Co-delivery of doxorubicin and quercetin with Angiopep-2 modified graphene oxide for targeted chemo-photothermal therapy of glioblastoma

Abstract

Glioblastoma multiforme (GBM) is an aggressive brain tumor with limited treatment options due to poor drug penetration across the blood-brain barrier (BBB). As combined treatment with doxorubicin and quercetin exhibited synergistic cytotoxic effects against U87 glioblastoma cells, graphene oxide (GO) can be employed to deliver these drugs, thereby enhancing their bioavailability during chemotherapy. GO can also serve as an efficient photothermal agent under near-infrared (NIR) irradiation. In this study, we functionalize GO with the Angiopep-2 peptide ligand to prepare a GOA nanocarrier. After co-loading quercetin and doxorubicin, the GOA/Q/D nanoplatform can enable BBB crossing/tumor targeting, pH-responsive drug delivery, and efficient NIR-mediated photothermal conversion for targeted chemo-photothermal therapy of GBM. Physicochemical characterization confirmed successful nanoparticle fabrication, favorable stability, and efficient drug loading and controlled drug release. Confocal microscopy and flow cytometry analyses indicate enhanced cellular uptake of GOA by U87 cells and intracellular doxorubicin accumulation. The integration of synergistic chemotherapy and NIR-induced photothermal therapy with GOA/Q/D resulted in significantly increased apoptosis and tumor cell death. In vivo studies confirmed Angiopep-2-mediated brain accumulation in healthy and tumor-bearing nude mice after intravenous administration. Therapeutic evaluation further demonstrated significantly inhibited tumor growth and prolonged survival in glioma-bearing mice treated with GOA/Q/D through intravenous injection plus NIR laser irradiation at the tumor site. Collectively, the integration of targeted delivery, synergistic dual-drug chemotherapy, and photothermal activation resulted in markedly improved therapeutic efficacy, underscoring the potential of chemo-photothermal therapy for the targeted treatment of glioblastoma.

Graphical abstract

1. Introduction

Glioblastoma multiforme (GBM), which accounts for up to 50% of all gliomas, is the most prevalent and aggressive primary brain tumor. Even with advancements in the current standard of care, the outcome for patients is still often fatal, with a median overall survival that ranges from 14.6 to 20.5 months (Wen and Reardon, 2016). Although chemotherapy remains the primary treatment modality, its clinical efficacy is frequently hampered by two major challenges: the restrictive nature of the blood-brain barrier (BBB) and the systemic toxicity associated with anticancer medications. The BBB is a highly selective barrier formed by tightly connected endothelial cells that limits the entry of the therapeutic agent into the brain. While this serves as a protective mechanism for normal brain tissue, it also limits the accumulation of chemotherapeutic agents at the tumor site, compromising treatment outcomes (Kim et al., 2015). The BBB can shield the brain from toxins and infections and preserve neural homeostasis under normal physiological conditions. However, its integrity may be jeopardized in pathological conditions like cancer, trauma, or neuroinflammation (Ter Linden et al., 2024). The structure and function of the BBB can be significantly altered in GBM due to tumor growth and related changes in the microenvironment.

Drug delivery using nano-sized carriers has been a successful, transformative platform in cancer therapy. By capitalizing on the unique pathophysiological features of tumor tissues, these nanocarriers may overcome key limitations, including poor solubility, low bioavailability, and severe off-target toxicity in conventional chemotherapy for GBM treatment (Liu et al., 2024). To improve therapeutic precision and reduce off-target effects, specific surface receptors that are overexpressed on glioblastoma cells and absent or expressed at much lower levels in normal brain tissue can be targeted for selective drug delivery to reduce side effects on healthy cells (Sun et al., 2023). Another significant characteristic of these systems is their capacity to provide stimulus-responsive drug release. Drugs can be released locally and under control within the tumor microenvironment by engineering nanocarriers to react to tumor-specific stimuli like enzymatic activity or pH changes. This targeted release lowers systemic exposure and maintains drug concentration at the tumor site (Wang et al., 2024). Furthermore, combination therapy is possible due to the nanocarrier's ability to co-deliver multiple therapeutic agents to concurrently target different molecular pathways in GBM growth and resistance (Yalamarty et al., 2023). This approach can enhance treatment outcomes and get around the drawbacks of traditional monotherapies.

Graphene oxide (GO), an oxygen-rich graphene derivative, has a honeycomb-lattice architecture. It is highly biocompatible and provides a large surface area for loading a wide range of drugs and biomolecules. In 2008, GO was used for the first time in the biomedical field as a novel nanocarrier for delivering poorly water-soluble anticancer drugs like doxorubicin (DOX) and SN-38 (Liu et al., 2008). With its two-dimensional sheet-like structure, GO provides an ideal platform for the co-delivery of multiple therapeutic agents, allowing for synergistic treatment strategies. A key advantage of GO is its high surface-to-volume ratio, which allows for high loading of multiple chemotherapeutic drugs through π–π stacking, hydrogen bonding, and electrostatic interactions with the numerous oxygen-containing functional groups on GO for chemotherapy (CT) (Karki et al., 2020). Additionally, the pH-sensitive release profiles of these drugs from drug-loaded GO in acidic intracellular compartments such as endosomes enable efficient drug delivery after intracellular uptake to maximize cytotoxicity toward cancer cells (Dash et al., 2021). The GO also provides facile surface functionalization ability by PEGylation to increase its suspension ability, which can be further conjugated with targeting ligands for active targeting of cancer cells. Furthermore, with the intrinsic photothermal property offered by GO, it can be used for photothermal therapy (PTT). These characteristics have made functionalized GO a promising nanoplatform for targeted combination cancer therapy (Shi et al., 2026).

Doxorubicin (DOX) is known to primarily exert its anticancer activity by intercalating into DNA base pairs, thereby disrupting macromolecular synthesis (Pommier et al., 2010). Topoisomerase II, an enzyme that relieves DNA supercoiling during transcription and replication, is a primary target of DOX. By stabilizing the topoisomerase II-DNA cleavage complex, DOX inhibits DNA replication by preventing DNA strands from re-ligating, which ultimately results in apoptosis (Tacar et al., 2013). However, acting as critical components of the BBB, efflux transporters are abundantly expressed on the brain capillary endothelial cells, which can actively pump foreign substances like DOX out of the central nervous system (Wu et al., 2023). Therefore, the therapeutic concentration of DOX in the brain is considerably decreased by these efflux mechanisms. Quercetin (3,3′,4′,5,7-pentahydroxyflavone) is a naturally occurring flavonoid from the flavonol subclass, widely present in common dietary sources such as tea, red grapes, onions, apples, broccoli, and various berries. Known for its low toxicity, quercetin exhibits a broad spectrum of pharmacological activities, including antioxidant, anti-inflammatory, and anticancer effects (Hashemzaei et al., 2017). Its anticancer efficacy has been demonstrated in multiple human cancer cell lines, including glioma, leukemia, breast, colon, prostate, cervical, hepatoma, and osteosarcoma cells (Atashpour et al., 2015). Mechanistically, quercetin exerts its antitumor activity through multiple pathways: it induces apoptosis by modulating pro- and anti-apoptotic proteins such as Bax and Bcl-2, triggers cytochrome c release, and activates caspase-9 and caspase-3. It also inhibits cell proliferation by modulating signaling cascades and arresting the cell cycle, while regulating key proteins like p21^WAF1 and p27^Kip1 (Zhao et al., 2019). Furthermore, quercetin functions as a multidrug resistance modulator by downregulating efflux transporters such as MRP1, BCRP, and P-gp, and inhibiting the drug-metabolizing enzyme CYP3A4. These actions enhance the therapeutic efficacy and bioavailability of co-administered chemotherapeutics, even at non-toxic doses (Azizi et al., 2022). Recent research has highlighted that dual-drug regimens, particularly those combining conventional chemotherapeutics with natural chemosensitizers, can further enhance treatment effectiveness and reduce side effects (Deng et al., 2025). Several studies indicate the advantages of co-delivery of DOX and quercetin for treating various cancers in vitro and in vivo (Chang et al., 2018; Liu et al., 2020; Lv et al., 2016; Pan et al., 2024; Ramasamy et al., 2017; Sun et al., 2020; Tzankova et al., 2024; Voycheva et al., 2023). Together, these studies demonstrate that co-delivery of DOX and quercetin in well-designed nanocarrier systems can provide synergistic effects and improve antitumor efficacy by overcoming resistance mechanisms in different cancer models in vitro and in vivo. Building on these insights, our work exploits this dual-drug combination in U87 glioblastoma cells to maximize anti-tumor potency against GBM.

While nanocarriers have shown significant promise in facilitating drug delivery across the BBB, non-targeted nanocarriers often suffer from limited accumulation in tumor tissue and widespread distribution in non-target organs, which can reduce therapeutic efficacy and increase systemic toxicity. In contrast, targeted delivery systems are designed to recognize and bind to specific receptors overexpressed on BBB endothelial cells or tumor cells, thereby enhancing the selectivity, uptake, and retention of therapeutic agents at the disease site. Among the various targeting strategies, ligand-mediated targeting has emerged as a powerful method to achieve brain- and tumor-specific delivery with the ligand's ability to bind specifically to tumor cells or overexpressed receptors on the BBB (Sethi et al., 2022). Angiopep-2, a 19-amino acid peptide that is derived from the Kunitz domain of aprotinin, has a strong affinity for low-density lipoprotein receptor-related protein 1 (LRP1), which is scarce in normal brain tissue but highly expressed on BBB endothelial cells and markedly overexpressed in glioblastoma and other brain metastases (Miida and Hirayama, 2009; Papademetriou et al., 2018). Angiopep-2 exhibits better tumor targeting and transcytosis efficiency than traditional ligands like transferrin and lactoferrin, which makes it a prime candidate for LRP1-mediated dual-targeting nanocarrier systems in glioma therapy (Jiang et al., 2012). In comparison to its parent molecule, aprotinin, which is also recognized by LRP1 for transport across endothelial barriers, Angiopep-2 showed noticeably higher BBB transcytosis and brain tissue accumulation. Many studies have highlighted the significance of Angiopep-2 in mediating tumor-specific uptake in GBM as well as BBB penetration (Cai et al., 2024; Endo-Takahashi et al., 2016; Kadari et al., 2018; Shi et al., 2020; Wei et al., 2012; Xu et al., 2021).

