repair enzymes downstream of HIF-1α. (A and B) FL-SDT-induced DNA damage, as indicated by the comet assay and its quantification. (C) FL-SDT induced the upregulation of γ-H2AX and p-ATR. (D) FL-SDT reduced the protein level of HIF-1α. (E) FL-SDT had little effect on the HIF-1α mRNA level. (F) The top 18 enriched KEGG pathway terms of downregulated genes in GL261 cells exposed to SDT mediated by FL under UI (1 MHZ, 1 W/cm2, 1 min). (G and H) FL-SDT significantly reduced mRNA levels of VEGFA and CA9. (I) FL-SDT reduced the protein levels of LIG4 and SMUG1. (J) FL-SDT suppressed the induction of HIF-1α, LIG4, and SMUG1 proteins caused by CoCl₂. (K and L) mRNA levels of Smug 1 (K) and LIG 4 (L) when CoCl2 was used as a control treatment. (The values are the means ± SD (n = 3, *p < 0.05, **p < 0.01, ***p < 0.001). Quantitative analysis of the Western blot data in C, D, I & J are shown in Figures S-11–14 in the supplementary information. As FL-mediated SDT decreased the level of the HIF-1α protein without significantly affecting its gene transcription (Figure 8D, E), we suspected that FL-mediated SDT might promote HIF-1α degradation. HIF-1α is primarily degraded via the ubiquitin‒proteasome pathway under normoxic conditions and via the autophagy‒lysosome pathway under hypoxic conditions (Chen et al. 2020). In our study, we observed a significant build-up of ubiquitinated HIF-1α amidst an increase in total ubiquitin in GBM cells following FL-SDT (Figure 9A, S-9). At the same time, we identified substantial co-localization of HIF-1α with ubiquitin and the proteasomes (Figure 9B, C). Furthermore, bortezomib (PS-341), a proteasome inhibitor that disrupts the ubiquitin-proteasome pathway, was shown to counteract the reduction in HIF-1α caused by FL-SDT (Figure 9B, C). Meanwhile, the GBM cells exhibited elevated autophagic flux following FL-SDT, as indicated by a build-up of autophagosomes (Figure 1J), increased LC3 II (Figures 10A and S-9), and decreased P62 (Figure 10A). We also observed substantial co-localization of HIF-1α with autophagosomes and lysosomes in GBM cells following FL-SDT (Figure 10B, C). Ritonavir (RTV), an autophagic flux blocker that inhibits autophagolysosomal maturation and autophagic proteolysis (Cheney et al. 2021), was found to counteract the reduction in HIF-1α caused by FL-mediated SDT (Figure 10D). These findings suggest that FL-SDT may promote HIF-1α degradation by activating both the ubiquitin‒proteasome system and the autophagy‒lysosome pathway. FL-SDT promoted HIF-1α degradation via the ubiquitin‒proteasome system. (A) Co-immunoprecipitation (CO-IP) analysis of ubiquitinated HIF-1α. (B) Immunofluorescence co-localization of HIF-1α and ubiquitin. (C) Immunofluorescence co-localization of HIF-1α and proteasome subunit beta (PMSB). The Pearson correlation coefficient (r) is used to measure the degree of linear correlation between two variables, with larger absolute values indicating stronger correlation. (D) Bortezomib mitigated the FL-SDT-induced decrease of HIF-1α. Quantitative analysis of the Western blot data in D are shown in Figure S-15 in the supplemental information. FL-SDT promoted HIF-1α degradation via the autophagy‒lysosome pathway. (A) Expression of the autophagy markers Beclin 1, P62, and LC3 II/I. (B) Immunofluorescence co-localization of HIF-1α and LC3-II, a marker of autophagosomes. (C) Immunofluorescence co-localization of HIF-1α and lysosome-associated membrane protein 1 (LAMP1). The Pearson correlation coefficient (r) is used to measure the degree of linear correlation between two variables, with larger absolute values indicating a stronger correlation. (D) Retonavir mitigated the FL-SDT-induced decrease of HIF-1α. Quantitative analysis of the Western blot data in A&D is shown in Figure S-16 in the supplemental information. Discussion Recent studies have identified FL as a safe, low-cost and efficient sonosensitizer. However, the systemic administration of FL for SDT of GBM has rarely been explored, largely due to its poor distribution in the GBM following