Time-staggered chemo-immunotherapy via engineered nanofiber resists postoperative dynamic immunosuppression in glioblastoma

Abstract

Glioblastoma (GBM) almost inevitably recurs after surgical resection owing to residual infiltrative tumor cells and postoperative immunosuppression. However, delayed initiation of adjuvant therapy fails to restrain rapid tumor regrowth during the early postoperative period. Here, we elucidate postoperative dynamic immune pathology characterized by early explosive tumor proliferation (Ki67⁺ > 38.5%) and progressive protumoral macrophage polarization, and provide a time-staggered chemo-immunotherapy strategy to promptly remedy the postoperative therapeutic gap. To implement the adaptive intervention, we design an engineered nanofiber that enables immediate chemotherapy followed by dynamic immune modulation tailored to the evolving postoperative pathology. Structurally, the tunable nanofiber composition enables time-staggered release of doxorubicin (DOX) to induce immunogenic cell death and BLZ945 to suppress protumoral macrophage programs and abundance. After surgical resection of orthotopic GL261 tumors, a single implantation achieves 83.9% tumor inhibition and establishes immune memory. This study explores the evolving GBM relapse process to guide engineered nanofiber design with time-staggered drug therapy against postoperative recurrence.

After surgical resection of glioblastoma, the evolution of postoperative tumor proliferation still causes severe progression even death to patients. Here this group reports an engineered nanofibers enabling the time-staggered release of doxorobucin and BLZ945 suppressing post-surgical tumor recurrence.

Introduction

Glioblastoma (GBM) is an aggressive primary intracranial neoplasm characterized by a dismal prognosis with a high recurrence rate of nearly 100% and a low median overall survival of approximately 16 months^1,2. In the clinic, maximal safe resection for GBM followed by adjuvant radiation and systemic chemotherapy remains the mainstay treatment strategy^3,4. However, the above adjuvant treatments start at least 14 days after surgery to allow for wound healing and systemic recovery, but generate a therapeutic gap⁵. Unfortunately, such a therapeutic gap creates a time window for the rapid proliferation of residual tumor satellites left behind after surgery, which is further exacerbated by a progressively immunosuppressive microenvironment^6,7.

After surgical resection of GBM, the evolution of postoperative tumor proliferation and the immune microenvironment is complex, resulting in dynamic immune pathology^8–10. In the early postoperative stage, residual tumor cells that remain in the resection margin undergo rapid proliferation, which is stimulated by a decrease in growth pressure and an increase in pro-tumorigenic cytokines¹¹. Concurrently, recent studies have revealed that the postsurgical niche undergoes a dynamic evolution of immunosuppression, marked by an acute inflammatory response immediately after surgery that gradually shifts into a profoundly immunosuppressive milieu^12–14. This transition is characterized by excessive infiltration of protumoral macrophages and upregulation of immunosuppressive molecules, which collectively promote extracellular matrix remodeling, angiogenesis, and T-cell exclusion, creating a tumor-supportive niche that accelerates tumor recurrence^15,16. Therefore, the dynamic and time-dependent immune evolution underscores the need for a therapeutic strategy that synchronizes timely tumor cell eradication with resistance to the evolving immunosuppressive environment to bridge the postoperative therapeutic gap in GBM.

Nonetheless, the standard clinical regimen for postoperative GBM is systemic chemotherapy, which faces blood-brain barrier (BBB) restriction and requires high doses and repeated administration, resulting in unsatisfactory dynamic therapeutic demands, systemic adverse effects, and impaired immune surveillance^17,18. As an alternative, the resection cavity offers a valuable opportunity for local drug delivery that bypasses the BBB and enables in situ controllable administration^19–21. To meet the requirements for postoperative GBM intervention, an elaborate engineered drug reservoir is needed to supply time-staggered drug therapy to inhibit early tumor proliferation and subsequent dynamic immunosuppressive responses and maintain sufficient mechanical flexibility to the irregular brain cavity without causing secondary damage²². However, conventional implantable systems, such as wafers, microspheres and hydrogels, have been widely explored for local drug delivery but face challenges, including uncontrollable degradation processes for multidrug release and insufficient mechanical adaptability for biomedical applications^23–25. In contrast, electrospinning technology in hierarchical or coaxial configurations can remedy such deficiencies by fabricating fibers with tunable structures and compositions and facilitating localized and sustained multidrug delivery tailored to satisfy the postoperative dynamic immune pathology of GBM^26,27.

In this work, we elucidate the dynamic immune microenvironment in a GBM surgical resection model, where antitumoral macrophages are recruited and subsequently converted into the protumoral phenotype. Critically, this phase involves the rapid proliferation of residual tumor satellites and the subsequent polarization of macrophages, which synergistically form the dynamic immune pathology (Fig.1a). Compared with the 14-day latency required for clinical chemotherapy, we provide a time-staggered chemo-immunotherapy strategy to promptly resist the progression of dynamic immunosuppression. To implement this strategy, an electrospinning technique is introduced to fabricate an engineered nanofiber that comprises a core-shell structure whose tunable composition enables dual-drug shielding and time-staggered release of DOX in the shell and BLZ945 in the core (Fig. 1b). After surgical cavity implantation, DOX is released within 3 days to induce immunogenic cell death (ICD) of residual tumor cells, followed by sequential release of BLZ945 to suppress protumoral macrophage abundance and programs (Fig. 1c), resisting the dynamic postoperative immunosuppression. Overall, this study presents a therapeutic strategy that aligns with the temporally dynamic immune pathology of GBM relapse. By integrating time-staggered drug delivery with precise local administration, we offer a clinically translatable platform for suppressing postoperative recurrence in GBM.

Fig. 1. Schematic illustration of engineered nanofiber of DOX/BLZ945-CNF to inhibit postoperative GBM recurrence.

a The process of tumor recurrence in an orthotopic GL261 glioma model. b Preparation process and characterization of the fiber structure of DOX/BLZ945-CNF (coaxial nanofiber). c Time-staggered drug release of DOX and BLZ945 from DOX/BLZ945-CNF to induce a cascade of ICD and suppression of protumoral macrophage abundance and programs to inhibit postoperative relapse. DC, dendritic cell; Treg, regulatory T cell; Tcm, central memory T cells and Tem, effector memory T cells. Created in BioRender. Mj, S. (2026) https://BioRender.com/0jals9q (applies to a–c).

Results

Rapid tumor proliferation and dynamic immunosuppression of postoperative glioblastoma

To validate the clinical relevance of macrophage infiltration and subtype polarization in the prognosis of gliomas, tumor samples were collected from 15 human glioma patients, who were classified by the World Health Organization (WHO) as having LGG (diffuse astrocytoma, DA) or HGG (anaplastic astrocytoma, AA; GBM). Immunofluorescence analysis (Fig. 2a) revealed a progressive increase in macrophages with immunosuppressive, protumoral features, reflected by increased Iba1⁺CD206⁺ and Iba1⁺TREM2⁺ populations, together with a corresponding decrease in antigen-presentation-associated macrophage features, indicated by reduced Iba1⁺HLA-DR⁺ and Iba1⁺CD86⁺ populations, with increasing glioma grade. The results indicated that the degree of immunosuppression in the tumor environment was dependent on the degree of malignancy. Compared with LGG, GBM patients presented 5.11-fold and 7.25-fold increases in the number of protumoral macrophages, quantified as Iba1⁺CD206⁺ and Iba1⁺TREM2⁺ cells, respectively (Supplementary Fig. 1). To further augment the sample size for more comprehensive analysis, immune cell infiltration levels were further analyzed by a scRNA-seq dataset (GSE182109) from the Gene Expression Omnibus (GEO) database. As shown in Fig. 2b, c, significant differences in immune cell infiltration were observed between the LGG and HGG groups, of which the HGG groups exhibited reduced enrichment of antitumoral macrophage signatures and increased enrichment of protumoral macrophage and Treg signatures. Dendritic cell populations exhibited subset-specific changes, characterized by a reduction in LAMP3⁺ dendritic cells, a migratory and mature subset associated with antigen presentation. These results indicate a more serious immunosuppressive microenvironment in HGG. In addition, based on data from 338 glioma patients, survival analysis showed that higher enrichment of the protumoral macrophage gene signature was associated with worse clinical outcomes (Fig. 2d).

Fig. 2. The process of rapid tumor regrowth and dynamic immunosuppression in postoperative GBM and the preparation of DOX/BLZ945-CNF.

a Immunofluorescence staining of macrophage populations in human glioma specimens, showing macrophages enriched for antigen-presentation-associated, antitumoral features (Iba1/HLA-DR/CD86) and immunosuppressive, protumoral features (Iba1/CD206/TREM2). b UMAP plots of cells from HGG and LGG brain tumor datasets colored by immune cell type. c Stacked bar chart showing the relative frequencies of different immune cell subsets in the HGG and LGG brain tumor datasets. d Kaplan-Meier disease-free survival analysis stratified by the protumoral macrophage program (SPP1/APOE/TREM2/GPNMB). e Schematic illustration of the proliferating tumor cells and macrophage functional evolution after surgery in GBM-bearing mice. f Ki67 immunofluorescence staining of tumor slices and quantitative analysis of the relative percentage (Ki67⁺) on different postoperative days. g Variation curve of macrophage populations across postoperative days, showing antigen-presentation-associated macrophages (MHC II⁺/CD80⁺) and protumoral macrophages (CD206⁺/TREM2⁺), quantified by flow cytometry. All four macrophage-associated markers (CD80, MHC II, CD206, and TREM2) showed statistically significant changes on postoperative day 2 compared with day 1 (P < 0.001). h ELISA for IL-4 and IL-13 expression in brain tumor tissue on different postoperative days. i SEM and STED images of coaxial electrospinning nanofiber. Scale bars: 2 μm (SEM) and 500 nm (STED). j Schematic illustration of time-staggered drug release properties. k Release profiles of DOX from the DOX/BLZ945-CNF with different ratios of PVA and SF. l Release profiles of BLZ945 from the DOX/BLZ945-CNF with different grades of PLGA. m The release profiles of DOX and BLZ945 from DOX/BLZ945-CNFs. The data are presented as the mean ± S.D (n = 5 independent samples in (k–m); n = 5 mice in f–h). Survival analysis was performed by the log-rank (Mantel-Cox) test (d). Statistical significance was calculated by one-way ANOVA with Tukey’s post hoc test (f,h). Three independent experiments were performed with similar results and representative results are shown. Source data are provided as a Source Data file. Created in BioRender. Mj, S. (2026) https://BioRender.com/8syvqg6 (applies to e).

Although our analysis of patient samples and single-cell transcriptomic data suggests that increased protumoral macrophage-associated features are associated with poor prognosis in GBM, they failed to characterize the dynamic changes in macrophages after surgery. Hence, we established a GBM surgical resection model. GL261 cells were intracranially injected into C57/BL6 mice to generate an orthotopic GL261 tumor model, followed by surgery to establish a GBM surgical resection model. According to the hematoxylin-eosin (HE) staining results shown in Supplementary Fig. 2, GL261 tumors became more invasive and infiltrated into the brain parenchyma, with an increase in tumor cell islands observed at 14 days post-implantation, indicating an appropriate time window for surgery. Moreover, luciferase-transfected GL261 (Luc-GL261) cell-bearing mice were also established to evaluate surgical debulking efficacy. As detected by bioluminescence signal (BLI) changes in Luc-GL261 cells before and after surgery in Supplementary Fig. 3a, a tumor resection efficiency of approximately 91% was obtained, confirming the establishment of the GBM surgical resection model. However, HE staining of the surgical cavity revealed that residual tumor satellites remained in the cavity and even infiltrated the normal brain parenchyma after surgery, indicating that the seeds were planted for GBM recurrence (Supplementary Fig. 3b).