The GO can be used in GBM therapy for targeted drug delivery or to increase the effectiveness of photothermal and photodynamic therapies (Einafshar et al., 2026). Cyclodextrin-modified GO has been used for loading DOX/Temozolomide (Einafshar et al., 2024) or selenium (Kazemian Kakhki et al., 2024) for glioblastoma treatment. Protoporphyrin IX-conjugated GO was used for photodynamic therapy of cervical cancer (Alikhanzadeh et al., 2025). With its strong near-infrared (NIR) absorbance and high photothermal conversion efficiency, GO is widely used as a photothermal agent for cancer treatment (Báez, 2023). It can be employed in PTT, using localized hyperthermia to destroy tumor cells. The heat generated by GO during PTT can disrupt cellular structures, denature proteins, and trigger apoptosis or necrosis (Han and Choi, 2021). This allows for the selective ablation of tumor cells while sparing surrounding healthy tissues, thereby improving therapeutic efficacy. Several studies have indicated that PTT alone often fails to eliminate tumor cells because of uneven heat distribution within the tumor and the limited penetration depth of light, which can result in tumor recurrence. To overcome these limitations, a combined approach involving CT and PTT has been suggested as an effective strategy to enhance cancer cell eradication and reduce recurrence risk. This combined treatment not only enhances the responsiveness of tumors to CT but also facilitates better drug uptake by cancer cells and enables controlled drug release (Liu et al., 2021). A GO nanoplatform loaded with DOX exhibited strong absorption in the NIR region and generated localized hyperthermia upon 808 nm laser irradiation (Zhang et al., 2011). The heat from PTT not only induced direct tumor cell death but also enhanced cellular uptake and cytotoxicity of DOX by increasing cell membrane permeability. In vitro and in vivo experiments demonstrated that the combined therapy significantly outperformed either CT or PTT alone in suppressing tumor growth. However, it should be noted that mild PTT operated at a lower, more controlled temperature level (< 50 °C) can provide enhanced safety and reduce severe tissue damage and necrosis side effects. Most importantly, mild heat can normalize tumor mechanics and enhance blood perfusion, increasing the accumulation of nanomedicines within the tumor rather than destroying the blood vessels that deliver them (Li et al., 2024). Mild PTT with GO as a photothermal agent can provide a highly effective therapeutic modality by combining with CT, which is expected to overcome the limitations of single-modality therapies, such as drug resistance and tumor hypoxia (He et al., 2023).

Herein, we designed a multifunctional GO-based nanoplatform that integrates Angiopep-2-mediated BBB crossing and tumor targeting for dual-drug (DOX and quercetin) CT and PTT. This integrated platform is designed to overcome BBB-associated delivery barriers, improve therapeutic selectivity, and enhance anti-tumor efficacy through a combination of targeted CT/PTT (Scheme 1).

Scheme 1.

(A) GO was functionalized with Angiopep-2 and loaded with doxorubicin and quercetin to prepare GOA/Q/D. GOA/Q/D can enable tumor targeting (b), pH-responsive drug delivery (c), and NIR-mediated photothermal heating (d) for targeted chemo-photothermal therapy of glioblastoma (e).

2. Materials and methods

2.1. Materials

The graphene oxide was obtained from CL Technology Co., Ltd. Doxorubicin (HCl salt) was obtained from LC Laboratories. Quercetin, Cy5.5-NH₂, and 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) were obtained from Sigma-Aldrich (St. Louis, MO, USA). Pierce Immobilized TCEP Disulfide Reducing Gel and bicinchoninic acid (BCA) protein assay kit were obtained from Thermo Fisher Scientific (Waltham, MA, USA). For cell culture studies, Dulbecco's modified Eagle's Medium (DMEM), fetal bovine serum (FBS), and 0.25% trypsin-EDTA and penicillin-streptomycin were used. Angiopep-2 (TFFYGGSRGKRNNFKTEEY-Cys) was supplied by BioTools (Taiwan). 1,2-Distearoyl-sn-glycero-3-phosphoethanolamine-N-[maleimide(polyethylene glycol)-2000] (DSPE-PEG₂₀₀₀-Mal) was obtained from Avanti. Annexin V- FITC/Propidium iodide (PI) apoptosis detection kit and LysoTracker green DND-26 were obtained from Invitrogen. The U87 human primary glioblastoma cell line, which was transfected with the firefly luciferase gene, was used for the in vitro and in vivo studies.

2.2. Preparation of GOA

Angiopep-2 was conjugated to GO via thiol-maleimide chemistry to prepare GOA (Awad et al., 2024). To introduce maleimide groups, GO was PEGylated using DSPE-PEG₂₀₀₀-Mal via hydrophobic interactions (Dash et al., 2021). Briefly, GO (1 mg/mL) was mixed with 0.5 mg/mL DSPE-PEG₂₀₀₀-Mal (mass ratio = 2:1) to a volume of 2 mL in water and sonicated for 30 min to attach the DSPE moiety in DSPE-PEG₂₀₀₀-Mal to the GO surface. The PEGylated GO was centrifuged and redispersed in 0.5 mL of water for conjugation with the -SH group in the cysteine residue in the peptide. To break possible disulfide bonds between the peptide chains, 100 μL of 1 mg/mL Cys-modified Angiopep-2 peptide solution was mixed with an equal volume of immobilized TCEP resin and incubated at 4 °C for 1 h under gentle rotation. The mixture was centrifuged, and the supernatant containing reduced Angiopep-2 (with free –SH groups) was collected for conjugation. The reduced Angiopep-2 solution (100 μL) was added to the prepared PEGylated GO in water (mass ratio = 1:5), made up to 1 mL, and allowed to react for 1 h at 4 °C under gentle mixing to enable the thiol-maleimide click reaction. To avoid possible interference from residual maleimide groups, TCEP-treated L-cysteine (mass ratio of L-cysteine to nanoparticles = 100:1) was added to the prepared nanoparticles, which can provide excess –SH groups to block residual maleimide groups on nanoparticles. Finally, Angiopep-2-conjugated GO (GOA) was recovered by centrifugation and dispersed in phosphate-buffered saline (PBS).

2.3. Preparation of GOA/Q/D

0.5 mg of 500 μL GOA dispersion in PBS was taken in an Eppendorf tube. Different volumes of quercetin solution (in ethanol) were added to make a total volume of 1 mL. The mixture was rotated at 4 °C in the dark for 24 h to allow drug adsorption. After incubation, the mixture was centrifuged, and the absorbance of the collected supernatant was measured at 373 nm to determine the amount of free quercetin. Based on this, the loading efficiency and loading content were calculated. After fixing the optimal quercetin concentration, DOX was loaded onto the quercetin-loaded GOA (GOA/Q) using the same method. Following 24 h of rotation at 4 °C, the supernatant was collected and measured at 490 nm to quantify the free DOX. The loading efficiency (LE) and loading content (LC) were calculated as:

$$\text{Loading Efficiency}\left(\mathrm{LE}\right)\left(\%\right)=\frac{\text{Weight of loaded drug}\left(\mathrm{mg}\right)}{\text{Total weight of drug}\left(\mathrm{mg}\right)}\times 100$$ (1)

$$\text{Loading Content}\left(\mathrm{LC}\right)\left(\frac{\mathrm{mg}}{\mathrm{mg}}\right)=\frac{\text{Weight of loaded drug}\left(\mathrm{mg}\right)}{\text{Total weight of}\mathrm{GOA}\left(\mathrm{mg}\right)}$$ (2)

2.4. Characterization techniques

The GO, GOA, and GOA/Q/D solutions were prepared at 1 mg/mL and added dropwise to a copper grid. The solution was slowly absorbed using filter paper, and the prepared sample was analyzed by high-resolution transmission electron microscopy (HR-TEM, JEOL JEM-2100Plus) at 80 kV to characterize particle morphology. Energy-dispersive X-ray spectroscopy (EDS) was performed using Oxford INCA X-Max between 5 kV and 15 kV to analyze the composition of carbon, oxygen, nitrogen, and sulfur elements. Both point analysis and elemental mapping were conducted to determine the presence and spatial distribution of the elements within the nanoparticles. The hydrodynamic size and zeta potential of GO, GOA, and GOA/Q/D were measured using a dynamic light scattering (DLS) analyzer (Zetasizer, Nano ZS 90, Malvern, UK). The concentrations of GO, GOA, and GOA/Q/D were diluted to 100 μg/mL in PBS and sonicated for 1 h. Each measurement was performed at 25 °C, and results were recorded as the average of three independent measurements. The Fourier Transform Infrared (FTIR) spectra of GO, GOA, and GOA/Q/D were recorded using an FTIR spectrometer (Bruker Tensor 27) in the range of 4000–400 cm⁻¹. The powders of each sample were mixed with potassium bromide (KBr) and compressed to form pellets. Blank KBr was made as a pellet and used as a control. Solutions of GO, GOA, and GOA/Q/D were made into powder by freeze-drying, and Raman spectra were measured using a Raman spectroscopy system (UniDRON, UniNano Tech) with a 532 nm wavelength laser at 25 W. X-ray diffraction (XRD) patterns of GO, GOA, and GOA/Q/D were recorded to confirm structural characteristics and functionalization. The measurements were performed on an X-ray diffractometer (Bruker D2 PHASER) equipped with a Cu Kα radiation source (λ = 1.5406 Å). Samples were analyzed in the 2θ range of 10°–70° at a scanning rate of 0.02°/s. All samples were air-dried and mounted on glass sample holders before measurement. GO, GOA, and GOA/Q/D solutions were placed in a quartz cuvette for UV–Vis spectroscopy measurement using a GENESYS 150 (Thermo Scientific) UV–Vis spectrophotometer in the wavelength range of 200–800 nm. The thermal stability and decomposition behaviour of the samples were assessed using thermogravimetric analysis (TGA) on a TGA Q50 analyzer (TA Instruments, New Castle, DE, USA). The measurements were carried out in a nitrogen atmosphere, with the temperature increasing at a rate of 10 °C/min up to 700 °C.

2.5. Drug loading and release

0.5 mg GOA/Q/D was dispersed in 1 mL PBS at pH 5.0 (endosomal pH) or pH 7.4 (physiological pH). The dispersions were shaken at 100 rpm and 37 °C. At predetermined time points (1, 2, 3, 4, 8, 12, and 24 h), the samples were centrifuged at 25000 g for 10 min, and the supernatant was collected to measure the released drug concentration. After removing the supernatant, the same amount of fresh PBS was added back to the sample to keep the total volume constant, and the incubation was continued for further time points. The cumulative release of quercetin and DOX was quantified by measuring the absorbance at 373 nm for quercetin and 490 nm for DOX.

$$\text{Cumulative drug release}\left(\%\right)=\frac{\text{Weight of released drug (mg)}}{\text{Weight of loaded drug}\left(\mathrm{mg}\right)}\times 100$$ (3)

2.6. Photothermal response

A 100 μL aliquot of GO, GOA, or GOA/Q/D (2 mg/mL) was placed in an Eppendorf tube and irradiated from the top with an 808 nm NIR laser at a power density of 1.5 W/cm² for 3 min. After 3 min of irradiation, for the samples to reach the maximum temperature in the heating cycle, the laser was turned off, and the samples were cooled down to room temperature during the cooling cycle. Thermal images were captured at different time intervals from the side of the tube using a thermal imaging camera. The time-dependent temperature profiles were recorded and plotted to assess the photothermal performance.

