Engineering an AIEgen-based platform integrating CRISPR/Cas9 to remodel the tumor microenvironment and reinforce photo-immunotherapy against glioblastom

Abstract

Glioblastoma remains one of the most lethal brain tumors. Although immunotherapy and other therapeutic modalities has achieved significant therapeutic success in several malignancies, its efficacy in glioblastoma remains limited primarily due to the complex tumor microenvironment (TME) and physiological barriers such as the blood–brain barrier (BBB). In this context, nanomedicine and gene editing have emerged as promising strategies due to their unique ability to cross the BBB and protect therapeutic agents through intrinsic physicochemical properties. To overcome the physiological barriers for better therapeutic outcomes. Here, a novel aggregation-induced emission luminogen (AIEgen), NDA-DPE, was synthesized, exhibiting NIR-I to NIR-II fluorescence and dual photothermal (PTT) and photodynamic (PDT) properties through restricted intramolecular motion. Bone-derived neutrophil–based biomimetic nanoparticles (bNe@AIE/Cas9-CD73) were then prepared by integrating NDA-DPE with CRISPR/Cas9-mediated CD73 gene silencing. The neutrophil encapsulation enabled efficient BBB penetration and targeted accumulation in glioblastoma tissue. CRISPR/Cas9-CD73 downregulated CD73 expression, disrupted the ATP–adenosine axis, and reshped the immunosuppressive TME into an immuno-supportive one, increasing the therapeutic sensitivity of tumor cells. Under NIR-II excitation, bNe@AIE/Cas9-CD73 achieved fluorescence-guided PTT and PDT, inducing immunogenic cell death (ICD), stimulating immune-cell recruitment, and activating systemic antitumor immunity. bNe@AIE/Cas9-CD73 demonstrated a potent gene–photothermal–photodynamic–immune synergistic effect, significantly inhibiting glioblastoma growth and establishing a promising nanoplatform for effective and targeted glioblastoma treatment.

Introduction

Glioblastoma is the most prevalent and lethal malignant brain tumor in humans, characterized by significant heterogeneity, invasiveness, and aggressiveness, with a current median overall survival (OS) of less than 15 months.^1,2 Over recent decades, extensive efforts have been directed toward developing effective therapeutic modalities, including surgery, chemotherapy, radiotherapy, and tumor-targeted therapies. Despite significant progresses in these approaches, glioblastoma remains the most treatment-resistant malignancy of the central nervous system (CNS), primarily due to its immunosuppressive tumor microenvironment (TME)³ and physiological barriers such as the blood–brain barrier (BBB) and blood–brain tumor barrier (BBTB). While the BBB protects the brain from toxins, drugs, and pathogens, it also restricts the penetration of therapeutic agents.⁴ Under these conditions, developing more effective and safer therapeutic strategies is both urgent and imperative.

Immunotherapy has gained significant attention as a promising novel therapeutic strategy to improve glioblastoma management,⁵ encompassing modalities such as immune checkpoint blockade (ICBs), vaccines, oncolytic viruses, and CAR-T cell therapy. These approaches aim to activate, recruit, and mature immune responses against glioblastoma cells to inhibit or limit tumor growth. However, due to the unique immune-privileged nature of the CNS, clinical outcomes have remained limited in the past fundamental and clinical trials. The brain’s inherently quiescent immune state, combined with intrinsic and extrinsic resistance, immunosuppression, myeloid cell activity, and adaptive resistance mechanisms, results in more than 50% of patients deriving little benefit from immunotherapy.⁶ Current strategies to convert “cold” tumors into “hot” tumors have shown promise for improving the efficacy of glioblastoma immunotherapy.^7,8 For example, engineered bacterial surface platforms and bacteria-based photothermal therapy (PTT) can stimulate a positive feedback loop among cancer cells, M1 macrophages, and T cells, remodeling the immunosuppressive TME and improving the sensitivity of glioblastoma to anti-PD-1 therapy.⁹ However, despite these advances, none of the immunotherapy-based strategies have achieved significant improvements in progression-free (PFS) or OS in glioblastoma patients, underscoring the ongoing need for more efficient therapeutic approaches.

In recent years, the rapid emergence and development of gene editing technologies, particularly clustered regularly interspaced short palindromic repeats and associated proteins (CRISPR/Cas), have attracted considerable attention as a promising strategy to improve immunotherapy against glioblastoma through precise genetic manipulation to convert “cold” tumors into “hot” tumors, improving therapeutic efficacy.^10,11 For example, a CRISPR/Cas-based platform, oncolytic virus–T cell chimera (ONCOTECH), was established in which a CRISPR/Cas9-engineered oncolytic adenovirus was coated with a biomembrane expressing a specific antigen to bind T cells via their surface receptors. Following modulation of PD-L1 expression, this approach reshaped the TME, simultaneously enhancing the effects of oncolytic virus therapy and adoptive T-cell therapy in solid tumors.¹² Among the therapeutic targets, CD73, a hypoxia-induced ectonucleotidase (ecto-5′-nucleotidase) encoded by the NT5E gene, has recently gained much attention as an immunotherapy target due to its association with both an immunosuppressive TME and glioblastoma malignancy. Its expression can be induced by aberrant signaling of transforming growth factor-β (TGF-β), interferons (IFNs), tumor necrosis factor (TNF), interleukin-1β (IL-1β), and prostaglandin E₂. Under physiological conditions, intracellular ATP and ADP released into the extracellular environment are hydrolyzed by CD39 to generate AMP, which is then converted by CD73 into adenosine and phosphate. Adenosine and ATP then bind to G protein-coupled P1 receptors (A1, A2A, A2B, A3) and P2 receptors (P2X, P2Y), promoting an immunosuppressive TME formation by reducing immune cell infiltration, inducing angiogenesis, and facilitating tumor progression, metastasis, and treatment resistance.¹³ Moreover, T-cell proliferation is suppressed that regulated by extracellular vesicles with CD73 expression in glioblastoma tissues, further contributing to immune evasion.¹⁴

Accordingly, CD73 is regarded as a critical combinatorial and immunological target for glioblastoma immunotherapy.^15–18 Downregulation of CD73 in macrophages following anti-CTLA-4 and anti-PD-1 treatment has been shown to significantly extend the survival of glioblastoma-bearing mice.¹⁶ Similarly, CAR-NK-CD73 cells inhibit adenylate production, promote NK-cell infiltration, and inhibit tumor growth.¹⁹ CD73 also represents an effective therapeutic target for combination regimens with PD-1 and CTLA-4 blockade across multiple tumor types.^20–22 It has been reported that the inhibition of ATP metabolism through the CD73 inhibitor AB680 disrupts the ATP–adenosine axis, activating immune responses and significantly inhibiting tumor proliferation.²³ In parallel, CD73 inhibition induces an inflammatory TME that favors macrophage maturation,²⁴ and nano-enzyme-mediated enhancement of the Fenton reaction to suppress CD73 expression alleviates hypoxia and achieves synergistic antitumor activity with radiotherapy.²⁵ Furthermore, small-molecule inhibitors, such as 35j and XC-12, have shown therapeutic potential,^26,27 and more than 20 CD73-targeted agents, including Mupadolimab, Oleclumab, and Quemliclustat, are currently under clinical investigation (clinicaltrials.gov). However, the potential side effect of small molecule drugs bring the uncentainty for broader application. Thus, the precise application of CRISPR/Cas9 technology targeting CD73 holds potential to improve the efficacy of glioblastoma immunotherapy, and it is eager to establish a new and effective strategy that target CD73 for better therapeutic outcomes.

The rapid progresses of nanomedicine has established a milestone in cancer immunotherapy by transforming “cold” tumors into “hot” ones, remodeling the immunosuppressive microenvironment through multiple pathways, and promoting a robust T-cell response.²⁸ Moreover, by optimizing morphology, surface topology, and chemical composition, engineered nanomaterials can traverse the BBB and BBTB, enabling efficient brain delivery for synergistic therapeutic effects.^29–31 Nanomedicine provides an alternative approach for noninvasive therapy with reduced systemic toxicity and improved immunotherapeutic efficacy compared with conventional modalities, including phototherapies such as PTT, sonodynamic therapy (SDT), and photodynamic therapy (PDT), and so on.^32–34 Considering the requirements for precise, localized treatment with high spatial selectivity and minimal side effects, phototherapy, particularly PTT and PDT, offers a promising approach for glioblastoma management by inducing hyperthermia and generating cytotoxic reactive oxygen species (ROS).^35,36 However, most existing nanoagents can not simultaneously serve as deep-tissue-excitable photosensitizers (PSs) and photothermal agents (PTAs) while providing effective imaging, limiting the therapeutic depth and precision of phototherapy. Among available nanomaterials, aggregation-induced emission luminogens (AIEgens) have emerged as ideal candidates for bioimaging-guided phototherapy. Unlike conventional fluorophores, which suffer from aggregation-caused quenching (ACQ) due to planar and extended π-conjugated structures and hydrophobicity, AIEgens exhibit enhanced luminescence upon aggregation, offering significant potential for glioblastoma and other brain disorders.^37–39 Although AIEgens have been successfully applied in glioblastoma therapy,^40–43 strategies that employ longer-wavelength near-infrared (NIR) excitation and multimodal AIEgen-based phototherapy are needed to further improve therapeutic efficacy.