After the GBM resection model was established, we further elucidated the dynamic immune pathology underlying GBM recurrence by analyzing tumor cell proliferation and macrophage functional states using immunofluorescence and flow cytometry (Fig. 2e). As shown in Fig. 2f, the percentage of proliferating tumor cells (Ki67⁺) increased to a peak of 30% at postoperative Day 3, followed by a sharp decrease. Hence, DOX administration within 3 days might be the optimal choice in light of both its efficacy and safety. After surgery, the infiltrating macrophage compartment was enriched for antigen-presentation-associated features, as indicated by a predominance of MHC II⁺ and CD80⁺ cells ( > 40%), consistent with an acute postoperative inflammatory milieu (Fig. 2g and Supplementary Fig. 4). However, accompanied by the wound healing process, the anti-inflammatory cytokines interleukin-4 (IL-4) and interleukin-13 (IL-13) coincided with a gradual attenuation of antigen-presentation-associated macrophage features and the emergence of protumoral macrophage program (Fig. 2h). As a result, the population of CD206⁺ protumoral macrophage surpassed CD80⁺ antitumoral macrophage and the number of CD206⁺ and TREM2⁺ macrophage gradually increased beginning on the second day. These results define the postoperative dynamic immune pathology as a two-phase process: an early phase marked by rapid residual tumor cell proliferation and enrichment of antigen-presentation-associated, antitumoral macrophage features, followed by a later phase characterized by the emergence of immunosuppressive, protumoral macrophage programs. This comprehensive temporal characterization provides essential guidance for designing a time-staggered therapeutic strategy to counter dynamic immunosuppression and prevent GBM recurrence.

Design, preparation, and characterization of engineered nanofibers

To match the postoperative dynamic immune pathology, engineered nanofibers should be designed for flexible drug loading and time-staggered drug release. For this purpose, the coaxial electrospinning technique was employed to prepare an engineered nanofiber comprising a core-shell structure with polyvinyl alcohol/silk fibroin (PVA/SF) as the shell and polylactic acid-glycolic acid (PLGA) as the core for hierarchical DOX and BLZ945 loading, termed DOX/BLZ945-CNF. For visualization by scanning electron microscopy (SEM) and stimulated emission depletion (STED) microscopy, the fluorescent dye DiO was used to replace BLZ945, thereby preparing an engineered nanofiber (DOX/DiO-CNF). The shells were prepared with different ratios of PVA/SF (6:1, 5:1, 4:1, 3:1, 2:1 and 1:1), termed DOX/DiO-CNF_6:1, DOX/DiO-CNF_5:1, DOX/DiO-CNF_4:1, DOX/DiO-CNF_3:1, DOX/DiO-CNF_2:1, and DOX/DiO-CNF_1:1, respectively. As shown in Fig. 2i, smooth surfaces and uniform diameters of the nanofiber were observed via SEM. When the proportion of SF increased, the average diameter of the nanofiber decreased due to the decreased viscosity of the spinning solution. As depicted in the STED images, the green fluorescent signals of DiO were uniformly distributed inside the DOX fluorescence along the whole fiber, which indicated that both cargos were successfully wrapped inside the fiber matrix and formed a uniform core-shell structure.

The time-staggered drug release of DOX during the initial three days and the rapid proliferation of residual tumor cells and remodeling of the macrophage phenotype by BLZ945 from the second day were desirable for postoperative GBM therapy (Fig. 2j). For the optimal DOX release profile, DOX/BLZ945-CNF with different ratios of PVA/SF was prepared to elucidate the relationship between the degree of DOX release and the ratio of PVA/SF. As shown in Fig. 2k, DOX release accelerated with increasing SF addition ratio due to the decreased diameter of the nanofiber and changes in the PVA/SF properties. Moreover, increasing the SF content in the nanofiber resulted in a higher dissolution rate and a lower swelling ratio, collectively contributing to an accelerated DOX release profile (Supplementary Fig. 5). When the ratio of PVA/SF was 2:1, more than 80% of the DOX was released within 3 days for timely inhibition of tumor proliferation during GBM recurrence. Hence, through rational regulation of the SF/PVA composition, DOX/BLZ945-CNF_2:1 presented desirable DOX release profiles.

Moreover, the release of BLZ945 was controlled by PLGA, which is available in different grades and offers tunable drug-release kinetics. Hence, various nanofibers were prepared using (PVA/SF)_2:1 as the shell and PLGA grades with different LA/GA ratios (85/15, 75/25, and 50/50) as the core. As the LA content increased, the initial release time was prolonged due to the higher hydrophobicity and slower polymer degradation (Fig. 2l). When the PLGA grade was 85/15 (LA/GA), BLZ945 was gradually released from the second day. Hence, the terminal DOX/BLZ945-CNF was optimized with a PVA/SF shell (2:1) and a PLGA core with a grade of 85/15. As shown in Fig. 2m, the terminal DOX/BLZ945-CNF nanofiber promoted the time-staggered drug release of DOX within 3 days, and BLZ945 from the second day to match the dynamic immune pathology of residual tumor cell proliferation and macrophage phenotype transition. Together, batch-to-batch reproducibility and sustained chemical stability of DOX and BLZ945 support the robustness and translational potential of the nanofiber platform (Supplementary Fig. 6, 7).

DOX/BLZ945-CNF nanofiber induced dynamic immune response in vitro

To match the dynamic immune pathology, the engineered nanofiber DOX/BLZ945-CNF was designed with time-staggered release of DOX and BLZ945 to induce a dynamic immune response in ICD and modulate macrophage functional programs. To better mimic the interaction between the nanofiber and tumor cells in vivo, GL261 cells were incubated with the nanofiber by a Transwell co-culture system (Fig. 3a). Before investigation, the drug-free nanofiber (blank CNF) presented negligible cytotoxicity toward GL261, BV-2, and SH-SY5Y cells (Supplementary Fig. 8). After drug loading, the IC₅₀ value of DOX/BLZ945-CNF against GL261 cells was 0.227 µM (Supplementary Fig. 9). In contrast, primary T cells, dendritic cells, and bone-marrow-derived macrophages (BMDMs) displayed 3.22-5.04-fold higher IC₅₀ values, maintaining >80% viability at 0.5 µM DOX. As shown in Fig. 3b and Supplementary Fig. 10a, an obvious green signal of the translocation of calreticulin (CRT) and nearly 70% of CRT-positive dying cancer cells were observed in different DOX loading preparations, including free DOX, DOX-CNF, and DOX/BLZ945-CNF. The results indicated that DOX in the nanofiber could effectively enhance the exposure of the “eat me” signal of CRT to promote the phagocytosis of dying cancer cells by macrophage. In contrast to that in the blank CNF-treated group, the concentration of high-mobility group box 1 (HMGB1) in the DOX/BLZ945-CNF group increased from 0.15 ± 0.06 ng/mL to 4.15 ± 0.39 ng/mL, which could be recognized as a “danger” signal to promote the maturation of antigen-presenting cells (APCs) (Supplementary Fig. 10b). The results above demonstrated that the ability of DOX in the nanofiber was similar to that of free DOX to induce ICD in GL261 cells as an “in situ vaccine” for priming an anti-tumor immunity.

Fig. 3. Dynamic immune response and mechanism of DOX/BLZ945-CNF in vitro.

a Schematic representation of ICD induced by different DOX preparations. b Representative immunofluorescence images of CRT in GL261 cells after treatment with different DOX formulations. The cell nuclei and CRT were stained with DAPI and a FITC-conjugated anti-CRT antibody, respectively. Scale bar = 20 μm. c Schematic representation of the phenotype and phagocytosis assay in macrophage treated with different BLZ945 formulations. d Quantification of CD80⁺, CD206⁺, MHC II⁺, and TREM2⁺ macrophages across different treatment groups. e Representative flow cytometric plots (left) showing the level of phagocytosis of DiO-labeled GL261 debris by macrophage and the corresponding quantitative analysis (right). f Schematic diagram of protein interactions associated with the CSF-1R signaling pathway in macrophage. g WB analysis of pCSF-1R, tCSF-1R and related downstream signaling pathways of macrophage treated with different formulations. Red hatched bars indicate the relative normalized band intensity used for densitometric quantification. h Heatmap of IL-10, TGF-β, and TNF-α secretion profiles in macrophage after treatment with different formulations determined by ELISA. i Schematic representation of DC maturation and phagocytosis assays. DiO-labeled GL261 cells were incubated with DOX to induce ICD, followed by collection of GL261 debris in the lower chamber. Meanwhile, macrophage was incubated with the resulting nanofiber for 24 h, and then collected to the upper chamber. j Representative flow cytometric plots (left) of DC maturation (CD80⁺CD86⁺) and the corresponding quantitative analysis (right). k Representative flow cytometric plots (left) showing the level of phagocytosis of DiO-labeled GL261 debris by DCs and the corresponding quantitative analysis (right). The data are presented as mean ± S.D (n = 3 independent samples in d,e,h,j,k). Statistical significance was calculated by one-way ANOVA with Tukey’s post hoc test (d,e,j,k). Three independent experiments were performed with similar results and representative results are shown. Source data are provided as a Source Data file.

Protumoral macrophage was established by coincubation of macrophage with the cytokines IL-4 (10 ng/mL) and IL-13 (10 ng/mL), as evidenced by the increase in the expression of CD206 from 20.2 to 49.8% (Supplementary Fig. 11). To further clarify the crosstalk between the “in situ vaccine” produced by DOX and protumoral macrophage in the immune microenvironment, protumoral macrophage was incubated with DOX-induced DiO-labeled GL261 debris, and the resulting nanofiber that was collected after DOX release (Fig. 3c). Flow cytometry analysis demonstrated that treatment with DOX/BLZ945-CNF significantly altered macrophage marker expression profiles (Fig. 3d). Compared with PBS and blank CNF controls, DOX/BLZ945-CNF treatment resulted in an approximately 1.5-fold increase in the proportion of CD80⁺ macrophages, accompanied by a ~ 3.5-fold reduction in CD206⁺ macrophages, leading to a pronounced elevation of the CD80⁺/CD206⁺ ratio. In parallel, DOX/BLZ945-CNF increased MHC II⁺ macrophages while reducing the proportion of TREM2⁺ macrophages, consistent with suppression of protumoral macrophage programs and enhancement of antigen-presentation-associated features. This shift is attributed to the combined effects of BLZ945-mediated inhibition of CSF1R signaling and exposure to DOX-induced immunogenic tumor debris, which together attenuate protumoral macrophage programs while promoting antigen-presentation-associated features. Moreover, both the BLZ945-loaded nanofiber (BLZ945-CNF) and the dual-drug-loaded nanofiber (DOX/BLZ945-CNF) presented similar macrophage-modulating abilities as free BLZ945. Following treatment with different BLZ945 preparations, the phagocytic capacity of macrophages increased approximately 4.9-fold (Fig. 3e), which is consistent with the observed suppression of protumoral macrophage programs and enhanced antigen-presentation-associated features.

From the mechanism, macrophage colony-stimulating factor (M-CSF), which is secreted by cancer cells (Supplementary Fig. 12), can bind to the tyrosine kinase CSF1R on macrophages, thereby promoting protumoral macrophage programs at the expense of antitumoral immune functions²⁸. As shown in Fig. 3f, BLZ945 effectively inhibited the phosphorylation of CSF1R into pCSF1R, which could regulate the cytokine signaling 3 (SOCS3)/cytokine signaling 1 (SOCS1) ratio via the protein kinase B (AKT)-mTOR-p70S6K pathway. SOCS3 is associated with antitumor activity, whereas SOCS1-expressing macrophages support tumor survival. As shown in Fig. 3g, BLZ945 and BLZ945-loaded nanofiber decreased the pCSF1R/total CSF1R (tCSF1R) ratio and inhibited AKT activity. Moreover, by regulating the AKT-mTOR-p70S6K pathway, BLZ945 and BLZ945-loaded nanofiber improved SOCS3-expressing macrophage to achieve robust antitumor activity. As a result, DOX/BLZ945-CNF effectively reduced the levels of the anti-inflammatory cytokines IL-10 and TGF-β by 1.57-fold and 1.85-fold, respectively, in the control group (Fig. 3h). In contrast, the pro-inflammatory cytokine TNF-α was substantially increased in the BLZ945 groups, which could promote further DC maturation and T-cell activation for antitumor immunity.