The photothermal conversion efficiency (η) was calculated using the following equation:

$$\mathrm{\eta }=\frac{\mathrm{hs}\left({\mathrm{T}}_{\max }-{\mathrm{T}}_{\mathrm{sur}}\right)-{\mathrm{Q}}_{\mathrm{dis}}}{\mathrm{I}\left(1-{10}^{-{\mathrm{A}}_{808}}\right)}$$ (4)

where h is the heat transfer coefficient, S is the surface area of the container, T_max and T_sur are the maximum and surrounding temperatures, respectively, ${\mathbf{Q}}_{\mathbf{dis}}$represents the heat dissipated by the solvent (assumed to be 0), I represents the laser power density (1.5 W/cm²), and A₈₀₈ is the sample absorbance at 808 nm.

The dimensionless parameter θ was calculated as:

$$\mathrm{\theta }=\frac{{\mathrm{T}}_{\max }-\mathrm{T}}{{\mathrm{T}}_{\mathrm{sur}}-{\mathrm{T}}_{\max }}$$ (5)

The cooling time constant (τ_s) was obtained from:

$$-\ln \left(\mathrm{\theta }\right)=\frac{\mathrm{t}}{{\mathrm{\tau }}_{\mathrm{s}}}$$ (6)

The heat transfer coefficient and surface area product hS was determined by:

$$\mathrm{hS}=\frac{{\mathrm{m}}_{\mathrm{d}}\times {\mathrm{c}}_{\mathrm{d}}}{{\mathrm{\tau }}_{\mathrm{s}}}$$ (7)

where ${\mathbf{m}}_{\mathbf{d}}$ is the mass of the solution (0.2 g), and ${\mathbf{c}}_{\mathbf{d}}$ is the specific heat capacity of water (4.2 J/g).

2.7. Confocal microscopy

Cy5.5-labeled GO and GOA were prepared using EDC/NHS coupling chemistry. Briefly, GO or GOA dispersion (1 mg/mL) was treated with 4 mg N-(3-dimethylaminopropyl)-N-ethylcarbodiimide hydrochloride (EDC) and 4 mg N-hydroxysuccinimide (NHS) for 30 min at room temperature under gentle stirring to activate the carboxyl groups on GO. Subsequently, Cy5.5-NH₂ dye (0.5 mg/mL) was added to the activated GO or GOA solution, and the reaction mixture was incubated for 24 h at room temperature in the dark. After centrifugation, the Cy5.5-labeled GO and GOA were resuspended in PBS and stored at 4 °C. Confocal microscopy was used to qualitatively assess the nanoparticles' intracellular uptake (Dash et al., 2024a). U87 cells (2 × 10⁴) were seeded onto confocal μ-dishes in a medium composed of 90% DMEM and 10% FBS and were incubated overnight at 37 °C with 5% CO₂. The medium was then replaced with new DMEM containing Cy5.5-labeled GO or GOA (red fluorescence, excitation/emission = 675/695 nm). After 4 h, cells were washed with PBS and stained with LysoTracker Green DND-26 (green fluorescence, excitation/emission = 504/511 nm) to detect lysosomes. The cells were then fixed in 4% paraformaldehyde and permeabilized in PBS containing 0.1% Triton X-100. Nuclear staining was carried out with Hoechst 33342 (1 μg/mL; blue fluorescence, excitation/emission = 405/460 nm). Imaging was conducted using a Zeiss LSM 780 confocal laser scanning microscope.

2.8. Flow cytometry analysis

To conduct flow cytometry analysis, U87 cells (5 × 10⁵) were cultured in medium composed of 90% DMEM and 10% FBS and were incubated overnight in a T-25 flask. The cells were then treated with Cy5.5-labeled GO or GOA dispersed in DMEM and incubated for 4 h. Following incubation, the medium was removed, and the cells were harvested via trypsinization. The collected cells were resuspended in 500 μL binding buffer and then analyzed with an Attune NxT flow cytometer. To assess intracellular accumulation of DOX, cells were similarly treated with free DOX, GO/Q/D, and GOA/Q/D (10 μg/mL DOX). After 4 h, the cells were collected and analyzed by flow cytometry based on the intrinsic red fluorescence of DOX, allowing for a comparative evaluation of drug uptake across different formulations. To evaluate apoptosis and necrosis induced by GOA, GOA (L+), GOA/Q, GOA/D, GOA/Q/D, and GOA/Q/D (L+), 7 × 10⁵ U87 cells were prepared in cell culture medium (DMEM +10% FBS) on T-25 flasks and incubated at 37 °C for 24 h. After treatment with the corresponding nanoparticles and an additional 24 h incubation, the GOA (L+) and GOA/Q/D (L+) groups were subjected to 808 nm NIR laser irradiation at 1.5 W/cm² for 3 min. The cells were harvested by trypsinization and centrifuged. The resulting cell pellet was then incubated with Annexin V-FITC for 30 min to stain live and early apoptotic cells. Subsequently, Propidium Iodide (PI) was used to label late apoptotic and necrotic cells. Finally, fluorescence emissions at 530 nm (FITC) and 570 nm (PI) were measured using a flow cytometer.

2.9. Cytotoxicity

To assess the cytotoxicity of various nanoparticle formulations, including GOA, GOA (L+), GOA/Q, GOA/D, GOA/Q/D, and GOA/Q/D (L+). U87 cells were incubated with each formulation for 24 h, followed by NIR laser irradiation (808 nm, 1.5 W/cm² for 3 min) for the GOA (L+) and GOA/Q/D (L+) groups. Cell viability was evaluated using the MTT assay. The cell culture medium was removed after 24 h, and cells were extensively washed with medium to remove adsorbed nanoparticles, followed by adding 10 μL of MTT solution (5 mg/mL). After incubating for 2 h at 37 °C, the MTT solution was removed, and 100 μL of DMSO was added. After shaking at 100 rpm for 10 min, the cell viability was determined from solution absorbance using an ELISA microplate reader at 570 nm (OD₅₇₀). The relative cell viability (%) was calculated using the equation.

$$\text{Relative cell viability}\left(\%\right)=\frac{{\mathrm{OD}}_{570}\text{of sample}}{{\mathrm{OD}}_{570}\text{of control}}⨯100$$ (8)

2.10. Synergistic effect of DOX and quercetin

The potential synergy between DOX and quercetin (Q) in U87 glioblastoma cells was assessed from cell viability treated with different drug combinations. U87 cells were seeded in 96-well plates at a density of 5 × 10³ cells per well and allowed to adhere overnight. The cells were then treated sequentially with varying DOX/Q mass ratios (0:1, 1:4, 1:1, 4:1, and 1:0) across a range of total drug concentrations (10, 20, 40, 60, 80, and 100 μg/mL), followed by a 24 h incubation period. After the drug exposure, the medium was replaced with MTT reagent, and the cells were incubated for an additional 2 h. The absorbance of the resulting solution, indicating cell viability, was measured at 570 nm (OD₅₇₀) using an ELISA microplate reader.

2.11. Hemocompatibility

The blood compatibility was analyzed by evaluating possible hemolysis induced by GOA. Briefly, a whole blood sample collected from mice was diluted with PBS, and 0.1 mL of the diluted blood sample was mixed with different concentrations (0.1, 0.2, 0.4, 0.6 mg/mL) of GOA and incubated at 37 °C for 2 h. The diluted blood sample treated with deionized water and PBS was used as the positive and negative controls, respectively. The mixture was centrifuged, and the absorbance of the supernatant was recorded using a UV–Vis spectrophotometer.

$$\text{Hemolysis ratio}\left(\%\right)=\frac{\left({\mathrm{OD}}_{540}\text{of sample}-{\mathrm{OD}}_{540}\text{of}\mathrm{PBS}\right)}{\left({\mathrm{OD}}_{540}\text{of water}-{\mathrm{OD}}_{540}\text{of}\mathrm{PBS}\right)}\times 100$$ (9)

To evaluate the biocompatibility of the drug-free nanocarrier, 5 × 10³ cancer cells (B16F10 melanoma) or normal cells (3T3 fibroblasts) were cultured using cell culture medium (DMEM+ 10% FBS) in 96-well plates. After incubation for 24 h, different concentrations of GOA were added. At 24 and 48 h, the cell culture medium was removed, and 10 μL of MTT solution (5 mg/mL) was added and incubated for 2 h at 37 °C. Then the MTT solution was removed, and 100 μL DMSO was added and shaken for 10 min. The cell viability from solution absorbance was determined (OD₅₇₀), and the relative cell viability (%) was calculated as before.

2.12. Brain targeting and biodistribution

All in vivo procedures were conducted in accordance with protocols approved by the Institutional Animal Care and Use Committee (IACUC) of Chang Gung University (IACUC No. CGU110–202). The mice were housed in standard laboratory conditions with 5 mice in ventilated polycarbonate cages on a light:dark cycle. Room temperature was maintained at 22–24 °C with 50–60% relative humidity. Corn cob bedding and nesting material were provided, with food and water available ad libitum. To establish an orthotopic brain tumor xenograft model, 3 μL of U87 cell suspension containing 3 × 10⁵ cells was injected into the right intracranial region of each female BALB/c nude mouse (aged 3–4 weeks) at a depth of 3 mm (Kruse et al., 2024). To validate the role of Angiopep-2 in facilitating BBB penetration and promoting accumulation in brain tumors, biodistribution studies were conducted in both healthy mice and U87 tumor-bearing mice. For tumor-bearing mice, the progression of intracranial U87 glioblastoma was monitored by assessing cell-associated bioluminescence signal intensity through bioluminescence imaging. This was performed using an in vivo imaging system (IVIS) following the intraperitoneal injection of 100 μL D-luciferin solution prepared in saline. Once the bioluminescence signal surpassed 1 × 10⁶ photons/s, Cy5.5-labeled GO or GOA was administered via tail vein injection. After 24 h, fluorescence imaging in the brain area was carried out using IVIS, allowing comparison of the brain targeting efficiency attributed to Angiopep-2. To assess biodistribution, mice were sacrificed 24 h post-injection, and major organs (heart, liver, lung, spleen, and kidney) were collected for ex vivo fluorescence imaging using IVIS. The fluorescence signal intensity in each organ was quantitatively analyzed by calculating the percentage contribution relative to the total signal intensity in all major organs.