peripheral administration. To address this challenge, we have devised an ultrasound-responsive, FL-loaded platelet (FL@plt), which, when used alongside low-intensity sonication, achieved adequate FL accumulation and effective SDT in mouse GBM models. The key to this strategy lies in ultrasound-responsive FL@plt, which can be readily prepared by incubating isolated platelets with a mixture of FDA and probenecid in PBS at room temperature for 30 min. The ease with which FL@plt is prepared is a major advantage of this study. As demonstrated by the results, FL@plt remains viable, stable, and functional for at least 8 h after preparation, allowing a wide time window for subsequent procedures. It should be noted that FDA is frequently used to assess cell viability, and the stable loading of FL itself is an indicator of cellular wellness, as only healthy cells have the necessary esterase activity to convert FDA into FL in terms of quantity and the plasma membrane integrity to prevent leakage of intracellular FL (Armour et al. 2008). However, platelets can still extrude intracellular FL, probably via the transporter protein MRP1 in the cell membrane. This is where probenecid comes into play. Probenecid is an approved drug for treating gout, and it competitively inhibits MRP1, blocking its efflux activity and ATPase function (Zhou et al. 2008). In our study, the addition of probenecid indeed markedly slowed the spontaneous release of FL from FL@plt (Figure 2D) but did not affect sonication-induced FL discharge from FL@plt (Figure 2G). More importantly, MRP1 is also constitutively expressed in BBB endothelial cells and often overexpressed in GBM cells, contributing to the intrinsic chemoresistance of GBM (Gomez-Zepeda et al. 2019). We therefore supposed that probenecid would help to trap the delivered FL in the GBM. As expected, probenecid was found to delay the release of FL by GBM cells (Figure 1D) and enhance FL accumulation in GBM cells (Figures 3, 5), leading to enhanced sonotoxicity to GBM cells in vitro and in vivo (Figures 1, 2, and 4) and improved SDT efficacy (Figures 6, 7). Note that probenecid was used at a dosage of 0.18 mmol/kg b.w. in our study, which is well below the dosage (0.35–0.90 mmol/kg b.w.) used for treating gout in mouse models. We have demonstrated effective delivery of FL and subsequent sonodynamic damage in both subcutaneous and intracranial GBM models mediated by FL@plt. LIFU was used for the subcutaneous tumors because of their shallow location and defined size, while LIDU was used to cover the intracranial tumors, as its exact location and extent, particularly in the early stages, is difficult to identify. Similar results were obtained in both settings, despite the different sonication methods used. Both subcutaneous and intracranial GBM grafts exhibited substantial accumulation of FL following sonication, albeit with slightly different distribution kinetics. FL turnover appeared to be faster in the subcutaneous tumors. Notably, there was no residual FL in the intracranial GBM at the end of therapy. This contrasts with our previous studies utilizing sonosensitizer nanoparticles, in which a significant residue of the nanoparticles persisted in the intracranial GBM until the end of therapy (Xu et al. 2022). Additionally, our approach showed no signs of systemic adverse effects (Figures S-8, S-24, and 25), and FL is rapidly cleared, primarily via hepatic and renal pathways (Figure S-23). From a clinical translation perspective, faster and more complete tissue clearance may necessitate a higher treatment frequency but would afford better safety. In clinical practice, adopting LIFU under MRI guidance could further increase therapeutic precision and safety. Hypoxia is a common feature of the TME, which arises because rapid tumor growth outpaces the development of adequate vasculature, leading to regions with insufficient oxygen supply. Hypoxia in tumors generates ROS, which are highly reactive molecules that can damage cellular components, including DNA (Zhao et al. 2023). ROS-induced