Hence, to further investigate this dynamic interaction, BLZ945-treated macrophages were cocultured with DCs in a Transwell insert (Fig. 3i). Thereafter, DiO-labeled GL261 debris was added to evaluate the phagocytic ability of the DCs. As shown in Fig. 3j, the DOX/BLZ945-CNF group provided 27.1% DC maturation, which was 3.80-fold greater than that of the PBS group. The percentage of DCs that phagocytized DiO-labeled GL261 debris in the DOX/BLZ945-CNF group was 5.38-fold greater than that in the PBS group (Fig. 3k), benefiting from the pro-inflammatory cytokines secreted by polarized antitumoral macrophages. In a macrophage-conditioned DC-T cell-GL261 co-culture assay, increased LDH release indicated that antitumoral macrophage-associated cues induced by DOX/BLZ945-CNF enhanced DC-mediated T-cell cytotoxicity (Supplementary Fig. 13). These findings revealed that BLZ945-mediated modulation of macrophage programs not only improved the capability for antigen capture, pro-inflammatory cytokine release but also promoted DC maturation and downstream T-cell-mediated antitumor immune responses.

Engineered DOX/BLZ945-CNF inhibited GBM regrowth with excellent safety in vivo

For initial investigations of biocompatibility, the nanofiber was implanted onto the surface of the incisional wound in the mouse brain, and the infiltration of astrocytes and microglia was observed. As shown in Fig. 4a and Supplementary Fig. 14, no obvious difference in the number of migrated astrocytes or microglia was observed near the incision site between the sham group and the CNF-implanted group, indicating that CNF provided safety profiles for brain implantation with no obvious host tissue response. After implantation, the nanofiber formed a soft and porous scaffold with a compliant mechanical profile, exhibiting a Young’s modulus of 2.62 kPa, which falls within the stiffness range of brain parenchyma (Supplementary Fig. 15). In vivo degradation analysis further demonstrated a progressive, time-dependent degradation of the PVA/SF-based nanofiber, with no detectable residual bulk fragments at later time points. Consistently, macroscopic and histological evaluation after tissue healing showed restoration of normal brain morphology without dense fibrotic encapsulation or abnormal tissue overgrowth, indicating that nanofiber degradation proceeds in concert with tissue repair and does not induce excessive scarring (Supplementary Fig. 16).

Fig. 4. Suppression of tumor recurrence by the engineered DOX/BLZ945-CNF in vivo.

a Quantification of astrocytes (GFAP⁺) and microglia (Iba1⁺) at different time points in the sham (red) and blank CNF implantation (blue) groups. Lines: median, boxes: 25–75th percentiles, whiskers: min to max. b Surgical debulking of tumors and cavity implantation in GL261 tumor-bearing mice on Day 14 after inoculation. The white arrow in the image indicates the edges of the GL261 tumors formed in the mice. c Experimental process: C57/BL6 mice were inoculated intracranially with Luc-GL261 cells, and the tumors were removed 14 days later prior to the implantation of the formulations in situ. The therapeutic efficacy was evaluated during and after the 18-day administration period. d In vivo bioluminescence images of GL261 tumor-bearing mice after different postoperative treatments on Days 0, 9, and 18. Mice treated with saline were used as the control group. e In vivo T2W-MRI images of GL261 tumor-bearing mice after different postoperative treatments. The white dashed circles point to the tumor site. The percentage variations in tumor volume were calculated. Scale bar = 2 mm. f HE staining of brain tissue sections obtained from GL261 tumor-bearing mice after different treatments. The numbers represent the ratio of tumor areas/brain tissue areas. Scale bar = 1 mm. The data are presented as mean ± S.D (n = 5 mice in a; n = 3 mice in d). Statistical significance was calculated by two-tailed unpaired t-test (a). Three independent experiments were performed with similar results and representative results are shown. Source data are provided as a Source Data file. Created in BioRender. Mj, S. (2026) https://BioRender.com/cdguyti (applies to c).

At 14 days post-implantation, tumor resection was carried out to establish a GBM resection mouse model, and 5 mm diameter of the nanofiber containing DOX (0.02 mg) and BLZ945 (0.1 mg) was implanted into the cavity under the guidance of a stereoscope (Fig. 4b). No neurological impairment or low mortality rate after surgery was observed, indicating that function-preserving radical resection of GBM was achieved. To further investigate the impact of time-staggered drug release and desirable DOX release periods on anti-GBM efficacy, nanofibers that possessed different DOX release periods were prepared and termed DOX/BLZ945-CNF_S (DOX released in 1 day), DOX/BLZ945-CNF (DOX released in 3 days), and DOX/BLZ945-CNF_L (DOX released in 12 days). In accordance with the therapeutic protocols, the growth of residual tumor cells was monitored through bioluminescence (BLI) signals from Luc-GL261 cells (Fig. 4c). As shown in Fig. 4d, irrepressible tumor growth was observed in both the control and blank CNF groups, highlighting the necessity of timely treatment. However, no increased but slightly decreased BLI signals were observed in the DOX/BLZ945-CNF-treated group. In contrast, both DOX/BLZ945-CNF_S and DOX/BLZ945-CNF_L slightly increased the BLI signals by 2.47-fold and 1.61-fold, respectively, from Day 0 to Day 18 (Supplementary Fig. 17). The results further revealed that better anti-tumor efficacy was achieved by appropriate release profiles of DOX, which could suppress the rapid proliferation of tumor cells in a timely manner without obvious damage.

On the 18th day post-administration, T2-weighted magnetic resonance imaging (T2W-MRI) of GL261 tumor-bearing mice brain was performed. As shown in Fig. 4e, compared with control treatment, DOX/BLZ945-CNF treatment substantially inhibited tumor recurrence, with a tumor volume of only 10.41%. Moreover, DOX/BLZ945-CNF_S and DOX/BLZ945-CNF_L inhibited tumor regrowth to different extents, providing 35.52% and 21.34% of the tumor volume of the control group, respectively. As shown by the HE staining results in Fig. 4f, extensive invasion of tumor cells was observed in the control group, indicating that residual tumor cells could rapidly proliferate for recurrence. In addition, the ratios of tumor areas/brain tissue areas in the control, blank CNF, BLZ945-CNF, DOX-CNF, DOX/BLZ945-CNF_S, DOX/BLZ945-CNF_L, and DOX/BLZ945-CNF groups were 0.45, 0.46, 0.32, 0.28, 0.23, 0.08, and 0.04, respectively. The significant anti-recurrence capability of DOX/BLZ945-CNF was attributed to the time-staggered drug release profile, which could induce a dynamic immune response to resist postoperative immunosuppression.

In addition, all the mice in the control group died within 34 days, with a median survival time of 25 days. In contrast, the mice that received DOX/BLZ945-CNF treatment presented a nearly 100% survival rate until Day 57, with the median survival time not reached during the observation period, and showed no obvious body weight loss, revealing the satisfactory anti-recurrence efficacy and safety profiles of DOX/BLZ945-CNF (Supplementary Fig. 18). Moreover, no significant damage to major organs or alterations in serum biochemical indicators were observed (Supplementary Fig. 19, 20), which indicated that intracranial implantation of the nanofiber could provide robust antitumor efficacy with desirable safety profiles.

Engineered DOX/BLZ945-CNF boosted dynamic immune response in vivo

After surgical debulking of GBM and implantation of DOX/BLZ945-CNF, time-staggered DOX and BLZ945 release was designed to increase the dynamic response of “in situ vaccine” generation and remodel macrophage-associated immune programs, matching the dynamic immune pathology (Fig. 5a). To investigate the dynamic immunomodulation capability, the brain tumor tissues surrounding the resection cavity were harvested for further analysis. As shown in Fig. 5b and Supplementary Fig. 21, a significant increase in immune cell infiltration of 26.62% was observed in the DOX/BLZ945-CNF group. In contrast, decreased lymphocyte enrichment of 12.97% was observed in the DOX/BLZ945-CNF_L group, which indicated that prolonged DOX release might cause potential damage to immune cells in the sacrificed immune response. Moreover, DOX/BLZ945-CNF_S moderately increased immune cell infiltration, which might be due to the tumor cell debris produced by limited DOX release. The results further confirmed that the optimized DOX release schedule was crucial for both inhibiting residual tumor cell proliferation and recruiting lymphocytes for robust immune responses.

Fig. 5. Dynamic immune modulation efficacy and mechanism of DOX/BLZ945-CNF in vivo.

a Schematic illustration of the in situ vaccine induced by the released DOX and the macrophage phenotype remolded by the released BLZ945 to reverse impaired immune surveillance and increase antitumor immunity. b The percentages of lymphocytes (CD45⁺) in brain tumor tissues were analyzed by flow cytometry after treatment with different formulations. c-f Quantification of CD80⁺, CD206⁺, MHC II⁺, and TREM2⁺ macrophages across different treatment groups, expressed as the percentage of marker-positive cells among total live cells. g WB analysis of pCSF1R, tCSF1R and related downstream signaling pathways in the brain tumor tissues of the mice treated with different formulations. Red hatched bars indicate the relative normalized band intensity used for densitometric quantification. h Heatmap of the profiles of IL-10, TGF-β, and TNF-α secretion in the brain tumor tissues of the mice treated with different formulations. i The percentages of DC maturation (CD80⁺CD86⁺) in the brain tumor tissues of the mice treated with different formulations. j The percentages of CD4⁺ T cells (CD3⁺CD4⁺) in the brain tumor tissues of the mice treated with different formulations. k Percentages of CD8⁺ T cells (CD3⁺CD8⁺) in the brain tumor tissues of the mice treated with different formulations. l The percentages of Tregs (CD25⁺FoxP3⁺) in the brain tumor tissues of the mice treated with different formulations. The data are presented as mean ± S.D (n = 5 mice in b–f,i–l; n = 3 mice in h). Statistical significance was calculated by one-way ANOVA with Tukey’s post hoc test (b-f, i-l). Three independent experiments were performed with similar results and representative results are shown. Source data are provided as a Source Data file.

In addition to increased immune cell infiltration, the regulation of protumoral macrophages into antitumoral macrophages by the subsequent release of BLZ945 is crucial for robust antitumor immunity and immune memory. As shown in Fig. 5c–f and Supplementary Figs. 22, 23, multiparametric flow cytometry was used to analyze macrophage subsets based on their identity, abundance, and function-associated programs. DOX/BLZ945-CNF treatment did not significantly alter the abundance of tissue-resident microglia, indicating that resident myeloid cells were largely preserved. This partial preservation may be attributed to survival signals within the brain tumor microenvironment that sustain microglia despite CSF1R inhibition, thereby limiting their complete depletion. In contrast, a significant reduction was observed in Ly6C⁺CCR2⁺ infiltrating monocytes, consistent with sustained inhibition of CSF1R-dependent monocyte recruitment and maintenance within the tumor microenvironment. Differentiated macrophages exhibited only a modest decrease, in line with their partial dependence on continuous monocyte influx. Beyond changes in cell abundance, DOX/BLZ945-CNF induced marked functional remodeling of the remaining myeloid populations. This remodeling was characterized by enhanced antigen-presentation–associated features, as evidenced by increased expression of MHC II and CD80, together with attenuation of protumoral and immunosuppressive programs, as reflected by reduced expression of TREM2 and CD206. Notably, tissue-resident microglia similarly exhibited decreased TREM2 expression and increased MHC II expression, indicating a shift toward antigen-presentation-associated functional states. Collectively, these results demonstrate that DOX/BLZ945-CNF reshapes the myeloid landscape through selective reduction of Ly6C⁺CCR2⁺ infiltrating monocytes and functional reprogramming of residual macrophages and microglia. WB results further confirmed that the phosphorylation of CSF1R was effectively inhibited in the DOX/BLZ945-CNF group (Fig. 5g). Compared with that in the DOX/BLZ945-CNF group, a greater pCSF1R/tCSF1R ratio was observed in both the DOX/BLZ945-CNF_S and DOX/BLZ945-CNF_L groups even when the same dose of BLZ945 was loaded. This difference might be due to the sacrificed tumor inhibition effect of DOX/BLZ945-CNF_S and DOX/BLZ945-CNF_L. Notably, the SOCS3/SOCS1 ratio was markedly elevated in the DOX/BLZ945-CNF group, further supporting a transcriptional shift toward inflammatory macrophage programs. Consistent with macrophage-associated functional modulation, increased pro-inflammatory cytokines (TNF-α) and decreased anti-inflammatory cytokines (IL-10 and TGF-β) were observed in the DOX/BLZ945-CNF group (Fig. 5h), which might contribute to facilitating DC maturation and enhancing T-cell activation and proliferation. And DOX/BLZ945-CNF treatment significantly reduced IL-4 and IL-13 levels in postoperative brain tissues, indicating attenuation of the immunosuppressive cytokine milieu (Supplementary Fig. 24). Collectively, these results demonstrate that DOX/BLZ945-CNF enables sustained CSF1R inhibition to reprogram protumoral macrophage states toward antitumoral immune functions, thereby reinforcing durable antitumor immunity.