Given the intrinsic advantages and limitations of both gene therapy and PTT/PDT, their combination is expected to yield better tumor eradication and synergistic therapeutic effects.⁴⁴ The integration of PTT with gene therapy can improve treatment outcomes through several mechanisms: (i) increasing cell membrane fluidity and permeability to facilitate nanocarrier endocytosis; (ii) enabling controlled release of genetic payloads at spatially and temporally defined sites; (iii) activating transcription via heat-sensitive promoters; (iv) regulating thermoresistance-related protein expression; and (v) remodeling the TME to favor PTT⁴⁵ and PDT.⁴⁶ Furthermore, the combination of gene therapy and PDT can achieve cascaded and sequential tumor suppression.⁴⁷ These findings suggest that integrating gene therapy with PTT and PDT represents a promising multimodal therapeutic strategy; however, no studies have yet explored this approach in glioblastoma.

Herein, a gene-photo-immunotherapy theranostic nanoplatform, termed bNe@AIE/Cas9-CD73, was designed by integrating novel AIEgen (NDA-DPE) as both PTA and PS with CRISPR/Cas9-CD73 (Scheme 1). Bone-derived neutrophils conferred the ability of bNe@AIE/Cas9-CD73 to cross the BBB and target glioblastoma tissue with high efficiency and specificity. CRISPR/Cas9-CD73 downregulated CD73 expression, inhibiting the conversion of ATP into adenosine and phosphate and reshaping the TME, improving therapeutic sensitivity. Upon 808 nm laser irradiation, the NIR-II fluorescence of NDA-DPE enabled tumor localization while simultaneously inducing PTT and PDT through hyperthermia and ROS generation, leading to immunogenic cell death (ICD) and immune activation against orthotopic glioblastoma. Both in vitro and in vivo studies confirmed that bNe@AIE/Cas9-CD73 produced strong gene–photothermal–photodynamic–immune synergistic effects, indicating that combining gene therapy with PTT and PDT represents a promising therapeutic approach with significant potential for advancing glioblastoma treatment.

Scheme 1.

Illustration of NDA-DPE-based NIR-II FLI imaging-guided PTT- and PDT-immunotherapy reinforced by nanoCRISPR of CD73 manipulation

Results

Synthesis and characterization of AIEgen

Conventional multimodal fluorescent molecules have been widely used in biological and medical sciences, including neurobiology, due to their high molar extinction coefficients (ε) that enable both imaging and therapeutic applications. However, their practical use is often limited by several drawbacks, including short emission wavelengths in the NIR-I region, ACQ, low photoluminescent quantum yield (Φ_PL) and complicated designation, etc., and complex structural design, which collectively limit their biomedical potential. Twisted aggregation-induced emission (AIE) molecules incorporating donor–acceptor interactions and π-conjugated (D–π–A) backbones can significantly increase Φ_PL and redshift emission into the NIR-II region,⁴⁸ while exhibiting desirable photothermal⁴⁹ and photodynamic⁵⁰ properties.

Following the synthetic route illustrated in Supplementary Fig. 1, a novel AIEgen, NDA-DPE (C₈₈H₁₀₄N₄O₄S₄, MW = 1408.694), was obtained with a yield of 42%. NDA-DPE possesses a typical D–π–A configuration. As shown in Fig. 1a, a naphthalenediimide-fused 2-(1,3-dithiol-2-ylidene)acetonitrile (NDA) moiety with large π-conjugation and strong electron-withdrawing capacity served as the acceptor. Based on previous research using tetraphenylethylene (TPE) as a donor, NDA-TPE showed strong photothermal activity, driven by efficient non-radiative decay enabled by active molecular motion in the solid state.⁵¹ Although this property favored PTT, incorporating a simultaneous photodynamic function could offer more therapeutic advantages. To enhance radiative decay within the D–π–A system, two [(E)-2-phenylethenyl]benzene units were conjugated symmetrically on both sides of the NDA core via Suzuki coupling, replacing TPE to yield NDA-DPE. The product was confirmed by ¹H NMR and ¹³C NMR and MALDI-TOF MS analyses with satisfactory results (Supplementary Figs. 2-4).

Fig. 1.

Theoretical simulation and photophysical property. a Molecular structure of NDA-DPE. b View of dihedral angle of NDA-DPE. c Illustration of HOMO/LUMO and the corresponding energy gap of NDA-DPE. d Nanofabrication of AIE dots and its application in NIR-III FLI imaging-guided cancer treatment with phototherapy. e UV-Vis spectra of NDA-DPE in THF solution and film. f Fluorescence spectra of NDA-DPE in solid film excited at 750 nm. g Fluorescence photographs of the NDA-DPE with different bandpass filters (LP1000, LP1150, and LP1300 nm). h PL spectra of NDA-DPE in H₂O/THF mixtures with different H₂O fractions (f_w). i Plot of the relative fluorescence emission intensity (I/I₀) versus the composition of the aqueous mixture of NDA-DPE. I₀ is the PL intensity in THF. j Photostability evaluation of NDA-DPE and ICG

To elucidate the underlying mechanism of the synthesized AIEgen, density functional theory (DFT) calculations were performed. As shown in Fig. 1b, the optimized geometry of NDA-DPE exhibited a dihedral angle of 40.4° between the donor and acceptor moieties, indicating efficient molecular conjugation and a favorable environment for strong absorption. Electrostatic potential (ESP) mapping displayed distinct negative regions attributed to cyan and carbonyl groups on the conjugated backbone, suggesting pronounced electron-withdrawing behavior that could restrict excited-state molecular motion via enhanced intermolecular interactions. The lowest unoccupied molecular orbital (LUMO) was localized on the NDA acceptor, while the highest occupied molecular orbital (HOMO) was confined to the stilbene donor unit (Fig. 1c). The clear spatial separation of HOMO and LUMO confirmed the intramolecular charge-transfer (ICT) nature of NDA-DPE.

To evaluate the photophysical characteristics of NDA-DPE, UV–Vis absorption and photoluminescence (PL) spectra were recorded. In tetrahydrofuran (THF) solution, two absorption peaks were observed at 350 nm and 650 nm, with a maximum at 650 nm. Upon aggregation (film state), a substantial redshift (~100 nm) to 750 nm was observed (Fig. 1e). The absorbance at 808 nm was higher than that of indocyanine green (ICG) at the same concentration (Supplementary Fig. 5), indicating better suitability for PTT and PDT. Similarly, fluorescence emission spectra revealed two peaks at 1018 nm (weak) and 1110 nm (strong) under 750 nm excitation (Fig. 1f), with a tail extending to 1400 nm, similar to NDA-TPE.⁵² These findings demonstrated strong redshift and aggregation-state emission, confirming NDA-DPE as an excellent NIR-IIa nanofluorophore for in vivo fluorescence imaging. The NIR-II fluorescence of NDA-DPE was further validated in vitro using band-pass filters (LP1000, LP1150, and LP1300 nm), showing intense emission across all channels (Fig. 1g). The AIE behavior of NDA-DPE was examined by gradually increasing the water fraction (f_w) in water/THF mixtures from 0% to 99%. As shown in Fig. 1h, fluorescence intensity increased sharply (22.4-fold) with higher water content due to restriction of intramolecular motion (RIM) upon aggregation (Fig. 1i). Increasing water content also slightly blue-shifted the emission maximum from 1110 nm to 1018 nm, likely due to reduced solvent polarity. When f_w > 80%, intense AIE emission was observed, confirming aggregation and potential for biological applications. Photostability testing under continuous 808 nm laser irradiation for 30 min revealed stable fluorescence from NDA-DPE, whereas ICG showed a sharp decay, confirming high photostability (Fig. 1j). Finally, NDA-DPE monomers were encapsulated into nanoparticles (“AIE dots”) using amphiphilic DSPE-PEG₁₀₀₀ (Supplementary Fig. 6) via a nanoprecipitation method for subsequent studies (Fig. 1d), the size of prepared AIE dots was average 114.06 nm (PDI = 0.209) (Supplementary Fig. 7), which is capable for biomedical application. These results verified the successful synthesis of NDA-DPE with strong NIR-I to NIR-II emission, robust AIE characteristics, and high photostability, highlighting its potential as a multifunctional agent for bioimaging and phototherapy.

Phototherapy properties evaluation of NDA-DPE in vitro

The combination of PTT and PDT is superior to either therapy alone because it can overcome the inherent limitations of single-modality treatment. PTT and PDT together achieve synergistic therapeutic effects, as PTT increases local oxygen concentration by enhancing blood flow rate, facilitating PDT activity, while PDT sensitizes and eliminates heat-resistant tumor cells that facilitating PTT efficiency.⁵³ In this study, NDA-DPE was evaluated as an ideal NIR-II-guided phototherapeutic candidate.

