Mannosylated graphene oxide nanotherapeutics co-delivering docetaxel and a STING agonist reprogram myeloid cells and potentiate antitumor immunity

Abstract

In triple-negative breast cancer (TNBC), tumor-associated macrophages (TAMs) and tumor-infiltrating dendritic cells (TIDCs) often exhibit an immunosuppressive phenotype. This myeloid-dominant milieu reinforces an immune-cold state, promotes metastasis and contributes to immunotherapy resistance. In this study, we developed a mannose-functionalized graphene oxide nanocarrier (GO-EDM) with favorable biocompatibility and high NIR photothermal conversion efficiency. Docetaxel and vadimezan (DMXAA), a prototypical STING agonist, were co-loaded into GO-EDM to generate GO-EDM-DTX-Vad. In vitro, the nanocomposite promoted M2-to-M1 repolarization, thereby attenuating pro-tumor paracrine signaling from M2 macrophages and limiting TNBC cell proliferation, migration and invasion. Concomitantly, it promoted dendritic cell maturation and induced immunogenic cell death (ICD) in tumor cells, collectively enhancing antigen presentation and T-cell activation. In vivo, both intratumoral administration and systemic intravenous administration of GO-EDM-DTX-Vad effectively suppressed primary tumor growth. In a lung metastasis model, when combined with 808 nm NIR irradiation, the pulmonary metastatic burden was significantly reduced. Specifically, this treatment reduced the accumulation of M2 macrophages, MDSCs and Tregs within tumors, while promoting DC maturation and increasing intratumoral infiltration of M1 macrophages, CD8⁺ T cells and memory T cells. Taken together, GO-EDM-DTX-Vad leverages passive tumor accumulation and mannose receptor–guided dual targeting of TAMs and TIDCs to integrate DTX-based chemotherapy, STING-mediated immune activation and mild NIR photothermal therapy. This integrated chemo–photothermal–immunotherapeutic design couples direct tumor cell killing with myeloid reprogramming and immune activation, offering a unified strategy for metastatic TNBC.

Graphical abstract

(A) Synthesis and stimuli-responsive drug release of GO-EDM-DTX-Vad. (B) Schematic illustration of GO-EDM-DTX-Vad for precision drug delivery, immunosuppressive tumor microenvironment reprogramming and cancer therapy.

1. Introduction

Triple-negative breast cancer (TNBC) is an aggressive subtype characterized by rapid progression and a high propensity for lung metastasis. Chemotherapy remains the clinical mainstay, yet its efficacy is limited by restricted tumor specificity, systemic toxicity and acquired resistance [1]. Moreover, an immunosuppressive tumor microenvironment (TME) enriched in M2-like tumor-associated macrophages (TAMs), myeloid-derived suppressor cells (MDSCs), T regulatory cells (Tregs) and tumor-infiltrating dendritic cells (TIDCs) further blunts therapeutic responses and promotes metastatic relapse [1,2]. In particular, M2-like TAMs drive immune suppression and angiogenesis, whereas dysfunctional TIDCs provide insufficient antigen presentation [3]. Accordingly, integrated strategies are needed to eliminate tumor cells while reprogramming the immunosuppressive TME into a therapy-responsive state.

Accumulating evidence implicates the STING pathway as a key bridge between innate and adaptive immunity. STING activation induces type I interferons and pro-inflammatory cytokines, promotes dendritic cell maturation and expands tumor-specific CD8⁺ T cells [4,5]. However, free STING agonists are often rapidly cleared, distribute nonselectively and may trigger systemic inflammation, which limits clinical translation [6,7]. These limitations have motivated the development of nanomedicine platforms that improve intratumoral retention and enable spatiotemporally controlled combination therapy. Such platforms can co-deliver chemotherapeutics, immunomodulators and phototherapeutic agents to both tumor cells and myeloid subsets [8,9]. Mannose-functionalized nanocarriers are particularly attractive because the mannose receptor (MR, CD206) is highly expressed on M2-like TAMs and TIDCs, enabling MR-mediated active targeting and preferential uptake by these immunosuppressive myeloid populations [10,11]. Nevertheless, most reported systems focus on a single modality, and relatively few achieve coordinated regulation of the tumor myeloid compartment and highly metastatic TNBC cells.

Graphene oxide (GO) offers a versatile drug-delivery scaffold owing to its large surface area, π-conjugated framework and abundant oxygen-containing groups [12,13]. These features enable covalent functionalization with hydrophilic ligands and high-capacity loading of aromatic drugs via noncovalent interactions. GO also exhibits strong near-infrared (NIR) absorbance, enabling photothermal therapy (PTT) [13,14]. Surface functionalization with mannose derivatives can further improve aqueous stability and biocompatibility, thereby facilitating receptor-mediated uptake by immune cells [[15], [16], [17]]. Despite these advantages, mannosylated GO remains underexplored as a dual TAM/TIDC-targeting platform that integrates chemotherapy, immunotherapy and photothermal modulation of the TNBC microenvironment [18].

In this study, we designed an ethylenediamine–mannosylated GO nanocarrier (GO-EDM) and co-loaded docetaxel (DTX) and the STING agonist vadimezan to generate a multifunctional nanocomposite, GO-EDM-DTX-Vad. Mannose functionalization increased tumor accumulation after systemic administration and enriched the formulation within MR-high M2-like TAMs and TIDCs in the tumor microenvironment. Docetaxel induces tumor cell apoptosis and immunogenic cell death (ICD), while vadimezan activates the STING signaling pathway to stimulate innate immune responses. These combined effects further promoted dendritic cell maturation and robust activation of effector T cells. Concomitantly, the nanocomposite remodels the tumor immune microenvironment by reprogramming M2-like macrophages toward an M1-like phenotype, thereby reducing pro-tumor paracrine support for TNBC cell proliferation, migration and invasion. Meanwhile, GO enabled mild and controllable photothermal therapy under 808-nm irradiation. This photothermal component synergized with chemotherapy and immune activation to improve primary tumor control and suppress lung metastasis.

In summary, we developed a mannosylated GO-based chemo–photothermal–immunotherapy platform. By integrating dual targeting of TAMs and TIDCs with chemotherapy and photothermal therapy, the nanocomposite couples direct tumor cell killing with effective remodeling of the tumor immune microenvironment. This study provides a feasible therapeutic strategy and mechanistic rationale for immune reprogramming in metastatic TNBC.

2. Materials and methods

2.1. Materials and cell lines

D-Mannose and vadimezan (DMXAA, abbreviated as Vad) were purchased from Sigma (St. Louis, MO, USA). Ethylenediamine (EDA), 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC) and N-hydroxysuccinimide (NHS) were obtained from Aladdin (Shanghai, China). Docetaxel, sulfo-Cy5.5, the Cell Counting Kit-8 (CCK-8) and rhodamine B were purchased from MedChemExpress (Monmouth Junction, NJ, USA). Graphene oxide was supplied by Nanjing XFNANO Materials Tech Co., Ltd. (Nanjing, China). Calcein-AM/PI Live/Dead staining kits were obtained from Servicebio (Wuhan, China). Annexin V-FITC/PI apoptosis detection kits were purchased from Elabscience Biotechnology Co., Ltd. (Wuhan, China). The EdU cell proliferation kit was purchased from Beyotime Biotechnology (Shanghai, China). Matrigel matrix and Transwell chambers were purchased from Corning (Corning, NY, USA). The DC2.4 dendritic cell line and RAW264.7 macrophage cell line were obtained from Procell Life Science & Technology Co., Ltd. (Wuhan, China). The 4T1 and MDA-MB-231 cell lines were purchased from Boster Biological Technology (Wuhan, China).

2.2. Synthesis of mannose-functionalized graphene oxide

Following the procedure reported by Sousa et al. [15], we synthesized mannosylated ethylenediamine (EDM) and subsequently covalently conjugated it to GO to obtain GO-EDM. Briefly, ethylenediamine was added to methanol at a volume ratio of 1:6 (v/v), and a small amount of iodine was introduced as a catalyst. The mixture was subjected to ultrasonication for 1 h and then allowed to react further at room temperature overnight to ensure completion of the reaction. The resulting precipitate was repeatedly washed with methanol and ethanol in a vacuum separation funnel to remove excess iodine and by-products.