The dynamic generation of an “in situ vaccine” and reduction in anti-inflammatory cytokines could increase T-cell activation, proliferation, and cytotoxic activity, fostering a robust anti-tumor immune response. Hence, the contents of matured DCs, CD8⁺ cytotoxic T cells, CD4⁺ helper T cells, and Tregs within the microenvironment were further detected. Compared with the control group, the DOX/BLZ945-CNF group presented a substantial 2.94-fold increase in the DC maturation ratio (Fig. 5i and Supplementary Fig. 25). Furthermore, the DOX/BLZ945-CNF group presented elevated percentages of helper T cells and cytotoxic T cells (Fig. 5j, k, and Supplementary Fig. 26), with increases of 3.77-fold and 4.31-fold, respectively, compared with those of the control group. Moreover, the ratio of immunosuppressive Tregs was significantly decreased by approximately 1.80-2.77-fold in the DOX/BLZ945-CNF group (Fig. 5l and Supplementary Fig. 27). In summary, the results indicated that DOX/BLZ945-CNF treatment not only enriched helper T cells and cytotoxic T cells to kill tumor cells but also decreased the number of immunosuppressive Tregs to remodel the immunosuppressive microenvironment. Functionally, in vivo depletion of either CD4⁺ or CD8⁺ T cells significantly attenuated, and combined depletion nearly abolished, the tumor control and survival benefit conferred by DOX/BLZ945-CNF, demonstrating a T cell-dependent antitumor mechanism (Supplementary Fig. 28). By aligning with the evolving immune landscape during the postoperative period, this time-staggered chemo-immunotherapy orchestrated a dynamic and sustained antitumor immune response, ultimately preventing GBM relapse.

To further validate the generalizability of therapeutic efficacy, an additional immunocompetent GBM model was established using CT-2A cells (Supplementary Fig. 29). Consistent with the GL261 model, postoperative treatment with DOX/BLZ945-CNF significantly suppressed tumor progression, as evidenced by reduced bioluminescence signals, decreased tumor volume on MRI, and diminished tumor burden on HE staining. DOX/BLZ945-CNF treatment also prolonged overall survival without inducing body weight loss. Immune profiling revealed increased infiltration of CD4⁺ and CD8⁺ T cells, accompanied by a shift toward antitumoral macrophage features, with increased CD80⁺ and MHC II⁺ populations and reduced CD206⁺ and TREM2⁺ macrophages (Supplementary Fig. 30). These results confirm that the therapeutic and immunomodulatory effects of DOX/BLZ945-CNF are reproducible across distinct immunocompetent GBM models.

Cavity implantation induced immune memory to inhibit GBM relapse

Robust antitumor immunity induced by the dynamic immune response of the nanofiber is crucial for the differentiation of T cells into central memory T (Tcm) cells and effector memory T (Tem) cells, activating memory immunity against tumor recurrence. Hence, to assess the immunological memory effect, mice that survived for 26 days after primary tumor resection and nanofiber implantation were rechallenged with Luc-GL261 cells in the opposite hemisphere (Fig. 6a). The mice in all the groups received no further treatment. Tumor growth progression, survival time, and immune-related indicators were recorded and examined. As shown in Fig. 6b, almost all the mice in the saline, blank CNF, BLZ945-CNF, and DOX-CNF groups presented obvious recurrence of the tumor in the contralateral brain and died within 36 days, with median survival times ranging from 29 to 34 days. In contrast, negligible recurrence of secondary tumors was observed in the mice in the DOX/BLZ945-CNF group. Moreover, the long-term survival rate of DOX/BLZ945-CNF-treated mice was 80% until Day 50 after primary tumor resection. In contrast, no mice in the DOX/BLZ945-CNF_S or DOX/BLZ945-CNF_L groups survived to the same time point, with median survival times of approximately 41 days (Supplementary Fig. 31). Brain tumor tissues from the surviving mice were harvested, and HE staining was conducted on the tumor-bearing brain sections, followed by imaging (Supplementary Fig. 32). The results demonstrated that the mice treated with DOX/BLZ945-CNF exhibited observable secondary tumor regression, with the tumor areas/brain tissue areas decreasing 70.2-fold compared with those in the control group, which aligns with the bioluminescence imaging results.

Fig. 6. DOX/BLZ945-CNF inhibited postoperative GBM recurrence after tumor rechallenge.

a Schematic illustration of Luc-GL261 cells rechallenge in mice subjected to primary tumor resection followed by treatment with different formulations. b In vivo bioluminescence images of the mice rechallenged with Luc-GL261 cells from different treatment groups. c Schematic illustration of the collection and analysis of Tcm and Tem cells from the collected spleen and peripheral blood. d Representative flow cytometric plots of the spleen and peripheral blood of Tcm (CD44⁺CD62L⁺) and Tem (CD44⁺CD62L^-) cells gated on CD3⁺CD8⁺ T cells after rechallenging the mice with Luc-GL261 cells. e Corresponding quantitative analysis of Tem and Tcm cells after rechallenging the mice with Luc-GL261 cells. f Flowchart of lymphocyte activity and heatmap of IFN-γ secretion profiles in lymphocytes from the collected spleens after they were restimulated with Luc-GL261 debris and co-incubated with (i) saline, (ii) protumoral macrophage, (iii) the resulting DOX/BLZ945-CNF-remodeled macrophage for 24 h. The data are presented as the mean ± S.D (n = 3 mice in b, f; n = 5 mice in d,e). Statistical significance was calculated by one-way ANOVA with Tukey’s post hoc test (e). Source data are provided as a Source Data file. Created in BioRender. Mj, S. (2026) https://BioRender.com/moocd00 (applies to a–c and f).

Moreover, on the 10th day following rechallenge, the spleen and peripheral blood of the mice were collected to assess the memory response of T cells with Tcm and Tem cells quantified by flow cytometry (Fig. 6c and Supplementary Fig. 33). As shown in Fig. 6d, e, obvious increases in both the percentages of Tcm and Tem cells in the spleen and blood were observed in the DOX/BLZ945-CNF group. The increased number of Tem cells could provide an immediate response, rapidly inhibiting tumor cell proliferation. Concurrently, Tcm cells proliferate and differentiate into new effector T cells and memory T cells to ensure a sustained and robust immune response, which accounts for the significant resistance to tumor relapse. In contrast, compared with those in the control, blank CNF, BLZ945-CNF, and DOX-CNF groups, the Tcm and Tem cells in the spleen and blood of the DOX/BLZ945-CNF_S and DOX/BLZ945-CNF_L groups increased by 1.11-7.62-fold. The results demonstrated that both the combination of DOX and BLZ945 and the time-staggered release profile aligned with the dynamic immune response were essential for enhancing a robust antitumor memory capable of preventing tumor relapse. Additionally, this enhanced memory response is unlikely due to nonspecific immunostimulation from nanofiber degradation, as no similar expansion of memory T cells was observed in the blank CNF group.

The reactivation and proliferation of Tem cells into cytotoxic T cells are key in immunological memory, of which antigen restimulation is a prerequisite and the immunosuppressive tumor microenvironment is a hindrance. Hence, to further illustrate the DOX/BLZ945-CNF-induced immunological memory, the lymphocytes extracted from the spleens of the mice in all groups were restimulated with GL261 debris in vitro. Thereafter, saline, protumoral macrophage and the remodeled macrophage were introduced for further incubation (Fig. 6f). When the Tem cells of lymphocytes are reactivated and proliferate into cytotoxic T cells, they can secrete interferon-γ (IFN-γ) to indicate Tem cell reactivation. The results revealed that lymphocytes extracted from DOX/BLZ945-CNF-treated mice secreted more IFN-γ than those in the other groups subjected to the same treatment in vitro. Moreover, when comparing IFN-γ secretion by lymphocytes from the same mice, lymphocytes co-incubated with DOX/BLZ945-CNF-remodeled macrophage presented the highest IFN-γ level of 1157.91 ng/mL, which was 1.12-fold and 1.20-fold greater than that of lymphocytes co-incubated with saline and protumoral macrophage, respectively. These results indicate that the time-staggered chemo-immunotherapy delivered by DOX/BLZ945-CNF can establish robust immunological memory, which can be effectively reactivated by GL261 tumor debris and immune-promoting macrophages.

Discussion

As the most aggressive intracranial neoplasms, GBM presents critical clinical challenges including a high recurrence rate, a dismal median survival, and limited therapeutic options. The standard initial clinical intervention for GBM patients involves maximal surgical resection followed by at least a two-week recovery period prior to initiating adjuvant chemotherapy. Owing to the infiltrative nature of GBM and functional brain constraints, disseminated residual tumor satellites persist and rapidly proliferate during the treatment window. Worse still, the postoperative microenvironment promotes macrophage infiltration and progressive protumoral polarization, creating an immunosuppressive niche to drive GBM recurrence. The dynamic interplay between residual tumor cells and the progressively protumoral macrophages underlies the postoperative immunosuppressive environment driving GBM relapse. Hence, we established a GBM resection model to clarify the dynamic immune pathology characterized by early explosive tumor proliferation (Ki67⁺ > 38.5%) during the initial three days and progressive protumoral macrophage infiltration from the second day (Fig. 2e–h).

Unlike the treatment of other tumors, GBM therapy also faces a dilemma in terms of drug delivery, which is hampered by the BBB and safety issues. Hence, treatment strategies for postoperative GBM have increased the requirements for drug dosage, the schedule of administration and drug delivery. Hence, precise postoperative GBM treatment depends on whether the therapeutic regimens match the dynamic immune pathology. Hence, we designed an engineered nanofiber comprising a core-shell structure for DOX and BLZ945 loading to induce “in situ vaccine” and remodel macrophage-associated immune programs, respectively. By adjusting the composition and preparation technology, the obtained nanofiber (DOX/BLZ945-CNF) presented time-staggered drug release profiles of DOX release during the initial three days and BLZ945 release from the second day (Fig. 2i–m). In addition, the flexible nanofiber implant conforms to irregular postoperative cavities and enables passive drug release without rigid interfaces or injection-induced pressure fluctuations, making it well-suited for the mechanically vulnerable early postoperative brain environment. Importantly, the material-controlled release of the nanofiber is expected to be robust to differences in postoperative cavity size and location, supporting its potential clinical applicability across heterogeneous GBM lesions. Moreover, the obtained nanofiber was highly scalable and reproducible, compatible with the electrospinning technique, and holds potential for large-scale production.

In vitro analysis revealed that DOX/BLZ945-CNF could effectively induce the ICD of GL261 cells as “in situ vaccines”, with the translocation of CRT as an “eat me” signal and the release of HMGB1 as a “danger” signal (Fig. 3a, b). Moreover, the resulting nanofiber effectively remodeled protumoral macrophage into antitumoral macrophage by blocking the CSF-1R signaling pathway (Fig. 3c–k). To further investigate the antitumor efficacy, immunostimulatory activity and mechanism of DOX/BLZ945-CNF in vivo, a GBM resection model was established. After tumor resection, DOX/BLZ945-CNF was implanted into the cavity under a stereoscope. Compared with the antitumor efficacy of nanofibers that possessed various DOX release time, we found that DOX/BLZ945-CNF (DOX release in 3 days) could provide optimal antitumor efficacy and satisfactory safety profiles, including a significant tumor inhibition of 83.9%, ignorable increased BLI signals, a 100% survival rate until Day 57 with no obvious body weight loss, etc. (Fig. 4). In addition, DOX/BLZ945-CNF implantation increased robust antitumor immunity, increased immune cell infiltration, decreased protumoral macrophages and Tregs, and increased CD8⁺ cytotoxic T cells and CD4⁺ helper T cells (Fig. 5). Moreover, memory immunity against tumor recurrence with increased Tcm and Tem cells was achieved by DOX/BLZ945-CNF implantation, resulting in nearly 100% relapse-free survival with an 80% survival rate within 50 days of rechallenge (Fig. 6). In summary, this study introduces a rationally engineered, cavity-implantable nanofiber that enables a dynamic immunopathology-guided drug release strategy. The time-staggered chemo-immunotherapy sequentially addresses distinct stages of postoperative recurrence, and the nanofiber allows adaptation to other malignancies via tunable material parameters.