To assess its photothermal and photodynamic performance, NDA-DPE contained solutions of various concentrations were irradiated with an 808 nm laser at a power density of 0.5 W/cm² for 10 min. The temperature increase showed a clear dose dependence (Supplementary Fig. 8). In a 100 µM NDA-DPE solution, typical power-dependent heating curves reached 88.6 °C at 1.0 W/cm² and 71.6 °C at 0.5 W/cm² (Fig. 2a, b). Both temperature levels are sufficient for effective tumor ablation. Rapid temperature elevation was observed across all tested power intensities, confirming the strong PTT potential of NDA-DPE (Fig. 2b). Given the potential thermal side effects of high-power laser exposure in biological systems, 0.5 W/cm² was selected as the standard laser power for subsequent experiments. NDA-DPE maintained excellent photothermal stability over three consecutive irradiation cycles, whereas the commercial photothermal agent IR780 exhibited significant attenuation (Fig. 2c and Supplementary Fig. 9). This photostability supports its suitability for both in vitro and in vivo applications.

Fig. 2.

PTT and PDT property of NDA-DPE in vitro. a Thermal images of NDA-DPE after NIR laser exposure. b Temperature change curves of NDA-DPE with different NIR laser power intensity. c Temperature change curves of IR780 and NDA-DPE irradiated by 808 nm NIR laser with three on/off cycles. d Fluorescence spectra of DCFH-DA in NDA-DPE contained solution. e The plot of fluorescence intensity at 525 nm of NDA-DPE and PBS solution. f ¹O₂ generation detected by ESR technique. g UV-Vis spectra of DPBF after NDA-DPE irradiated with 808 nm NIR laser at 0.5 W/cm² for different timepoints. h The plotted quantitative results of temperature of mouse brain tissues with different thicknesses (skull, 1, 2, 3, 4, and 6 mm). i Thermal images of NDA-DPE contained solutions sheltered by brain tissues with different thicknesses upon 808 nm laser irradiation

The photodynamic activity of NDA-DPE was then investigated. As shown in Fig. 2d, e, the fluorescence intensity of DCFH-DA increased in NDA-DPE solutions following 808 nm laser irradiation at different time points, indicating efficient ROS generation and strong PDT efficacy. Furthermore, electron spin resonance (ESR) spectra revealed distinct singlet oxygen (¹O₂) signals after laser exposure, confirming the formation of ¹O₂ compared with deionized water and non-irradiated NDA-DPE solution (Fig. 2f). ROS characterization using 1,3-diphenylisobenzofuran (DPBF) and methylene blue (MB) assays further validated these findings. Upon 808 nm irradiation, absorbance at 410 nm for DPBF declined progressively (Fig. 2g and Supplementary Fig. 10a), demonstrating 85.64% degradation within 5 min, consistent with a high ¹O₂ quantum yield. A similar trend was observed in MB degradation assays, indicating hydroxyl radical (•OH) generation (32.52% degradation in 5 min) (Supplementary Fig. 10b, c). However, no superoxide anion formation was detected by nitroblue tetrazolium (NBT) assay (Supplementary Fig. 10d). These findings confirmed that NDA-DPE efficiently produces ¹O₂ and •OH, validating its potent PDT functionality.

As light penetration strongly influences tumor phototherapy, where the incident optical energy drops to ~37% within a few millimetres,⁵⁴ the tissue-penetration properties of NDA-DPE were also examined. C57BL/6 J mouse brain tissues were dissected, sliced into thicknesses of 1, 2, 3, 4, and 6 mm, and placed over quartz cuvettes containing NDA-DPE solution, the uncovered sample served as the 0 mm control. Then, the photothermal capability of deep tissue was further determined under 808 nm laser irradiation (0.5 W/cm²) for 5 min, the temperature changes after penetration through brain tissues of increasing thickness were measured (Fig. 2h, i). The plotted quantitative attenuation showed that although temperature decreased progressively with increasing tissue thickness, the 808 nm laser effectively penetrated several millimeters of brain tissue, producing measurable heating (Fig. 2h). These results demonstrate that NDA-DPE nanoparticles possess high photothermal-to-heat conversion efficiency, strong photostability, excellent photodynamic reactivity, and substantial tissue-penetration capacity, confirming their suitability for in vivo theranostic applications.

Synthesis and characterization of bNe@AIE/Cas9-CD73

Although PTT and PDT have been proved to be effective modalities against tumor. However, the complexity of TME, including oxygen content, pH value, glutathione concentration, the expression of heat-shock protein (HSP) and extracellular matrix (ECM), etc., provides serious hinderance on the therapeutic outcomes of PTT and PDT.⁵⁵ Thereof, TME remodeling before the implementation of PTT and PDT might be a new option for better PTT and PDT. As discussed earlier, CD73 plays a key role in maintaining an immunosuppressive TME and promoting tumor progression.¹³ Notably, NT5E expression was significantly upregulated in glioblastoma compared to other cancers such as adenoid cystic carcinoma and acute myeloid leukemia (Supplementary Fig. 11a). It is significantly overexpressed in glioblastoma tissues (3.67 times over normal tissue, Supplementary Fig. 11b) but shows no significant correlation with OS (Supplementary Fig. 11c) or PFS (Supplementary Fig. 11d) according to GEPIA database analysis (http://gepia.cancer-pku.cn/). Meanwhile, previous studies have demonstrated that CRISPR/Cas9-mediated gene editing provides an effective alternative to improve immunotherapy across various tumors.

In this study, three single-guide RNAs (sgRNAs) (Supplementary Table 1) targeting the NT5E gene was cloned into the px458 plasmid (Supplementary Fig. 12) at the BbsI restriction site. The constructed plasmids were verified by agarose gel electrophoresis and Sanger sequencing (Supplementary Fig. 13a, b). The px458 plasmid carried an EGFP tag coupled to the NT5E gene, allowing direct visualization of transfection efficiency following PEI40000-mediated transfection. Three px458-CRISPR/Cas9-CD73 plasmid candidates were screened and validated by fluorescence microscopy, RT-PCR, and western blotting analysis. Strong fluorescence signals were observed in all three plasmids and the px458 control at 48 h post-transfection (Supplementary Fig. 14a), indicating high transfection efficiency. RT-PCR and western blotting results (Supplementary Fig. 14b, c) confirmed a significant reduction in CD73 mRNA and protein expression. Among the three px458-CRISPR/Cas9-CD73 plasmids, the second construct showed the highest knockdown efficiency and was selected for subsequent experiments.

Inflammation is a hallmark of tumors, facilitating immunosuppression and therapy resistance. As neutrophils display intrinsic chemotaxis toward inflammatory microenvironments and the capability of avoiding immune surveillance, they have been widely investigated as biological carriers for targeted drug and gene delivery, offering improved delivery efficiency and therapeutic precision.⁵⁶ Combined with the need for BBB permeability, neutrophils have been regarded as an effective delivery tool due to their extremely high tropism for the inflammatory environment.^57,58 Based on these properties, bone marrow-derived neutrophils (bNe) were isolated from C57BL/6 J mice via density-gradient centrifugation and employed to construct the bNe@AIE/Cas9-CD73 nanoplatform by grafting neutrophils onto the surface of CRISPR/Cas9-CD73–PEI40000–DNA complexes and AIE dots. The isolated bNe were characterized by Wright–Giemsa staining (Supplementary Fig. 15a) and immunofluorescence labeling of CLM-5 and Ly-6G/Ly-6C (Supplementary Fig. 15b), confirming successful isolation.

Then, AIE-Cas9-CD73 complexes was gained by mixing AIE dots and CRISPR/Cas9-CD73–PEI40000–DNA complexes. To evaluate plasmid condensation, agarose gel retardation assays were performed using different mass ratios (0.5, 1, and 2 µg plasmid mixed with 50 µg AIE dots). The strongest DNA-retarded band was observed at a 2 µg ratio (Supplementary Fig. 16), confirming efficient plasmid encapsulation. The loading efficiency of CRISPR/Cas9 on AIE dots was calculated as 41.60%, indicating a high gene-delivery capacity. The AIE dots/plasmid-PEI40000 complexes were then incubated with neutrophils for 60 min at various mass ratios (5, 10, 20, 50, 100, 150, and 200 µg per 1 × 10⁵ bNe cells), followed by PBS washing and centrifugation to obtain bNe@AIE/Cas9-CD73. The overall loading efficiency was determined to be 54.28% (Supplementary Fig. 17).