An aqueous GO solution (2 mg/mL) was sequentially mixed with mannose-grafted ethylenediamine (6 mg/mL) and EDC/NHS as coupling agents, and the reaction was allowed to proceed under magnetic stirring for 24 h. After completion, the product was washed repeatedly until the pH of the supernatant approached neutrality. The suspension was dialyzed (MWCO 3.5 kDa) against deionized water for 72 h to remove residual small molecules and salts. The purified product was then lyophilized to obtain GO-EDM powder for subsequent physicochemical characterization and drug loading.

2.3. Cytotoxicity and biocompatibility evaluation of the nanocarriers

2.3.1. In vitro cytotoxicity

DC2.4 and RAW264.7 cells were seeded in 96-well plates. Cells were treated with GO or GO-EDM (0–100 μg/mL) for 24 h. In a time-course experiment, cells were treated with GO or GO-EDM (10 μg/mL) for 0–72 h. After treatment, CCK-8 reagent (KGA1606, KeygenBioTECH) was added, and the plates were incubated for 2 h. Cell viability was normalized to the untreated control group and defined as 100%.

2.3.2. Plasma coagulation

The whole blood from healthy female BALB/c mice was collected in sodium citrate tubes to prepare platelet-poor plasma (PPP). PPP was incubated with GO or GO-EDM (1–100 μg/mL) at 37 °C for 30 min. The activated partial thromboplastin time (APTT), prothrombin time (PT) and fibrinogen (FIB) were measured using an automated coagulation analyzer (STA Compact Max, STAGO, France).

2.3.3. Hemolysis assay

The heparinized whole blood from healthy female BALB/c mice was processed to isolate red blood cells (RBC). RBC suspensions were incubated with GO-EDM (1–50 μg/mL) for 1 h. PBS and deionized water served as the negative and positive controls. Supernatant absorbance was measured by UV–vis spectrophotometry to quantify hemolysis.

2.4. Optimization of mass ratio, drug loading and in vitro release

2.4.1. Optimization of the carrier-to-drug mass ratio

The molar ratio of DTX to vadimezan was fixed at 1:1 based on preliminary biological activity to ensure effective STING activation while maintaining formulation stability. To optimize the carrier-to-drug composition, nanocomposites were prepared at different GO-EDM: (DTX + Vad) mass ratios of 2:1, 3:2, 1:1, 2:3 and 1:2. GO-EDM was mixed with the corresponding amount of DTX and Vad. The mixture was stirred at room temperature in the dark for 72 h. The dispersion was then dialyzed (MWCO 8–14 kDa) against deionized water for 72 h.

2.4.2. Determination of encapsulation efficiency and drug loading

Encapsulation efficiency (EE) and loading efficiency (LE) were determined by HPLC (Agilent 1260 Infinity, Agilent Technologies, USA). 10 mg GO-EDM-DTX-Vad was extracted in methanol (50 mL). The separation was performed on a TC-C18 column (4.6 × 250 mm, 5 μm) and the mobile phase was acetonitrile/water (70:30, v/v) at 1.0 mL/min. DTX and vadimezan were detected at 227 nm and 345 nm. Drug content was calculated using external calibration curves.

2.4.3. In vitro drug release study

In vitro drug release was evaluated using a dialysis method. GO-EDM-DTX-Vad (10 mg) was dispersed in release medium (5 mL). The sample was sealed in a dialysis bag (MWCO 8–14 kDa) and immersed in 50 mL medium at pH 7.4, 6.5 or 5.5. At predetermined time points, 0.5 mL of the external medium was collected. The same volume of fresh medium was added. For NIR-triggered release, samples were irradiated with an 808 nm laser (1.0 W/cm², 5 min) before each sampling.

2.5. Characterization of nanocarriers and nanocomposites

Morphology was examined by SEM (SU8010, Hitachi, Tokyo, Japan) and TEM (JEM-1200EX, JEOL, Tokyo, Japan). Structural and compositional features were analyzed by XRD (Rigaku SmartLab SE, Japan), FTIR (Nicolet iS20, Thermo Fisher Scientific, USA), XPS (Nexsa, Thermo Fisher Scientific, USA) and Raman spectroscopy (LabRAM HR Evolution, Horiba, Japan). Hydrodynamic diameter and PDI were measured using a Zetasizer Nano ZS90 (Malvern Panalytical, UK). TGA was performed on an STA200 thermal analyzer (Hitachi, Japan). DSC was performed using a Q20 calorimeter (TA Instruments, USA). Liquid-state ¹³C NMR spectra were acquired on an Avance III HD 500 MHz spectrometer (Bruker, Germany).

2.6. In vivo targeting and biodistribution of nanocarriers

4T1 cells (1 × 10⁶) were injected subcutaneously into the right flank of female BALB/c mice. When tumors reached 150–200 mm³, mice were randomized into two groups (n = 3). They received an intravenous injection of 100 μL Cy5.5-GO or Cy5.5–GO-EDM-DTX-Vad. The whole-body fluorescence images were acquired under isoflurane anesthesia from 0 to 48 h post-injection using an IVIS Spectrum system (PerkinElmer, USA). At 48 h, mice were euthanized. Major organs and tumors were collected for ex vivo fluorescence imaging.

2.7. Uptake of nanocomposites by dendritic cells

To assess uptake by DC2.4 cells, GO-EDM-DTX-Vad was labeled with rhodamine B (RhoB). For competition assays, DC2.4 cells were pretreated with mannose (1 mg/mL) for 1 h. Mannose was maintained during incubation. For photothermal treatment, cells were irradiated with an 808 nm laser (1 W/cm², 5 min). After treatment, cellular uptake was quantified by flow cytometry (DxFLEX, Beckman Coulter, USA) and visualized by confocal laser scanning microscopy (Nikon A1, Nikon, Japan).

2.8. In vitro activation and maturation of dendritic cells

According to the methods described by Wang and Pei et al. [19,20], DC2.4 cells were seeded in 24-well plates and incubated with the indicated nanocomposites for 24 h at 37 °C For photothermal studies, cells were additionally irradiated with an 808 nm NIR laser (1 W/cm², 5 min). Fc receptors were blocked by pre-incubation with TruStain FcX™ PLUS (anti-mouse CD16/32, BioLegend). The cells were then stained with anti-mouse CD40 (APC, clone 3/23), CD80 (APC, clone 16-10A1) and CD86 (PE, clone A17199A) (BioLegend) and analyzed by flow cytometry.

2.9. In vitro polarization of M2-like to M1-like macrophages

2.9.1. Western blot analysis

Proteins were extracted from treated M2-polarized macrophages using a phosphorylated protein extraction kit (BC3730, Solarbio, Beijing, China). Protein concentrations were determined using a BCA protein assay kit (ZJ101, Epizyme, Shanghai, China) and normalized to equal amounts. Equal amounts of protein were separated by 10% SDS–PAGE and subsequently transferred onto PVDF membranes. After blocking with 5% BSA in TBST for 1 h at room temperature, the membranes were incubated overnight at 4 °C with primary antibodies against TBK1 (3504, CST), phospho-TBK1 (R30260, Zenbio, China), IRF3 (ET1612-14, HuaBio), phospho-IRF3 (HA722772, HuaBio) and GAPDH (ET1601-4, HuaBio). The membranes were then washed and incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies for 1 h at room temperature. The immunoreactive proteins were detected using an ultra-high-sensitivity enhanced chemiluminescence kit (YS0123, Immunoway, San Jose, CA, USA). Protein expression levels were quantified using ImageJ software.

2.9.2. Repolarization of pre-established M2-like macrophages

RAW264.7 cells were seeded in 12-well plates at 1 × 10⁶ cells per well. Cells were stimulated with IL-4 and IL-13 (20 ng/mL) for 24 h to induce M2-like polarization. The medium was replaced with complete medium containing the indicated nanocomposites for 24 h. After treatment, supernatants were collected, and TGF-β, IL-10 and TNF-α were measured by ELISA kits (Boster, Wuhan, China).