Methods

Ethics statement

All research was conducted in accordance with relevant ethical regulations. The use of human glioma tissue samples was approved by the Medical Ethics Committee of Wannan Medical University under approval number 2021-21. Human brain tumor specimens were obtained from patients who underwent surgical resection at the First Affiliated Hospital of Wannan Medical University. Written informed consent was obtained from all patients for the use of their samples in scientific research. Tumor classification and grading were performed according to the WHO criteria for central nervous system neoplasms by a pathologist and a surgeon. All animal experiments were approved by the Institutional Animal Care and Use Committee of China Pharmaceutical University under approval number 2023-06-004 and were performed in accordance with institutional guidelines for animal care and use. Animals were euthanized immediately if they exhibited any signs of distress, including but not limited to severe weight loss, ulceration, or a moribund state.

Materials

Polyvinyl alcohol (PVA, 1799) was procured from Shanghai Aladdin Reagent Co., Ltd. Poly(lactic acid/glycolic acid) (PLGA, 85/15, 75/25, 50/50) was obtained from Shanghai Ashland Inc (Shanghai, China). Doxorubicin hydrochloride (DOX·HCl) ( > 98%) was provided by Dalian Meilun Biotechnology Co., Ltd. (Beijing, China). 3,3’-Dioctadecyloxacarbocyanine Perchlorate (DiO), BCA protein detection kit and RIPA Lysis Buffer were supplied by Beyotime Biotechnology Co., Ltd. (Shanghai, China). DNase I was purchased from Jiangsu KeyGEN Biotechnology Co., Ltd. (Nanjing, China). IL-4 and IL-13 cytokines were bought from Proteintech Group, Inc. (Chicago, USA). Protease Inhibitor Cocktail and Phosphatase Inhibitor Cocktail I were purchased from TargetMol, USA. The enzyme-linked immunosorbent assay (ELISA) kits for IL-4, IL-13, HMGB1, IL-10, TGF-β, TNF-α and M-CSF were obtained from Yifeixue Biotechnology Co. Ltd. (Nanjing, China). D-Luciferin sodium salt was purchased from Macklin Biochemical Technology Co., Ltd. (Shanghai, China). IFN-γ ELISA kit was bought from Quanzhou Ruixin Biological Technology Co., Ltd. (Quanzhou, China).

Cell lines and animals

The mouse glioma cell line GL261, mouse glioma cell line CT-2A, mouse microglia cell line BV-2 and human neuronal cell line SH-SY5Y were obtained from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). Cell line authentication was performed by the provider using short tandem repeat (STR) profiling where applicable before use. All cell lines were routinely tested for mycoplasma contamination using a PCR-based mycoplasma detection assay, and only mycoplasma-negative cells were used in this study. The cells were cultured in DMEM supplemented with 10% FBS at 37 °C in a humidified environment with 5% CO₂. Six-week-old male C57BL/6 mice were purchased from Qinglongshan Animal Center (Jiangsu, China). The mice were housed and maintained in a regular 12 h light/dark environment. All animal experiments were approved as described in the Ethics statement. A tumor volume of 2000 mm³ was defined as the humane endpoint euthanasia and was not exceeded in any experiment. Animals were euthanized immediately if they exhibited any signs of distress, including but not limited to severe weight loss, ulceration, or a moribund state.

Clinical patient samples analysis

Brain tumor specimens were obtained from patients who underwent surgical resection at the First Affiliated Hospital of Wannan Medical University, as described in the Ethics statement. Tumor classification and grading were performed according to the WHO criteria for central nervous system neoplasms by a pathologist and a surgeon. Fifteen glioma samples were selected to analyze macrophage phenotypes by immunofluorescence. Immunofluorescence staining was conducted on paraffin-embedded human glioma sections that were 5 μm thick. After deparaffinization, rehydration, and antigen retrieval using citrate buffer (pH 6.0), sections were blocked with 5% normal goat serum for 1 h at room temperature. Primary antibodies, such as anti-Iba1 (Abcam, Cat# ab178846, 1:500, USA), anti-CD86 (BioLegend, Cat# 305401, 1:200, USA), anti-HLA-DR (BioLegend, Cat# 327002, 1:200, USA), anti-CD206 (BioLegend, Cat# 321101, 1:500, USA), and anti-TREM2 (Zenbio, Cat# 510482, 1:100, China) were left to incubate overnight at 4°C. After washing with PBS, the sections were exposed to HRP-conjugated secondary antibodies, including goat anti-mouse IgG-HRP (Servicebio, Cat# GB23303, China) and goat anti-rabbit IgG-HRP (Servicebio, Cat# GB23302, China), followed by signal amplification using a TSA kit (Servicebio, Cat# G1236, China) according to the instructions. Specifically, iF488-Tyramide, iF555-Tyramide, and iF647-Tyramide were sequentially used to visualize CD86, CD206, and Iba1, respectively. Additional staining for HLA-DR and TREM2 was performed in separate sections or using additional rounds of TSA amplification. DAPI (1 μg/mL) was utilized to counterstain the nuclei, and the sections were mounted using an antifade mounting medium. Fluorescence images were captured using a Digital Pathology Slide Scanner (VS200, Olympus, Japan) and analyzed with ImageJ software.

Bioinformatics analysis

Single-cell transcriptomic data from low-grade glioma (LGG) and high-grade glioma (HGG) samples (GSE182109) were processed using Scanpy (v1.9.3). Cells with <200 genes, > 20% mitochondrial content, and genes expressed in <3 cells were removed. Data were normalized to 10,000 counts per cell and log-transformed. Highly variable genes were selected (min_mean = 0.0125, min_disp = 0.5), followed by PCA (top 50 PCs). Batch correction was performed using BBKNN (neighbors_within_batch = 6, n_pcs = 50). UMAP embedding (min_dist = 0.3, spread = 1.0) and Leiden clustering (resolution = 0.5) were applied. Cell types were annotated manually using canonical marker genes collected from CellMarker 2.0. Myeloid populations were further classified according to transcriptional programs rather than conventional M1/M2 nomenclature. Specifically, macrophages were categorized into antitumoral macrophages enriched for inflammatory and interferon-associated genes (CXCL9, CXCL10, CXCL11, STAT1, ISG15) and protumoral macrophages corresponding to the SPP1-associated program (SPP1, APOE, TREM2, GPNMB). Dendritic cell subsets were further annotated based on canonical marker genes. Specifically, cDC1 cells were identified by the expression of CLEC9A, XCR1, CADM1, WDFY4, and IRF8; cDC2 cells were defined by CD1C, FCER1A, CLEC10A, CD1E, and CLEC4A; and LAMP3⁺ DCs, representing a mature migratory dendritic cell state, were characterized by LAMP3, CCR7, FSCN1, CCL19, and IDO1. Survival analyses were conducted using the GEPIA platform (https://gepia3.bioinfoliu.com/) based on TCGA LGG and GBM cohorts. Kaplan-Meier survival curves were generated by stratifying patients into high- and low-risk groups according to the expression score of the protumoral macrophage program defined by the SPP1/APOE/TREM2/GPNMB gene set, using the median value as the cutoff. Overall survival was used as the clinical endpoint.

GBM surgical resection model

C57/BL6 male mice, six weeks old, were used to establish the GBM surgical resection model. Avertin (1.25%, 20 μL/g) was administered to mice via intraperitoneal injection for anesthesia. To expose the skull, a small incision was made in the scalp, and then a burr hole was drilled at the preferred location on the right frontal lobe, approximately 1.8 mm lateral from bregma and at a 3.0 mm depth. Luc-GL261 cells (1 × 10⁵ cells/mouse) were then inoculated into the brain tissue at the same site using a Hamilton syringe. After that, the incision was closed using bone wax. Mice were euthanized at day 10, 12, 14, and 16 postinoculation to monitor tumor progression. Brains from the euthanized mice were collected, fixed in a 4% paraformaldehyde solution in PBS, and then stained with hematoxylin and eosin (HE). Using an optical microscope (Olympus IX53, Japan), the HE-stained brain sections were examined to analyze the growth of the tumor and the extent of local infiltration at the edges.

According to HE staining result, tumors at 14 days postinoculation exhibited pronounced local infiltration into the surrounding brain parenchyma with multiple tumor cell islands, suggesting this is an optimal time point for surgical intervention. Therefore, at 14 days after Luc-GL261 cells implantation, surgical resection was performed. The previous incision was reopened to expose the tumor, and microsurgical techniques and a stereomicroscope were used for careful resection of the tumor tissue while minimizing damage to the surrounding healthy tissue. Hemostasis was achieved by applying gel foam thorough irrigation and closing the incision. The surgical operation lasted approximately 30 min, and the mortality rate was below 5%, mainly occurring within 2 days after tumor resection. Mice received an intraperitoneal injection of D-Luciferin sodium salt at a dose of 150 mg/kg, and bioluminescence was detected 10 min after injection using an in vivo imaging system (VISQUE, Vieworks Co., Ltd., Gyeonggi-do, Korea). Following surgery, mice were euthanized on the second day postresection, and whole brains were carefully harvested and fixed in 4% paraformaldehyde at 4 °C. The fixed brain tissues were dehydrated, embedded in paraffin, and sliced into 5μm thick sections using a microtome. These sections were placed on glass slides, deparaffinized, and rehydrated with a series of graded ethanol solutions before being stained with HE. Stained slides were scanned using a Digital Pathology Slide Scanner (VS200, Olympus, Japan), and high-resolution images were analyzed to evaluate the presence of residual tumor cell foci at the resection margin.

Determination of tumor proliferation rate after surgery

At various postoperative intervals (Day 1, 2, 3, 5, 7, 9 and 11), the brain tissues of the tumor-bearing mice were carefully excised and fixed in a 4% paraformaldehyde solution in PBS. After being fixed, the tissue was embedded in paraffin and then cut into sections in the coronal plane. Heating the sections in sodium citrate buffer (pH 6.0) was used to perform antigen retrieval. Once cooled, slides were washed in PBS and treated with 5% normal goat serum in PBS for 1 h at room temperature to minimize nonspecific binding. Following this, the sections were exposed to a primary anti-Ki67 antibody (Abcam, ab15580, 1:600, UK) overnight at 4°C, followed by a FITC-labeled secondary antibody for 30 min incubation at room temperature in the dark. DAPI (1 μg/mL) was applied for nuclear counterstaining. The stained sections were visualized and captured using a DeltaVision™ PD imaging system (GE Life Sciences, MA, USA). Using Image J software, the Ki67 positive cells were quantified as a percentage of the total cells, providing insight into the proliferation rate.