2.13. In vivo antitumor efficacy

The in vivo antitumor efficacy was evaluated in U87 tumor-bearing mice by randomly assigning them into five groups (n = 6 per group), each receiving an intravenous injection of 100 μL of the corresponding formulation via the tail vein. The groups include PBS, GOA, GOA (L+), GOA/Q/D, and GOA/Q/D (L+). The dosage of nanoparticles is 10 mg/kg, DOX is 2.8 mg/kg, and quercetin is 3.4 mg/kg body weight, respectively. Injections were administered on days 8, 12, and 15 after tumor implantation. One hour after nanoparticle administration, the tumor area in each mouse in the GOA (L+) and GOA/Q/D (L+) groups was irradiated with an 808 nm NIR laser at 1.5 W/cm² for 3 min. Bioluminescence imaging with IVIS was performed on days 7, 9, 13, and 16 to monitor tumor progression. Throughout the study, the body weight of the mice was recorded. Mice were euthanized if they lost more than 20% of their initial body weight or exhibited neurological symptoms such as hemiparesis, hunchback posture, or seizures. Survival curves were generated for each treatment group. Tumor bioluminescence signal intensity was quantified on days 9, 13 and 16, and the data were normalized by dividing the bioluminescence intensity (BLI) at each time point by the BLI measured on day 7 (approximately 1 × 10⁷ p/s) (Dash et al., 2024a).

2.14. Hematological analysis

To assess systemic toxicity, blood samples were collected from all mice using BD Microtainer® tubes containing K2EDTA (Becton, Dickinson and Company, USA) for hematological analysis. The parameters measured included white blood cell count (WBC), red blood cell count (RBC), hemoglobin (HGB), hematocrit (HCT), mean corpuscular volume (MCV), mean corpuscular hemoglobin (MCH), mean corpuscular hemoglobin concentration (MCHC), red cell distribution width–coefficient of variation (RDW-CV), and platelet count (PLT).

2.15. Histological analysis

To assess the safety of the treatments and their potential effects on major organs, hematoxylin and eosin (H&E) staining was performed on tissue sections from the main organs of animals in all experimental groups. Tumor tissues were collected, embedded in paraffin, and sectioned at a thickness of 5 μm onto glass slides. These slides were stained with H&E and further analyzed by immunohistochemistry (IHC). The IHC staining was conducted to analyze the expression of Ki-67 for cell proliferation, cleaved caspase 3 (CC3) for cell apoptosis, and HSP70 for thermal resistance. For IHC staining, slides were stained with anti-Ki-67 antibody (Cell Signaling, 1:200 dilution), anti-cleaved caspase 3 (anti-CC3) antibody (Cell Signaling, 1:400 dilution), or anti-HSP70 antibody (Abcam, 1:100 dilution). The nucleus was counterstained with hematoxylin after incubation with the secondary antibody. Protein expression was quantified semi-quantitatively using ImageJ software by measuring the percentage of immunoreactive areas within a defined region of interest (ROI) (n = 4).

2.16. Statistical analysis

The data are expressed as mean ± standard deviation (SD), and statistical analysis was performed to compare different groups using one-way analysis of variance (ANOVA). The homogeneity of variance was confirmed before statistical comparison using Levene's test. The Tukey HSD post-hoc test was used to identify differences between groups. A value of p < 0.05 was considered statistically significant. The log-rank Mantel-Cox test was performed for survival comparison.

3. Results and discussion

3.1. Preparation and characterization of nanoparticles

The GOA/Q/D was prepared by PEGylation of GO using DSPE-PEG₂₀₀₀-MAL to provide reactive maleimide groups for peptide attachment. Subsequently, Angiopep-2 with an SH groups on the C-terminal cysteine residues was conjugated to the maleimide on PEGylated GO through thiol–maleimide chemistry to form GOA. This structural design is expected to facilitate efficient BBB penetration and brain tumor targeting via LRP1 receptor-mediated transcytosis. The amount of Angiopep-2 conjugated to GO is 0.049 ± 0.001 mg/mg after determining the peptide concentration with the BCA protein assay kit. This high peptide loading will enable efficient receptor-mediated targeting. To achieve synergistic therapeutic effects, DOX and quercetin were co-loaded onto the GOA through π–π stacking to form GOA/Q/D. From dynamic light scattering (DLS), GO demonstrated a comparatively narrow size distribution with a polydispersity index (PDI) of 0.27 ± 0.03 and an average particle size of 100.2 ± 14.5 nm (Fig. 1a). The average size increased to 142.0 ± 11.0 nm, and the PDI slightly decreased to 0.25 ± 0.06 after conjugation with Angiopep-2 in GOA. After drug loading, the particle size was 176.6 ± 17.5 nm, and the PDI decreased to 0.20 ± 0.05 in GOA/Q/D, suggesting a more uniform particle size distribution. Nonetheless, it should be noted that using DLS to measure the particle size of GO-based nanomaterials may be inaccurate, as they have a highly non-spherical shape, polydispersity, and 2D planar structure (Amaro-Gahete et al., 2019).

Fig. 1.

The particle size distribution (a), zeta potential distribution (b), high-resolution transmission electron microscopy (HR-TEM) images (c), and energy dispersive X-ray spectroscopy (EDS) analysis (d) of GO, GOA, GOA/Q/D nanoparticles. Bar = 100 μm.

The GO exhibited a zeta potential of −32.9 ± 1.4 mV (Fig. 1b), which can be attributed to the abundant electronegative functional groups introduced on the graphite lattice during oxidation (Krishnamoorthy et al., 2013). The zeta potential increased to −27.9 ± 0.9 mV in GOA, due to surface modification with positively charged Angiopep-2 (Lu et al., 2017). Following co-loading of DOX and Q, the zeta potential further increased to −24.4 ± 1.7 mV [59]. The reduction in negative charge is primarily attributed to masking of the negatively charged GO surface by cationic DOX, while Q contributed minimally to this shift (Fang et al., 2018). The characteristics of GO, GOA, and GOA/Q/D nanoparticles are summarized in Table 1. From HR-TEM, thin, translucent sheet-like structures, typical of exfoliated GO layers, were visible in the pristine GO sample (Fig. 1c). The size increases successively for GOA and GOA/Q/D, with less transparent GO sheets (Fig. 1c). Using energy-dispersive X-ray spectroscopy (EDS) to investigate the elemental composition and distribution, carbon (84.8 wt%) and oxygen (15.2 wt%) are present in GO (Fig. 1d). For GOA, sulfur (0.7 wt%) and nitrogen (1.1 wt%) appear in addition to oxygen (29.7 wt%) and carbon (68.4 wt%). This indicates that Aniopep-2, which has amino acid residues and sulfur from cysteine residues, was successfully conjugated. The carbon content in GOA/Q/D increased to 74.3 wt% due to the loading of Q and DOX (Fig. 1d). A noteworthy increase in nitrogen content to 1.8 wt% further confirmed the presence of DOX, which contains nitrogen in its chemical structure. At 0.7 wt%, the sulfur content did not change, suggesting that Angiopep-2 is still present.

Table 1.

The average particle size and polydispersity index (PDI) from dynamic light scattering analysis, and the average zeta potentials (n = 4).

Particles	Size (nm)	PDI	Zeta Potentials (mV)
GO	100.2 ± 14.5	0.27 ± 0.03	−32.9 ± 1.4
GOA	142.0 ± 11.0α	0.25 ± 0.06α	−27.9 ± 0.9α
GOA/Q/D	176.6 ± 17.5α, β	0.20 ± 0.05α, β	−24.4 ± 1.7α, β

^αp < 0.05 compared with GO,

^βp < 0.05 compared with GOA.

From UV–Vis absorption spectra, GO shows an absorption peak at ∼230 nm due to the π–π* transitions of aromatic C C bonds, which retains and shows no appreciable shift in GOA and GOA/Q/D (Fig. 2a) (Dash et al., 2022). A small characteristic peak appears at ∼280 nm for GOA from Angiopep-2, and two additional peaks were visible at ∼373 nm and ∼490 nm for GOA/Q/D from Q and DOX, respectively. The FTIR spectra of GOA and GOA/Q/D confirm the surface functionalization and drug loading to GO (Fig. 2b). With the presence of oxygen-containing functional groups, GO exhibited characteristic peaks at 3417 cm⁻¹ (O—H stretching), 1735 cm⁻¹ (C O stretching of carbonyl/carboxyl groups), 1395 cm⁻¹ (O—H), and 1056 cm⁻¹ (C—O stretching from epoxy or alkoxy groups) in addition to C C bonding at 1624 cm⁻¹ (Dash et al., 2021). The GOA reveals new peaks at 1506 cm⁻¹ (amide II), 1465 cm⁻¹ (C—H deformation), and 1234 cm⁻¹ (P O stretching) after conjugation with Angiopep-2 (Lebedenko et al., 2023). The GOA/Q/D spectrum shows new peaks at 3327 cm⁻¹ (N—H stretching) and 805 cm⁻¹ (C H bending), indicating the effective incorporation of drugs without compromising the structural integrity of GO. The XRD spectra of GO exhibited a sharp diffraction peak at 2θ ≈ 10.5°, corresponding to the (001) plane (Rahimi et al., 2023) (Fig. 2c). This peak represents a characteristic of the layered structure of GO, due to the interlayer spacing caused by oxygen-containing functional groups. This peak remains in GOA and GOA/Q/D, indicating that GO preserves this layered, graphitic-like structure. The crystalline nature was calculated using the Debye-Scherrer equation. After conjugation with Angiopep-2, the full width at half maximum (FWHM) increases from 3.10° to 4.98°, indicating peak broadening due to disruption of stacking and increased defects. This results in a much smaller crystallite size, from 2.57 nm for GO to 1.60 nm for GOA. Interestingly, when DOX and quercetin were co-loaded, the FWHM decreased to 3.64°, leading to a partial increase in crystallite size to 2.19 nm. The Raman spectra show D and G bands from the sp2 vibration of C C double bond disorder (Fig. 2d). The I_D/I_G value, which is the intensity ratio of the D band (1346 cm⁻¹) to the G band (1586 cm⁻¹), is 1.0 for GO (Andrade et al., 2021). GOA had a nearly similar I_D/I_G value to GO, indicating that functionalization with Angiopep-2 did not influence the aromatic structures of GO. In the case of GOA/Q/D, a slight decrease in the I_D/I_G ratio was observed, suggesting that drug loading may have neutralized surface defects through noncovalent π–π interactions without introducing additional disorder (Sadeghipour et al., 2021). Thermogravimetric analysis (TGA) and derivative thermogravimetric (DTG) analysis were performed to assess the thermal stability and compositional changes of samples (Fig. 2e and f). All samples exhibited a characteristic multi-step degradation pattern typical of graphene-based systems. The initial weight loss below 100 °C corresponds to the removal of adsorbed water molecules. The major weight loss between 150 and 250 °C is associated with the decomposition of oxygen-containing functional groups present in GO sheets (Dash et al., 2024b). Upon functionalization with Angiopep-2, GOA reveals little shift in the peak decomposition temperature in the DTG curve without significant structural disruption. In contrast, the GOA/Q/D displayed a more pronounced change, where the main degradation peak shifted toward higher temperatures (180–250 °C). This shift can be attributed to the decomposition of loaded drugs, where DOX reveals a major weight loss from 210 to 280 °C and another weight loss from 300 to 400 °C, while quercetin shows a decomposition peak around 320 °C. As shown before, at low drug loading levels (<70 wt%), the GO remains the dominant thermal component, and the thermal decomposition behaviour is only marginally affected by incorporated drugs (Zhang et al., 2013).