2.9.3. RT-qpcr

Total RNA was extracted using the NcmSpin Cell/Tissue Total RNA Kit (NCM Biotech, China) and reverse-transcribed into cDNA with the RevertAid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific, USA). RT-qPCR was performed using PowerUp™ SYBR™ Green Master Mix (Thermo Fisher Scientific, USA), with GAPDH as the internal control. Relative expression of iNOS, CD86, CD206 and IL-10 was calculated using the 2⁻ΔΔCt method. The primer sequences are provided in Table S1. PCR primer synthesis was performed by Beijing Tsingke Biotech Co., Ltd.

2.9.4. Phagocytosis assay and T cell activation assay

According to the methods described by Rong et al. [21], dead 4T1 cells were labeled with Annexin V-FITC. RAW264.7 macrophages were polarized to the M2 phenotype and then repolarized by treatment with the nanocomposites. After co-incubation for 4 h, cells were stained with anti-mouse F4/80 (Clone: BM8) to identify macrophages, and phagocytosis was quantified by flow cytometry as the percentage of F4/80⁺FITC⁺ cells. As for T cell activation assay, T cells were isolated from the spleens of 8-week-old female BALB/c mice. M1-like macrophages obtained after treatment with the nanocomposites were first co-cultured with killed 4T1 cells for 4 h, followed by the addition of splenic T cells. After 24 h of co-culture, T cell activation was evaluated by flow cytometry by measuring the proportion of CD8⁺CD69⁺ T cells.

2.9.5. Inhibition of de novo M2 polarization during IL-4/IL-13 conditioning

RAW264.7 cells were seeded in 12-well plates at 1 × 10⁶ cells per well. IL-4 and IL-13 (20 ng/mL) were added together with the indicated nanocomposites. Cells were incubated for 48 h to assess inhibition of M2 polarization during induction. The percentages of CD86⁺ and CD206⁺ cells were determined by flow cytometry.

2.9.6. Uptake by macrophages of different phenotypes

Nanocomposite uptake was assessed as described above. M0 and M1-like macrophages were included as controls. M1-like macrophages were generated using IFN-γ (20 ng/mL, Novoprotein, Shanghai, China, C746) and LPS (100 ng/mL,Sigma) for 24 h. Unstimulated RAW264.7 cells were defined as M0 macrophages.

2.10. In vitro effects of nanocomposites on M2 macrophage–mediated tumor cell functions

2.10.1. Preparation of macrophage-conditioned media (CM)

RAW264.7 cells were polarized to M1-like (LPS/IFN-γ) or M2-like (IL-4/IL-13) phenotypes for 24 h. For M2 reprogramming, polarized M2 macrophages were further incubated with the indicated nanocomposites for 24 h. After treatment, cells were washed thoroughly and cultured in fresh cytokine-free medium for 48 h to minimize carryover of polarization stimuli and formulations. The conditioned media (CM) were collected and stored at −80 °C.

2.10.2. Colony formation

4T1 and MDA-MB-231 cells (500 cells per well) were seeded in 6-well plates and cultured in macrophage CM for 14 days. The colonies were fixed and stained with crystal violet.

2.10.3. Wound healing

The confluent 4T1 and MDA-MB-231 monolayers were scratched and cultured in macrophage CM. Images of the same wound regions were captured at 0 h and 24 h using an inverted microscope. Wound closure was quantified as the percentage reduction in wound width relative to 0 h.

2.10.4. EdU proliferation assay

4T1 and MDA-MB-231 cells were cultured in macrophage CM for 48 h, pulsed with EdU (10 μM) and analyzed by flow cytometry to determine the EdU positive fraction.

2.10.5. Transwell migration and invasion

For Transwell assays, 4T1 or MDA-MB-231 cells (3 × 10⁴) were suspended in macrophage CM and seeded in the upper chambers (725321, NEST Biotechnology, Wuxi, China). Medium containing 10% FBS (BaiDi Biotechnology Co., Ltd.(BDBIO, Shanghai, China)) was placed in the lower chambers. After 24 h, migrated cells were fixed and stained with crystal violet. The invasion assays were performed using Matrigel-coated inserts under the same conditions.

2.11. In vitro antitumor activity

2.11.1. CCK-8 assay and IC₅₀ determination

4T1 and MDA-MB-231 cells were seeded in 96-well plates and treated with free DTX or GO-EDM-DTX-Vad for 48 h. At the end of each treatment, CCK-8 solution was added and incubated for 2 h. Dose–response curves were fitted to calculate IC₅₀ values.

2.11.2. Apoptosis and live/dead staining

4T1 and MDA-MB-231 cells were treated with the indicated nanocomposites for 24 h. The apoptosis was quantified by Annexin V-FITC/PI staining (Elabscience, Wuhan, China) followed by flow cytometry, and cell viability was visualized by Calcein-AM/PI live/dead staining (Servicebio, Wuhan, China) using CLSM.

2.12. In vitro ICD assessment and DC maturation

2.12.1. ICD-associated markers

4T1 and MDA-MB-231 cells were treated with the indicated formulations for 8 h. The treated cells were incubated overnight with anti-HMGB1 and anti-CRT antibodies (Proteintech, Wuhan, China), followed by labeling with a secondary antibody. The images were captured by CLSM.

2.12.2. Transwell co-culture for DC maturation

DC2.4 cells were seeded in the lower chamber, and 4T1 cells were seeded in 0.4-μm Transwell inserts (Corning, USA) and treated with the indicated formulations for 24 h. DC2.4 cells were processed for flow-cytometric analysis of maturation markers as described in Section 2.8.

2.13. Animal ethics

All animal experiments in this study were approved by the Animal Experimentation and Welfare Committee of Capital Medical University (approval number: AEEI-2025-1048). Female BALB/c mice (4–6 weeks of age) were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. and housed under standard specific pathogen-free (SPF) conditions.

2.14. Establishment of the 4T1 subcutaneous breast cancer model and intratumoral treatment

Female BALB/c mice were inoculated subcutaneously in the right flank with 4T1 cells (1.0 × 10⁶). Body weight and tumor size were recorded every 2 days. Tumor length (a) and width (b) were measured using calipers and tumor volume was calculated as V = a × b²/2. When tumors reached approximately 200 mm³, mice were randomized into six groups (n = 5): PBS, GO-EDM, GO-EDM-DTX, GO-EDM-Vad, GO-EDM-DTX-Vad, and GO-EDM-DTX-Vad + NIR. The baseline body weights were comparable across groups, with no significant differences observed (Fig. S20). Treatments were administered intratumorally every 3 days for three doses. The injection volume was 100 μL. DTX-containing formulations were dosed at 20 mg/kg DTX equivalent. For GO-EDM-DTX-Vad + NIR, tumors were irradiated at 808 nm (1 W/cm²) for 10 min at 30 min after each injection. Mice were euthanized 8 days after the last dose. Tumors were excised and weighed.

2.15. Establishment of the 4T1 lung metastasis model and intravenous treatment

The subcutaneous 4T1 model was established as described above. Besides, 4T1 cells (1.0 × 10⁶) were injected slowly via the tail vein to induce experimental lung metastasis. When tumors reached approximately 200 mm³, mice were randomized into five groups (n = 5): PBS, GO-EDM, free DTX + Vad, GO-EDM-DTX-Vad and GO-EDM-DTX-Vad + NIR. Treatments were administered intravenously every 3 days for three doses. The injection volume was 100 μL. DTX-containing regimens were normalized to 20 mg/kg DTX equivalent. For GO-EDM-DTX-Vad + NIR, the primary tumor was irradiated at 808 nm (1 W/cm²) for 10 min at 8 h after each injection. Mice were sacrificed 8 days after the last dose. Lungs and tumors were collected for gross nodule counting and H&E staining. Spleens and tumors were processed into single-cell suspensions for flow cytometry. Serum IFN-γ (EMC101g, Neobioscience Technology Co, Ltd.), TNF-α and IL-10 were quantified by ELISA. Paraffin-embedded tumor sections were analyzed by immunofluorescence for CD8⁺ T cells, CD11c⁺ dendritic cells, MDSCs and M1/M2 macrophages.

2.16. Analysis of infiltrating immune cells

Single-cell suspensions from the spleen were prepared by mechanical dissociation and filtration according to a standard protocol (Bio-protocol, USA). Fc receptors were blocked by incubating cells with anti-CD16/32 antibodies at room temperature for 15 min, followed by staining with Zombie NIR™ Fixable Viability Kit to discriminate live and dead cells.