Determination of macrophage phenotype and inflammatory cytokine expression after surgery

The tumor-bearing mice were perfused with cold PBS at different times after surgery (Days 1, 2, 3, 5, 7, 9, and 11) to eliminate circulating blood cells from their organs, thereby minimizing contamination of brain-resident immune cells. Brain tissues were then rapidly dissected on ice and all cells were extracted through mechanical and enzymatic dissociation using 500 μg/mL collagenase I and 50 U/mL DNase I, incubating at 37 °C on an incubator shaker at 100 rpm for 30 min to facilitate digestion. Following incubation, the digested mixture was filtered through a 70 μm nylon cell strainer to eliminate debris, resulting in a single-cell suspension which was then washed twice with PBS containing 1% FBS. Fixable Viability Dye eFluor 780 (eBioscience, 65-0865-14, USA), anti-CD45-PerCP-Cyanine5.5 (eBioscience, 45-0451-82, 1:160, USA), anti-CD11b-FITC (eBioscience, 11-0112-81, 1:100, USA), anti-CD80 (B7-2)-Super Bright 436 (eBioscience, 62-0801-82, 1:300, USA), anti-CD206 (MMR)-APC (eBioscience, 17-2061-80, 1:80, USA), anti-MHC Class II (I-A/I-E)-PE-Cyanine7 (eBioscience, 25-5321-80, 1:80, USA), anti-TREM-2-PE (BioLegend, 824805, 1:80, USA) were used to characterize macrophage activation states and functional programs, followed by FACS analysis (FACS Celesta, Becton, Dickinson, USA). For the test of anti-inflammatory cytokines (IL-4, IL-13), the brain tissues were ground using a tissue grinder (TL2010S, DHS Biological Technology Co., Ltd., Tianjin, China) at 1500 rpm for ten rounds (working for 60 s per round with 10 s interval). The supernatant was collected after centrifuging at 10,000 g at 4 °C for 20 min, and levels of IL-4 and IL-13 in brain tissue homogenates were quantified using ELISA kits. In short, 100 μL of standards or samples were added to each well of a 96-well plate that had been pre-coated with capture antibodies. This was followed by sequential incubation with detection antibodies and enzyme conjugates. After the substrate reaction and termination, the absorbance was measured at 450 nm using a microplate reader (Varioskan LUX, Thermo Fisher Scientific, Inc., MA, USA). Cytokine concentrations were determined using a standard curve created with known concentrations of recombinant IL-4 and IL-13.

Fabrication of the DOX/BLZ945-CNF

The nanofibers were fabricated by utilizing a coaxial metal needle. The outer needle had a diameter of 0.9 mm, while the inner needle had inner and outer diameters of 0.33 mm and 0.63 mm, respectively. The outer solution contained a 12% (w/v) combination of PVA and SF in de-ionized water, while the inner solution consisted of a 15% (w/v) PLGA in a mixture of acetone/DMF (1:1, v/v). A microinjection pump (ET-2535H, Beijing Ucalery Industry Technology Development Co., Ltd., China) was used to maintain a consistent flow with a fixed flow rate of 0.15 mL/h for the outer fluids and 0.03 mL/h for the inner fluids. The distance between the needles and the collector (aluminum foil) was kept at 14 cm, with a constant electric potential of 15 kV. To produce a DOX/BLZ945-loaded coaxial nanofiber, the outer solution contained DOX (10 mg/mL) and the inner solution contained BLZ945 (10 mg/mL). All electrospinning procedures were conducted at a consistent temperature of 25 °C and a humidity level of 60%. To fine-tune the degradation rates and drug release profiles of the core-shell structured nanofibers, different mass ratios of PVA to SF were investigated, specifically 6:1, 5:1, 4:1, 3:1, 2:1, and 1:1. Additionally, different grades of PLGA with varying lactic acid (LA) to glycolic acid (GA) molar ratios were employed in the inner core solution, including PLGA (85:15 LA/GA), PLGA (75:25 LA/GA), and PLGA (50:50 LA/GA).

Characterization of the DOX/BLZ945-CNF

For visualization, the fluorescently labeled nanofibers (DOX/DiO-CNF) were fabricated by incorporating DOX (1 mg/mL) into the outer solution containing PVA/SF (6:1, 5:1, 4:1, 3:1, 2:1, and 1:1) dissolved in deionized water. Meanwhile, the DiO (80 μg/mL), a lipophilic green fluorescent dye, was added to the inner solution containing PLGA in a mixture of acetone and DMF (1:1, v/v). The nanofibers were electrospun under the previously described conditions, producing a core-shell structure with DOX in the shell and DiO in the core. The core-shell composition of the fluorescently labeled nanofibers (DOX/DiO-CNF) was examined using Stimulated Emission Depletion (STED) microscopy (STELLARIS 8, Leica Microsystems Co., Ltd., Germany). DOX was illuminated at a wavelength of 480 nm, resulting in emission detected at 590 nm, while DiO was excited at 484 nm and emitted fluorescence at 501 nm.

The surface morphology of DOX/BLZ945-CNF was analyzed using Field Emission Scanning Electron Microscopy (FE-SEM) (JSM-7001F, Phenom Pure Ltd., China). Briefly, the nanofibers were fabricated by a coaxial electrospinning technique under consistent processing parameters. The outer solution comprised DOX (10 mg/mL) dissolved in a PVA/SF (6:1, 5:1, 4:1, 3:1, 2:1, and 1:1), while the inner solution consisted of BLZ945 (10 mg/mL) dissolved in PLGA. For SEM imaging, the nanofiber mats were sputter-coated with a thin layer of gold to improve electrical conductivity. SEM imaging was performed at an accelerating voltage of 5 kV. The diameter distribution of DOX/BLZ945-CNF was analyzed using ImageJ software (version 1.46 h, NIH, USA). The stability of DOX/BLZ945-CNF was evaluated by storage in PBS at 4 °C, storage in PBS at 37 °C, and storage in mouse brain tissue homogenate at 37 °C for 8 weeks. At the indicated time points (0, 2, 4, 6, 8 weeks), samples were collected and the amounts of DOX and BLZ945 were quantified, with drug concentrations normalized to the initial values at time 0. The compressive mechanical properties of DOX/BLZ945-CNF were characterized by uniaxial compression testing, and the Young’s modulus was calculated from the slope of the stress-strain curve in the linear elastic region.

Release profiles of DOX from DOX/BLZ945-CNF in vitro

To optimize the ratio of PVA and SF for the outer shell, the release profile of DOX was evaluated from nanofibers prepared with different mass ratios of PVA/SF (6:1, 5:1, 4:1, 3:1, 2:1, and 1:1), designated as DOX/BLZ945-CNF_6:1, DOX/BLZ945-CNF_5:1, DOX/BLZ945-CNF_4:1, DOX/BLZ945-CNF_3:1, DOX/BLZ945-CNF_2:1, and DOX/BLZ945-CNF_1:1, respectively. The nanofiber, weighing 10 mg, was submerged in 20 mL of PBS and placed in an incubator shaker programmed to operate at 37 °C and 100 rpm. To maintain sink conditions, the 100 μL of release medium was collected at specific time points of 0.5, 1, 2, 3, 4, 6, 8, 10, 14, 18, 22 and 26 days. An equal volume of fresh PBS was added after each collection. The released DOX from the nanofibers was quantified using HPLC, with absorbance readings taken at 480 nm. The analysis took place on an Agilent Zorbax Eclipse Plus C18 column (100 × 4.6 mm, 3.5 μm particle size) at 25 °C using a mobile phase consisting of acetonitrile with 0.1% TFA and water with 0.1% TFA in a 3:1 ratio, using isocratic elution over 20 min. The flow rate was set to 1.2 mL/min, with an injection volume of 10 μL.

Release profiles of BLZ945 from DOX/BLZ945-CNF in vitro

To investigate the effect of different PLGA compositions on drug release, BLZ945-loaded nanofibers were fabricated using PLGA with different LA/GA molar ratios: 85/15, 75/25, and 50/50, designated as DOX/BLZ945-CNF_85/15, DOX/BLZ945-CNF_75/25, DOX/BLZ945-CNF_50/50. The release experiments were conducted similarly by immersing the nanofiber samples (10 mg) in 20 mL PBS at 37 °C with shaking at 100 rpm. At set time intervals, the release medium was fully removed and replaced by an equal amount of fresh PBS to ensure sink conditions were maintained. BLZ945 concentrations were determined by HPLC under the same chromatographic conditions as for DOX, with detection at 272 nm.

Biocompatibility evaluation of blank CNF in vitro and in vivo

For in vitro biocompatibility evaluation, GL261, SH-SY5Y, and BV-2 cells were plated in 96-well plates with a density of 1 × 10⁴ cells per well in DMEM supplemented with 10% FBS. After 12 h of incubation for cell attachment, the old medium was replaced with fresh DMEM containing different amounts of blank CNF (0.3, 0.5, 1.0, 2.0, 4.0, and 6.0 mg). The blank CNF was created using coaxial electrospinning with drug-free inner and outer solutions under the same conditions as the drug-containing nanofiber. The cells were then incubated with blank CNF for 48 h at 37 °C in a 5% CO₂ humidified environment. Afterward, MTT solution (0.5 mg/mL) was added to each well and incubated for 4 h at 37 °C. The medium was then removed, DMSO was added to dissolve the crystals, and the plate was shaken for 10 minutes at room temperature. Using a microplate reader, the absorbance of treated and untreated cells was measured at 490 nm. Cell viability was calculated using the formula:

$$\text{Cell}\mathrm{viability}\left(\%\right)=\left({\text{OD}}_{\text{sample}}/{\text{OD}}_{\text{control}}\right)\times 100\%$$

where${\text{OD}}_{\text{sample}}$ and ${\text{OD}}_{\text{control}}$ represent the absorbance values of treated and untreated control cells, respectively. For in vivo biocompatibility evaluation, materials were intracranially implanted into mice and retrieved at predefined time points (0, 1, 2, 4, 6 weeks). At each time point, animals were euthanized and residual nanofibers were carefully isolated from the implantation sites, photographed and weighed. Meanwhile, the implantation sites were photographed to document macroscopic changes.

Cytotoxic effects of DOX/BLZ945-CNF on immune cells and tumor cells

Single-cell suspensions were prepared from murine spleen, and T cells were purified using a Pan T Cell Isolation Kit (Miltenyi Biotec, Cat. No. 130-095-535, Germany), while DCs were enriched using CD11c MicroBeads (Miltenyi Biotec, Cat. No. 130-108-338, Germany). Bone marrow–derived macrophages (BMDMs) were generated from bone marrow cells harvested from femurs and tibias and differentiated in the presence of recombinant mouse macrophage colony-stimulating factor (M-CSF; PeproTech, Cat. No. 315-02, USA). All procedures were performed following the manufacturer’s protocols, and the purity of isolated cells was confirmed by flow cytometry prior to subsequent experiments. Cells were seeded into 96-well plates with a density of 1 × 10⁴ cells per well and incubated for 48 h. After that, the old medium was replaced with fresh medium, which was the eluate of DOX/BLZ945-CNF containing different amounts of DOX (0.001, 0.01, 0.05, 0.1, 0.25, 0.5, 1, 2, 4 and 8 μM). Subsequently, MTT reagent was used to evaluate cell viability according to the manufacturer’s instructions, and absorbance at 490 nm was measured using a microplate reader. Cell viability was normalized to untreated controls. Dose–response curves were fitted and half-maximal inhibitory concentration (IC₅₀) values were calculated.

Immunogenic cell death (ICD)

GL261 cells were placed on cell slides in the lower chambers of 6-well Transwell plates with a pore size of 0.4 µm at a concentration of 1 × 10⁵ cells/well and grown in DMEM with 10% FBS at 37 °C in a humidified atmosphere with 5% CO₂. After 12 h of attachment, the culture medium was replaced with fresh DMEM and the upper chamber was added different treatments, including PBS (negative control), blank CNF, Free DOX, DOX-CNF, and DOX/BLZ945-CNF with the DOX concentration of 0.5 μM and blank CNF matched to the carrier concentration of drug-loaded groups. After a 48 h co-incubation period, the cells underwent two washes with PBS, were fixed with 4% paraformaldehyde for 20 min at room temperature, and were subsequently blocked with 5% bovine serum albumin (BSA) for 30 min. Following this, the cells were exposed to a primary anti-calreticulin (CRT) antibody (BioLegend, Cat# 871401, 1:300, USA) overnight at 4 °C. After washing, a FITC-conjugated secondary antibody (Servicebio, Cat# GB22303, 1:400, China) was applied to the cells for 30 min in the dark at room temperature. Nuclei were stained with DAPI, and CRT surface exposure was assessed using the DeltaVision™ PD imaging system (GE Life Sciences, MA, USA). For the analysis of HMGB1 secretion, GL261 cells were subjected to the same treatment for 24 h. The culture supernatants were collected and centrifuged at 10,000 g for 10 min to eliminate cellular debris. HMGB1 levels were measured using an HMGB1 ELISA kit following the provided instructions. The absorbance was read at 450 nm using a microplate reader (Varioskan LUX, Thermo Fisher Scientific, Inc., MA, USA).