Comprehensive characterization of bNe@AIE/Cas9-CD73 was conducted using transmission electron microscopy (TEM), confocal laser scanning microscopy (CLSM), UV–Vis spectroscopy, and fluorescence spectroscopy. TEM imaging revealed spherical morphology of NDA-DPE and clear accumulation of NDA-DPE/plasmid-PEI40000 complexes within the neutrophil cytoplasm (Fig. 3a). CLSM analysis further confirmed the internalization of NDA-DPE/plasmid complexes into neutrophils (Fig. 3b). The UV–Vis and fluorescence spectra of the conjugates corroborated these findings (Fig. 3c, d; Supplementary Fig. 18). Similar to lipid-based transfection, the manner of AIE dots-PEI/Cas9 poly-complexes co-loading have been conducted with the inhibitor of following inhibitors: methyl-β-cyclodextrin (MβCD, inhibitor of lipid raft-mediated endocytosis), sucrose (inhibitor of clathrin-mediated endocytosis), or 5-(N,N-dimethyl)-amiloride hydrochloride (amiloride, inhibitor of macropinocytosis, or low-temperature treatment (4 °C), respectively, and evaluated via measuring the fluorescence intensity of NDA-DPE. As expected, the introduction of endocytosis inhibitors inbihited the increase of the cell fluorescence intensity, suggesting the GL261 cell endocytosis efficiency was blocked (Supplementary Fig. 19). Importantly, the MβCD (27.18%) inhibited the cell endocytosis efficiency over sucrose (67.66%), 5-(N,N-dimethyl)-amiloride hydrochloride (70.26%), or 4 °C (63.46%), suggesting the AIE dots-PEI/Cas9 poly-complexes were co-loaded by neutrophils via lipid raft-mediated endocytosis. Also, with the assistance of the manner of the release of neutrophil extracellular traps (NETs), degranulation, apoptosis and phagocytosis, contact-dependent transmission, as well as the stimulation response release, the cargoes could be released from neutrophils effectively to gain therapeutic outcomes. Under this circumstance, the drug release behaviors from neutrophil were determined under different pH value and inflammatory condition. As depicted in Supplementary Fig. 20, acidic and inflammatory environment facilitated the release of co-loaded cargoes, which enhances the tumor targeting capatibility and avoids side-effect significantly.

Fig. 3.

Construction and characterization of bNe@AIE/Cas9-CD73. a TEM images of NDA-DPE, bNe and bNe@AIE/Cas9-CD73. b CLSM results of bNe@AIE/Cas9-CD73, scale bar: 50 μm. c UV-Vis spectra of NDA-DPE, bNe and bNe@AIE/Cas9-CD73. d Fluorescence spectra of NDA-DPE, bNe and bNe@AIE/Cas9-CD73 excited by 750 nm. e, f Western blotting images and quantitative results of CD73 expression treated with bNe@AIE/Cas9-CD73 in GL261 glioblastoma cells. Statistical significance was calculated by one-way ANOVA with Tukey’s multiple comparisons test, *p < 0.05, **p < 0.01, ***p < 0.001, ns, not significant

To verify gene-delivery efficiency, we first evaluated the potential toxicity of PEI40000 in GL261 glioblastoma cells and isolated bone marrow neutrophils using Cell Counting Kit-8 (CCK-8). As shown in Supplementary Fig. 21, PEI40000 showed good safety, over 80% cells viability was observed when 7.5 mg/mL was incubated with cells for 48 h, which is higher than the applied concentration in this study. Then, the knockout efficiency of NT5E gene manipulation (gene disruption) by bNe@AIE/Cas9-CD73 was determined using DNA electrophoresis combined with PCR, T7 endonuclease I cleavage (T7EI) assay and flow cytometry analysis. The results of DNA electrophoresis of PCR product from GL261 cells genomic DNA showed that there was an obvious reduction of DNA band when GL261 cells were treated with bNe@AIE/Cas9-CD73 (Supplementary Fig. 22a, c), where the gray intensity of DNA band in bNe@AIE/Cas9-CD73 group was 2.71-fold lowere than that in the PEI40000 group (5.37% vs 14.56%), indicating that bNe@AIE/Cas9-CD73 caused indels at the targeted CD73 loci to disrupt the expression of CD73 effectively, verifying the CRISPR-mediated CD73 blockade. T7EI assays was further employed to gene disruption, the results of DNA electrophoresis demonstrated the indels with cleaved band after treated bNe@AIE/Cas9-CD73 (Supplementary Fig. 22b). When compared with control GL261 cells without treatment, no cleaved bands were noticed. Then, GL261 cells were incubated with bNe@AIE/Cas9-CD73 for 48 h, followed by western blotting analysis. As shown in Fig. 3e, f, CD73 expression was significantly reduced in treated cells, confirming effective intracellular delivery and successful CRISPR/Cas9-mediated gene disruption efficiency (~25.09% of control group). Moreover, to measure the gene disruption efficiency by bNe@AIE/Cas9-CD73, the CRISPR/Cas9 plasmid (px458) that contains the same sgRNA sequence with the EGFP tag was employed to prepare bNe@AIE/Cas9-CD73 to knockout CD73 in vitro and in vivo. After treated for 72 h, the ratio of EGFP positive GL261 cells was 52.90% (Supplementary Fig. 23), suggesting the high gene disruption efficiency in GL261 cells. These results demonstrated that nanodelivery of CRISPR/Cas9 via bNe@AIE efficiently manipulated CD73 expression, providing a promising means to remodel the TME and potentiate combined therapeutic outcomes.

Cell uptake, subcellular localization, cytotoxicity, and inflammation-active penetration across the BBB in vitro

After establishing the phototherapeutic properties of NDA-DPE, the cellular uptake, subcellular localization, cytotoxicity, and inflammation-responsive penetration of bNe@AIE/Cas9-CD73 across the BBB were investigated in vitro. The px458-CRISPR/Cas9-CD73 plasmid contained a green EGFP tag that allowed visualization of cell internalization. GL261 glioblastoma cells were incubated with different nanoparticle formulations for 6 h, washed, and further cultured for 48 h. Transfection with px458-CRISPR/Cas9-CD73 using PEI40000 served as the positive control. As shown in Fig. 4a, b, no significant difference in fluorescence intensity was observed among the px458-transfected, px458-CRISPR/Cas9-CD73-AIE-dot, and bNe@AIE/Cas9-CD73 groups, indicating comparable uptake efficiency. However, time-dependent of fluorescence intensity after treated with Cy5 label- bNe@AIE/Cas9-CD73 revealed the time-dependent endocytosis (Supplementary Fig. 24). To assess inflammation-mediated targeting, formyl-methionyl-leucyl-phenylalanine (fMLP) was used to induce inflammation in GL261 glioblastoma cells before nanoparticle treatment. Under these conditions, significantly stronger fluorescence signals were observed in the bNe@AIE/Cas9-CD73 group compared with controls (Fig. 4a, b), confirming that the intrinsic inflammatory tropism of neutrophils increased targeted cellular uptake in glioblastoma cells. The subcellular localization of NDA-DPE was then analyzed using organelle-specific probes and NDA-DPE fluorescence in GL261 glioblastoma cells. Colocalization studies with Lyso-Tracker Green and Mito-Tracker Green revealed that NDA-DPE accumulated primarily in lysosomes and mitochondria (Fig. 4c), with Pearson’s correlation coefficients of 0.99 and 0.81, respectively. These findings indicate that NDA-DPE acts as a multi-organelle-targeted theranostic agent, using the biological importance of these organelles for glioblastoma cells survival.

Fig. 4.

Cell uptake, subcellular localization, cytotoxicity and inflammation-active penetration across BBB in vitro. a Fluorescent images of GL261 glioblastoma cells after received different treatment. b Mean fluorescence intensity of (a). c CLSM images of NDA-DPE subcellular localization and co-localization analysis with different organelles. d CCK-8 analysis results of AIE dots in bNe, GL261 and BV2 cells. e Schematic system of BBB model in vitro. f Mean fluorescence intensity of NDA-DPE in lower chamble after treatments

Biocompatibility was next assessed, given the potential insolubility and cytotoxicity of NDA-DPE, as well as the biosafety requirements for nanoagents for further clinical translation of bio-nanomedicines. CCK-8 assays were conducted in GL261 glioblastoma cells, normal murine microglial Bv2 cells, and isolated neutrophils exposed to varying concentrations of AIE dots under dark conditions. As shown in Fig. 4d, cell viability remained above 80% across all tested cell types at 100 µg/mL, confirming excellent in vitro compatibility of the AIE dots.

To evaluate the ability of bNe@AIE/Cas9-CD73 to traverse the BBB, a primary and inevitable obstacle for glioblastoma treatment, a well-established in vitro murine brain microvascular endothelial (bEnd.3) monolayer model was employed (Fig. 4e). Under physiological conditions, all AIE-dot formulations exhibited minimal permeability through the endothelial layer (Fig. 4f). bNe@AIE/Cas9-CD73 produced a pronounced increase in fluorescence signal intensity at the basolateral chamber, reaching 3.53-fold higher intensity after 6 h compared with AIE dots alone (Fig. 4f). When inflammatory conditions were induced using fMLP-stimulated glioblastoma cells, penetration of bNe@AIE/Cas9-CD73 further increased by 1.81-fold relative to the non-inflammatory setting (Fig. 4f), consistent with neutrophil-mediated tropism toward inflamed tissues. These results demonstrate that bNe@AIE/Cas9-CD73 exhibits efficient cellular uptake, lysosomal–mitochondrial localization, excellent cytocompatibility, and higher inflammation-activated transport across the BBB, underscoring its potential for targeted gene-photoimmunotherapy in glioblastoma.