Fig. 2.

The UV–Vis spectra (a), Fourier-transform infrared (FTIR) spectra (b), X-ray diffraction (XRD) spectra (c), Raman spectra (d), thermogravimetric analysis (TGA) (e), and derivative thermogravimetric (DTG) curves (f) of different samples.

The colloidal stability of GOA/Q/D was evaluated in both phosphate-buffered saline (PBS) and in DMEM +10% FBS over a 48-h period. Direct visual observation revealed no significant change in solution colour due to particle aggregation in both solutions, indicating excellent stability (Fig. 3a). To quantitatively assess stability, DLS was used to monitor changes in zeta potential, particle size, and PDI at predetermined time intervals (Fig. 3b). The zeta potential of GOA/Q/D remained consistently negative in both media with no significant change, and the values remained within a range that typically supports colloidal stability (below −20 mV). Particle size analysis indicates GOA/Q/D can maintain its average particle size in PBS and 10% serum, with an apparent increase in particle size observed only at 24 h and 48 h, but still below the 200 nm threshold for intravenous administration (Öztürk et al., 2024). The PDI values remained low (below 0.3) throughout the observation period, further confirming the narrow size distribution and stable dispersion of the nanoparticles. These findings demonstrate that GOA/Q/D exhibits good colloidal stability in both physiological buffer and serum-containing medium, supporting its suitability for biological applications.

Fig. 3.

The colloidal stability of GOA/Q/D (0.5 mg/mL) in DMEM +10% FBS and in PBS by gross observation (a), zeta potential (b), average particle size (c), and polydispersity index (d). ^αp < 0.05 compared with PBS.

For targeted drug delivery using nanoparticles, a protein corona may form on the nanoparticle as soon as it enters a biological fluid, impacting the targeting efficiency (Xiao and Gao, 2018). Proteins in the corona can wrap around the nanoparticle, physically concealing the targeting ligands (like peptides or antibodies) attached to its surface. This often prevents the nanoparticle from binding correctly to its intended cells, significantly decreasing functional targeting ability. Certain proteins in the corona may also act as opsonins, which flag the nanoparticles as foreign bodies, leading to premature clearance, reducing off-target accumulation, and lowering drug delivery at the targeted site. Furthermore, cells may interact with the newly formed protein rather than the engineered nanoparticle, resulting in an altered biological identity. In our preparation, GO was PEGylated by mixing with DSPE-PEG₂₀₀₀-Mal. The Mal residue in PEGylated GO was conjugated with the -SH group of the cysteine residue in Angiopep-2 to prepare GOA. This PEGylation can evade rapid mononuclear phagocyte system clearance and minimize off-target effects (Suk et al., 2016).

The photothermal response of GO, GOA, and GOA/Q/D was evaluated under 808 nm NIR laser irradiation at 1.5 W/cm². All samples exhibited a time-dependent increase in temperature, confirming their ability to convert NIR light energy into heat. The thermal images (Fig. 4a) and time-lapse peak temperature curves (Fig. 4b) demonstrated that GO, GOA, and GOA/Q/D reached similar maximum temperatures within 180 s, indicating good photothermal response in all three formulations. A single laser on-off cycle was used to calculate the photothermal conversion efficiency (η) (Fig. 4c). The linear regression relationship between ln (θ) and cooling time (t) was used to determine the time constant (τ_s) (Fig. 4d). The calculated photothermal conversion efficiencies (η) were 37.7% for GO, 37.8% for GOA, and 36.5% for GOA/Q/D. Therefore, the overall photothermal performance remained consistent across all samples after Angiopep-2 conjugation and drug loading.

Fig. 4.

The thermal images (a) and time-dependent peak temperature profiles (b) after NIR laser (808 nm, 1.5 W/cm²) irradiation of 100 μL GO, GOA, and GOA/Q/D sample (2 mg/mL) for 180 s. The heating and cooling curves (c) and time constants (τ_s) during a NIR laser on/off cycle for calculating the photothermal conversion efficiency by irradiating 100 μL GO, GOA, and GOA/Q/D sample (2 mg/mL) with NIR laser (808 nm, 1.5 W/cm²).

Using 0.5 mg GOA for drug loading, increasing the amount of quercetin from 0.1 to 0.4 mg resulted in a consistent increase in loading content (LC) from 0.06 to 0.17 mg/mg. In contrast, the loading efficiency (LE) reduced from 51% to 21% due to surface saturation and limited binding interactions at higher doses (Fig. 5a). Choosing 0.3 mg quercetin to load 0.5 mg GOA, various concentrations of DOX were used to load GOA/Q to prepare GOA/Q/D. Fig. 5b shows that increasing the DOX input from 0.1 to 0.4 mg resulted in higher LC up to 0.14 mg/mg but also decreased LE from 45% to 15%. Based on the drug loading analysis, 0.3 mg of DOX was determined to be an ideal amount for drug loading, and GOA/Q/D used in the following studies used 0.5 mg GOA to successively load 0.3 mg of Q and 0.3 mg of DOX. The pH-dependent release of DOX and quercetin from 0.5 mg/mL GOA/Q/D was tested over a 24 h period under physiological (pH 7.4) and acidic (pH 5) conditions. Overall, under acidic conditions, there was a quick initial burst release during the first six hours for both drugs, which was followed by a slower, longer-lasting release phase (Dash et al., 2021). The release of DOX showed a strong pH-responsive behaviour (Fig. 5c). After 24 h, cumulative DOX release at pH 5.0 was roughly 78%, while at pH 7.4, it was only 26%. A quicker drug release was observed in an acidic environment than in a neutral environment, most likely because the functional groups of GO are protonated, and its interactions with DOX will be weakened at lower pH values. Likewise, when quercetin was released at pH 5 as opposed to pH 7.4, the cumulative release was noticeably higher and faster (Fig. 5d). About 72% of quercetin was released at pH 5 after 24 h, whereas only 30% was released at pH 7.4. Tumor microenvironments are generally more acidic (6.5–7.2) than normal tissue (7.4). The pH-responsive release mechanism of quercetin and DOX from GO relies primarily on the disruption of non-covalent molecular interactions in acidic environments. Under physiological conditions, both drugs remain tightly bound to the GO matrix, minimizing premature leakage. When the nanoformulation enters an acidic tumor microenvironment (pH 5.0–6.0) or intracellular endosomes/lysosomes, a controlled release is triggered through three primary mechanisms: protonation and attenuation of hydrogen bonding at pH 7.4, disruption of π-π stacking via electrostatic repulsion, and increased hydrophilic solubility in acidic media (Khakpour et al., 2023). The pH-responsive release of chemotherapeutic drugs specifically within this tumor area can therefore reduce off-target toxicity for tumor-targeted therapy. After intracellular uptake, GOA/Q/D will end up in lysosomes, organelles with an acidic environment (4.5–5.5). The acid-triggered drug release will allow the drugs to be released directly where they are needed, improve bioavailability, minimize side effects in healthy tissues, and enhance therapeutic efficacy for targeted chemotherapy (Nunziata et al., 2025).

Fig. 5.

The loading efficiency (LE) and loading content (LC) of quercetin (a) and doxorubicin (DOX) (b). 0.5 mg of GOA was mixed with different amounts of quercetin to prepare GOA/Q. GOA/Q prepared with 0.3 mg quercetin was used to load different amounts of DOX to prepare GOA/Q/D. The release of quercetin (c) and DOX (d) from GOA/Q/D (0.5 mg/mL) in pH 5 and pH 7.4 solutions at 37 °C.

3.2. In vitro studies

The intracellular uptake and subcellular localization of Cy5.5-labeled GO and GOA in U87 glioblastoma cells were assessed using confocal laser scanning microscopy (Fig. 6a). The cells were co-stained with Hoechst 33342 for nuclear staining and LysoTracker Green for lysosome marking. Red fluorescence (Cy5.5) was seen inside the cytoplasm after internalization of Cy5.5-labeled GO or GOA by U87 cells. Significantly greater red fluorescence intensity was observed in cells treated with GOA than in cells treated with GO. This suggests that Angiopep-2 functionalization improves intracellular uptake in U87 glioblastoma cells, most likely by targeting low-density lipoprotein receptor-related protein-1 (LRP-1), which is overexpressed in U87 cells, triggering highly efficient receptor-mediated endocytosis. By binding to LRP-1, which is highly expressed on both the BBB and U87 cells, this peptide can mediate enhanced glioma internalization in addition to BBB traversal, facilitating targeted delivery to brain tumor areas. The overlapped red and green signals in merged images produced areas of yellow fluorescence, suggesting that after endocytosis, the internalized GO and GOA were transported into lysosomes. These findings validate the involvement of Angiopep-2 in enhancing targeted cellular internalization in glioblastoma cells by confirming the increased uptake and lysosomal accumulation of GOA in comparison to GO. Flow cytometry was used to quantitatively assess the cellular uptake efficiency of Cy5.5-labeled GO and GOA (Fig. 6b). Cells treated with GOA exhibited a 3.5-fold increase in mean fluorescence intensity (MFI) compared to GO-treated cells, supporting the confocal microscopy observation that Angiopep-2 modification significantly enhanced cellular uptake (Fig. 6c). The presence of GOA is expected to enhance cell-membrane penetration and intracellular accumulation of loaded drugs. This is confirmed by confocal microscopy observation, where cells treated with GOA/Q/D appear to show the highest intracellular DOX accumulation among different groups (Fig. 6d). Confocal microscopy indicates that once GOA/Q/D is internalized, it effectively releases DOX, allowing the drug to enter the nucleus to exert its DNA-intercalating anticancer effects. The intracellular accumulation of DOX in U87 cells was also compared quantitatively from flow cytometry analysis of its red fluorescence signal (Fig. 6e). Cells treated with GO/Q/D and GOA/Q/D had higher DOX uptake than cells treated with free DOX. In comparison to free DOX, GO/Q/D leads to 1.7-fold higher MFI (Fig. 6f). DOX uptake by cancer cells can be inefficient. This can originate from several factors, including the presence of drug efflux pumps (like P-glycoprotein) and other intrinsic resistance mechanisms within cancer cells (Duan et al., 2023). Cancer cells can express high levels of ATP-binding cassette transporter proteins, such as P-glycoprotein (P-gp), which actively pump DOX out of the cell, preventing it from reaching its target and exerting its cytotoxic effect. Some cancer cells may have inherent cellular mechanisms that reduce DOX uptake or increase its detoxification, making them inherently resistant to the drug (Swain et al., 2025). This highlights the benefit of using a GO as a nanocarrier for DOX delivery. Most importantly, with enhanced intracellular uptake, GOA/Q/D treatment notably increased MFI by an additional 1.5-fold compared with GO/Q/D, which can further enhance cytotoxicity toward U87 cells (Fig. 6f). Ultimately, this demonstrates that GOA can improve both the transport efficiency across cellular membranes and the localized, intracellular concentration of loaded drugs.