M-CSF secretion by GL261 cells

GL261 cells were seeded in 24-well dishes at a concentration of 5 × 10⁴ cells per well and cultured for 0, 6, 12, 24 and 48 h. The supernatants were harvested, centrifuged at 1,000 g for 10 min at 4 °C to remove cellular debris, and the clarified supernatants were stored at −80 °C until analysis. The concentration of M-CSF in the conditioned media was quantified using a mouse M-CSF ELISA kit according to the manufacturer’s instructions.

Macrophage phenotype remolding in vitro

RAW 264.7 cells were plated in 24-well dishes at a concentration of 1 × 10⁵ cells per well and left to incubate overnight in DMEM with 10% FBS. Subsequently, the cells were treated with IL-4 and IL-13 at a dose of 10 ng/mL each for 24 h in order to stimulate protumoral macrophage polarization. The successful establishment of the protumoral phenotype was verified by staining with anti-CD206-PE antibody (eBioscience, 12-2061-80, 1:160, USA) and anti-CD80-APC antibody (eBioscience, 17-0801-81, 1:300, USA), followed by flow cytometric analysis (CytoFLEX LX, Beckman Colter, USA).

To investigate macrophage reprogramming, protumoral macrophage was continuously cultured in GL261-conditioned medium, which was obtained by collecting the supernatant from GL261 cells after 12 h of incubation, followed by filtration through a 0.22 µm filter. Meanwhile, the resulting nanofiber was prepared by collecting the nanofiber-released medium after 3 days of DOX release. The protumoral macrophage was then treated with resulting nanofiber for 24 h, including PBS (negative control), blank CNF, Free BLZ945, DOX-CNF, and DOX/BLZ945-CNF (with a BLZ945 concentration of 2.5 μM), and that the blank CNF group matched the carrier concentration of the drug-loaded groups. After treatment, the macrophage was divided into two aliquots for flow cytometric analysis. One aliquot was stained with anti-CD206-PE antibody (eBioscience, 12-2061-80, 1:160, USA) and anti-CD80-APC antibody (eBioscience, 17-0801-81, 1:300, USA). The other aliquot was stained with anti-MHC class II (I-A/I-E)-PE-Cy7 antibody (eBioscience, 25-5321-80, 1:80, USA) and anti-TREM-2-PE antibody (BioLegend, 824805, 1:80, USA). Stained cells were subsequently analyzed by flow cytometry as previously described. The supernatant was collected after centrifuging at 10,000 g for 20 min at 4 °C and analyzed using ELISA kits to measure the secretion of cytokines, including IL-10, TNF-α, and TGF-β.

Phagocytosis capability of the remodeled macrophage in vitro

RAW264.7 cells were seeded in 24-well plates at a density of 1 × 10⁵ cells per well and induced to protumoral macrophage as described above. To facilitate qualitative analysis of phagocytosis of the remodeled macrophage, GL261 cells were labeled with the fluorescent dye DiO according to the manufacturer’s instructions. Briefly, GL261 cells were incubated with DiO (1 μM) for 30 min at 37 °C. After labeling, the GL261 cells were exposed to a DOX concentration of 0.5 μM for 48 h to induce ICD. Following incubation, the GL261 cells were centrifuged at 5000 g for 20 min to collect the cell debris. The GL261 debris was then resuspended in GL261-conditioned medium and co-incubated with the remolding RAW264.7 cells placed in the lower chamber for 24 h at 37 °C in a 5% CO₂ atmosphere. Meanwhile, PBS (negative control), blank CNF, Free BLZ945, DOX-CNF, and DOX/BLZ945-CNF (with a BLZ945 concentration of 2.5 μM) were placed in the upper chamber. After the 24 h co-incubation, the cells were washed with PBS to remove any non-phagocytosed debris and then stained with anti-F4/80-PE antibody (eBioscience, Cat# 12-4801-81, 1:100, USA) to label RAW264.7 cells. The cells were subsequently analyzed by flow cytometry using a CytoFLEX LX flow cytometer (Beckman Colter, USA). To determine the phagocytosis capability, the percentage of DiO⁺ cells within the F4/80⁺ RAW264.7 population was quantified.

Western blot for RAW264.7 cells

In 6-well plates, 4 × 10⁵ RAW264.7 cells were planted on each well and remodeled as described above. After being treated, the modified macrophage was lysed in an ice bath with 200 μL of RIPA lysis buffer per well, which included a Protease Inhibitor Cocktail (1:100) and Phosphatase Inhibitor Cocktail I (1:100). The total protein concentration was determined using a bicinchoninic acid assay (BCA). Subsequently, the proteins were separated using 10% SDS-PAGE and transferred onto a polyvinylidene difluoride (PVDF) membrane. The membrane was then blocked with 5% BSA for 2 h at room temperature. Primary antibodies, including anti-pCSF1R (Zenbio, 310272, 1:1000, China), anti-CSF1R (Zenbio, 160600, 1:800, China), anti-pAKT (Zenbio, R381555, 1:500, China), anti-AKT (Zenbio, 342529, 1:1000, China), anti-SOCS1 (Zenbio, 160748, 1:1000, China), anti-SOCS3 (Zenbio, 251981, 1:1000, China), and β-actin (Servicebio, GB11001, 1:2000, China) were incubated overnight at 4 °C. After being washed with TBST buffer, the cells were incubated with HRP-conjugated secondary antibodies (HRP-labeled sheep anti-rabbit IgG and HRP-labeled sheep anti-mouse IgG) for 1 h at room temperature. Subsequently, the membranes were stained with enhanced chemiluminescence (ECL) and visualized using a gel imaging system (Tanon 5200, Tanon Technology Co., LTD., Shanghai, China).

DC2.4 cells maturation, phagocytosis capability and T cell killing assay in vitro

To evaluate the influence of ICD and macrophage functional states on the maturation and phagocytic capacity of DC2.4 cells, macrophages collected after different treatments (PBS as negative control, blank CNF, free BLZ945, DOX-CNF, and DOX/BLZ945-CNF; BLZ945 concentration: 2.5 μM), representing treatment-induced functional states of the overall macrophage population without flow cytometric sorting into predefined subsets, were first seeded into the upper chambers of a 24-well transwell system (pore size: 0.4 µm) at a density of 1 × 10⁴ cells per insert and allowed to adhere overnight. Meanwhile, DC2.4 cells were seeded into the lower chambers at a density of 1 × 10⁵ cells per well and cultured overnight. DiO-labeled GL261 debris was prepared by incubating GL261 cells with 0.5 μM DOX for 48 h to induce ICD, followed by centrifugation at 5000 g for 20 min and resuspension in fresh medium. After 24 h of co-culture, during which soluble factors secreted by the remodeled macrophage could modulate the DC2.4 cells through the semipermeable membrane, the cells in the lower chambers were harvested. The collected cells were stained with anti-CD11c-PE antibody (eBioscience, 12-0114-81, 1:40, USA) to identify DC2.4 cells. For maturation marker analysis, DC2.4 cells were further stained with anti-CD80-APC (eBioscience, 17-0801-81, 1:100, USA) and anti-CD86-FITC (eBioscience, 11-0862-81, 1:400, USA) at 4 °C for 30 min in the dark. After staining, cells were washed and resuspended in PBS. The expression levels of CD80 and CD86 were used to evaluate the activation status of DC2.4 cells. Flow cytometry (CytoFLEX LX, Beckman Colter, USA) was performed to assess phagocytosis, and the percentage of DiO⁺ within the CD11c⁺ cell population was quantified as an indicator of the phagocytic activity of DC2.4 cells. To evaluate the tumor-killing ability of T cells activated by DC2.4 cells, T cells were isolated from murine spleens using the method described above and subsequently activated by co-culture with DC2.4 cells and the remodeled macrophage for 48 h. Activated T cells were then collected and co-cultured with GL261 cells at a ratio of 3:1 for 24 h. After that, supernatants were harvested, and the level of lactate dehydrogenase (LDH) released into the culture medium was quantified using an LDH cytotoxicity detection kit (Beyotime Biotechnology, C0016, China) according to the manufacturer’s instructions.

The Biocompatibility evaluation of blank CNF in vivo

C57BL/6 mice were anesthetized and subjected to a cortical injury by making a 5-mm incision on the brain surface using a sterile surgical blade. Thereafter, saline as the sham control and blank CNF (0.5 mg) was implanted into the wound site. Mice were sacrificed at designated time points (1 day, 1 week, 2 weeks, and 4 weeks postimplantation), and the brain tissues surrounding the implant site were harvested for subsequent analysis. All tissue sections for immunohistochemistry (IHC) were deparaffinized in xylene and then hydrated with a graded ethanol series. To retrieve the antigens, the sections underwent microwave treatment with an epitope retrieval solution (0.01 M citrate buffer, pH 6.0) for 20 min. In addition, then immersed in 0.3% hydrogen peroxide for 10 min with the purpose of antagonizing endogenous peroxidase. Subsequently, the sections were incubated with a primary antibody: either a mouse monoclonal anti-Iba1 antibody for macrophage (Abcam, Cat# ab178846, 1:500, USA) or a rabbit polyclonal anti-GFAP antibody for astrocytes (Abcam, ab7260, 1:1000, USA) in Dako REAL antibody diluent for the mouse model. The HRP-Polymer anti-Mouse/Rabbit IHC Kit (MaxVision, 5010) was employed for the secondary antibody and the avidin-biotinylated peroxidase system. For analysis, images were captured using Digital Pathology Slide Scanner (VS200, Olympus, Japan). Positive staining areas were quantified using ImageJ software by setting a uniform threshold for DAB intensity.

Postoperative GBM progression monitoring

To establish the tumor model, C57BL/6 mice were anesthetized by intraperitoneal injection of Avertin and were intracranially implanted with Luc-GL261 cells (1 × 10⁵ cells per mouse) or Luc-CT-2A cells (6 × 10⁴ cells per mouse) as described above. On day 14 after tumor inoculation, mice underwent tumor resection surgery. Under Avertin anesthesia (1.25%, 20 μL/g), the scalp was incised, and a craniotomy was carefully performed to expose the tumor area. Visible tumor tissues were excised as thoroughly as possible under a surgical microscope to simulate postoperative residual disease. Following tumor removal, different treatment formulations (equivalent to 1 mg/kg DOX), blank CNF, or saline were directly implanted into the tumor cavity. Postoperative monitoring of tumor recurrence and progression was performed by bioluminescence imaging (BLI). At postoperative day 0, day 9 and day 18, mice were intraperitoneally injected with D-Luciferin sodium salt (150 mg/kg), and imaged using an in vivo imaging system (VISQUE, Vieworks Co., Ltd., Korea). Quantitative ROI analysis was conducted to measure the luminescence intensity of intracranial Luc-GL261 cells or Luc-CT-2A cells. The body weights and survival periods of the mice were recorded during treatment.

On postoperative day 18, mice underwent magnetic resonance imaging (MRI) for further evaluation of intracranial tumor progression. For MRI scanning, mice were anesthetized using isoflurane and fixed in a prone position within the animal coil. The scanning parameters are as follows: TR = 500 ms, TE = 10.67 ms, Fov = 100 mm, FA = 152.9°, slice thickness = 1 mm, number of layers = 10, matrix = 256 × 256. And the tumor volumes were calculated with the formula as follows:

$$V=\frac{d_{\mathrm{cc}}\times d_{\mathrm{ap}}\times d_{\mathrm{l}}\times \pi }{6}$$

where $d_{\mathrm{cc}}$, $d_{\mathrm{ap}}$, and $d_{\mathrm{l}}$ represent the craniocaudal diameter, anteroposterior diameter, and the largest lateral diameter, respectively.

On postoperative day 18, mice were sacrificed, and brain tissues were harvested for histopathological analysis. Brain samples were fixed in 4% paraformaldehyde, embedded in paraffin, sectioned, and subjected to HE staining and visualized by a Digital Pathology Slide Scanner (VS200, Olympus, Japan).