Efficient targeted phototherapy against glioblastoma cells in vitro

Building on the confirmed phototherapeutic properties of NDA-DPE, we further evaluated the efficacy of bNe@AIE/Cas9-CD73-mediated phototherapy in glioblastoma cells in vitro. To assess ROS generation, which is key to PDT, 2′,7′-dichlorodihydrofluorescein diacetate (DCFH-DA) was used as an indicator. As shown in Fig. 5b, high intracellular green fluorescence was observed in GL261 glioblastoma cells following incubation with AIE dots and exposure to an 808 nm NIR laser (0.5 W/cm², 5 min). CRISPR/Cas9-mediated downregulation of CD73 further increased the ROS-generating capacity of NDA-DPE in GL261 glioblastoma cells compared with AIE-dots-based PDT alone, indicating that gene manipulation increased photodynamic activity. The elevated ROS disrupted mitochondrial and lysosomal integrity, ultimately inducing cancer cell death. Cell apoptosis was then evaluated using the CCK-8 assay and flow cytometry. Eight treatment groups were established: (1) Control, (2) bNe@CRISPR/Cas9-CD73, (3) bNe@AIE dots, (4) bNe@AIE/Cas9-CD73, (5) Control + Laser (L), (6) bNe@CRISPR/Cas9-CD73 + L, (7) bNe@AIE dots + L, and (8) bNe@AIE/Cas9-CD73 + L. As shown in Fig. 5c, phototoxicity assays revealed a concentration-dependent reduction in cell viability. Consistent with this, flow cytometry analysis demonstrated significant apoptosis in the bNe@AIE/Cas9-CD73 + L group (Fig. 5d and Supplementary Fig. 25), and also the Calcein AM/PI co-staining (Supplementary Fig. 26a, b) and CCK-8 cell counting kit analysis (Supplementary Fig. 26c). Importantly, the inflammatory condition induced by the addition of PMA would caused more cell death (Supplementary Fig. 26), which was attributed to the higher delivery efficiency due to the tropism of neutrophils. The manipulation of CD73 significantly increased the cytotoxicity of NDA-DPE-based phototherapy, indicating synergistic gene–photo treatment effects against glioblastoma cells.

Fig. 5.

In vivo antitumor treatment. a Schematic of the mechanism of bNe@AIE/Cas9-CD73-based phototherapy and its immunological regulation. b Fluorescent CLSM images of ROS after different treatments. c GL261 cells viability was analyzed with CCK-8 analysis after treated with different nanomaterials with/without 808 nm laser irradiation. d GL261 glioblastoma cells viability was analyzed with flow cytometry after treated with different nanomaterials with/without NIR laser irradiation. e Infiltration of mature DCs after different treatments with flow cytometry. f, g The concentration analysis of TNF-α and IFN-γ by ELISA after different treatments. Evaluation of released ATP content (h) and adenosine content (i) of GL261 glioblastoma cells treated with the different materials with/without NIR laser irradiation. Statistical significance was calculated by one-way ANOVA with Tukey’s multiple comparisons test, *p < 0.05, **p < 0.01, ***p < 0.001, ns, not significant

Both PTT and PDT are known to induce apoptosis, necrosis, and ICD, releasing tumor-associated antigens that can activate dendritic cells (DCs) and initiate tumor-specific immune responses.⁵⁹ The main immune cells reside in the blood and migrate into the brain tissue when they receive a stimulus. To determine whether bNe@AIE/Cas9-CD73-triggered phototherapy could elicit immune activation, the induction of ICD were examined using western blotting and immunofluorescence. As shown in Supplementary Fig. 27, the increase of cGAS, STING, high mobility group protein 1 (HMGB1) and Calreticulin, indicated the occurrence of ICD in bNe@AIE dots + L, bNe@CRISPR/Cas9-CD73, bNe@CRISPR/Cas9-CD73 + L, bNe@AIE/Cas9-CD73, and bNe@AIE/Cas9-CD73 + L goups. Moreover, a Transwell coculture system was employed (Supplementary Fig. 28). As shown in Fig. 5e and Supplementary Fig. 29, macrophage activation was evident in bNe@AIE dots + L, bNe@CRISPR/Cas9-CD73, bNe@CRISPR/Cas9-CD73 + L, and bNe@AIE/Cas9-CD73 + L groups relative to control, control + L and AIE-dot-only groups. Among all groups, bNe@AIE/Cas9-CD73 + L group exhibited the highest proportion of polarized, tumor-supportive macrophages. These results suggest that phototherapy with bNe@AIE/Cas9-CD73 promotes macrophage polarization, enhancing tumor-directed immune responses. To further assess adaptive immune activation, isolated T cells were cocultured with pretreated GL261 cells, and the concentrations of TNF-α and IFN-γ in the supernatants were quantified using ELISA. Without NIR laser irradiation, both cytokines were significantly elevated in the bNe@CRISPR/Cas9-CD73 and bNe@AIE/Cas9-CD73 groups (Fig. 5f, g), indicating that CD73 inhibition alone could activate T-cell responses. Upon NIR laser exposure, TNF-α and IFN-γ secretion increased further in bNe@AIE dots and bNe@AIE/Cas9-CD73 groups. Cytokine (TNF-α and IFN-γ) levels were highest in the bNe@CRISPR/Cas9-CD73, bNe@AIE/Cas9-CD73, and bNe@AIE/Cas9-CD73 + L groups, suggesting both neutrophil-mediated delivery and gene-silencing-induced immune. These results confirm that bNe@AIE/Cas9-CD73-based phototherapy can induce and potentiate immune activation, facilitating immune cell infiltration across the BBB.

As ICD stimulation by PTT and/or PDT can increase the release of intracellular ATP, which is sequentially metabolized by CD39/CD73 into adenosine and phosphate (ATP → ADP → AMP → adenosine + phosphate), excessive adenosine formation can increase immunosuppression and hinder therapeutic outcomes.^13,60 Disruption of the CD39/CD73 axis, however, reverses this suppression and restores antitumor immunity.⁶¹ Also, previous studies have shown that chemotherapy, PTT, PDT, SDT, RT, and other treatments can increase ATP levels, which does not benefit therapeutic outcomes. To explore this molecular mechanism, the experimental design is illustrated in Fig. 5a. The concentrations of ATP, ADP, AMP, adenosine, and phosphate were quantified using commercial assays following various treatments. As shown in Fig. 5h, Supplementary Figs. 30–32, and Fig. 5i, significant increases in ATP, ADP, and AMP were observed in bNe@CRISPR/Cas9-CD73, bNe@AIE/Cas9-CD73, bNe@AIE dots + L, bNe@CRISPR/Cas9-CD73 + L, and bNe@AIE/Cas9-CD73 + L groups compared to control, while adenosine and phosphate were significantly decreased in bNe@CRISPR/Cas9-CD73, bNe@AIE/Cas9-CD73, bNe@CRISPR/Cas9-CD73 + L, and bNe@AIE/Cas9-CD73 + L groups, but not in Control + L, bNe@AIE dots, and bNe@AIE dots + L groups, compared with Control group. These findings confirmed that NT5E (CD73) knockdown inhibited ATP → ADP → AMP → adenosine and phosphate conversion pathway, limiting adenosine and phosphate accumulation and reshaping the suppressive TME into a pro-immunogenic state. Moreover, inhibition of CD73 expression may activate the cGAS–STING signaling pathway, further increasing immune activation and promoting antitumor responses, which was confirm as shown in Supplementary Fig. 27.⁶² These results demonstrate that the combination of gene therapy and phototherapy not only modulates ATP metabolism but also increases ICD-driven immune stimulation, providing an innovative strategy for effective glioblastoma treatment.