Fig. 6.

The confocal laser scanning microscopy (a) and flow cytometry analysis (b and c) of intracellular uptake of Cy5.5-labeled GO and GOA by U87 cells in 4 h (n = 3). The lysosomes and nucleus are labeled with LysoTracker Green and Hoechst, respectively. The control is PBS. Bar = 20 μm. ^αp < 0.05 compared with control, ^βp < 0.05 compared with GO. The confocal laser scanning microscopy (d) and flow cytometry analysis (e and f) of intracellular DOX accumulation in U87 glioblastoma cells (n = 3). Bar = 20 μm. ^αp < 0.05 compared with control, ^βp < 0.05 compared with GOA, ^γp < 0.05 compared with DOX, ^δp < 0.05 compared with GOA/Q/D. The fluorescence intensity is based on 10,000 cells used in flow cytometry analysis. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

To determine whether doxorubicin (DOX) and quercetin (Q) exhibit synergistic, additive, or antagonistic interactions, U87 cells were treated with various drug combinations at a fixed total concentrations (10, 20, 40, 60, 80, and 100 μg/mL) while varying the DOX:Q ratios (1:1, 1:4, and 4:1). After 24 h treatment period, cell viability was determined using the MTT assay, and the percentage viability was calculated relative to untreated controls. DOX has significantly higher cytotoxicity than quercetin at all tested concentrations because of its strong DNA-intercalating and topoisomerase II inhibitory activity, which directly inhibits cell proliferation (Fig. 7a). However, when the two drugs were combined, cytotoxicity increased significantly compared to DOX alone, especially at 1:1 and 4:1 DOX:Q ratios, indicating a strong synergistic effect. This enhancement can be attributed to quercetin's ability to inhibit efflux pumps (such as P-glycoprotein) and modulate apoptotic pathways, thereby improving intracellular retention of DOX. Interestingly, the combination with a high quercetin proportion (1:4) exhibited viability values comparable to DOX alone at lower drug concentrations (10–40 μg/mL), but at 100 μg/mL, the same ratio showed enhanced cytotoxicity, indicating a concentration-dependent effect. To quantify drug interactions, the Chou-Talalay method was used to calculate the combination index (CI) (Chou, 2010). Based on the Chou-Talalay method, the CI is plotted against the fraction affected (Fa) to quantify drug interactions. The Fa value represents the fraction of cells affected (Fa = 1 - cell viability fraction). The Fa value was determined for each treatment using CI values computed from the following equation:

$$\mathrm{CI}=\frac{{\mathrm{D}}_1}{{\mathrm{D}}_{\mathrm{x}1}}+\frac{{\mathrm{D}}_2}{{\mathrm{D}}_{\mathrm{x}2}}$$ (10)

where D₁ and D₂ represent doses in combination that inhibit x% of cells, and D_x1 and D_x2 are doses of each drug alone that produce the same effect (Liu et al., 2020). The CI values were plotted against Fa to visualize the nature of interaction, with CI < 1 indicating synergism, CI = 1 indicating an additive effect, and CI > 1 indicating antagonism. As shown in Fig. 7b, the CI value is below 1 for all drug concentration combinations tested. This indicates that the DOX/Q combination exhibited synergistic behaviour where the combined effect is greater than the sum of individual effects. This supports the co-delivery of DOX and quercetin with GOA/Q/D for synergistic glioblastoma chemotherapy.

Fig. 7.

(a) Cell viability by doxorubicin (DOX) and quercetin (Q) treatment with single drug or combination treatments at different DOX:Q weight ratios (1:1, 1:4, 4:1) across a range of concentrations (n = 5). (b) Combination index (CI)–fraction affected (Fa) plot generated using CompuSyn software. ^αp < 0.05 compared with quercetin, ^βp < 0.05 compared with DOX, ^γp < 0.05 compared with DOX:Q (1:1), ^δp < 0.05 compared with DOX:Q (1:4).

However, it should be noted that two synergistic drugs with optimized in vitro therapeutic ratios are administered in vivo; their differing release kinetics, metabolic clearance, and half-lives are rarely preserved in the tumor microenvironment (TME). The therapeutic ratio is generally compromised due to factors like pharmacokinetic mismatches, TME barriers (interstitial fluid pressure, heterogeneous permeability, and metabolic gradients), and cellular internalization and action (Browning et al., 2017). Pharmacokinetic mismatches may skew the ratio and compromise efficacy. With different clearance rates and plasma half-lives in vivo, one drug may wash out, metabolize, or be excreted much faster than the other, creating a fluctuating, suboptimal ratio (Aparna et al., 2026). Co-delivering both drugs within a single nanoscale vehicle (e.g., GOA) may force them to circulate together and protect them from premature metabolic clearance.

The cytotoxic of GOA/Q, GOA/D, GOA/Q/D, GOA (L+), and GOA/Q/D (L+) treatment toward U87 cells demonstrated a concentration-dependent decrease in cell viability (Fig. 8a). Nevertheless, GOA/Q/D demonstrated the strongest cytotoxic effect, lowering cell viability to less than 20% at 100 μg/mL. Furthermore, the dual-drug-loaded GOA/Q/D exhibited higher cytotoxicity compared to the single-drug formulations GOA/Q and GOA/D, confirming the synergistic effect of DOX and quercetin (Pan et al., 2024). Furthermore, with NIR irradiation, the GOA/Q/D (L+) group showed the maximum cytotoxicity with less than 10% cell viability at 100 μg/mL. Undoubtedly, GOA serves both as a photothermal agent and a drug carrier for targeted dual-drug CT/PTT to significantly elevate anticancer efficacy against U87 cells in vitro. To assess cell apoptosis and necrosis, flow cytometry analysis was performed following various treatments using Annexin V-FITC/Propidium Iodide (PI) staining (Fig. 8b). The percentage of apoptotic cells in each group corresponded well with the trends observed in the MTT cytotoxicity assay (Table 2). The PBS (control) and GOA (vehicle) groups showed minimal total apoptosis (early apoptosis in Q4 + late apoptosis in Q2), with values of 4.7% and 8.1%, respectively. The total apoptosis rate in the GOA (L+) group rose to 16.6% (early apoptosis: 16.4%, late apoptosis: 0.2%) after being exposed to NIR laser irradiation, indicating the impact of PTT. Targeted delivery of quercetin or DOX by GOA elicits pronounced cytotoxicity due to CT, with a 20.6% total apoptosis rate (early: 16.5%, late: 4.1%) in the GOA/Q group (Niazvand et al., 2019), and 27.8% total apoptosis in the GOA/D group (early: 9.7%, late: 18.1%) (Dash et al., 2024b). The apoptotic rate was considerably increased by the combination formulation GOA/Q/D, which decreased viable cells to 43.6% and increased early and late apoptotic populations to 14.6% and 34.9%, respectively. A synergistic interaction between DOX and quercetin is demonstrated by this heightened effect in the GOA/Q/D group, which surpasses the apoptotic induction observed in either GOA/D or GOA/Q alone. Their co-delivery through the GOA nanocarrier successfully initiates intrinsic and extrinsic apoptotic pathways, supporting the rationale for dual-drug therapy in the treatment of glioblastoma. A remarkable rise in late apoptotic cells (Q2: 49.4%), a moderate amount of early apoptosis (Q4: 7.5%), and a decrease in cell viability to 33.8% were noted in the GOA/Q/D (L+) group. The results demonstrate that co-loading DOX and quercetin onto GOA, in conjunction with NIR-induced PTT, increases apoptosis in U87 glioblastoma cells in a synergistic manner.

Fig. 8.

(a) In vitro cytotoxicity toward U87 cells by treating with GOA, GOA (L+), GOA/Q, GOA/D, GOA/Q/D, and GOA/Q/D (L+) (n = 5). The L+ groups are treated with the NIR laser for 3 min. The concentration of GOA is 1 mg/mL in all samples. ^αp < 0.05 compared with GOA, ^βp < 0.05 compared with GOA (L+), ^γp < 0.05 compared with GOA/Q, ^δp < 0.05 compared with GOA/D, ^εp < 0.05 compared with GOA/Q/D. (b) Quantification of cell apoptosis and necrosis by flow cytometry with Annexin V-FITC/PI staining (n = 4). Q3: live; Q4: early apoptosis; Q2: late apoptosis; Q1: necrosis.

Table 2.

The quantification of cell apoptosis and necrosis by flow cytometry analysis (n = 4).

Group	Live cells (Q3)	Early Apoptosis (Q4)	Late Apoptosis (Q2)	Necrosis (Q1)
Control	95.3 ± 3.3	4.7 ± 3.3	0.3 ± 0.1	0.2 ± 0.1
GOA	94.5 ± 4.0	8.1 ± 1.3	0.2 ± 0.1	0.2 ± 0.0
GOA (L+)	83.4 ± 0.5α	16.4 ± 0.5	0.2 ± 0.0	0.3 ± 0.1
GOA/Q	77.5 ± 3.3α, β	16.5 ± 5.8	4.1 ± 1.8	0.2 ± 0.0
GOA/D	63.8 ± 6.4α, β, γ	9.7 ± 1.5	18.1 ± 5.6	8.4 ± 3.1
GOA/Q/D	43.6 ± 1.3α, β, γ, δ	14.6 ± 0.4	34.9 ± 3.1	7.0 ± 1.7
GOA/Q/D (L+)	33.8 ± 1.3α, β, γ, δ, ε	7.5 ± 0.7	49.4 ± 2.3	9.4 ± 0.6

^αp < 0.05 compared with GOA

^βp < 0.05 compared with GOA (L+)

^γp < 0.05 compared with GOA/Q

^δp < 0.05 compared with GOA/D

^εp < 0.05 compared with GOA/Q/D.