Tumor tissues were digested with collagenase IV, hyaluronidase and DNase I (MedChemExpress, USA) at 37 °C for 1 h. After digestion, intratumoral single cells were enriched by Percoll density gradient centrifugation (Solarbio, Beijing, China). Cells were then incubated with anti-CD16/32 antibodies at room temperature for 15 min to block Fc receptors, followed by Zombie NIR™ staining for viability and labeled with Alexa Fluor® 700–CD45 to identify CD45⁺ immune cell populations.

The following fluorochrome-conjugated antibodies (all from BioLegend) were used: Alexa Fluor® 700–CD45 (I3/2.3), PE–CD3 (17A2), FITC–CD4 (RM4-5), APC–CD8a (53-6.7), PE–CD44 (IM7), BV421–CD62L (MEL-14), APC–CD25 (PC61), APC–F4/80 (BM8), PE–CD86 (A17199A), PE/Cyanine7–CD206 (C068C2), FITC–CD49b (DX5), FITC–CD11c (N418), APC–CD80 (16-10A1), FITC–CD11b (M1/70), and APC–Gr-1 (RB6-8C5). For intracellular transcription factor staining, the True-Nuclear™ Transcription Factor Buffer Set and PE–Foxp3 (MF-14) were used according to the manufacturer's instructions.

2.17. In vivo photothermal performance

The subcutaneous 4T1 model was established as described above. When tumors reached 150–200 mm³, mice were randomized into an intratumoral group or an intravenous group (n = 6). Mice received 100 μL of PBS or GO-EDM-DTX-Vad via intratumoral or intravenous administration. Tumors in the intratumoral cohort were irradiated 30 min after dosing, whereas those in the intravenous cohort were irradiated 8 h after dosing, the time selected based on in vivo biodistribution to coincide with maximal tumor accumulation. Near-infrared (NIR) irradiation was applied at 808 nm (1 W/cm²) for 10 min. Tumor surface temperatures were monitored using an infrared thermal imaging system.

2.18. Statistical analysis

All quantitative results are reported as mean ± SD from at least three independent experiments. Statistical analyses were performed using SPSS 26.0. Assumptions of normality and homogeneity of variance were evaluated before hypothesis testing. Two-group comparisons were analyzed using an unpaired two-tailed Student's t-test. For three or more groups, differences were assessed by one-way ANOVA followed by Tukey's post hoc test. Statistical significance was defined as P < 0.05. Significance levels are indicated as ∗P < 0.05, ∗∗P < 0.01, and ∗∗∗P < 0.001.

3. Results and discussion

3.1. Synthesis and characterization of the GO-EDM nanocarrier

The mannose-functionalized nanocarrier GO-EDM was synthesized using a modified synthetic route [17,22]. Briefly, ethylenediamine was first conjugated to the reducing end of D-mannose to generate mannosylated ethylenediamine (EDM). EDM was then covalently grafted onto graphene oxide (GO) via EDC/NHS-mediated amidation between its amine groups and the carboxyl groups on GO, thereby yielding mannosylated ethylenediamine–modified GO (GO-EDM, Figs. S1 and S2).

FTIR and ¹³C NMR consistently supported glycosylamine formation between D-mannose and ethylenediamine. FTIR showed a broad O–H/N–H stretching band (3281 cm⁻¹) and a band at 1397 cm⁻¹ consistent with C–N stretching (Fig. 1A). In ¹³C NMR, the mannose-region resonances changed and the ethylenediamine signal split at 42.96 ppm (Fig. 1B) [17]. Together, these features indicate covalent coupling at the reducing end of mannose.

Fig. 1.

Characterization and biocompatibility of mannosylated graphene oxide. (A) FTIR spectra of D-mannose and mannosylated ethylenediamine. (B) ¹³C NMR spectra of ethylenediamine, D-mannose and mannosylated ethylenediamine. (C–H) XPS survey and high-resolution spectra of GO and GO-EDM (C1s, O1s and N1s). (I) Raman spectra of GO and GO-EDM. (J, K) DC2.4 cell viability after 24 h exposure to GO or GO-EDM (0–100 μg/mL). (L) UV–vis spectra of supernatants from erythrocytes incubated with GO-EDM. (M − O) Coagulation assays of platelet-poor plasma after exposure to GO or GO-EDM: APTT, PT and FIB. Data are mean ± SD (n = 3). ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001; ns, not significant.

XPS corroborated the effective functionalization of the GO surface. Comparison with GO, the survey spectrum of GO-EDM exhibited a distinct N1s signal (Fig. 1C). The emergence of an O=C–N component in the C1s spectrum (288.6 eV), together with an N1s peak (399.0 eV), supports the formation of amide linkages and covalent grafting of EDM (Fig. 1D–H). Raman spectroscopy showed red shifts of the D and G bands after functionalization, indicating successful introduction of oxygen- and nitrogen-containing moieties (Fig. 1I). Accordingly, the FTIR spectrum of GO-EDM displayed two new FTIR bands at 1644 and 1567 cm⁻¹, corresponding to amide I (C=O stretching) and amide II (N–H bending). The intensified band near 1405 cm⁻¹ (C–N/C–O) and accompanying changes in the O–H/N–H, C–H and C–O regions further support amidation of carboxyl groups on the GO surface (Fig. 2C).

Fig. 2.

Formulation Optimization, physicochemical characterization, in vitro drug release and in vivo biodistribution of GO-EDM-DTX-Vad. (A, B) Encapsulation efficiency and loading efficiency DTX (A) and vadimezan (B) at different mass ratios. (C) FTIR spectra. (D) XRD patterns. (E) SEM and TEM images of GO, GO-EDM and GO-EDM-DTX-Vad (SEM scale bar, 1.00 μm; TEM scale bar, 200 nm). (F) Zeta potential. (G) hydrodynamic size. (H) DSC thermograms. (I) TGA curves. (J, K) Cumulative release of DTX (J) and vadimezan (K) at pH 5.5, 7.4 and 8.4 with or without NIR irradiation. (L) In vivo fluorescence imaging of 4T1 tumor-bearing mice after tail-vein injection of Cy5.5-GO or Cy5.5-GO-EDM-DTX-Vad. (M) Ex vivo imaging of major organs and tumors at 48 h. Data are mean ± SD (n = 3).

3.2. Biocompatibility and biosafety Evaluation of the GO-EDM nanocarrier

We first assessed the in vitro biocompatibility and biosafety of GO and GO-EDM in DC2.4 and RAW264.7 cells. GO induced a dose-dependent decrease in cell viability. By contrast, in the GO-EDM groups, the viability of both cell lines generally remained above 80%, even after 24 h of high-dose treatment or 72 h of medium-dose exposure (Fig. 1J, K and Fig. S3–5). This level is generally considered indicative of acceptable cytocompatibility and is consistent with previous reports that surface functionalization can attenuate the cytotoxicity of GO [23].

To assess hemocompatibility, we examined the hemolytic effect of GO-EDM by measuring hemoglobin absorbance at 540/570 nm [16,24]. Distilled water treatment caused severe hemolysis and produced pronounced absorption peaks at 540 and 570 nm. In contrast, the absorbance for the GO-EDM group was comparable to that of the PBS negative control, indicating minimal disruption of erythrocyte membrane integrity and a low risk of hemolysis (Fig. 1L).

To further evaluate the effects on plasma coagulation, we measured coagulation parameters in platelet-poor plasma [25,26]. At medium to high concentrations, both GO and GO-EDM prolonged the APTT to varying degrees, whereas no significant changes were observed for PT (Fig. 1M and N). And both materials reduced plasma FIB levels at higher doses (Fig. 1O). Notably, GO-EDM induced a weaker prolongation of APTT and a less pronounced reduction in FIB than unmodified GO at equivalent mass concentrations. These results indicate that graphene-based carriers may interact with coagulation proteins and disrupt coagulation homeostasis in a dose-dependent manner. However, ethylenediamine–mannose functionalization partially attenuated these effects, with minimal impact observed at the lower doses relevant to biomedical applications [27].