Evaluation of bNe@AIE/Cas9-CD73 tumor targeting

Leveraging the intrinsic inflammatory-targeting properties of neutrophils, the engineered biohybrid nanoplatform bNe@AIE/Cas9-CD73 was expected to display strong BBB penetration and precise tumor-targeting capability in vivo. The targeting and distribution of bNe@AIE/Cas9-CD73 was evaluated in orthotopic GL261-Luc xenograft–bearing C57BL/6 J mice using a small-animal in vivo imaging system (Series II 900/1700-H; 0.3 W). Tumor-bearing mice received intravenous injections of either AIE dots or bNe@AIE/Cas9-CD73, and their biodistribution was tracked by NIR-II fluorescence. As shown in Fig. 6a, distinct tumor localization was observed in the bNe@AIE/Cas9-CD73 group, consistent with efficient BBB penetration and the enhanced permeability and retention (EPR) effect mediated by neutrophils, suggesting passive targeting of the as-prepared nano-therapeutic agents. AIE-dots-treated mice showed minimal fluorescence accumulation at the tumor site, reflecting limited BBB permeability and the lack of active targeting capability by DSPE-PEG₁₀₀₀. The NIR-II fluorescence (1000–1100 nm) from NDA-DPE in bNe@AIE/Cas9-CD73-treated brains progressively increased over time from 1 h to 48 h post-injection and declined at 96 h (Fig. 6a, b). In the AIE-dot group, fluorescence intensity remained weak at the tumor sites at all timepoints, confirming neutrophil-mediated active targeting and prolonged retention of bNe@AIE/Cas9-CD73 at the glioblastoma tiuus site. Ex vivo biodistribution analysis of AIE dots and bNe@AIE/Cas9-CD73 in isolated tumors and major organs using NIR-II fluorescence imaging in GL261-bearing mice at 96 h post-injection, revealed that both AIE dots and bNe@AIE/Cas9-CD73 primarily accumulated in the liver and lungs (Fig. 6c, d). The AIE-dot group also showed strong fluorescence in the spleen, confirming its role in nanomaterial clearance. Tumor fluorescence intensity in bNe@AIE/Cas9-CD73 group far exceeded that in AIE-dot group, demonstrating better tumor-targeting efficiency. The homologous neutrophil coating also reduced nonspecific uptake by the reticuloendothelial system (liver and spleen), improving in vivo delivery to glioblastoma. To verify gene-editing efficacy in vivo, orthotopic GL261-Luc-xenograft–bearing C57BL/6 J mice were anesthetized, and the tumor tissues were harvested 3 days after injection, homogenized, and analyzed by western blotting. As shown in Fig. 6e, f, CD73 expression was significantly reduced in tumors from bNe@AIE/Cas9-CD73–treated mice compared with controls. At same time, the high heterogeneity of glioblastoma provides lethal hinderance to therapeutic response, and the drugs is prone to be eliminated by immune cells in vivo, combined with the high expression of CD73 in macrophage, we also evaluated the gene manipulation efficiency of CD73 caused by bNe@AIE/Cas9-CD73. The tumor tissues were harvested three days post-injection, macrophage were isolated using Macrophage Isolation Kit after dissociated, the purity of isolated macrophage was identified with flow cytometry analysis, over 98% of isolated cells were labeled with F4/80 antibody (Supplementary Fig. 33a), suggesting the high purity of isolated macrophages. Then, the gene manipulation efficiency of CD73 were determined by detecting EGFP positive cells. As shown in Supplementary Fig. 33a, c, no EGFP positive macrophage were detected in Control group, and 3.94% of EGFP positive macrophage were determined, indicating the low gene manipulation efficiency of CD73 of macrophage in vivo. Meanwhile, the results of western blotting indicated that bNe@AIE/Cas9-CD73 could down-regulate the expression of CD73 with no obvious difference in macrophage in vivo (Supplementary Fig. 33b, d). These results confirmed the successful CRISPR/Cas9-mediated gene silencing in vivo.

Fig. 6.

Determination of the targeting, biodistribution, and metabolism of bNe@AIE/Cas9-CD73 in vivo. a NIR-II fluorescence images of AIE dots and bNe@AIE/Cas9-CD73 treated mice in vivo at different timepoints. b NIR-II mean fluorescence intensity in glioblastoma cancer cells-bearing mice at different timepoints. c Fluorescence images of major organs and tumors obtained 96 h post-injection. He: heart, Lv: liver, Sp: spleen, Lu: lung, Ki: kidney, Br: brain; (d) NIR-II mean fluorescence intensity corresponding to (c). e, f Western blotting and quantitative results of brain tumor tissue after treated with bNe@AIE/Cas9-CD73. g Pharmacokinetic profiles of NDA-DPE, AIE dots and bNe@AIE/Cas9-CD73 after tail intravenous administration. n = 3, statistical significance was calculated by one-way ANOVA with Tukey’s multiple comparisons test, *p < 0.05, **p < 0.01, ***p < 0.001

Pharmacokinetic analysis in C57BL/6 J mice further demonstrated that neutrophil-assisted NDA-DPE nanoparticles exhibited extended blood circulation compared with unmodified NDA-DPE and DSPE-PEG₁₀₀₀-encapsulated AIE dots (Fig. 6g). The half-life of NDA-DPE, AIE dots, and bNe@AIE/Cas9-CD73 were 0.93 h, 2.23 h, and 5.92 h, respectively. DSPE-PEG₁₀₀₀ improved the systemic biocompatibility of NDA-DPE, whereas the homologous neutrophil surface reduced self-elimination and immune clearance, prolonging the half-life of injected nanomaterials. Protein-corona interactions between plasma proteins and the nanoparticle surfaces may have further contributed to biological retention, while decreased blood levels of bNe@AIE/Cas9-CD73 were attributed to active tumor tropism and physiological excretion. These findings demonstrate that neutrophil-assisted bNe@AIE/Cas9-CD73 exhibits prolonged systemic circulation, effective BBB penetration, and higher glioblastoma-targeting capacity, increasing tumor-site accumulation and impvoing the therapeutic potential of gene-photoimmunotherapy in vivo.

Evaluation of bNe@AIE/Cas9-CD73 amplified synergistic antitumor effect in vivo

Based on the intrinsic NIR absorption, high light-to-heat conversion efficiency, and strong ROS-generating capability of NDA-DPE, the bNe@AIE/Cas9-CD73 nanoplatform was further evaluated as a phototherapeutic agent for combined PTT and PDT against glioblastoma. Given the promising results in vitro showing efficient ICD induction and immune activation, the synergistic antitumor performance was systematically investigated in orthotopic GL261-Luc xenograft–bearing C57BL/6 J mice (Fig. 7a). Orthotopic glioblastoma models were established and randomly assigned to eight treatment groups: (1) Control (PBS), (2) bNe@CRISPR/Cas9-CD73, (3) bNe@AIE dots, (4) bNe@AIE/Cas9-CD73, (5) Control + Laser (L), (6) bNe@CRISPR/Cas9-CD73 + L, (7) bNe@AIE dots + L, and (8) bNe@AIE/Cas9-CD73 + L. Forty-eight hours post-injection, groups 5–8 were irradiated with an 808 nm NIR laser (0.5 W/cm², 5 min). The corresponding infrared thermal images and tumor temperature changes at different time points were recorded simultaneously with an infrared thermal imaging camera (Fig. 7b, d). Under NIR laser irradiation, tumor-site temperatures in bNe@AIE dots– and bNe@AIE/Cas9-CD73–treated mice rapidly increased from approximately 32.1 °C to 55.9 °C and 59.8 °C, respectively, levels sufficient for tumor ablation. Control and bNe@CRISPR/Cas9-CD73 groups exhibited only minimal temperature elevation (32.0 °C to 34.3 °C), indicating negligible photothermal response. PDT efficacy was assessed 24 h after 808 nm NIR laser irradiation. As shown in Supplementary Fig. 34, mouse brain slices from groups (7) and (8) demonstrated pronounced ROS generation, with the bNe@AIE/Cas9-CD73 group showing the strongest fluorescence. This improved PDT effect was due to CD73 inhibition, which alleviated hypoxia and facilitated ROS generation, confirming efficient in vivo PDT activity.

Fig. 7.

In vivo evaluation of antitumor immunotherapy. a Schematic illustration of therapeutic profile of the orthotopic glioblastoma tumor model. b In vivo photothermal images after 808 nm laser irradiation at different timepoints. c Representative in vivo bioluminescence images of orthotopic glioblastoma at different time points after bNe@AIE/Cas9-CD73-based synergistic therapy. d In vivo photothermal temperature curve. e Semi-quantitative analysis of in vivo bioluminescence signals at different timepoints in the brain tumor site after different treatment. f Survival curves of mice in the orthotopic glioblastoma tumor model with different treatments. g H&E staining images of mice brain after diffent treatment. Data are presented as the mean ± SEM. P values were calculated using one-way ANOVA. *p < 0.05, **p < 0.01, and ***p < 0.001

Tumor progression was monitored by bioluminescence imaging and body weight tracking every 3 days (Fig. 7a). Higher bioluminescence intensity correlated with larger tumor size and faster growth. As shown in Fig. 7c, e, compared to the Control group, both bNe@CRISPR/Cas9-CD73 and bNe@AIE/Cas9-CD73 treatments significantly inhibited glioblastoma tumor growth, whereas bNe@AIE dots induced only moderate inhibition, likely due to mild neutrophil-mediated inflammation resolution.⁶³ No growth differences were observed between the Control and Control + L groups or between bNe@CRISPR/Cas9-CD73 and bNe@CRISPR/Cas9-CD73 + L groups, confirming that the NIR laser alone had no therapeutic effect on glioblastoma. Meanwhile, there were differences between the Control group and the bNe@AIE dots + L and bNe@AIE/Cas9-CD73 + L groups under phototherapy, especially in the bNe@AIE/Cas9-CD73 + L group, which not only inhibited glioblastoma progression to the highest extent but also reduced tumor size, as indicated by tumor growth curves and reduced luminescence intensity. Body-weight analysis (Supplementary Fig. 35) revealed significant weight loss in Control and Control + L groups, consistent with advanced tumor burden. Only mild fluctuations were observed in mice treated with bNe@CRISPR/Cas9-CD73, bNe@AIE dots, bNe@AIE/Cas9-CD73, or bNe@CRISPR/Cas9-CD73 + L, indicating low systemic toxicity and good biosafety. These findings confirm that bNe@AIE/Cas9-CD73 achieves robust synergistic therapeutic efficacy through combined gene, photothermal, and photodynamic mechanisms.