A hemolysis assay was used to assess the hemocompatibility of GOA before in vivo application. Visual inspection and UV–Vis spectra of the supernatant were used to assess the hemolytic effect of diluted red blood cells (RBC) when treated with different doses of GOA. The positive control was deionized water, while the negative control was phosphate-buffered saline (PBS). GOA induces negligible RBC damage, as evidenced by minimal change in colour and UV–Vis spectra up to 0.6 mg/mL when comparable to the PBS control (Fig. 9a). In contrast, RBC treated with water resulted in a red-colored solution from gross observation, along with two noticeable absorbance peaks at 540 and 580 nm associated with oxyhemoglobin liberated from lysed RBC. The hemolysis ratio was calculated from OD₅₄₀ and compared among groups (Fig. 9b). GOA demonstrated excellent hemocompatibility, with the hemolysis ratio staying well below 1% for concentrations up to 0.6 mg/mL, confirming that GOA induced minor hemolysis. Besides hemocompatibility, the cytocompatibility of GOA is also important. This was assessed from the relative cell viability of a cancer cell line or a normal cell line by incubating cells with GOA at different concentrations for 24 and 48 h. As shown in Figs. 9c and d, GOA does not elicit cytotoxicity toward either cell line, and the relative cell viability was higher than 95% across all GOA concentrations, supporting the high cytocompatibility of GOA.

Fig. 9.

The hemocompatibility evaluation from UV–Vis absorbance spectra of released hemoglobin and gross view (A) and the hemolysis ratio (n = 4) (B) by incubating red blood cells with different concentrations of GOA. The absorbance at 540 nm (OD₅₄₀) was used to calculate the hemolysis ratio. Water is the positive control, and PBS is the negative control. The cytocompatibility of GOA tested with B16F10 cancer cells (C) and 3T3 fibroblasts (D). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3.3. In vivo studies

Using IVIS fluorescence imaging, time-dependent brain accumulation of Cy5.5-labeled GO and GOA was first assessed in healthy nude mice (Fig. 10a). Imaging at 1 h, 2 h, and 4 h post-injection revealed markedly higher fluorescence intensity for GOA compared to GO at all time points, indicating improved BBB penetration. Quantitative analysis revealed that GOA achieved a 4.8-fold increase and a 2.5-fold increase at 1 and 2 h, respectively, over GO, indicating enhanced and faster BBB penetration mediated by Angiopep-2 (Fig. 10b). The overall biodistribution of GO and GOA was assessed in tumor-bearing mice. After 24 h of injecting Cy5.5-labeled GO or GOA via the tail veins into nude mice bearing an intracranial U87 tumor, IVIS fluorescence imaging was performed to assess the effectiveness of tumor targeting. As shown in Fig. 11a, mice treated with GO or GOA show higher fluorescence signal intensity in the brain tumor region for GOA. Although the BBB may be partially damaged in tumor-bearing mice, better accumulation of nanoparticles in the brain area supports that Angiopep-2 is a highly effective targeting peptide for glioma treatment, which can enhance both BBB penetration and accumulation at tumor sites in vivo to exert a higher anti-cancer effect. This can be support from the IVIS images of ex vivo fluorescence imaging of excised brains (Fig. 11b). The fluorescence intensity per gram of brain tissue is 1.29 ± 0.14 $\mathbf{\times }{\mathbf{10}}^{\mathbf{9}}$and 2.70 ± 0.16 $\mathbf{\times }{\mathbf{10}}^{\mathbf{9}}$ per gram of brain tissue for GO and GOA, respectively. The enhanced accumulation of GOA may be attributed to potential Angiopep-2/LRP1-mediated transport processes. GOA showed increased brain accumulation; however, the direct involvement of Angiopep-2/LRP1-mediated transcytosis warrants further validation. Future competitive inhibition studies, such as the co-administration of free Angiopep-2 peptide or anti-LRP1 antibodies, will be necessary to definitively confirm this specific uptake mechanism.

Fig. 10.

The representative in vivo fluorescence images in the brain region (a) and total fluorescence intensity (b) using an in vivo real-time imaging system 1 h, 2 h, and 4 h post-injection of Cy5.5-labeled GO or GOA to healthy nude mice via the tail veins (n = 3). ^αp < 0.05 compared with GO.

Fig. 11.

The representative in vivo fluorescence images in the brain region (a), ex vivo fluorescence imaging of excised brain (b) or major organs (c), and the percentage of nanoparticles (%) (d) or percentage of nanoparticles per g of tissue (e) in major organs. An in vivo real-time imaging system was used 24 h post-injection of Cy5.5-labeled GO or GOA to tumor-bearing nude mice via the tail veins (n = 3). ^αp < 0.05 compared with GO.

To examine biodistribution, major organs were excised for ex vivo fluorescence imaging (Fig. 11c). Based on biodistribution studies, the liver was the main location for the accumulation of GO and GOA (Fig. 11d). However, when based on per gram of tissue, lungs and livers show comparable distribution ratios (Fig. 11e). This is due to the much smaller weight of the lung compared to the liver, as the liver is the largest internal organ, while the lung primarily consists of delicate, air-filled alveolar tissue, resulting in a lower absolute weight. The liver is widely recognized as the primary organ for the accumulation and sequestration of nanoparticles following systemic administration, as nanoparticles are cleared from the blood by the mononuclear phagocyte system (MPS) located in the liver (Kupffer cells) and via liver sinusoidal endothelial cells (He et al., 2024). It should be noted that 2D planar fluorescence imaging may not faithfully reflect 3D biodistribution. We have assessed the use of 2D fluorescence imaging data to represent the accumulation of intravenously administered fluorescent gold nanoparticles in major organs and tumors in healthy nude mice and nude mice bearing orthotopic U87 brain tumors. The biodistribution trend from 2D fluorescence imaging using IVIS is consistent with that from the 3D distribution of Au content measured by inductively coupled plasma optical emission spectroscopy (ICP-OES) (Kruse et al., 2024). Furthermore, although Cy5.5 was covalently bound to the nanocarriers to prepare Cy5.5-labeled GO and GOA, we cannot rule out Cy5.5 leakage from the nanocarriers in biological environments, which might lead to artifacts and inaccurate biodistribution data.

To evaluate the in vivo photothermal efficacy of the nanocarrier, U87 tumor-bearing nude mice were randomly divided into PBS (L+), GOA (L+), and GOA/Q/D (L+) groups, and each mouse received intravenous injections of 100 μL of the sample solution. One hour post-injection, the tumor area was subjected to NIR laser irradiation for 3 min at a power density of 1.5 W/cm². An infrared thermal camera was used to capture the tumor site temperature (Fig. 12a). The PBS (L+) group showed only a slight temperature increase, reaching approximately 39 °C, likely due to direct laser-induced heating. However, the tumors in the GOA (L+) and GOA/Q/D (L+) groups showed notable temperature increases, rising to around 45 °C and 46 °C, respectively, after 180 s of laser irradiation (Fig. 12b). There was no significant difference in peak temperature between the GOA and GOA/Q/D groups, indicating that the addition of drugs did not adversely affect photothermal conversion efficiency in vivo. This temperature can provide mild PTT for tumor ablation without damaging the skin (Zhang et al., 2019). The temperature range achieved for mild PTT offers significant advantages over traditional high-temperature PTT (> 50 °C) by providing effective tumor treatment with fewer side effects, such as reduced damage to healthy tissues. Mild PTT also enhances drug delivery by improving vascular perfusion and remodelling the tumor microenvironment to boost immune responses (Gao et al., 2021).

Fig. 12.

The representative thermal images acquired from a thermal camera (a) and the time-lapsed peak temperature profiles at the tumor site in U87 tumor-bearing mice (n = 4). 100 μL of PBS, GOA, or GOA/Q/D (10 mg/kg of GOA) was injected through the tail veins to tumor-bearing mice, and the tumor area was irradiated with 808 nm NIR laser for 3 min (1.5 W/cm²) 1 h post-administration.

The brain is exceptionally sensitive to thermal damage due to the dense, interconnected nature of neurons and the delicate BBB (Li et al., 2026). While hyperthermia from 42.5 to 45 °C promotes apoptosis in tumor cells, temperatures exceeding this range risk permanent neurological deficits (Wang et al., 2022). Exceeding these limits can degrade the BBB, resulting in edema, hemorrhaging, and damage to surrounding healthy microvessels. Mild-temperature PTT can effectively eradicate cancer cells while preserving the adjacent vasculature. Rather than high-power rapid ablation, utilizing lower laser powers with extended exposure durations helps control heat dissipation. Procedures like MRI-guided laser interstitial thermal therapy are critical as they can provide real-time thermometry, allowing neurosurgeons to actively monitor temperature gradients and halt laser emission before thermal damage (Chen et al., 2008). Employing GOA for targeted PTT can ensure that light-to-heat conversion is strictly localized to the tumor, vastly reducing collateral damage. Delivering nanoparticles directly into the tumor interstitium under continuous-flow infusion can also provide a higher payload of photothermal agents at the target site, which may lower the required laser power (Lu et al., 2021). Alternatively, using nanoparticles responsive to higher wavelengths (1000–1700 nm) can also reduce safe power limits and achieve deeper penetration with less superficial brain heating compared to an 808 nm NIR laser (Zhao et al., 2021).

On days 8, 12, and 15 following the implantation of U87 cells in nude mice in the xenograft tumor model, we randomly assigned the mice to five groups, and each mouse received a 0.1 mL injection of GOA or GOA/Q/D solution via the tail veins (n = 6 per group). PBS was used as the control group. One hour after injection, the GOA (L+) and GOA/Q/D (L+) groups were exposed to 3 min of 808 nm NIR laser irradiation at 1.5 W/cm². Luciferin was injected intraperitoneally, and IVIS imaging was performed to measure the bioluminescence intensity (BLI) (Fig. 13a). This method is a validated, highly sensitive, and non-invasive imaging technique for longitudinal monitoring and quantification of intracranial tumor growth in rodent models using luciferase-tagged U87 glioblastoma cells in this study. Studies have demonstrated a strong correlation between the BLI signal and tumor volumes determined by magnetic resonance imaging (Szentirmai et al., 2006). On days 9, 13, and 16, the BLI was measured and normalized to the BLI value on day 7 (Fig. 13b). The PBS and GOA groups exhibited comparable and fast tumor growth rates based on the normalized BLI values (Fig. 13c). The mild PTT in the GOA (L+) group can slow down tumor growth on day 16. This implies that NIR can induce localized hyperthermia from accumulated GOA in the brain tumor area for PTT in vivo. The suppression of tumor growth can be further improved in the GOA/Q/D group on days 13 and 16 from the synergistic chemotherapeutic action of quercetin and DOX. Nonetheless, the GOA/Q/D (L+) group exhibited the highest level of therapeutic efficacy compared with all groups, with significantly lower normalized BLI after day 13 from CT and PTT, producing a potent synergistic effect. Safety evaluation of the treatments by monitoring animal body weights during the treatments indicates no significant changes among all groups (Fig. 13d). Nonetheless, both the PBS and GOA groups showed a drop in body weight after day 9, most likely because of aggressive intracranial tumor growth. The GOA/Q/D and GOA/Q/D (L+) groups, on the other hand, maintained relatively stable body weights, indicating a lower tumor burden and improved health. The animal survivals were tracked in order to assess the treatment's therapeutic efficacy in more detail (Fig. 13e). The median survival times aligned with the bioluminescence imaging: PBS (15.0 days) $\mathbf{\cong }$GOA (15.0 days) < GOA (L+) (17.0 days) < GOA/Q/D (20.0 days) < GOA/Q/D (L+) (23.0 days). By calculating the average survival times (mean ± SD), the GOA/Q/D (L+) group shows significantly longer survival times (22.7 ± 1.0 days) than all other groups (Fig. 13f). As non-targeted GO/Q/D or a group receiving free DOX + quercetin at equivalent doses were not used, we cannot rule out the observed benefits arising from passive accumulation and/or nanocarrier-mediated pharmacokinetics rather than receptor-specific targeting. In the future, active targeting versus passive EPR accumulation as well as therapeutic advantage provided by the nanocarrier platform over conventional co-administration of free drugs should be studied.