3.3. Optimization of drug loading ratio and physicochemical characterization

To optimize drug-loading performance, we evaluated different carrier-to-total drug mass ratios (GO-EDM:DTX + Vad = 2:1, 3:2, 1:1, 2:3 and 1:2) and quantified encapsulation efficiency (EE) and loading efficiency (LE) (Fig. 2A, B and Fig. S6). With increasing drug input, the LE of DTX increased from 24.56% to 39.60%, whereas its EE decreased from 97.41% to 56.53%, vadimezan showed a similar trend. When the carrier-to-drug ratio exceeded 1:1, gains in LE became attenuated while EE declined more rapidly, consistent with progressive saturation of available binding sites [25]. Accordingly, a 1:1 mass ratio was selected for subsequent studies. At this ratio, the LE and EE of DTX were 35.35% and 88.47%, respectively, whereas those of vadimezan were 11.37% and 85.37%.

The FTIR spectrum of GO-EDM-DTX-Vad retained drug-associated bands at 1487 cm⁻¹ and 1329 cm⁻¹, which correspond to the characteristic signals observed for the free drugs, supporting successful co-loading onto GO-EDM (Fig. 2C). Minor peak shifts relative to the free drugs suggest noncovalent association of DTX and vadimezan with GO-EDM.

Morphologically, SEM showed that GO exhibited a wrinkled, sheet-like structure, whereas functionalization and drug loading increased the density of surface-associated features and induced mild local folding of the nanosheets. TEM further confirmed that the samples retained a two-dimensional lamellar framework with only limited local aggregation and wrinkling (Fig. 2E).

DLS measurements indicated that GO-EDM-DTX-Vad had a hydrodynamic diameter of 188.4 nm (PDI = 0.136) (Fig. 2F and Fig. S7) and the ζ-potential was about −23.6 mV (Fig. 2G). These physicochemical parameters fall within ranges commonly reported for intravenously administered nanocarriers and passive tumor accumulation. To probe environmental effects on colloidal stability, GO-EDM-DTX-Vad was incubated in NaCl, urea, and surfactant solutions [28]. The distinct changes in particle size and colloidal stability suggest that electrostatic interactions, hydrogen bonding, hydrophobic and π–π interactions collectively regulate its supramolecular organization. These cooperative interactions may contribute to the nanocomposite's environmental responsiveness and structural robustness (Fig. S8).

XRD patterns showed that the characteristic crystalline peaks of free DTX and vadimezan were absent or markedly attenuated in GO-EDM-DTX-Vad, suggesting that both drugs were present predominantly in an amorphous or highly dispersed state on the carrier (Fig. 2D) [29,30]. Consistently, the melting endotherms of free DTX and vadimezan were not observed in the DSC thermogram of the composite (Fig. 2H and Fig. S9) [29,31]. In the TGA thermogram, GO-EDM-DTX-Vad showed a more gradual weight-loss trajectory than the free drugs. This behavior is consistent with enhanced apparent thermal stability due to confinement within the carbon framework and drug–carrier interactions (Fig. 2I) [29].

3.4. In vitro release profiles of DTX and vadimezan

In vitro release studies revealed a rapid initial burst release of both drugs, followed by a slower phase that approached a sustained-release plateau. At pH 5.5, the cumulative release at 72 h reached 58.39% for DTX and 55.45% for vadimezan, and increased to approximately 67.22% and 64.14% upon NIR irradiation. At pH 7.4, both drugs showed intermediate release, whereas alkaline conditions yielded the lowest release (Fig. 2J and K).

These release profiles indicate that the GO-EDM carrier is dually responsive to acidic pH and NIR irradiation. Under tumor-relevant acidic conditions, protonation of oxygen-containing groups and grafted amines increases electrostatic repulsion and hydrophilicity. This protonation may weaken drug–carrier interactions, thereby accelerating release [30]. NIR irradiation produces a local photothermal effect that enhances molecular motion and weakens noncovalent interactions, further promoting drug diffusion [32,33]. Together, these stimuli enhance tumor-relevant release and enable improved spatiotemporal control of drug delivery [34].

3.5. In vivo biodistribution and targeting of the nanocomposite

We compared the in vivo biodistribution and clearance profiles of intravenously injected Cy5.5-GO and Cy5.5–GO-EDM-DTX-Vad. Cy5.5–GO showed broad systemic distribution at early time points (0–4 h). The fluorescence signal declined rapidly thereafter. By 8 h, no appreciable tumor accumulation was observed. In contrast, Cy5.5–GO-EDM-DTX-Vad showed a similar initial systemic distribution, with transient signal enrichment in the upper abdomen. Tumor fluorescence increased over time and peaked at 4–8 h post-injection. The tumor signal remained detectable at 24 and 48 h, consistent with prolonged circulation and sustained tumor retention (Fig. 2L).

Ex vivo imaging at 48 h showed higher tumor fluorescence in the Cy5.5–GO-EDM-DTX-Vad group, with detectable signals in reticuloendothelial organs (liver and spleen) (Fig. 2M). These findings suggest that this delivery system achieves effective passive tumor accumulation while maintaining a moderate level of hepatic uptake. In addition, surface mannose ligands may enhance receptor-mediated uptake by CD206⁺ TAMs and TIDCs in the tumor microenvironment. This mechanism may promote local retention and intratumoral distribution of the nanocomposite.

3.6. In vitro photothermal performance of the nanocomposite

Owing to its broadband optical absorption and highly delocalized π-electron structure, graphene oxide efficiently converts near-infrared (NIR) light into heat. Photothermal heating profiles under 808 nm irradiation were evaluated as a function of laser power density (0.5–2.0 W/cm⁻²) and nanocomposite concentration (0.2–1.0 mg/mL) [35]. The resulting temperature–time curves showed a clear dependence on both laser power density (Fig. 3A and B) and nanocomposite concentration (Fig. 3C and D). The heating and cooling cycles exhibited excellent stability and reproducibility (Fig. 3F). This favorable photothermal robustness supports the use of GO-EDM-DTX-Vad as a candidate for subsequent in vivo photothermal therapy (PTT) and chemo-photothermal combination studies.

Fig. 3.

Photothermal performance and GO-EDM-DTX-Vad–induced dendritic cell uptake and maturation. (A–D) Infrared thermal images and temperature–time curves of GO-EDM-DTX-Vad under 808 nm irradiation at varying power densities (0.5–2.0 W/cm², 1.0 mg/mL) or concentrations (0.2–1.0 mg/mL, 2.0 W/cm²). (E) CLSM images of DC2.4 cells incubated with RhoB-labeled nanocomposites (nuclei, blue; nanocomposites, red; scale bar, 50 μm). (F) Photothermal stability over five laser on/off cycles. (G, H) Flow cytometry quantification and representative histograms of DC2.4 uptake of RhoB–GO-EDM-DTX-Vad. (I–K) Frequencies of CD40⁺, CD80⁺ and CD86⁺ DC2.4 cells after stimulation. Data are mean ± SD (n = 3). ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001; ns, not significant.

3.7. Targeted uptake of the nanocomposite by dendritic cells and their activation

Dendritic cells (DCs) express high levels of the mannose receptor (MR, CD206) on their surface, which continuously recycle between the plasma membrane and endosomes and mediate ligand-dependent endocytosis [20,35]. This property has been widely exploited to design mannose-functionalized delivery systems for DC targeting. In cellular uptake assays, flow cytometry and CLSM images showed that intracellular fluorescence from RhoB-labeled GO-EDM-DTX-Vad increased over time (Fig. 3G and H). Preincubation with free mannose markedly reduced intracellular fluorescence, supporting MR-mediated endocytosis as a major uptake pathway (Fig. 3E and Fig. S10). Additionally, 808 nm NIR irradiation further increased intracellular fluorescence intensity. This effect may reflect photothermal enhancement of membrane permeability and endocytic uptake.

We next examined whether the encapsulated STING agonist modulated DC activation. Flow cytometry showed that GO-EDM-DTX-Vad increased surface expression of CD40, CD80 and CD86 on DCs (Fig. 3I–K) [20]. This effect was further enhanced by NIR irradiation, consistent with a more mature phenotype and STING pathway activation.

Collectively, these data indicate that GO-EDM-DTX-Vad promotes DC maturation through MR-mediated targeting, photothermal enhancement of uptake and STING-dependent immune activation.