To further evaluate the therapeutic benefit of bNe@AIE/Cas9-CD73-based therapy in mice, survival rates were recorded throughout the treatment period (Fig. 7f). Mice in the control and control + L groups showed the lowest survival, whereas those in bNe@AIE/Cas9-CD73 + L and bNe@CRISPR/Cas9-CD73 + L groups displayed significantly prolonged survival, confirming improved treatment efficacy. Post-treatment, mice were euthanized, and tumor tissues were collected for histopathological analysis. Hematoxylin and eosin (H&E) staining (Fig. 7g) of tumor slices revealed dense, compact tumor morphology in Control and Control + L groups, whereas tissues from groups (2), (3), (4), (6), (7), and particularly (8) exhibited disrupted structure, necrosis, and significantly reduced tumor area, suggesting the excellent therapeutic effects of synergistic theapeutic effects of bNe@AIE/Cas9-CD73.

To further elucidate the immune activation induced by bNe@AIE/Cas9-CD73–based synergistic therapy, tumor-infiltrating immune cells within glioblastoma tissues were analyzed by flow cytometry. As shown in Fig. 8a, e, treatment significantly increased the population of mature DCs and pro-inflammatory M1-type macrophages. Compared with the control group (10.82%), the percentage of tumor-associated M1 macrophages (CD11c⁺-gated) increased to 17.23% in the bNe@CRISPR/Cas9-CD73 group, 14.83% in the bNe@AIE dots group, 20.56% in the bNe@AIE/Cas9-CD73 group, 14.23% in the control + L group, 20.16% in the bNe@CRISPR/Cas9-CD73 + L group, 34.33% in the bNe@AIE dots + L group, and 41.90% in the bNe@AIE/Cas9-CD73 + L group. This shift confirmed significant polarization toward the M1 phenotype after treatment. Downregulation of CD73 reduced adenosine-mediated immunosuppression and promoted the development of an inflammatory TME favorable for DC maturation.²⁴ The proportions of cytotoxic CD8⁺ and helper CD4⁺ T cells in brain tumor tissues were also significantly increased following treatment. As shown in Fig. 8b, c, f, and g, CD8⁺ T cells increased from 0.50% in the control group to 3.83% in the bNe@AIE/Cas9-CD73 + L group, while CD4⁺ T cells increased from 0.29% to 5.28%, indicating improved adaptive immune activation. In parallel, natural killer cell populations were significantly elevated (Control: 3.16%; bNe@CRISPR/Cas9-CD73: 6.02%; bNe@AIE dots: 3.63%; bNe@AIE/Cas9-CD73: 6.11%; Control + L: 3.50%; bNe@CRISPR/Cas9-CD73 + L: 6.40%; bNe@AIE dots + L: 8.93%; bNe@AIE/Cas9-CD73 + L: 11.83%) (Fig. 8d, h), confirming robust immune activation. ELISA analyses of brain tumor tissues further supported these findings. TNF-α and IFN-γ levels increased by 2.73-fold and 3.90-fold, respectively, compared with the control group (Fig. 8i, j). Immunofluorescence staining of brain slices (Supplementary Fig. 36) corroborated extensive immune-cell infiltration in treated tumors. The inherent heterogeneity of glioblastoma often limits uniform immune-cell distribution; however, CD73 inhibition facilitated improved T-cell infiltration and activity within the tumor.²⁵ These results confirm that bNe@AIE/Cas9-CD73–mediated gene-phototherapy efficiently induces ICD and enhances immune responses in glioblastoma tissues.

Fig. 8.

In vivo immune activation evaluation. a–d Flow cytometry results of polarized macrophages, CD4⁺ T cells, CD8⁺ T cells and NK cells. e–h Quantitative results of immune cells responding to A-D. i, j ELISA results of TNF-α and IFN-γ. Evaluation of released ATP content (k), AMP content (l) adenosine content (m) in vivo. Data are presented as the mean ± SEM. P values were calculated using one-way ANOVA. *p < 0.05, **p < 0.01, and ***p < 0.001

To investigate the mechanism of immune activation, ATP–adenosine axis metabolites (ATP, ADP, AMP, adenosine, and phosphate) in GL261-bearing tumors were quantified using commercial kits. As shown in Fig. 8k–m and Supplementary Figs. 37–37, the ATP → adenosine + phosphate pathway was significantly inhibited in vivo, consistent with in vitro observations. In control tumors, the concentrations of ATP, ADP, AMP, adenosine, and phosphate were 0.19, 0.28, 0.19, 0.30, 0.21, and 0.31 µM, respectively. Following bNe@AIE dots + L and bNe@AIE/Cas9-CD73 + L treatments, ATP levels increased to 0.46 µM and 0.66 µM due to PTT/PDT-induced ICD and blockade of ATP metabolism. Similar trends were observed for ADP and AMP (Supplementary Fig. 37a and Fig. 8l). Adenosine and phosphate levels significantly decreased, from 48.84 µM and 11.46 mM in the control group to 18.25 pM and 5.87 mM, respectively (Fig. 8m and Supplementary Fig. 37b). These results confirm that bNe@AIE/Cas9-CD73-based treatment induced ICD in glioblastoma cells while concurrently reducing immunosuppressive adenosine production, demonstrating excellent therapeutic efficacy and great potential to remodel the immunosuppressive TME to favor antitumor immunotherapy. Modulation of the ATP–adenosine axis in conjunction with AIEgen-based phototherapy represents a potent mechanism to improve immunotherapy outcomes.

Finally, biosafety from different treatments was assessed to ensure translational applicability. In the frontline, due to the broad expression of NT5E in many tissues and cells, including liver and lung, the main organs of nanodrugs accumulation, of mouse and human body,⁶⁴ the potential side effect of gene disruption of NT5E were first evaluated. The GL261 cancer cells-bearing mice were established and distributed into different groups randomly, and the mice were injected with bNe@AIE/Cas9-CD73 for different timepoints (0, 1, 3, 7, 10 and 14 days), where health mice were set as positive control. To reach the goal, H&E staining, wet/dry (W/D) ratio, the index of blood routine and liver or lung lesion (ALT, AST, ALB, ALP, TBIL and TP) using blood biochemistry analysis, and also the ELISA analysis were performed. At different timepoints postinjection, photographs of liver and lung tissues demonstrated that there were no obvious pathological changes caused by NT5E gene disruption (Supplementary Fig. 38a). In addition, the results of W/D ratio of liver and lung tissues to gauge the extent of lesion showed that there no significant lesion between different timepoints after NT5E gene disruption when compared with healthy control (Supplementary Fig. 38b, c). The results of blood routine and blood biochemistry showed in Supplementary Fig. 39a and b also indicated that there were no significant alteration of blood content after NT5E gene manipulation. On ther other hand, the lung and liver tissues were also sectioned and stained by H&E method, there is no obviously pathological lesion after the reduced expression of NT5E (Supplementary Fig. 39c). At last, the proinflammatory cytokines, including IL-6, IL-1β and TNF-α, from lung tissue were also evaluated due to their biological function of CD73 in inflammation regulation. At the default timepoint, bron-choalveolar lavage fluid (BALF) was collected and processed for the quantification of proinflammatory cytokines utilizing ELISA kit, respectively.As shown in depicted in Supplementary Fig. 39d, no obvious significance of proinflammatory cytokines concentration were observed between health mice and treated mice at different timepoints. Collectively, the NT5E gene disruption showed negligibl side-effect to biological system in vivo, but also the excellent biocompatibility of bNe and AIE dots.

Meanwhile, the biosafety after therapeutic treatments were also determined. Body weight changes across all groups corresponded to tumor burden, with no abnormal fluctuations in treated mice. Routine hematology, serum biochemistry, and histopathological analyses of major organs (heart, liver, spleen, lung, and kidney) demonstrated no signs of systemic toxicity or tissue damage (Supplementary Figs. 40–41). H&E staining revealed intact organ architecture, and blood and biochemical parameters remained within normal ranges across all treatment groups. These results demonstrate that bNe@AIE/Cas9-CD73 not only elicits potent ICD-mediated immune activation and effectively reprograms the immunosuppressive TME by modulating the ATP–adenosine axis, but also maintains excellent biosafety, supporting its potential for safe and effective clinical translation.