Fig. 13.

The representative IVIS bioluminescence images on days 7, 9, 13 and 16 (a), the normalized bioluminescence intensity (BLI) value on days 9, 13 and 16 (b), the change of normalized BLI value with time (c), the change of animal body weight with time (d), the survival curve (e), and the average survival time (mean ± SD) (f) when nude mice bearing U87 brain tumors received different treatments (n = 6). Log-rank Mantel-Cox test was performed for survival comparison. ^αp < 0.05 compared with PBS, ^βp < 0.05 compared with GOA, ^γp < 0.05 compared with GOA (L+), ^δp < 0.05 compared with GOA/Q/D.

Following sacrifice, blood samples were taken from animals in each group, and hematological analysis was carried out to assess potential in vivo toxicity. Good systemic biocompatibility for the safety of nanoparticle administration and laser irradiation was concluded by analyzing the hematological parameters compared to the PBS control group, which showed no significant deviations (Fig. 14a). Major organs (heart, lung, liver, spleen, and kidney) of sacrificed mice were harvested at the end of the experiment, and tissue slices were prepared for H&E staining to assess possible systemic toxicity from the treatment. As shown in Fig. 14b, no apparent changes in tissue morphology were noted in mice subject to different treatments when compared with the PBS group. Furthermore, the cardiotoxicity from DOX can be considered minimal in GOA/Q/D. Our data indicate GOA/Q/D (L+) treatment was well-tolerated without observable adverse effects under the specific conditions tested. While intracranial NIR irradiation was well-tolerated in this cohort without acute thermal or histopathological toxicity at 1.5 W/cm², the small sample size and short follow-up period preclude definitive conclusions regarding long-term safety. Future chronic safety studies with larger cohorts are strictly required to affirm the safety of the treatments.

Fig. 14.

The hematological analysis of blood samples (n = 4) (a) and the representative H&E staining of excised major organs (b) from the sacrificed mice after different treatments. Bar = 50 μm.

Hematoxylin and eosin (H&E) staining was performed on tumor samples to assess cell morphology at the end of the study (Fig. 15a). Tumor cells have their nuclei stained purple. Higher cellular density is observed in the PBS and GOA groups due to aggressive tumor growth. The GOA (L+) group, on the other hand, showed a noticeable reduction in cell density in response to NIR laser-induced heat stress, supporting the efficacy of PTT. The cell density in the tumor section was further reduced in the GOA/Q/D and GOA/Q/D (L+) groups as expected. The expression of Ki-67, cleaved caspase-3 (CC3), and HSP70 proteins in the tumor section was also evaluated by immunohistochemical (IHC) staining (Fig. 15a). IHC staining of the proliferation marker Ki-67 in tumor tissues revealed a trend consistent with the H&E staining results. The Ki-67 expression is in the order PBS ≅ GOA > GOA (L+) > GOA/Q/D > GOA/Q/D (L+). The GOA/Q/D (L+) group shows the lowest Ki-67 expression, which supports it as the most effective treatment modality to suppress U87 glioma tumor growth in vivo. The staining of CC3, a validated cell apoptosis marker, shows a reversed trend as expected. Tumor sections in the GOA/Q/D (L+) group show the highest expression of CC3 due to cell apoptosis. The reduction in cell proliferation rate from Ki-67 expression coincides with the higher apoptosis rate from the CC3 expression. We observed highly upregulated HSP70 protein expression in the GOA (L+) group compared to the PBS, GOA, and GOA/Q/D groups due to NIR-induced heat stress. The higher HSP70 expression in the GOA/Q/D group over GOA may result from the fact that DOX can increase HSP70 levels in cancer cells as a protective mechanism against genotoxic stress and apoptosis [66]. Overexpression of HSP70 can significantly reduce the effectiveness of DOX by allowing cancer cells to escape DOX-induced G2/M cell cycle arrest, allowing cells to survive chemotherapy stress [67]. However, in GOA/Q/D (L+), the HSP70 expression was drastically downregulated to a level compared to GOA (L+). This is due to the presence of the HSP70 inhibitor quercetin in GOA/Q/D. Quercetin can inhibit heat-induced HSP70 expression by interfering with the activation of heat shock factor 1 (HSF1), the primary transcription factor responsible for HSP70 synthesis [68]. Quercetin acts as a transient natural inhibitor of HSP70 expression during acute cellular stress. It modulates the cellular stress response primarily by suppressing the HSF1 regulatory pathway, which prevents HSP70 upregulation and renders cells more vulnerable to stress-induced apoptosis. Quercetin downregulates HSP70 induction by inhibiting upstream kinases such as casein kinase 2 and calmodulin-dependent kinase II. By inhibiting these kinases, quercetin blocks the heat-induced phosphorylation of HSF1 (Wang et al., 2009). This prevents HSF1 from binding to the heat shock element on the DNA, effectively halting the transcription of HSP70. A semi-quantitative analysis of IHC staining indicates that the GOA/Q/D (L+) group shows a downregulation in Ki-67 expression, an upregulation in CC3 expression with significant differences from all other groups, and a downregulation of HSP70 expression compared to the GOA (L+) group (Fig. 15b).

Fig. 15.

(a) The H&E and immunohistochemical (IHC) staining of Ki-67, cleaved caspase 3 (CC3), and HSP70 in tumor sections when nude mice bearing U87 tumors were treated by intravenous administration of PBS, GOA, GOA/Q/D. The GOA (L+), GOA/Q/D (L+) groups were irradiated with NIR laser (1.5 W/cm²) for 3 min, one hour post-administration of the nanoparticles. Bar = 50 μm. (b) Semi-quantitative analysis of Ki-67, CC3, and HSP70 expression. The protein expression was quantified semi-quantitatively using ImageJ software by measuring the percentage of immunoreactive areas within a defined region of interest (ROI) (n = 4). ^αp < 0.05 compared with PBS, ^βp < 0.05 compared with GOA, ^γp < 0.05 compared with GOA (L+), ^δp < 0.05 compared with GOA/Q/D.

Taken together, GOA/Q/D provides superior drug-loading performance. Unlike conventional nanocarriers that suffer from low encapsulation efficiencies for hydrophobic agents, the GOA/Q/D platform leverages a highly optimized surface-to-volume ratio and multivalent chemical affinities. This architecture yields significantly higher loading capacities for both quercetin and DOX simultaneously, minimizing the carrier-associated matrix required for therapeutic efficacy (Lv et al., 2024). GOA/Q/D also has pH-responsive drug release behaviour. While standard formulations often exhibit premature cargo leakage in systemic circulation, GOA/Q/D demonstrates highly stable retention at physiological pH (∼7.4). Conversely, it triggers rapid, accelerated payload release exclusively within acidic endosomes/lysosomes and the acidic tumor microenvironment (TME), ensuring localized cytotoxicity while sparing healthy tissues (Yan et al., 2026). Concurrently, GOA/Q/D provides potent quercetin-derived HSP70 inhibition. A key limitation of standalone photothermal therapies is the development of thermotolerance in cancer cells via heat shock protein 70 (HSP70) overexpression. By effectively bypassing this resistance mechanism through the synchronized intracellular delivery of quercetin, GOA/Q/D can achieve robust, localized suppression of HSP70. This can sensitize the target cells to thermal stress and downregulate the downstream cytoprotective pathways (Ji et al., 2025). Lastly, GOA/Q/D can provide enhanced mild photothermal therapy outcomes. Traditional photothermal agents require high-temperature ablation, which frequently causes severe inflammatory responses and collateral damage to adjacent healthy tissues. By synergizing HSP70 inhibition with optimized near-infrared (NIR) absorption, GOA/Q/D (L+) treatment can achieve complete tumor eradication under mild photothermal conditions (< 50 °C). This significantly mitigates secondary inflammation and enhances the overall safety profile of the treatment (Lv et al., 2022).

4. Conclusions

A multifunctional Angiopep-2-functionalized graphene oxide nanoplatform (GOA/Q/D) was successfully developed for targeted chemo-photothermal therapy of glioblastoma. The nanosystem enabled efficient BBB penetration and tumor targeting, while providing pH-responsive co-delivery of DOX and quercetin together with effective NIR-triggered photothermal conversion. Angiopep-2 functionalization leads to increased cellular uptake and intracellular drug accumulation, resulting in pronounced synergistic cytotoxicity, apoptosis, and antitumor activity against U87 glioblastoma cells combined with NIR irradiation. Furthermore, GOA/Q/D exhibited excellent hemocompatibility and negligible hemolytic effects at therapeutic concentrations. In vivo investigations confirmed enhanced Angiopep-2-mediated brain accumulation, significant tumor growth inhibition, and prolonged survival in glioma-bearing mice. Overall, the integration of targeted delivery, dual-drug synergistic chemotherapy, and photothermal therapy highlights the potential of GOA/Q/D as a promising nanomedicine platform for the treatment of glioblastoma. Further investigations focusing on evaluating its long-term safety, pharmacokinetics, and therapeutic efficacy in clinically relevant orthotopic glioblastoma models are warranted to facilitate its translation toward clinical applications.

CRediT authorship contribution statement

Yu-Jen Lu: Validation, Supervision, Methodology, Formal analysis, Data curation, Conceptualization. Banendu Sunder Dash: Validation, Methodology, Formal analysis, Conceptualization. Karthika Suresh: Writing – original draft, Investigation, Formal analysis. Jyh-Ping Chen: Writing – review & editing, Supervision, Resources, Project administration, Funding acquisition, Formal analysis, Conceptualization.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Jyh-Ping Chen reports financial support was provided by National Science and Technology Council. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

Data will be made available on request.