3.8. In vitro repolarization of M2 macrophages by the nanocomposite

We evaluated the ability of the nanocomposite to modulate the phenotype of TAM-like macrophages in vitro. An M2-like macrophage model was established by stimulating RAW264.7 cells with IL-4/IL-13 for 24 h. The cells were then treated with the indicated formulations for an additional 24h. Flow cytometry further showed a reduced CD206⁺ fraction and an increased CD86⁺ fraction following nanocomposite treatment (Fig. 4A and Fig. S11). Consistently, analysis of polarization-related markers revealed that GO-EDM-DTX-Vad upregulated M1-associated markers (CD86 and iNOS) and downregulated M2-associated markers (CD206 and IL-10) (Fig. 4C).

Fig. 4.

Macrophage-targeted GO-EDM-DTX-Vad induces M2-to-M1 repolarization and limits TNBC cell proliferation and migration. (A) After incubation of M2-polarized macrophages with the nanocomposites for 24 h, CD206 and CD86 were quantified by flow cytometry. (B) After co-treatment of M0 RAW264.7 cells with IL-4/IL-13 and the nanocomposites for 48 h, CD206 and CD86 were quantified by flow cytometry. (C) RT-qPCR of CD86, iNOS, CD206 and IL-10 expression. (D) RAW264.7 immunoblots of STING axis including TBK1, phospho-TBK1, IRF3 and phospho-IRF3, with GAPDH as a reference. (E) CLSM images of RAW264.7 uptake of RhoB–GO-EDM-DTX-Vad (nuclei, blue; nanocomposites, red; scale bar, 100 μm). (F) ELISA of TGF-β, IL-10 and TNF-α in culture supernatants. (G) Schematic of the in vitro workflow for generating macrophage-conditioned media and assessing CM-driven tumor cell functions. (H, I) Colony formation (H) and EdU flow-cytometry analyses (I) of 4T1 and MDA-MB-231 cells after treatment with the indicated macrophage-CM. Groups: G1, PBS; G2, GO-EDM; G3, GO-EDM-DTX; G4, GO-EDM-Vad; G5, GO-EDM-DTX-Vad; G6, GO-EDM-DTX-Vad + NIR. Data are mean ± SD (n = 3). ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001; ns, not significant.

We next investigated whether this phenotypic modulation was associated with activation of the canonical STING signaling pathway. As shown in Fig. 4D and Fig. S13, the total protein levels of TBK1 and IRF3 remained unchanged. In contrast, the nanocomposite loaded with the STING agonist vadimezan markedly increased the phosphorylation levels of TBK1 and IRF3. These results indicate that the Vad-loaded nanocomposite effectively activated the STING signaling pathway, thereby contributing to the repolarization of macrophages from the M2 phenotype.

To further evaluate the functional consequences of macrophage repolarization, cytokine secretion in the culture supernatants was analyzed by ELISA. GO-EDM-DTX-Vad treatment significantly reduced the levels of immunosuppressive cytokines (TGF-β and IL-10) while markedly increasing the secretion of TNF-α (Fig. 4F), further confirming the induction of a pro-inflammatory macrophage phenotype.

Beyond repolarizing pre-established M2 macrophages, nanocomposite treatment during IL-4/IL-13–driven polarization of M0 macrophages suppressed de novo M2 polarization and promoted an M1-like phenotype. Together, these data indicate that the nanocomposites both reprogram established M2-like macrophages and intercept M2 skewing during induction (Fig. 4B and Fig. S12) [36,37]. This dual activity may help maintain a more immunostimulatory tumor milieu by re-educating suppressive macrophages and limiting ongoing M2 replenishment.

GO-EDM-DTX-Vad treatment enhanced the phagocytic capacity of macrophages that had been repolarized toward an M1-like phenotype. In addition, these M1-like macrophages exhibited an improved ability to activate T cells, as evidenced by a significantly higher proportion of CD8⁺CD69⁺ T cells following co-culture. This enhanced T-cell activation is likely attributable to the increased phagocytic activity and antigen-presentation capability of the repolarized macrophages. Collectively, these findings indicate that GO-EDM-DTX-Vad effectively repolarizes M2 macrophages toward an M1-like phenotype with enhanced phagocytic and T-cell-activating functions (Fig. S14).

The flow cytometry and confocal imaging assessed uptake of RhoB–GO-EDM-DTX-Vad in macrophages with distinct polarization states. The nanocomposite exhibited markedly higher uptake in M2 macrophages than in M0 cells. The enhanced uptake was competitively inhibited by free mannose, supporting MR (CD206)–mediated endocytosis as a major uptake route. (Fig. 4E and Fig. S15).

We also observed the uptake of the nanocomposite by M1 macrophages. Our results futher showed that treatment of LPS-induced M1 macrophages with GO-EDM-DTX-Vad increased CD86 expression and TNF-α secretion, indicating an additional enhancement of the pro-inflammatory phenotype (Fig. S16). Collectively, these findings suggest that the nanocomposite targets M2 macrophages through MR-mediated uptake while also being efficiently internalized by M1 macrophages due to their inherently high phagocytic activity, thereby potentially exerting broader immunomodulatory effects across macrophage subsets.

3.9. In vitro attenuation of M2-like macrophage–driven tumor cell proliferation, migration and invasion by the nanocomposite

Previous studies have shown that M2-type tumor-associated macrophages (M2-TAMs) secrete cytokines and chemokines that promote tumor cell proliferation, migration and invasion, thereby driving tumor progression and contributing to immunosuppression [38,39]. To determine whether the nanocomposite could reverse M2 macrophage–driven protumor activity, we performed tumor-cell functional assays using macrophage-conditioned medium (CM) (Fig. 4G). The results showed that M2-CM increased clonogenicity, wound closure, Transwell invasion, migration and the EdU-positive fraction (Fig. 4H, I and Fig. S17) in 4T1 and MDA-MB-231 cells. In contrast, CM from M2 macrophages treated with GO-EDM-DTX-Vad, particularly when combined with 808 nm NIR irradiation, substantially attenuated these protumor effects. These findings indicate that GO-EDM-DTX-Vad reprograms the secretory output of M2-like macrophages and indirectly suppresses tumor cell proliferation, migration, and invasion.

3.10. In vitro evaluation of the chemo-photothermal antitumor activity of the nanocomposite

To assess in vitro antitumor activity, we evaluated the cell death and apoptosis using Calcein-AM/PI staining and Annexin V/PI flow cytometry. GO-EDM carrier did not significantly affect the viability or apoptosis in 4T1 or MDA-MB-231 cells, indicating a well cytocompatibility within the tested dose range. In contrast, the GO-EDM + NIR group induced moderate cell death due to photothermal damage. The GO-EDM-DTX group significantly increased apoptotic cell populations, confirming the direct cytotoxic role of docetaxe, whereas the GO-EDM-Vad group showed minimal apoptosis, consistent with the limited direct cytotoxicity of vadimezan. GO-EDM-DTX-Vad nanocomposite reduced the viable fraction and increased early and late apoptotic populations. Upon 808 nm NIR laser irradiation, nanocomposite-induced cell death and apoptosis were further enhanced, suggesting a synergistic interaction between photothermal effects and chemotherapy (Fig. 5A–D).

Fig. 5.

GO-EDM-DTX-Vad combined with NIR induces TNBC cell death, apoptosis and immunogenic cell death. (A, B) Calcein-AM/PI live/dead staining of 4T1 (A) and MDA-MB-231 (B) cells (live, green; dead, red; scale bar, 200 μm). (C, D) Annexin V/PI apoptosis plots for 4T1 (C) and MDA-MB-231 (D) cells. (E, F) Quantification of viable and early/late apoptotic 4T1 cells. (H, I) Quantification of viable and early/late apoptotic MDA-MB-231 cells. (G, J) Dose–response viability curves after 48 h treatment with free DTX or GO-EDM-DTX-Vad in 4T1 (G) and MDA-MB-231 (J). (K, L) Immunofluorescence of CRT exposure and HMGB1 release after 24 h treatment in 4T1 (K) and MDA-MB-231 (L) cells (nuclei, blue; CRT/HMGB1, green; scale bar, 100 μm). (M) Schematic of the ICD-to-DC maturation assay. (N–P) Frequencies of CD40⁺, CD80⁺ and CD86⁺ DCs after co-culture. Data are mean ± SD (n = 3). ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001; ns, not significant.