Discussion

Glioblastoma remains a major clinical challenge among various tumors because of its aggressive biology, lack of early diagnostic and therapeutic targets, intrinsicly therapeutic resistance and BBB. Previous studies have revealed the importance of TME in regulating the progresses and the therapeutic response of glioblastoma.^65,66 Therefore, the precise targeting of key switch could reshape TME to enhance the therapeutic outcomes effectively.⁶⁷ Due to the importance of energy supply in shaping tumor- growth supportive TME and promoting the growth of cancer cells, which is prone to be the hottest topic in combating glioblastoma.⁶⁸ Among the various energy metabolism pathway, the direct disruption of ATP metabolism owns great potential in eliminating cancer cells.⁶⁹

Hypoxia in one of typical characteristics of glioblastoma that changes the way of energy supply and provides hinderance to radiotherapy. Interesting, a hypoxia-induced ectonucleotidase (ecto-5′-nucleotidase), CD73, an universal expression protein encoded by the NT5E gene, was identified. It also has been verified to be associated with immunosuppressive TME and glioblastoma malignancy.^13,16 In details, CD73, aberrantly overexpression in glioblastoma, catalyzes the generation of adenosine and phosphate from AMP after ATP and ADP are hydrolyzed by CD39, furthermore, adenosine binds G protein-coupled P1 receptors (A1, A2A, A2B, A3) and P2 receptors (P2X, P2Y) directly, which benefits for the formation of immunosuppressive TME, tumor progression, metastasis, invasion, and treatment resistance. Therefore, CD73 is promising target for glioblastoma. At present, various small molecule inhibitors, including AB-680, PSB-12379, LY3475070, antibody drugs, including Oleclumab, Mupadolimab, Uliledlimab, antibody-conjugated drugs, including BB-1709 and HB-0052, nanoantibody, oncolytic virus, and so on, have been proposed and transformed into clinical trials. However, the application of most of drugs are limited due to the low targeting capability and potential side effects. Recently, gene editing, including CRISPR/Cas9, TALEN, and so on, is proved to be an effective tool in tumor treatment,⁷⁰ which accelates the engagement of gene therapy combined with different treatment methods, therefore, targeted manipulation of NT5E expression using CRISPR/Cas9 represents a powerful approach for tumor therapy when compared with CD73 inhibitors. Here, NT5E-targeted gene manipulation were achieved, the transfection of CRISPR/Cas9 exhibited the down-regulated expression of CD73 in vitro, but not in vivo, which might attribute to the rapid degradation and clearance rate of biological system. The inhibition of CD73 expression blocked the hydrolysis of AMP into adenosine and phosphate that disrupt the ATP→ adenosine + phosphate axis, where the reduced adenosine not only reshapes TEM by inhibiting tumor angiogenesis, but also activating the anti-tumor T cell response via reducing the cellular concentration of adenosine to accelerate pyrimidine synthesis.⁷¹ Reduced phosphate content increases cellular pH value from acidic environment, which benefits for the polarization from M2 macrophage to M1 macrophage. In this study, we confirmed that CRISPR/Cas9 can manipulate CD73 expression (Supplementary Figs. 30–32) to block ATP→ adenosine + phosphate axis effectively (Fig. 5h, i and Supplementary Figs. 22 and 23). Collectively, gene manipulation of CD73 via CRISPR/Cas9 tool could regulate ATP metabolism to reshape TME effectively and evoke immune system. Similar to previous studies, the broad application of CRISPR-Cas9-based gene editing in tumor treatment is limited by the extremely delivery efficiency and targeting specificity. Therefore, it is necessary to design and construct a reasonable and effective delivery system to overcome unfavorable physicochemical properties of the CRISPR-Cas9-based gene editing and achieve strong anti-tumor efficacy.

In the past decades, the development of drug delivery system have achieved great progresses and applied in clinics successfully. The introduction nanomaterials have boosted the promising development of medicine by exploiting the physical and chemical properties of as-prepraed nanosystem. Importantly, the intrinsic properties of nanomaterials empower effective treatment efficiacy by enhancing outcomes of different therapeutic modalities, but also settling out the weakness of traditional methods. Among various therapeutic methods, PTT and PDT represent two effective treatment methods in brain diseases because of their high efficiency, non-invasiness, high spatial selectivity and minimal side effects.^72,73 More important, PTT and PDT are effective strategies for inducing ICD to stimulate antitumor immunity.⁷⁴ Alarmingly, due to the intrinsic limitation of glioblastoma tumor tissue, such as hypoxia for PDT, low penetration ablility of NIR laser for PTT and PDT, separate PTT or PDT is prone to failed in eliminating cancer cells effectively with non-negligible side-effects, the combination of PTT and PDT could achieve win-win results. The synthesis of multifunctional AIEgen, NDA-DPE, in our study exhibited excellent PTT and PDT performance (Fig. 2). Meanwhile, NDA-DPE owns excellent NIR-II fluorescence for bioimaging in vivo (Fig. 6a, b). Therefore, NDA-DPE-based nanoplatform is capable to achieve bioimaging-guided PTT/PDT treatment and gain the synergistic outcomes of “1 + 1 > 2”, which can kill cancer cells and inhibit the tumor growth efficiently (Fig. 7). Also. we demonstrated that NDA-DPE-based PTT/PDT could induce cancer cells into ICD, evoke and mature immune system, including the increase of CD4⁺ and CD8⁺ T cells, and NK cells (Fig. 8), to obtain immunotherapy. Collectively, this observation suggested a potential exploration of multifunctional AIEgen on glioblastoma treatment. Moreover, combined with the CD73 manipulation-reshaped TEM, these all provides a molecular rational for strong activation and maturation of immune system to gain immunotherapy.

On the other hand, BBB is an inevitable and unresolved challenge for glioblastoma and other brain diseases. Development of specialized delivery systems, such as the bNe@AIE/Cas9-CD73 in our study have the potential for addressing limited BBB acrossing, cellular internalization, tissue penetration, undesirable tissue specificity of CRTSPR/Cas9 and AIE dots. In our study, by taking the tumor-associated characteristics, inflammation, bone marrow-derived neutrophils was employed as nanodelivery platform to construct the hybrid nanosystem bNe@AIE/Cas9-CD73, enabling precise delivery of CRISPR/Cas9 and AIEgen components to glioblastoma tissues, due to its intrinsic chemotaxis toward inflammatory microenvironments and the capability of avoiding immune surveillance, which offers improved delivery efficiency and therapeutic precision.⁷⁵ With bNe@AIE/Cas9-CD73, the high efficiency crossed BBB, targeted specificity of nanoplatform toward glioblastoma lesions were reached (Fig. 6). Upon localized NIR laser irradiation, excellect PTT and PDT were realized to induce strong ICD, the release of CRISPR/Cas9-CD73 from neutrophils mediated downregulation of CD73 disrupted the ATP–adenosine axis, converting the immunosuppressive TME into an immuno-supportive one that improved both phototherapeutic and immune responses, which inhibited the tumor growth effectively. Although, our study exhibited that the downregulation of CD73 were achieved in tumor tissue (Fig. 6) and slight decrease in tumor-associated macrophage (Supplementary Fig. S33), suggesting the gene manipulation efficiency. Simultaneously, the potential for off-target effects remains a concern (Supplementary Fig. S39), bNe@AIE/Cas9-CD73 with specific tumor targeting and distribution may own the potential to solve the limitation. While our research results is not capable to draw the conclusion, it is urgent and promising to explore the potential of bNe@AIE/Cas9-CD73 in settling out off-target effects of CRISPR/Cas9-based gene editing and pre-clinical application. Furthermore, bone-marrow-derived neutrophils show expected efficacy in preclinical models, while the loading efficiency of plasmid and nanomaterials, and the short life cycle in vitro, human clinical trails, are other concerns for further and deeper research. The full investigation of these key puzzles through continued endeavor and innovation in brain-targeted delivery strategies and synergistic therapies will be vital for realizing the cure of glioblastoma and pave the broader way for innovative therapies.

In conclusion, our study demonstrates that the disruption of CD73 expression is a innovative strategy to block ATP→ adenosine + phosphate axis effectively to reshape TME that benefits for PTT, PDT, and immunotherapy. By linking the gene editing and PTT/PDT, novel AIEgen, NDA-DPE, was synthesized, and we provide a strong rationale for incorporating gene therapy, phototherapy (PTT and PDT), and immunotherapy, into precision treatment against glioblastoma, paving the way for precise and effective therapies in glioblastoma.

Author contributions

^†G. L, F.M, Y.Y., and C. Y contributed equally to this work. G.L., K.M., Z.Z., B.T., and Y.D. designed the project. G.L, F.M, Y.Y, C.Y, X.L, J.X, K.X., and P.C. performed the experiments. G.L. and F.M. wrote the paper, K.M., Z.Z., B.T., and Y.D. reviewed the paper. All authors have read, wed and approved the final version of the manuscript.

Data availability

The authors declare that all the data supporting the findings of this study are available within the article and Supplementary Information files, and are also available from the corresponding authors upon reasonable request.

Competing interests

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.

Supplementary information

The online version contains supplementary material available at 10.1038/s41392-026-02699-0.