T cell-mediated antitumor efficacy of DOX/BLZ945-CNF

GL261-bearing mice were randomized to five groups: Control, DOX/BLZ945-CNF, DOX/BLZ945-CNF + αCD4, DOX/BLZ945-CNF + αCD8, and DOX/BLZ945-CNF + αCD4 + αCD8. For T cell depletion, GL261-bearing mice were respectively treated with anti-CD4 (Selleck, Cat. A2101) and anti-CD8 monoclonal antibodies (Selleck, Cat. A2102) on days -7, -3, 0, +3, +7, and +10 relative to tumor challenge on day 0. Each depleting antibody was administered i.p. at 100 μg per mouse, and the depletion effect was confirmed by flow cytometry. Tumor progression was monitored by bioluminescence imaging, and survival was analyzed using Kaplan–Meier methods.

Western blot in vivo

Tumor tissues extracted from sacrificed animals on postoperative day 19 were immediately frozen in liquid nitrogen to protect protein quality. These frozen tissues were then homogenized in RIPA lysis buffer (200 μL/sample) with Protease Inhibitor Cocktail (1:100) and Phosphatase Inhibitor Cocktail I (1:100) on ice. After centrifugation at 10,000 g for 20 min at 4 °C, the supernatants were collected. The BCA assay was used to determine the total protein concentrations. The subsequent procedures, such as SDS-PAGE separation, PVDF membrane transfer, blocking, primary and secondary antibody incubation, chemiluminescent detection, and imaging, were carried out following the same procedure used for the RAW264.7 cell samples.

Immune stimulation in vivo

To further assess the antitumor immune response, five mice from each treatment group were euthanized on postoperative day 19. The brain tissues containing residual or recurrent tumor regions were carefully harvested under sterile conditions. Tissues were mechanically minced and enzymatically digested with 500 μg/mL collagenase I and 50 U/mL DNase I at 37 °C for 30 min to prepare single-cell suspensions. Following digestion, the suspensions were filtered through a 70 μm cell strainer and washed twice with cold PBS supplemented with 1% FBS. The obtained single-cell suspension was evenly divided into six parts for subsequent staining with various antibody panels targeting different immune cell subsets. Specifically, lymphocytes were labeled with anti-CD45-FITC (eBioscience, MA5-28679, 1:500, USA). For myeloid cell profiling, cells were stained with Fixable Viability Dye eFluor 780 (eBioscience, 65-0865-14, USA), anti-CD45-PerCP-Cyanine5.5 (eBioscience, 45-0451-82, 1:160, USA), anti-CD11b-FITC (eBioscience, 11-0112-81, 1:100, USA), anti-Ly-6C-eFluo 450 (eBioscience, 48-5932-80, 1:80, USA), anti-Ly-6G-APC (eBioscience, 17-9668-80, 1:40, USA), anti-TMEM119-PE-Cyanine7 (eBioscience, 25-6119-80, 1:20, USA), F4/80- Brilliant Violet 650 (eBioscience, 416-4801-82, 1:20, USA) and anti-CD192 (CCR2)-PE (BioLegend, 150609, 1:40, USA), which were used to identify tissue-resident microglia, differentiated macrophages, and Ly6C⁺CCR2⁺ infiltrating monocytes. Macrophage and microglia activation states and functional programs were further characterized using anti-CD80 (B7-2)-Super Bright 436 (eBioscience, 62-0801-82, 1:20, USA), anti-CD206 (MMR)-APC (eBioscience, 17-2061-80, 1:80, USA), anti-MHC Class II (I-A/I-E)-PE-Cyanine7 (eBioscience, 25-5321-80, 1:80, USA), and anti-TREM-2-PE (BioLegend, 824805, 1:80, USA). Dendritic cell (DC) maturation was stained with anti-CD11c-PE (eBioscience, 12-0114-81, 1:40, USA), anti-CD80-APC (eBioscience, 17-0801-81, 1:300, USA), and anti-CD86-FITC (eBioscience, 11-0862-81, 1:400, USA). T cells were stained with anti-CD3-FITC (eBioscience, 11-0032-80, 1:200, USA), anti-CD4-PE (eBioscience, 12-0041-82, 1:160, USA), and anti-CD8-APC (eBioscience, 17-0081-81, 1:160, USA). Regulatory T cells (Tregs) were identified by staining with anti-CD4-PE (eBioscience, 12-0041-82, 1:160, USA), anti-Foxp3-FITC (eBioscience, 11-5773-82, 1:50, USA), and anti-CD25-APC (eBioscience, 17-0251-82, 1:160, USA). The cells were washed twice with PBS containing 1% FBS and resuspended in staining buffer for acquisition after antibody incubation at 4 °C for 30 min. Flow cytometric analysis was carried out using a CytoFLEX flow cytometer (Beckman Colter, USA), and the data were analyzed with FlowJo software (Tree Star, USA). The ELISA kits were used to detect the levels of IFN-γ, IL-10, and TNF-α in brain tissues according to the manufacturer’s instructions. Brain tumor tissues were harvested, weighed, and homogenized in cold PBS (1:10, w/v) with protease inhibitors. The homogenates were centrifuged at 10,000 g for 10 min at 4 °C to remove debris, and the supernatants were collected for cytokine quantification. The absorbance at 450 nm was measured using a microplate reader (Varioskan LUX, Thermo Fisher Scientific, MA, USA).

Immunological memory response

To assess the induction of immunological memory following DOX/BLZ945-CNF treatment, long-term surviving mice were subjected to tumor rechallenge. Specifically, on day 26 after primary tumor resection, the surviving mice were intracranially re-inoculated with Luc-GL261 cells (1 × 10⁵ cells suspended in 5 μL PBS) into the contralateral hemisphere using a stereotactic apparatus. Two days after the re-inoculation, peripheral blood and spleens were collected for immune profiling. Peripheral blood mononuclear cells were isolated using a Mouse Peripheral Blood Mononuclear Cell Isolation Kit (Tianjin Haoyang Biological Manufacture Co., Ltd., China), according to the manufacturer’s protocol. Similarly, splenic lymphocytes were isolated using a Mouse Spleen Lymphocyte Isolation Kit (Tianjin Haoyang Biological Manufacture Co., Ltd., China). Briefly, peripheral blood was mixed in a 1:1 ratio with PBS, carefully layered onto the separation medium, and centrifuged at 800 × g for 20 min at room temperature using a swing-bucket rotor centrifuge, without braking during deceleration. The mononuclear cell layer was then collected with care, washed twice with PBS containing 1% FBS, and resuspended for further staining. Spleens were mechanically broken down into single-cell suspensions, filtered through a 70 μm cell strainer, and underwent similar density gradient centrifugation with a swing-bucket rotor to isolate splenic lymphocytes. The lymphocytes in the peripheral blood and spleens were stained with anti-CD3-FITC (eBioscience, 11-0032-80, 1:200, USA), anti-CD8-APC (eBioscience, 17-0081-81, 1:160, USA), anti-CD44-Perp Cy5.5 (eBioscience, 45-0441-82, 1:80, USA), anti-CD62L-PE (eBioscience, 12-0621-81, 1:160, USA). The staining was conducted for 30 min at 4 °C in the dark, followed by washing with PBS containing 1% FBS. Flow cytometric analysis was performed using a CytoFLEX flow cytometer (Beckman Colter, USA), and data were analyzed by FlowJo software (Tree Star Inc., USA). The CD3⁺CD8⁺CD44⁺CD62L⁺ were identified as the Tcm, and CD3⁺CD8⁺CD44⁺CD62L⁻ were identified as the Tem. The survival of the rechallenged mice was monitored, and tumor tissue slices were subjected to HE staining.

The isolated lymphocytes were then restimulated in vitro with GL261 tumor cell debris at a concentration of 20 μg/mL. Meanwhile, saline, protumoral macrophages, or DOX/BLZ945-CNF-remodelled macrophages were added to the cultures, respectively, to assess their effects on lymphocyte activation. After 48 h of co-culture, the culture supernatants were collected, and the secretion level of interferon-γ (IFN-γ) was measured using an ELISA kit according to the manufacturer’s instructions. IFN-γ levels were used as an indicator of the reactivation and cytotoxic differentiation of Tem.

Safety evaluation

To assess safety levels, we conducted a thorough evaluation by analyzing serum biochemical markers, including alanine aminotransferase (ALT), creatine kinase (CK), blood urea nitrogen (BUN), alkaline phosphatase (AKP), lactate dehydrogenase (LDH), and creatinine (CRE). We also assessed major organ damage through histopathological analysis using HE staining. Blood samples were collected before and after the treatment to measure ALT, CK, BUN, AKP, LDH and CRE levels. Additionally, liver, kidney, and spleen tissues were collected, fixed, and subjected to HE staining for examination. The resulting tissue sections were carefully analyzed under a microscope to detect any abnormalities or signs of organ damage. Statistical analysis was used to compare biochemical marker levels between the treatment and control groups. The results of this safety assessment provided valuable insights into the potential adverse effects of the treatment on serum biochemical markers and major organ tissues, enabling a comprehensive evaluation of safety parameters.

Statistics and reproducibility

All data were analyzed by SPSS software (version 26, SPSS Inc., Chicago, IL, USA) and the results were presented as mean ± S.D. For two-group comparisons, parametric data and non-parametric data were analyzed by independent t-tests and Mann-Whitney U tests, while multiple-group comparisons were performed using one-way ANOVA with Tukey’s post hoc test for parametric data and Kruskal-Wallis test with Dunn’s post hoc test for non-parametric data. Survival analysis was performed using the log-rank (Mantel-Cox) test. Experiments were independently repeated at least three times with similar results. Statistical significance was defined as P < 0.05. Error bars, p values, n numbers and statistical tests used are reported in the figures or figure legends. No data were excluded from the analyses. No statistical method was used to predetermine sample size. Samples were randomly allocated into experimental groups. The investigators were not blinded to allocation during experiments and outcome assessment.

Visualization

Graphs were generated using GraphPad Prism 8.0.2. Flow cytometry data were visualized using FlowJo v10.x. Microscopy images were processed using ImageJ v1.52 v. Figures were created and assembled using Adobe Photoshop 2025 and Adobe Illustrator 2025. BioRender-created elements, where applicable, were cited in the corresponding figure legends according to the BioRender publication license.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

M.S. designed the study, conducted experiments, and analyzed the data. Z.Z., R.C., and X.Z. contributed to data collection and analysis. Y.Y., Y.J., and Q.D. assisted with experimental procedures and provided technical support. J.Z. supervised the project and reviewed the manuscript. H.Z. and Y.D. provided overall project guidance and contributed to manuscript preparation. All authors reviewed and approved the final manuscript.

Peer review

Peer review information

Nature Communications thanks John S. Yu who co-reviewed with Keerthi Sadanala; Zihou Deng, Haifeng Xu and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

The authors gratefully acknowledge the National Natural Science Foundation of China (82372113 and 82373815), the Fundamental Research Funds for the Central Universities (No. 2632026TD05), the National Ten Thousand Talents Program for Young Top-notch Talents, and the Natural Science Foundation Outstanding Youth Fund of Jiangsu Province (BK20240096, China). In addition, the authors thank the Fourth Phase Construction Project of Superior Disciplines in Jiangsu Universities, and the Jiangsu 333 High-level Talent Training Project ([2022]3-16-190). The authors also thank Joint Fund of Zhejiang Provincial Natural Science Foundation of China under Grant (No. LKLY25H180014), the Lingang Laboratory, Grant No. LGL-2614-09, Youth Elite Scientists Sponsorship Program by CAST (NO. YESS20240291), Postgraduate Research & Practice Innovation Program of Jiangsu Province (KYCX25_1057) and National Innovation and Entrepreneurship Training Program for Undergraduate (No.202510316018).

Data availability

The main data supporting the results of this study are available within the paper and its Supplementary Information. The single-cell transcriptomic dataset used in this study is available in the Gene Expression Omnibus under accession code GSE182109. All other data supporting this work are available in the main article, Supplementary Information and Source Data files. Source data are provided with this paper.

Code availability

No custom code or mathematical algorithm central to the conclusions of this study was generated. Standard software and publicly available packages used for data analysis, including Scanpy v1.9.3 and BBKNN, are listed in the Methods and Reporting Summary.

Competing interests

The authors declare that they have no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-75065-x.