To quantitatively evaluate the enhancement of drug efficacy by nanodelivery, half-maximal inhibitory concentrations (IC₅₀) were determined by nonlinear regression of the dose–response curves. In 4T1 cells, the IC₅₀ decreased from 16.07 ng/mL (free DTX) to 7.276 ng/mL (GO-EDM-DTX-Vad). In MDA-MB-231 cells, the IC₅₀ decreased from 20.52 to 10.19 ng/mL, indicating that comparable cytotoxicity was achieved with 50% lower DTX exposure (Fig. 5G and 5J).

These results indicate that the GO-EDM nanodelivery system enhances the cytotoxic activity of DTX against breast cancer cells without detectable carrier-related toxicity, and that NIR-triggered photothermal irradiation further potentiates its antitumor effects.

3.11. In vitro evaluation of nanocomposite-triggered immunogenic cell death in tumor cells

We next evaluated whether the nanocomposite induces immunogenic cell death (ICD) in tumor cells and examined the functional consequence on dendritic cell (DC) maturation. ICD is classically characterized by surface exposure of calreticulin (CRT), nuclear-to-extracellular release of HMGB1 [40]. These ICD-associated danger signals can promote antigen uptake and maturation of DCs, thereby facilitating the initiation of adaptive immune responses.

Compared with the control and GO-EDM groups, GO-EDM-DTX-Vad increased cell-surface CRT exposure and decreased nuclear HMGB1 retention. Upon 808 nm NIR irradiation, these ICD hallmarks were further enhanced, consistent with amplification of the ICD response (Fig. 5K, L and Fig. S18, S19).

Building on these findings, we used a Transwell co-culture system to test whether soluble factors released from treated tumor cells promote DC maturation (Fig. 5M) [41]. Flow cytometry showed that DCs co-cultured with tumor cells pretreated with GO-EDM-DTX-Vad upregulated CD40, CD80 and CD86 (Fig. 5N–P). These results indicate that ICD-associated DAMPs and soluble mediators released from treated tumor cells are sufficient to drive DC maturation in a contact-independent manner.

3.12. Inhibition of subcutaneous tumor growth via intratumoral injection of the nanocomposite

To evaluate the in vivo antitumor efficacy of the nanocomposite formulations, we established a 4T1 subcutaneous tumor model in BALB/c mice and administered the formulations via intratumoral injection (Fig. 6A). As shown in Fig. 6B,C and Fig. S21, tumors in both the control and GO-EDM groups exhibited rapid and continuous growth throughout the observation period, with no significant difference between the two groups. In contrast, both GO-EDM-DTX and GO-EDM-Vad treatments delayed tumor growth to varying degrees, indicating that local chemotherapy or STING activation alone confers only a limited antitumor effect. Notably, in the GO-EDM-DTX-Vad group, tumor progression was further suppressed, and tumor volumes remained consistently lower throughout the treatment period. The most pronounced tumor growth inhibition was observed when GO-EDM-DTX-Vad was combined with 808 nm NIR laser irradiation (Fig. 6D–F).

Fig. 6.

GO-EDM-DTX-Vad suppresses primary tumor growth and lung metastasis after intratumoral or intravenous administration. (A) 4T1 subcutaneous tumor model and intratumoral dosing schedule. (B) Body-weight changes. (C) Tumor growth curves. (D) Representative excised tumors. (E) Infrared thermal images during NIR irradiation 30 min after intratumoral injection of PBS or GO-EDM-DTX-Vad. (F) Tumor weights. (G) 4T1 primary tumor–lung metastasis model and tail-vein dosing schedule. (H) Body-weight changes. (I) Primary tumor growth curves. (J) Representative excised tumors. (K) Infrared thermal images during NIR irradiation 8 h after intravenous injection of PBS or GO-EDM-DTX-Vad. (L) Lung surface metastatic nodule counts. (M) Representative lungs images (arrows, metastatic nodules). (N) H&E-stained lung sections (10 × , 20 × ). Data are mean ± SD (n = 5). ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001; ns, not significant.

We next monitored temporal changes in tumor temperature following intratumoral injection. In the PBS-injected control group subjected to laser irradiation, the local tumor temperature exhibited minimal increase, indicating that laser exposure alone at this power density was insufficient to induce effective photothermal damage. By contrast, after intratumoral injection of GO-EDM-DTX-Vad, irradiation under the same conditions resulted in a rapid temperature increase at the tumor site. The temperature reached approximately 43 °C within a few minutes and remained within the mild hyperthermia range of 40–45 °C (Fig. 6E and Fig. S22). Previous studies have suggested that this temperature window is sufficient to compromise tumor cell viability and enhance chemosensitivity, while posing a relatively low risk of thermal damage to surrounding normal tissues [42,43]. During treatment, only mild congestion and transient erythema were observed in the skin overlying the tumor, with no obvious burns or ulceration detected.

3.13. Intravenous nanocomposite therapy inhibits primary tumor growth

Given the highly invasive and strong propensity for distant metastasis of TNBC, we established a 4T1 subcutaneous tumor model with lung metastases in mice. GO-EDM-DTX-Vad was administered via tail-vein injection to evaluate its systemic antitumor activity (Fig. 6G). During tumor progression, mice in the control and GO-EDM carrier groups showed a marked decrease in body weight. In contrast, body weight remained largely stable in the free DTX + Vad, GO-EDM-DTX-Vad and GO-EDM-DTX-Vad + NIR treatment groups (Fig. 6H).

For primary tumor growth, treatment with free DTX + Vad induced only a modest delay in tumor progression compared with the control group. This limited efficacy is likely attributable to rapid systemic clearance and poor intratumoral accumulation of the free STING agonist following systemic administration. By contrast, GO-EDM-DTX-Vad significantly inhibited primary tumor growth. The tumor growth rate was reduced and final tumor weights were decreased, indicating that nanodelivery increased intratumoral drug exposure and improved therapeutic efficacy at the tumor site (Fig. 6I, J and Fig. S23, S24).The H&E staining revealed the GO-EDM-DTX-Vad group, particularly when combined with photothermal therapy, exhibited pronounced tumor cell reduction and extensive necrosis, indicating superior antitumor efficacy consistent with the tumor growth inhibition results (Fig. S25). Accordingly, the GO-EDM-DTX-Vad + NIR group exhibited the strongest tumor suppression, supporting an enhanced therapeutic benefit from combining chemotherapy, STING-mediated immune activation and photothermal therapy.

Previous studies have suggested that the optimal timing for laser irradiation in photothermal therapy should correspond to the time point at which nanomaterials achieve maximal accumulation at the tumor site [44]. In our study, the fluorescence intensity at the tumor site reached its maximum 8 h after intravenous injection, indicating the highest level of nanocomposite accumulation at this time point. Therefore, to assess the in vivo photothermal performance under systemic delivery conditions, mice received a tail vein injection of GO-EDM-DTX-Vad followed by 808 nm NIR laser irradiation 8 h later. Infrared thermal imaging revealed that the tumor temperature in the GO-EDM-DTX-Vad group increased to approximately 41–42 °C, whereas only a slight temperature elevation was observed in the PBS group (Fig. 6K and Fig. S26). These results demonstrate that the nanocomposite can effectively accumulate at the tumor site after intravenous administration and generate a mild yet therapeutically relevant photothermal effect under NIR irradiation.

3.14. Intravenous administration of the nanocomposite reduces pulmonary metastatic burden

As shown in Fig. 6L and M, numerous metastatic nodules were observed on the lung surfaces of mice in both the PBS and GO-EDM groups. Treatment with free DTX + Vad modestly reduced nodule number, whereas GO-EDM-DTX-Vad substantially suppressed pulmonary metastatic foci. When combined with NIR irradiation, only a few small and scattered tiny nodules remained visible.