oxidative DNA damage, such as base modifications and strand breaks, contributes to genomic instability. This damage is exacerbated during reoxygenation due to a surge in ROS levels (Bohr 2002). However, hypoxia, hypoxia-induced ROS, and DNA damage all activate HIF-1α, a key transcription factor that promotes angiogenesis, DNA repair, cancer cell survival, and migration (Rohwer et al. 2013; Hielscher and Gerecht 2015). SDT and PDT generate excessive amounts of ROS, disrupt the tumor vasculature, and deplete oxygen in the TME. These actions aggravate hypoxia in the TME and thereby stimulate further HIF-1α activation in tumor cells. In a previous study, we indeed found increased HIF-1α expression and activity in response to DNA damage in GBM cells (GL261) following PDT, and the activation of HIF-1α signaling led to increased DNA repair activity, such as heightened expression of SMUG 1 and LIG4 (Guo et al. 2025). However, in the present study, the GBM cells exhibited depressed expression of HIF-1α and SMUG 1 and LIG4 following FL-SDT, despite significant DNA damage (Figures 8, S-9). RNA-seq analysis additionally showed fluorescein sonodynamic therapy downregulated the HIF-1 pathway versus controls (Figure 8F). These findings highly suggest that the substantial DNA damage following the FL-SDT is due to a combination of direct disruption of DNA and impaired DNA repair capacity. Further results indicate that the FL-SDT promoted the elimination of HIF-1α rather than suppressing HIF-1α gene transcription (Figures 8–10). The regulation of HIF-1α involves several oxygen-dependent and oxygen-independent degradation routes. The oxygen-dependent ubiquitin-proteasome system (UPS) dominates under normoxia, while the autophagy-lysosome pathway (ALP) provides a major alternative route, especially in hypoxic or stressed conditions (Brahimi-Horn and Pouysségur 2005; Chen et al. 2020). In our study, the obtained data suggest that FL-SDT activated both the UPS and the ALP under normoxia (Figures 9 and 10). Notably, FL-SDT still promoted HIF-1α clearance when the UPS was blocked by CoCl₂, a hypoxia-mimicking agent (Figure 8J). These findings suggest that the ALP probably drives HIF-1α degradation under hypoxia following FL-SDT. It is worth mentioning that HIF-1α acts as a master switch in the cellular response to hypoxia and stress, regulating functions that extend far beyond DNA repair. Reduced HIF-1α expression or activity would diminish the overall capacity of tumor cells to survive and progress. As the FDA-derived FL exhibited nearly homogeneous intracellular distribution (Figure S-1A), it is unlikely that HIF-1α was the only protein affected by FL-SDT. The activation of the UPS and the ALP was probably a response to eliminate oxidatively damaged proteins at a global scale, rather than just HIF-1α. Impaired clearance of damaged proteins would wreak further disruption in the tumor cells, thus enhancing the cytotoxicity of FL-SDT. Therefore, it is expected that blocking the UPS and/or the ALP will enhance the efficacy of FL-SDT. The FL@plt approach can easily be extended to simultaneously deliver multiple therapeutic agents via a strategy demonstrated in our previous study. We are currently exploring this co-delivery strategy using bortezomib (a proteasome inhibitor) and ritonavir (an inhibitor of autophagic flux). Both drugs are approved for clinical use. Conclusions In summary, incubating FL@plt with a mixture of FDA and probenecid is a simple and convenient method of preparing ultrasound-responsive FL@plt. When applied with a low dose of probenecid and low-intensity sonication, FL@plt can effectively target GBM for SDT. The therapeutic mechanism of FL-SDT critically involves impairing HIF-1α-regulated DNA damage repair capabilities. This platelet-based platform represents a practical, safe strategy for glioblastoma with strong clinical translation potential. Supplementary Material Author_Checklist_E10_only__1_.pdf Supplementary Material Supplementary material20260519.docx Supplementary material20260519.docx