Histopathological examination of lung tissue sections by H&E staining further corroborated these findings (Fig. 6N) [45,46]. In the control and GO-EDM groups, the lungs contained numerous large metastatic nodules, some of which had coalesced to occupy much of the lung parenchyma, thereby severely disrupting the normal alveolar architecture. Free DTX + Vad decreased both the number and size of metastatic lesions, but a substantial burden of medium-sized tumor foci remained. In contrast, the GO-EDM-DTX-Vad group exhibited a pronounced reduction in intrapulmonary tumor nodules, which were mostly small and sparsely distributed. In the GO-EDM-DTX-Vad + NIR group, only a few small residual metastatic foci were detected, and most microscopic fields displayed largely preserved alveolar architecture. These results indicate that GO-EDM-DTX-Vad can significantly suppress breast cancer lung metastasis and mitigate damage to normal lung parenchyma.

Considering that STING agonists may potentially induce systemic inflammatory responses, histological analysis of major organs was further performed. H&E staining revealed that the heart, liver, spleen and kidney tissues in all treatment groups maintained normal architecture without obvious inflammatory infiltration, tissue necrosis or structural abnormalities (Fig. S27). In addition, serum biochemical parameters, including alanine aminotransferase (ALT), aspartate aminotransferase (AST), total bilirubin (TBIL), blood urea nitrogen (BUN) and creatinine (CREA), remained within normal physiological ranges, indicating no evident hepatic or renal dysfunction. These results suggest that GO-EDM-DTX-Vad did not induce noticeable systemic toxicity or systemic inflammation at the administered dose (Fig. S28).

3.15. Nanocomposite-mediated remodeling of systemic and intratumoral immune landscapes

As shown in Fig. 7A and B, mice in the control and GO-EDM groups displayed a typical immunosuppressive profile in both the spleen and tumor tissues. The proportions of M2-like macrophages, MDSCs and Tregs remained elevated.

Fig. 7.

GO-EDM-DTX-Vad remodels systemic and intratumoral immune landscapes. (A) Flow-cytometric quantification of splenic T-cell subsets (CD4⁺, CD8⁺, CD4⁺ TCM, CD8⁺ TCM), NK cells and Tregs. (B) Flow-cytometric quantification of tumor-infiltrating immune cells, including MDSCs, Tregs, M1-/M2-like macrophages and activated DCs (CD11c⁺CD80⁺ and CD11c⁺CD86⁺). (C) Representative immunofluorescence staining of tumor sections for immune infiltration and proliferation (CD8, CD11c, F4/80+iNOS, F4/80+CD206, Gr-1 and Ki-67; nuclei, DAPI; scale bar, 100 μm). Data are mean ± SD (n = 5). ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001; ns, not significant.

Treatment with free DTX + Vad partially alleviated this immunosuppressive state, whereas GO-EDM-DTX-Vad induced a much more pronounced and comprehensive immune reprogramming. Compared with the free drug group, the nanocomposite more effectively reduced the frequencies of M2 macrophages, MDSCs and Tregs in both the spleen and tumors, while increasing CD8⁺ T cells infiltration and expanding CD4⁺ TCM and CD8⁺ TCM populations [47]. When combined with NIR irradiation, these immunoregulatory effects were further strengthened. Serum cytokine profiling by ELISA further supported systemic immune activation, revealing elevated IFN-γ and TNF-α levels and reduced IL-10 levels after nanocomposite treatment (Fig. S29).

In summary, these results indicate that GO-EDM-DTX-Vad, particularly when combined with photothermal therapy, can reverse immunosuppression at both systemic and intratumoral levels. This treatment reshaped the immune landscape toward an effector- and memory T-cell–enriched state, which is favorable for mounting robust antitumor immune responses [48,49].

3.16. Multiplex immunofluorescence reveals nanocomposite-driven reprogramming of the tumor microenvironment

To evaluate therapeutic outcomes at the tissue level, we performed multiplex immunofluorescence of tumor sections from each treatment group. As shown in Fig. 7C, in the control group, tumors exhibited features of aggressive growth and profound immunosuppression. Ki-67 staining was diffuse and strongly positive, indicating high proliferative activity. In contrast, TUNEL⁺ cells were rare, consistent with low apoptotic activity (Fig. S24). Analysis of immune infiltration further revealed sparse cytotoxic T cells, antigen-presenting cells and M1-like macrophages, alongside marked accumulation of immunosuppressive myeloid cells. Together, these features defined a tumor microenvironment characterized by robust proliferation, limited apoptosis and pronounced immunosuppression [4,50].

Compared with the control group, intravenous administration of the nanocomposite partially mitigated these adverse pathological features. At the immune level, tumor tissues from nanocomposite-treated mice exhibited increased infiltration of CD8⁺ T cells, dendritic cells and M1-like macrophages, together with a marked decrease in M2-like macrophages and MDSCs. These changes are consistent with alleviated immunosuppression and enhanced effector immune engagement [51,52].

In summary, these histopathological findings indicate that GO-EDM-DTX-Vad, particularly under photothermal co-treatment, not only suppresses tumor proliferation and promotes apoptosis but also extensively remodels the tumor immune microenvironment. The nanocomposite shifts an immunologically “cold” tumor toward a “hot” phenotype, thereby providing a structural and cellular foundation for durable antitumor immunity.

In our study, PTT was integrated with chemotherapy and STING pathway activation, forming a multimodal therapeutic strategy. NIR-triggered photothermal heating synergistically enhanced the cytotoxic effect of docetaxel, promoted tumor antigen release and amplified STING-mediated immune activation. This combinational mechanism can effectively reduce residual tumor cells and strengthen systemic antitumor immune responses, thereby partially overcoming the intrinsic limitations associated with photothermal monotherapy. Such a design may improve the clinical applicability of NIR-triggered nanotherapeutic systems, particularly by leveraging synergistic therapeutic mechanisms rather than depending exclusively on direct thermal destruction. Nevertheless, for the treatment of deep-seated solid tumors or disseminated metastatic lesions, external NIR-I irradiation alone may not always achieve sufficient energy deposition. Future studies may therefore explore photothermal agents responsive to the second near-infrared window (NIR-II, 1000–1700 nm) or combine fiber-optic and endoscopic light delivery strategies to improve clinical applicability.

In addition, although this study observed a significant increase in splenic CD4⁺ and CD8⁺ central memory T cells, accompanied by elevated levels of effector cytokines such as IFN-γ and TNF-α. It should be noted that expansion of memory-phenotype T cells alone does not constitute definitive evidence of long-term protective immunity. Tumor rechallenge experiments and long-term survival analyses are generally considered essential functional assays for confirming durable antitumor immune memory. The memory-related immune responses reported here will require further validation through long-term functional studies in future investigations.

4. Conclusion

In this study, we developed a mannose-functionalized graphene oxide nanocarrier and co-loaded docetaxel and the STING agonist vadimezan to generate GO-EDM-DTX-Vad. Multimodal characterization confirmed successful mannosylation, stable drug association and favorable nanoscale properties. The nanocomposite exhibited pH and NIR responsive drug release and robust photothermal performance with high cycling stability. Mannose receptor–facilitated uptake promoted dendritic cell maturation, and the formulation both repolarized pre-established M2-like macrophages and suppressed de novo M2 programming, thereby reducing pro-tumor paracrine support. In tumor cells, chemo–photothermal treatment enhanced apoptosis and induced immunogenic cell death, thereby promoting downstream antigen presentation and T-cell activation. In vivo, GO-EDM-DTX-Vad suppressed primary tumor growth and markedly reduced lung metastasis, with the strongest efficacy observed under NIR irradiation. In parallel, it remodeled systemic and intratumoral immunity by decreasing immunosuppressive subsets and enriching effector and memory T cells. Overall, GO-EDM-DTX-Vad integrates chemotherapy, STING activation and photothermal therapy to achieve coordinated tumor control and immune remodeling.

CRediT authorship contribution statement

Zang Xin: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing – original draft. Liu Dezhi: Conceptualization, Data curation, Formal analysis. Zhang Jingyi: Conceptualization, Investigation, Methodology. Chao Xuyuan: Data curation, Methodology, Resources. Chen Xin: Data curation, Investigation, Methodology. Yang Shengjie: Resources, Software. Fang Yingying: Supervision, Validation. Li Yan: Validation, Visualization. Li Yinjuan: Visualization, Writing – original draft. Qi Lu: Project administration, Supervision. Wang Xinghe: Conceptualization, Funding acquisition, Project administration, Supervision, Writing – review & editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article:

Data availability

No data was used for the research described in the article.